This PDF file contains the front matter associated with SPIE Proceedings Volume 13512, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

In-situ testing refers to assessing the geotechnical properties of soil in its natural state, primarily using methods such as static cone penetration, dynamic probing, standard penetration tests, and vane shear tests. Each testing method measures different soil parameters. This paper presents a new design for an in-situ testing probe based on static cone penetration and vane shear technologies, aiming to comprehensively measure multiple mechanical parameters of submarine soils. The probe features a CPTU module and a vane shear module, enabling simultaneous measurement of several soil mechanical parameters, including cone tip resistance, sidewall friction, pore water pressure, and shear stress. These parameters reflect most geotechnical properties. The internal sensor of the static cone penetration probe is designed with a "three-cylinder" structure that offers a wide measuring range and high sensitivity. This design allows for the measurement of both soft and hard soils using a single probe, with sensor errors within 0.5% of full scale. The vane shear module's internal sensor is a non-contact torque sensor with a maximum relative error and repeatability of within 5%. Additionally, the probe utilizes ultrasonic wireless transmission for its internal sensors, avoiding the complications of machining and wiring issues typically associated with wired systems.

In recent years, soft grippers have gained significant attention due to their adaptability and safety. However, when performing horizontal object grasping, their limited structural stiffness prevents them from withstanding the torque generated by friction between the gripper and the object, leading to a loss of contact and limiting their horizontal load capacity. To address this limitation while preserving the adaptability of soft grippers, this paper proposes a gripper structure called the rigid-soft robotic hand, which incorporates lateral rigid connecting rods outside the soft joints. The paper analyzes the bending capacity of the rigid-soft joint using kinematic methods and investigates the influence of link design parameters on maximum bending angle. Subsequently, a finite element analysis is conducted to study the finger structure with the added connecting rods, aiming to optimize the link design parameters. Experimental results demonstrate the grasping capabilities of the rigid-soft gripper in both vertical and horizontal orientations. The findings show that the rigid-soft gripper significantly improves the load-carrying capacity during horizontal object grasping while maintaining the adaptability of the soft gripper.

This paper presents the design and implementation of an autonomous picking robot specifically for harvesting elongated fruits such as loofahs and bitter melons. The robot utilizes a tracked chassis to navigate field terrains, equipped with IMU and GNSS modules for navigation and positioning. Mounted on the chassis is a 6-degree-of-freedom robotic arm designed for picking actions, with a combination of a flexible gripper and electric scissors at its end to enhance picking efficiency and safety. Two RGBD cameras are used for global and local perception, while the YOLO model is employed for fruit recognition and localization. Experimental results demonstrate that the robot can complete a full picking cycle in an average of 20 seconds and temporarily store the harvested fruits in an onboard storage bin. Overall, this picking robot significantly reduces labor costs and improves the efficiency and quality of field operations and fruit harvesting.

The impact localization of spacecraft structures is a crucial element in ensuring their safe and stable operation during successful launch and in-orbit missions. In this paper, an improved triangulation impact location method based on fiber Bragg grating (FBG) network is proposed. Based on the maximum slope method to determine the arrival time of impact signals from FBGs, impact location was determined by adaptively constructing the positioning triangle. This method is verified by the titanium alloy plate model, which can simulate the generation of damage in reality. The experimental results show that the minimum location error is up to 2.01mm.

Based on a four-phase balanced method, a novel modulator circuit with four identical mixers and continuous orthogonal four-channel carrier signals was developed in this paper. The modulator can achieve high carrier suppression by inputting two anti-phase I-channel signals and two anti-phase Q-channel signals. Compared with conventional quadrature modulators, the four-phase balanced modulator avoids the effect of circuit's amplitude and phase unbalance to improve carrier suppression. Moreover, the four-phase balanced modulator can achieve a wider relative bandwidth of more than 20% without a baseband amplifier and RF 90 degree power bridge. The theoretical analysis and design methodology are introduced in this paper. The simulation results show the carrier suppression is over 36.562dBc at 26 GHz. The-four-phase balanced modulator can be potential applied in space microwave direct modulation transmitters.

Electroencephalography (EEG) emotion recognition is an important task in Human-Computer Interaction (HCI), which has critical applications in modern electronic devices. In recent years, significant advancements have been made in EEG emotion recognition through deep learning algorithms and the incorporation of attention mechanisms. However, EEG signals possess complex and multi-dimensional features, and there is a lack of a compact method with unified attention modules for EEG-based emotion recognition. In this paper, we propose a novel network called Joint-Dimension-Aware Transformer (JDAT) based on the novel squeezed Multi-head Self-Attention (MSA) mechanism for EEG emotion recognition. The adaptive Squeezed MSA is applied to multi-dimensional features, allowing JDAT to focus on diverse EEG information, including space, frequency, and time. Through joint-dimension attention, JDAT is able to capture complicated brain activities, such as signal activation, phase-intensity couplings, and resonance. The proposed JDAT is evaluated on the DEAP, DREAMER, and SEED datasets, and experimental results show that it outperforms state-of-the-art methods. The ablation study further illustrates the performance improvement brought by joint-dimension attention.

Despite substantial advancements in perovskite solar cells (PSCs), addressing surface defects in three-dimensional perovskite films remains a critical challenge. In this study, we introduce 2,6-diacetylpyridine (DAPyr) as effective molecular passivator for perovskite films, crucial for reducing carrier losses. We demonstrate that targeted molecular enhancements, notably the integration of nitrogen-containing (C-N) groups, significantly adsorb free lead ions, enhancing the film’s structural integrity. Furthermore, the incorporation of carbonyl (C=O) functionalities substantially increases carrier collection efficiency and achieves thorough defect passivation. Optimized with DAPyr, our PSCs reached a record efficiency of 24.82%. These advancements lay a robust groundwork for elevating the performance and extending the commercial potential for PSCs.

The extensive incorporation of renewable resources into the power network has become more common, with gridconnected converters being vital to this procedure. Notably, grid-forming converters, which offer voltage and frequency reinforcement to the grid, have attracted considerable interest. Nevertheless, grid-forming converters may experience temporary instability during grid malfunction circumstances. Current research mainly emphasizes the impact of active power circuits, while less emphasis is placed on reactive power circuits. This study particularly examines how the interaction between active and reactive power management circuits influences the temporary stability of grid-forming converters. The findings suggest that such interaction might impair the stability performance of these converters. This conclusion was confirmed via simulation experiments.

During long-term overload operation, the belt conveyor used in coal mines can withstand excessive tension. Once the belt breaks, the entire production process will be forced to interrupt, and coal transportation will be paralyzed. Therefore, a longitudinal tear control method for the belt conveyor used in coal mines is proposed. Build a collection system based on laser equipment to capture images of longitudinal belt tearing. Implement denoising and extract regions of interest (ROI) for images. Based on the ROI extraction results, perform edge detection and count the number of pixels in the torn area within the edge area, calculate the proportion of the torn area, and determine the degree of belt tearing. Make three different control strategies based on the degree of tearing. The results show that the proportion of tear area controlled by this method has only increased by 0.05%percnt; compared to before control, indicating that the degree of tear has been well controlled and proving the application effect of this method.

Based on the Capacitive and Inductive Loaded Substrate Integrated Waveguide (CILSIW) resonator, a novel fourthorder bandpass filter (BPF) is designed. The capacitive- and inductive-loaded elements of the proposed CILSIW resonator are realized by a floating metallic patch and a floating metallic spiral line, respectively. The floating metal patch and the metal helix, the metal helix and the ground plane are connected by two metal through-holes. Due to the both capacitive- and inductive-loaded elements, the resonance can be shifted to lower frequency with a large ratio. A filter with a center frequency of 0.5 GHz and a fractional bandwidth of 6% is designed, manufactured, and measured. The experiment showed that the measurement results were the same as the simulation results, which verified the scientificity. In addition, the filter has the advantages of miniaturization, good selection performance, and stopband performance.

MMC is suitable for high voltage application because of its modular structure, but it also makes the converter bulky and heavy, increasing the cost and space requirements for offshore converter stations when used in HVDC-based offshore wind farm. To address this, an improved MMC topology is proposed in this paper based on bridge-arm multiplexing, which reduces capacitance and improves MMC capacitance utilization through bridge-arm multiplex and innovative switching while maintaining DC fault interruption capability. Then, an efficient modulation strategy and power balance method applicable to this MMC topology are designed. Simulation verification shows that this topology has the above features.

Noise is currently a highly concerned index for High end automobiles. In the rapid development of the new energy vehicle field, Most of the existing automotive relays cannot meet the noise requirements of new energy vehicle customers, so optimizing the noise of automotive relays is urgently. This article uses the digital models combined with experimental testing to analyze the noise mechanism of automotive relays, and determine the sources and propagation paths of the noise. Then, optimize the noise of automotive relays by focusing on the noise source and propagation paths. During the optimization process, based on the electromagnetic model of the digital models, the impact of the optimization plan on the electrical performance parameters of automotive relays is analyzed. And based on the thermal model of the digital models and experimental verification, the impact of changed the noise propagation paths on the temperature rise of automotive relays is analyzed. Finally, a batch of physical noise testing experiments was conducted on automotive relays to verify that the optimization scheme can significantly improve the noise of automotive relays.

This paper reported a method of improving endometrium blood flow. Multi focused ultrasonic probe combined with vaginal cervix probe to stimulate the acupoint, pelvic muscle and cervix. The focused ultrasound module used the digital frequency generator technique and adaptive load technology to produce freguency (1MHz, 0.8MHz), and the electrical stimulation module used the programmable waveform constant current technology. This study used the device to produce focused ultrasound and electrical stimulation to treat the patient who got thin endometrium. The results show statistically significant difference in the level of VI, FI, VFI, PI, PS compared to prior treatment (P<0.05), and there was no statistically significant difference in the level of endometrium thickness, volume, shape and menstrual cycle (P>0.05). This project developed a new device for improving endometrial blood flow, which can effectively improve uterine blood circulation in infertile patients.

The chip morphology of indexable insert drills affects chip evacuation performance, which in turn influences the dimensional accuracy and surface quality of the workpiece. Simultaneously, the cutting force impacts the balance of radial forces, further affecting the dimensional accuracy and surface quality. Therefore, studying the effects of chip morphology and cutting force on the high stability applications of indexable insert drills is of significant importance. This paper proposes an optimized insert groove design for high-stability applications, focusing on cutting force and chip radius optimization. First, based on the cutting principles of indexable insert drills, insert groove parameters were selected, and an orthogonal experimental scheme was designed. Key parameters affecting chip radius and cutting force were determined through simulation experiments. Second, the parameters were ranked by their influence on chip radius and cutting force, and the optimal parameter combinations were identified. Finally, the feasibility of the optimized insert groove design was verified through experiments.

The design and implementation of a bird repeller based on the MLX90640 infrared sensor is proposed. The system can carry out real-time temperature measurement and accurate bird repellent. After the experimental verification of the ambient temperature, measurement distance and bird repelling effect, this paper sets the threshold of ambient temperature segmentation control algorithm to improve the real-time monitoring of the system and the ability of response speed; the system hardware and software design is carried out. The detection system applies Infrared thermometry technology to effectively improve the effect of bird repellent.

In order to improve the design to improve the performance of the bearing, in view of the influence of the friction torque on the rotation of the bearing when the high-speed bearing is running, the bearing test method and technology are used to carry out the bearing clearance, surface processing quality, sealing grease, working parameters and other factors on the friction torque. Research and discussion of the impact. The research results show that the roundness, surface roughness, grinding wire flow and broken wire of the bearing raceway will cause the friction torque to fluctuate proportionally. At the same time, the bearing's working load, working temperature, structural parameters, and grease characteristics also have a great influence on the friction torque of the high-speed ball bearing. The results have certain guiding significance for the development of the high-speed ball bearing.

The research and development of scientific research projects benefit from efficient management. PDCA cycle management mode is an effective measure of current scientific research project management. The PDCA cycle theory describes a continuous management process that encompasses four key stages: planning, doing, checking, and acting. It aims to enhance project execution efficiency, mitigate project risks, and optimize resource allocation. Through this system to carry out research project management work to achieve the purpose of improving the overall management effect. This paper thoroughly explores the application of the PDCA cycle management model in the operations of schools. It not only analyzes the foundation of its rationality, but also dissects its effectiveness in practice. By thoroughly examining each step of the PDCA (Plan-Do-Check-Act) cycle management model, this paper aims to uncover its significant role in ensuring the efficient and orderly progress of scientific research projects.

Automatic generation control (AGC) units may have the problem of insufficient frequency modulation capacity due to wind power infiltration, which in turn leads to the difficulty of interconnected AGC control. Therefore, a hierarchical control strategy for the interconnected AGC system is proposed, and the BAS-MPC controller is designed on the basis of the conventional MPC control, and the hierarchical and progressive control of the interconnected AGC system is carried out in combination with the fractional-order PID controller, so as to realize the rational allocation of control resources. In this paper, a three-region interconnected AGC system including wind farms is constructed and simulated. The results show that, compared with the traditional AGC control method, the proposed control strategy not only has more advanced control performance, but also optimizes the control cost.

This paper is aimed at long-distance high-power laser wireless power transmission and researches the theory of laser atmospheric transmission to establish a physical model for calculating the laser power transmission characteristics in different atmospheres, which realizes the quantitative solution of laser power loss and beam quality degradation in the transmission process and simplifies the engineering calculations of laser atmospheric transmission characteristics. Based on the theoretical model, a laser atmospheric transmission simulation system is developed to comprehensively evaluate the laser radiate property and transmission process in complex atmospheric environments, which is of high value for the engineering design and system evaluation of laser wireless power transmission.

The electric power industry is an important part of the national economy, and the application of electric power big data is also a popular field that has attracted much attention in recent years, while the assessment of electric power data assetization has become one of the issues that need to be studied urgently. Based on the real options theory and considering the market value risk volatility, the paper constructs a power big data asset value assessment model, and takes the data of China's BYD new energy automobile as an example, and the results of the example show that the data asset value assessment model constructed in the paper can be effectively applied to the actual practice, and plays a role in promoting the process of data assetization.

This study employs a hybrid recommendation model that integrates Attention, CNN, and LSTM for course suggestions. The LSTM component is utilized to discern the interdependencies within user rating data, thereby identifying user preference patterns. Additionally, CNN is incorporated to capture the local feature relationships of the courses. An Attention mechanism is also applied to refine the weighting process. Valuable information regions are given greater weights, and user and course features are extracted more accurately. The predicted score is calculated, and recommendations are generated based on the ranking results of the predicted score. This article conducts simulation experiments on the proposed model using publicly available datasets. To further assess the effectiveness of the proposed recommendation model, a comparative study was carried out using established recommendation models. The experimental findings show that the model introduced in this paper surpasses the performance of the other models evaluated.

When evaluating neural network target detection models based on traditional methods, most of the models are evaluated from a single perspective, such as the recognition accuracy or robustness of the model, and the evaluation conclusions have certain limitations. The article proposes to take the adversarial sample attack in the field of artificial intelligence security into consideration, and proposes to comprehensively evaluate the model from multiple perspectives, such as the recognition accuracy, robustness and recognition reliability of the model, firstly, constructing three types of test samples, such as unfamiliar samples, transformed samples, and adversarial samples, and then, through the statistical comparison and analysis of the prediction results and the labels of the test samples, combined with the indexes of recognition correctness, misdetection rate, and leakage rate, the model can finally be evaluated quantitatively. Finally, the model can be evaluated quantitatively, and give the model evaluation conclusions including four graded grades: fail, pass, good and excellent. This model evaluation method can provide certain reference value for the rational use of the model.

In recent years, due to the social network, information network, biological network, science and technology network and many other disciplines, there are a lot of interconnections, the formation of a variety of complex networks. Link prediction is an important topic in a complex network, its main purpose is to predict the existing data of the existing network, and apply it to the friend recommendation in social network, protein interaction prediction, network evolution mechanism and so on. Aiming at the shortcomings of similarity analysis by network topology, the method of K-shell decomposition and importance denoising of adjacent nodes is used to predict links. Taking the optimal noise reduction as the starting point, the K-Shell decomposition method and the degree measurement of neighbor nodes are used to comprehensively evaluate the importance of each node in the network, while the noise environment is set and the nodes judged as noisy is eliminated, then the link prediction is carried out. Experiments show that the prediction accuracy of KSDNN is better than that of conventional similar methods for six networks of different sizes.

Rate control is an important module in video coding. A rate control algorithm with the saliency features of video content based on the high efficiency video coding (HEVC) standard is proposed in this paper. This algorithm is to extract the content saliency feature of natural video and bitrate is allocated to the region with strong salience more effectively. Firstly, the Gaussian pyramid of brightness and direction of natural content video was constructed by Gaussian sampling method, and the luminance and direction feature factors of video sequence images were calculated. The complexity significance factors of luminance and direction were obtained by combining the feature factors of different scales. The pixel level feature factor is used to guide the coding tree unit layer for target bit allocation. Then, a new rate-distortion model is constructed based on inter-frame correlation. Experimental results show that, compared with the reference algorithm, the average bitrate mismatch of the proposed algorithm is reduced to 0.031% under the condition of low delay coding, and the accuracy of the bit rate control is improved. Under the condition of the same quality, the average bitrate of the algorithm is saved by 3.55%, and the rate distortion (RD) performance of the encoded video is improved.

As one of the more advanced algorithms, the gating convolutional neural network (Gated Convolutional Neural Network, GCNN), has the advantages of parallelized computing, sparsity, multi-layer feature extraction, gating mechanism, and so on. GCNN uses SmoothL1Loss (smooth L1 loss) as the loss function. To improve the smoothness of the loss function, the sample membership, namely SmoothL1Loss + L (based on the smoothness loss of sample membership L), should be used. However, SmoothL1Loss + L, whose membership sum is one, has normalization, with the problem of excessive contribution of noisy samples and too low contribution of high membership samples. Therefore, this paper uses the relaxed normalization condition to derive a new membership division method and proposes a new loss function SmoothL1Loss + L * (L * function to adjust the membership under the relaxed normalization condition). To facilitate the comparison of the effect before and after the loss function, the filtering model of SmoothL1Loss + L * is called GCNN _ L *; the filtering model of SmoothL1Loss + L is called GCNN _ L. Experiments on the same data set, the results show that the mean absolute error of GCNN _ L * model 3.8% higher than GCNN _ L, respectively, so the SmoothL1Loss + L * function is competitive.

In recent years, global climate change and human activities have intensified the frequency and severity of floods. Based on the big data of flood probability, combined with the index data of monsoon intensity, river management, deforestation, urbanization and climate change, this paper studies from the perspective of flood probability prediction. Firstly, this paper selects indicators and collects corresponding data, and processes the data. Then, the importance of index characteristics is evaluated by the random forest algorithm. Secondly, the number of final indicators is determined by kernel principal component analysis, and the final indicators are determined by combining the importance of features. Then, the number of classifications is determined by the AIC/BIC criterion, and the risk is classified into three categories: high, medium, and low, by GMM clustering, based on the EM algorithm. Finally, 60% of the data is used as the training set, 30% of the data as the verification set, and 10% of the data as the test set to divide the data set, and then it is brought into the neural network prediction model for prediction. Finally, by analyzing the clustering results and prediction results, it provides a scientific basis and a practical application scheme for disaster prevention and reduction.

Due to the difficulty of PID parameters and range adjustment, long response time and small measurement and control automation parameters, the design and research of temperature automatic measurement and control system of coke jujube production equipment based on PLC is proposed. The hardware unit of the design system includes temperature acquisition unit, PLC configuration unit and temperature control unit; the software module includes temperature signal processing module, PLC control program module and limit alarm design module. Through the design of hardware unit and software module, the operation of automatic temperature measurement and control system of flexible production line is realized. The simulation experiment results show that compared with the standard value, the response time of the design system is short, the measurement and control automation parameters are large, which meets the safety requirements of the current flexible production line, and fully shows that the temperature measurement and control performance of the design system is better.

Virtual power plant(VPP) is a fusion application of "Internet of Things +" and electric power industry, a new generation of smart grid control technology and interactive business model that aggregates and optimises "source, network, load and storage", which relies on modern advanced communication technology, carries out collaborative optimization control and market transactions, realizes multi-energy complementarity on the power side and flexible interaction on the load side, and provides forward-looking solutions for cracking the problem of clean energy consumption and low-carbon energy transformation. It relies on modern advanced communication technology to carry out cooperative optimisation control and market trading, realises multi-energy complementarity on the power side and flexible interaction on the load side, creates an active, convenient and sharing ecosystem for emerging market players, provides forward-looking solutions for solving the problems of clean energy consumption and low-carbon energy transformation, and helps the construction of "three types and two grids". This paper focuses on the problem of VPP, firstly, combined with the actual operation of the power grid, analyses the main needs in the process of power grid operation. Secondly, it analyses the characteristics of VPPs, the application scenarios of participating in grid operation, as well as the connection relationship with the existing grid scheduling and the market, and finally, on this basis, it looks forward to the operation prospect of VPPs in China and the problems that need to be solved.

In the software industry, Software Reliability Growth Models (SRGMs) with confidence intervals (C.I.) are frequently employed as valuable tools to assist manager to determine the optimal timing for software releases and enhancing the effectiveness of testing processes. However, many SRGMs lack clear explanations regarding the estimation of variance for cumulative software errors. This limitation can impede their ability to accurately derive C.I. for the mean value function (MVF), making it challenging for developers to evaluate potential risks and uncertainties within the software system. Consequently, the practical application of these models may be compromised. To address this issue, this paper proposes the utilization of Ohba’s Inflection S-shaped model and Yamada’s delayed S-shaped model to construct SRGMs with C.I., thereby providing developers with improved guidance for making optimal release time decisions.

The traditional signalling systems have a high degree of independence, leading to high costs in resources and labor. In order to reduce the number of servers and achieve unified management of infrastructure, the automatic train supervision (ATS) subsystem can be incorporated into the cloud platform. The deployment scheme for integrating the servers and workstations in the control center into the cloud platform is described in this paper, based on the characteristics and existing architecture of the signal ATS. On this basis, the deployment mode of each host and the interface scheme between ATS and other systems are described. To ensure the normal use of the system in the event of malfunctions, the ATS subsystem of signal adopts a multi-level redundancy strategy. The integration of the ATS system into the cloud platform can meet the functions of ATS as well as the requirements of high security and reliability.

As blockchain technology is widely applied across various fields, the performance requirements of blockchain systems, such as throughput and transaction latency, are increasing. In this context, blockchain systems have gradually evolved to adopt a two-phase consensus protocol that involves election and block generation. By electing a small number of nodes from a large-scale network to form a committee, a more efficient internal consensus protocol can be executed, thereby improving the overall performance of the blockchain system. This paper proposes a node performance detection method based on smart contracts, which sets performance thresholds for committee nodes without compromising the decentralization of public chains. This ensures high performance, low latency, and enhances the reliability of the consensus protocol.

The integrated satellite-terrestrial network represents a pivotal direction in the evolution of sixth-generation mobile communication. It integrates the low latency, high capacity, and large-scale machine access characteristics of fifthgeneration mobile network with the extensive coverage and robust stability of satellite systems. A key trend in the development of future integrated network systems will be the shared user of a unified frequency band between satellites and terrestrial networks. To address this, the paper considers a cognitive satellite-terrestrial network scenario, where satellite networks serve as the primary network and terrestrial networks as the secondary. The paper applies deep reinforcement learning for multidimensional resource optimization, aiming to enhance user service quality while alleviating frequency resource scarcity. This paper delves into resource allocation methods of cognitive integrated satellite-terrestrial network based on deep reinforcement learning. The paper jointly optimizes the beam patterns, bandwidth allocation, terrestrial base station power allocation, and user association for low-orbit satellites within the system, thereby improving network throughput and ensuring user delay fairness. Initially, this paper systematically models the cognitive satellite-terrestrial network and formulated optimization objectives encompassing throughput and user delay fairness. Then a multi-agent based deep Q-network method was introduced to achieve these objectives. Through simulations, this paper validates that the proposed algorithm outperforms traditional resource allocation algorithms in terms of throughput and user latency fairness.

The rapid advancement of science and technology has underscored the significant potential of intelligent driving systems in the fields of robotics and automotive industries. We focus on the design and implementation of intelligent trolley driving system based on Raspberry Pi driver in this paper, in which a system that utilizes Raspberry Pi as its central control platform, leveraging its robust computational capabilities and diverse interface resources. By integrating the OpenCV computer vision library, the proposed system performs real-time environmental image acquisition and processing, accurately identifying and locating obstacles. In this study the hardware configuration is meticulously detailed, encompassing the selection and integration of Raspberry Pi components and sensor arrangements. The software design process is thoroughly explored, covering image acquisition algorithms, obstacle recognition and localization techniques, and the development of driving strategies. The rigorous testing demonstrated the system's ability to make swift and precise decisions across various environmental conditions, implementing effective problem-solving measures with high stability and reliability. Finally, we expect that this approach offers a viable technical solution for applications in related fields, contributing to the development of autonomous systems and intelligent transportation technologies.

As the key subsystem in the signaling system for urban rail transit, the stable operation of the zone controller(ZC) is very important to the safe operation of urban rail transit. The health assessment system of zone controller based on gray prediction model is proposed to solve the problem of lack of preventive maintenance of ZC. Firstly, according to the actual deployment, historical operating status and expert experience of ZC, the health evaluation index system and index deduction standard are constructed. Secondly, the analytic hierarchy process is used to determine the index weight. Then, fault analysis is performed on the status data obtained in real time, and the health score and health grade of ZC are calculated based on the evaluation model. Then the health trend of ZCis calculated using the gray prediction model based on the historical health score. Finally, the fault alarm, health assessment, trend analysis and maintenance suggestions of ZC are displayed through the browser interface. The field application verifies that the system can reflect the working state of the equipment reasonably, and has a good guiding effect for maintenance personnel to maintain ZC.

Map generation based on Unmanned Aerial Vehicle (UAV) image stitching can fully utilize the convenient image acquisition capabilities of unmanned aerial vehicle (UAV), enabling the rapid acquisition of high-resolution remote sensing maps for large scenes, meeting the needs of applications such as scene reconstruction and map navigation. However, generating large-scale maps requires stitching a large number of images, which incurs significant computational resource costs and is prone to error accumulation. To address this, this paper proposes an image stitching and fusion method that integrates the geographic information inherent in UAV images with an improved Scale-Invariant Feature Transform (SIFT) algorithm. The proposed method uses geographic information to divide a large number of images into strips, then stitches and fuses the images within each strip, ultimately stitching and fusing multiple strips to create the map. When calculating the transformation matrix, geographic information is used to determine the range of overlapping areas to form a mask that reduces the probability of mismatches, thereby improving matching accuracy. Experimental results demonstrate that the algorithm is well-designed and achieves low processing time while ensuring good stitching quality.

The ‘coal-to-electricity’ project plays an essential role in the electrification of heating field in North China. For ‘coal-to-electricity’ load, which can maintain sufficient power demand at night during the heating season, it has spatiotemporal coupling with local renewable power generation. In this paper, we build the time series production simulation model to analyze the renewable enerfy accommodation effects from ‘coal-to-electricity’ quantitatively. Simulation results show that the project can provide additional renewable energy consumption space and change the peak-valley load of the grid, which will affect the conventional unit start-up mode, the total minimum technical output of thermal units and peak shaving capacity, which may squeeze the accommodation space of renewable energy. These two aspects converge to form the supportive effect of ‘coal-to-electricity’ load on renewable energy accommodation during the heating season.

As the demand for privacy protection continues to grow, ensuring effective financial data processing while preventing sensitive information leakage has become a pressing issue in the financial industry and related research fields. We propose a financial data processing method based on Bayesian networks and differential privacy, innovatively introducing an exponential mechanism to optimize the Bayesian network and adding Laplace noise to disrupt the data, thereby reducing the risk of sensitive information leakage. Through the evaluation of data usability using support vector machines, experimental results show that the processed synthetic data is highly similar to the original data in key metrics, demonstrating that this method maintains data usability while protecting privacy.

Through the analysis of patent efficacy phrases, a study is conducted to identify high-value patents in the Chinese digital economic domain. This research aims to provide targeted guidance for industrial development and innovation, leveraging patent resources to facilitate further growth and innovation in the field of digital economics. This study organizes existing related research and constructs a high-value patent evaluation index system covering three dimensions: "technology-economics-law." It utilizes patent utility phrases in patent texts and their connections with patents to complete the construction of a patent-utility phrase heterogeneous graph. Subsequently, a heterogeneous graph attention model is adopted to construct an automatic identification model for high-value patents. Through experimental analysis, this study demonstrates that the heterogeneous graph attention network model, based on the perspective of patent utility phrases, achieves better predictive results in accuracy, recall rate, and F1 score compared to previous high-value patent identification models, proving the model's effectiveness and reliability.

The accuracy of distribution network load supply capability assessment from a planning perspective is significantly affected by the uncertainty in the actual load distribution and its growth patterns, as well as changes in the topology of the distribution network. An innovative methodology for assessing the power supply capability of distribution networks is presented in this paper, which takes into account both load growth patterns and network reconfiguration strategies. By integrating the natural growth of existing loads with the installation applications of large-scale users, and leveraging the theory of sequential operations, a model is proposed that accurately forecasts load trends. Further-more, an optimization model is constructed, which not only contemplates the feasibility of network reconfiguration but also incorporates the aforementioned load growth patterns, aiming to effectively prevent power supply bottlenecks. In order to calculate the load supply capacity of a distribution network, an improved repetitive power flow method is employed. The effectiveness of the proposed approach is substantiated through case studies on the PG&E 69-bus system. Experimental results demonstrate that the methodology offers a more precise evaluation of the distribution network's power supply capability, exhibiting commendable practicality and adaptability across various scenarios.

Line loss not only directly affects the economic efficiency of power enterprises, but also is a key indicator for assessing the operational efficiency of power grids, and line loss management is an important work of power supply enterprises. In order to predict whether the effectiveness of line loss management in high-loss areas can reach the required target, it is necessary to analyse the main factors causing abnormal power loss in high-loss areas and assess the effectiveness of line loss management. This research report focuses on the methodology of the loss reduction effectiveness evaluation model for high-loss stations and lines. Through the analysis, the loss reduction effectiveness evaluation index system is determined from the economic and technical dimensions. Finally, using the entropy weight method, the loss reduction effectiveness assessment model method is proposed.

Materials with van der Waals (vdW) bonding in one, two, or three dimensions have attracted much attention due to their possibility to reduced dimensionality by mechanical exfoliation. In this work, by using the XGBoost algorithm, we successfully built the machine-learning (ML) models for formation energies and bandgaps for materials with vdW bonding, and the performances of the built ML models, i.e. R2 and MAE values, reach 0.98/0.02 eV/atom and 0.89/0.56 eV, respectively. Based on the analysis of feature importance and SHAP values, 20 important features are identified and further reduced by the correlation analysis. Our work not only demonstrates the effectiveness of the XGBoost method in modeling key properties of materials with vdW bonding, but also contributes to the understanding of the critical properties in these materials.

Large-scale distributed PV access to the low-voltage distribution network is prone to cause serious power back-feeding, resulting in PV distribution transformers in the distribution network reversing heavy overload and node voltage rise over the limit, exceeding the distributed PV carrying capacity in the distribution network. In response to the issue, based on the full consideration of the distribution network, the economic and safe operation of the energy storage system. We construct a two-layer optimization model of the distributed PV storage, considering the PV carrying capacity in the distribution network, the power grid's security, and the economy of the energy storage system. The simulation and analysis of selected actual PV heavy overload areas prove the effectiveness of the configuration method for managing the problem of PV heavy overload areas exceeding the carrying capacity and reducing the comprehensive cost.

Aiming at the problem of radio observation data quality degradation or invalid data caused by Radio Frequency Interference (RFI), we have studied the adaptive RFI algorithm and proposed an interference mitigation method based on the adaptive filter. An adaptive interference suppression model has been established to effectively counter complex and intense interference. This model includes a normalized factor, derived from the filter’s correlation function, which improves the performance of the adaptive filtering algorithm in reducing RFI. The adaptive filter algorithm flexibly adjusts the filter gain automatically, based on the correlation between the RFI reference signal and the observation signal, thereby effectively eliminating the RFI mixed with astronomical signals. Simulation results show that the algorithm can effectively eliminate radio frequency interference, improve the signal-to-noise ratio of astronomical signals and signal recognition rate. It also prevents the loss of signal bandwidth, ensuring the provision of reliable data for subsequent post-processing analysis.

Copper minerals, as a critical industrial raw material, are essential to global economic development. To explore the research status, hotspots, and frontiers in the field of copper mineral prediction, 472 Chinese articles and 152 English articles published between 1974 and 2024 were collected from the CNKI and Web of Science WoS core databases. Using the community structure analysis software CiteSpace, a visualization analysis was conducted on aspects such as collaborative authorship, research countries, research institutions, keyword clustering, and spatiotemporal keyword distribution maps. The results reveal an evolutionary trend in copper mineral prediction research from fundamental theory to technical applications: early studies focused primarily on basic issues such as metallogenic conditions, metallogenic processes, and geochemical anomalies, while recent studies have gradually shifted toward the application of intelligent technologies such as predictive models, deep mineral exploration, and spatial data analysis. Domestic research emphasizes resource development and precise prediction in specific regions, whereas international research focuses more on the introduction and promotion of intelligent methods. In the future, technologies such as data mining, machine learning, and 3D modeling are expected to further advance the efficient exploration and development of copper mineral resources This study provides an important reference for understanding research dynamics and directions in the field of copper mineral prediction.

As the public pays more and more attention to fire, the importance of fire prevention is becoming increasingly prominent. This paper focuses on the realization of fire alarm software that can alarm in real time. Through in-depth research on sensor technology, cloud computing technology, mobile application development framework and real-time data processing algorithm, the accurate monitoring of smoke concentration and timely alarm function are realized. Through cloud server communication technology, the software can be stably connected with the supporting hardware equipment to complete the alarm notification in real time. At the same time, by continuously querying the server data at a frequency of once per second, it ensures that the alarm notification is issued to the user as soon as the server alarm data is updated. In addition, the software has a simple user interface, which is convenient for users to set up and view historical data, and fully considers the convenience of use. After rigorous testing, this software has excellent performance in accuracy, real-time and stability, providing a convenient and efficient solution for improving fire prevention capabilities, and has broad market prospects for its application in fire protection.

Radar HRRP is easy to be obtained and contains rich target information. According to the scattering center theory, in the high frequency range, the target radar echo can be regarded as a superposition of several discrete and independent scattering center echoes. Extracting scattering center features from HRRP can be used for automatic target recognition. However, HRRP is generally complex value data, and traditional real-valued neural networks are difficult to process directly. To address this issue, this paper proposes a complex-valued neural network structure and designs the loss function based on the sparsity of the scattering center parameter space for HRRP scattering center extraction. Simulation experiments verify the effectiveness of the method.

Due to the complex structure and numerous electronic components, the reliability analysis of electronic systems has become more challenging. Therefore, we propose a reliability analysis method for electronic systems based on survival signature. The level of fault impact factors is evaluated based on the survival signature theory, and the weight of fault impact factors is evaluated through the analytic hierarchy process. This method simplifies the structural analysis process of complex systems, making reliability analysis more objective and efficient. The effectiveness of this method was verified through reliability analysis of the digital output control circuit.

The contradiction between large stroke drive and high-precision feed is difficult to solve in ultra precision special machining. This article takes three-dimensional structure machining as the object, constructs a prototype manufacturing system for large-scale three-dimensional structures, and proposes a solution to improve positioning accuracy. Through the control method of digital fuzzy control+PID combined with bidirectional static error compensation, the platform achieves sub micron level accuracy within a hundred millimeter stroke, ensuring the high response characteristics and low-speed stability of the servo system. The experimental results prove that the system runs well and can achieve high positioning accuracy, meeting the requirements of micro fabrication.

On the basis of the type-2 intuitionistic fuzzy sets and the fundamental operations, this new paper latest research the definitions of the new type-2 bipolar-valued fuzzy sets, nextly the paper define some fundamental operations of type-2 bipolar-valued fuzzy sets. At last, the paper also deeply study some properties of the fundamental operations.

The modeling of street trees is of great significance to the research of highway landscape design. However, it is very difficult to accurately model a complete and realistic street tree. This paper simplifies the tree model by only retaining the main branches and imagining the branches as cones that bent along a 3D spline curve with gradually decreasing radius. Based on this simplification strategy, the tree branches are modeled by the main trunk and sub-branches, and finally, a spline curve skeleton and its 3D model are obtained. After that, the point cloud of tree branches can be obtained by simulating laser scanning to sample points on the 3D model. This method of obtaining 3D tree models, point clouds, and skeletons is flexible, efficient, and low-cost, which can efficiently generate a large amount of training data for further AI model trainings.

Traffic flow data, as a new type of time-series data, the missing phenomenon is common and inevitable due to its own spatio-temporal features and the influence of complex external factors. This paper analyzes the characteristics of traffic flow data and the existing problems in imputation approaches, combines the idea of generative data from Generative Adversarial Network, and fuses the spatio-temporal correlation of traffic data, then proposes the spatio-temporal Generative Adversarial Network to impute missing traffic flow data (ST-IMGAN-ext). Experiments were performed on the real open source TaxiBJ GPS dataset. The experimental results were evaluated by Root Mean Square Error (RMSE). Compared with the traditional imputation methods, the method, proposed in this paper, is proved to be more accurate and applicability.

Study Finds Flu Virus Gene Sequences Differ in Each Outbreak. Viruses have been in a state of mutation during the evolutionary process, in which a certain segment of the sequence changes, but there is also a similar distribution of bases. Technological advances have allowed rapid access to coronavirus data, making it difficult to directly analyze large amounts of data. By visualizing the gene sequences into images and analyzing them with mature image processing techniques, we can effectively reveal the association between sequences and base distribution characteristics. In this paper, the visualization method for processing coronavirus gene data is based on the ideas of variational logic, bioinformatics and statistics, Firstly, the gene sequence was segmented, the base sequences were transformed accordingly and the number of bases in each segment was counted to calculate their corresponding probabilities; secondly, the probabilities were transformed using Polar Coordinate System, and then the Gramian matrix are used to map the polar coordinates into matrices, so that the one-dimensional sequences are processed into two-dimensional mapping, and the graphical results are analyzed at last. graphical results can be analyzed.

Vehicle detection is a crucial part of solving the congestion problem. Compared with the traditional video detection, geomagnetic detection has the characteristics of simple equipment, convenient signal processing and all-weather detection. A vehicle detection algorithm based on the geomagnetic adaptive threshold is proposed in this paper . Firstly, the magnetic dipole array model is introduced. According to the magnetic field characteristics of the measured vehicle, it is found that the magnetic field of the vehicle could be better simulated by using three magnetic dipoles with an interval of 2m. Then, considering the external environment interference, a detection algorithm based on the fluctuation of the magnetic field base value is studied. And the problem of magnetic shift caused by environment and temperature is solved by geomagnetic adaptive threshold. Finally, the parked state of the vehicle is judged by the state machine to prevent misjudgment. Compared with the traditional detection algorithm, the interference of the external magnetic environment can be well filtered. The detection of vehicles would be fast and accurate.

The increasing use of carbon fiber-wound Type III hydrogen storage tanks necessitates the development of effective nondestructive testing (NDT) methods to ensure their structural integrity. Traditional NDT techniques have limitations in detecting specific defects like debonding within these composite structures. This study explores the application of microwave-based NDT for identifying debonding defects in Type III hydrogen storage tanks. A Fermi antenna, known for its wide operational bandwidth and stable modes, was utilized as the microwave probe in conjunction with a vector network analyzer. The system was tested on tanks with debonding defects of 6cm and 9cm lengths. Frequency scanning identified an optimal operational band around 18.99GHz. Experimental results demonstrated that the Fermi antenna successfully identified the presence of debonding defects through variations in the reflection coefficient (S11) values. The study concludes that microwave-based NDT using a Fermi antenna is a promising method for detecting debonding defects in composite hydrogen storage tanks, providing a non-contact, efficient, and sensitive alternative to existing methods.

As an indispensable high-precision magnetic field measurement element in modern electronic technology, fluxgate sensors play an important role in space exploration, positioning and navigation, geological exploration and other fields. In practical application, due to machining technology and other reasons, three-axis magnetic sensors exhibit nonorthogonality errors, sensitivity inconsistency errors, and zero offset errors. In this paper, a calibration method of fluxgate sensor based on ellipsoid fitting is studied. Firstly, the error sources of fluxgate sensor are analyzed, including non-orthogonality, sensitivity and zero offset, and the corresponding error models are established. Then a calibration method based on ellipsoidal fitting is adopted. By collecting the magnetic field data of the sensor in different directions, nine parameters of the ellipsoid are fitted by the least square method. Afterwards, the three-axis fluxgate sensor's magnetic field data undergoes a calibration process to rectify and offset any sensor inaccuracies. Finally, in order to confirm the calibration method's efficiency, a MATLAB-based simulation test has been carried out. The magnetic measurement data of the fluxgate sensor calibrated by an ellipsoidal fitting method is significantly improved.

The key to solve the problem of low frequency noise in substation is to accurately grasp the low frequency spectrum contribution without weight of the main noise sources and the plant boundary. Taking 220kV substation as an example, the noise measurement of main equipment and plant boundary in the station was carried out, and the A-weight sound pressure level at each measuring point and the data in the weight frequency domain were recorded. It was calculated that the low-frequency noise line spectrum of 100Hz-500Hz for the main transformer accounted for 21.9% of the total energy of 100Hz-20000Hz. The 100Hz energy at the plant boundary accounts for 31.8% of the total energy of 100HZ-500Hz, and the analysis shows that the low-frequency line spectrum energy of substation noise accounts for a relatively high proportion, and the 100Hz line spectrum should be taken into account in the process of noise reduction and treatment in substation.

Using an X-ray detector to detect the fractures and damage of invisible steel cords on the surface of the conveyor belt; using an infrared detector to monitor dangerous heat sources on the conveyor belt to prevent open flames and avoid overheating of the belt and equipment caused by friction; using a foreign object identification camera to intelligently identify various debris on the conveyor belt, detect surface wear, and monitor belt misalignment; using a laser sensor to identify fine cracks and damage on the inner surface of the conveyor belt; using an ultrasonic sensor to detect severe cracks and damage on the inner surface of the conveyor belt under conditions of heavy smoke and water vapor. By collecting data through infrared, ultrasonic, machine vision, X-ray, and other sensors, and adopting multi-sensor information fusion technology, the system reduces false alarms and missed detections, improving overall reliability. Additionally, a conveyor belt status monitoring platform has been developed to achieve comprehensive monitoring, health management, visual monitoring, and maintenance guidance for the conveyor belt. The research results play an important role in promoting the diagnosis and prediction of longitudinal tear faults in conveyor belts, avoiding unplanned shutdowns, ensuring the efficient, reliable, and continuous operation of the transportation system, and improving production efficiency.

The theory of surface oxygen corrosion of Mn-Fe (111) surface is discussed deeply by studying the corrosion theory of Mn-Fe (111) surface. The results show that Mn atom doping can significantly improve the corrosion resistance of Fe (111) surface. By analyzing the calculated electronic structure, we find that Mn atom doping changes the electron state density distribution on the surface of Fe, which affects the adsorption and dissociation process of oxygen molecules on the surface.

Fabric defect detection is an essential process for maintaining product quality in the textile industry. In this work, we propose a novel deep learning network named DAP-YOLO for fabric defect detection. The key idea is to integrate dynamic convolution into the framework of YOLOv9s. First, the improved dynamic snake convolution C-DSC is introduced to recognize adaptively the subtle and intricate features of fabric surface defects. Second, the original kernel-based dynamic upsampling operator is replaced by a super-lightweight dynamic upsampling operator, aiming to enhance effectively the capability of identifying complex and irregular target edges. At last, the PIoUv2 loss function is adopted to improve the convergence and scale adaptability. The experimental results on self-produced datasets demonstrated that the DAP-YOLO model achieved superior performance of fabric defect detection to mainstream methods.

With the advancement of computer and human speech signal processing technology, the functionality and practicality of English pronunciation learning systems have been further enhanced. Gaussian Mixed Model (GMM) is a model that accurately quantifies things using Gaussian probability density function, decomposes a thing into several models based on Gaussian probability density function, and can directly approximate any probability distribution. GMM also combines the advantages of a single Gaussian density function and vector quantization, which can more effectively describe the distribution of speaker speech features. After extracting speech features using mel scale frequency cepstral coefficients (MFCC), the GMM algorithm is used for text-dependent English pronunciation learning system. The designed English pronunciation learning system is implemented on MATLAB software. This design collected the speech of 50 people with text-dependent for similarity English pronunciation, and after testing, the similarity reached over 93% under normal conditions.

Aiming at the problems such as complex structure and large number of parameters, a lightweight substation defect identification and detection model with improved YOLOv5s was proposed. Firstly, the depth-separable convolution and inverse residual structure of the MobileNet model are used for reference, and combined with the CBAM attention mechanism, a Bneck structure is formed to replace CBS convolution, construct a Bottleneck network, and replace the Bottleneck structure of the C3 structure, thus forming the C3-BNECK module. C3-bneck module is used to replace all C3 modules in the backbone network, and DWConv deep convolution is used to replace CBS convolution module, which greatly reduces the number of model parameters and realizes model lightweight. Introduce CBAM attention mechanism at the same time, improve detection precision of the model. Finally, K-Means ++ algorithm is used to cluster prior boxes to accelerate the convergence speed and reduce the error of clustering results. Compared with the original YOLOv5s model, mAP is increased by 3.9%, the number of parameters is reduced by 58.6%, the model volume is reduced by 34%, and the calculation amount is reduced by 76.8%. The detection speed reaches 102FPS, which is 29.7% higher than before the improvement. The experimental results show that the improved YOLOv5s model can effectively improve the accuracy and speed of defect detection of transformer equipment while ensuring lightweight, and is easy to deploy and meet the requirements of full power operation and maintenance field.

With the continuous iteration and development of computer vision technology, intelligent applications and devices centered around this technology are gradually playing an increasingly important role in human daily life and military operations. Among these, technologies based on object recognition and detection are widely applied in fields such as robotics, drones, and aircraft combat. These areas require precise localization information provided by object detection technology to achieve functions like navigation or attack. However, due to the characteristics of the targets being detected, their scale changes according to their distance, resulting in significant scale variations in images for the same target. Timely detection of distant targets is crucial for aircraft control and early warning, making small target recognition a key factor limiting the widespread application of recognition algorithms. To address this issue, this paper processes the original images based on color space level, global level, and pixel level, proposing a small target recognition method for RGB images. This method involves separately processing different color spaces of the RGB image, followed by information fusion, analysis of the entire image and local regions, and pixel-level filtering in local areas. Experimental results demonstrate that our algorithm can accurately recognize targets of varying sizes and observation angles.

Ice accumulation on cables can severely compromise the stability and safety of power transmission, making efficient detection and de-icing of icing cable critically important. This study proposes a method for cable icing detection and de-icing based on infrared technology. By integrating infrared detection and heating technologies, real-time icing detection and rapid de-icing are achieved. The icing state on the cable surface is analyzed using infrared imaging and image processing techniques. The accuracy of icing detection and the effectiveness of de-icing are validated by comparing infrared images and data distributions before and after de-icing. Experimental results demonstrate that the integrated infrared detection and de-icing system exhibits reliable performance in cable de-icing, providing a novel solution for cable maintenance in low-temperature environments.

Evaluating balance is crucial in gait rehabilitation. Traditional methods mainly depend on clinicians' subjective observations, which is complicated and time-consuming. While sensor-based research offers in-depth gait analysis, it lacks direct applicability to daily walking scenarios. This research introduces a novel method for assessing balance in patients, incorporating IMUs and pressure sensors. The study involved 19 hemiplegic patients and 6 healthy individuals, all of whom wore the sensors and walked along ten meters post-BBS testing. The data were analyzed to identify features. The SHAP model integrated with three classifiers, was used to determine the significance of different features. These were then applied as weights in calculating the weighted distance between every participant's gait features and the mean value of those in normal subjects. This computed distance is used to assess patients' balance abilities. Experimental findings revealed a linear correlation of -0.943 between the proposed index of patients’ balance ability in this research and the BBS scores given by therapists using the Pearson coefficient, suggesting that this new method could effectively replace the complex Berg balance scale.

This paper proposes a theme recognition method based on BERT-LDA model, which makes up for the problem of missing contextual semantic associations in LDA model and improves the theme extraction accuracy; secondly, K-means clustering is used to carry out semantic association clustering analysis on spliced vectors, and the recognition of the research theme is realised through the clustering results and not using the feature words within the clusters. Taking the expert group in the field of highly skilled personnel as an example for empirical research, the results show that the BERT-LDA model improves the theme coherence by 5.1%, 3.0% and 3.4% respectively compared with the traditional LDA model, TF-IDF and Word2Vec model in theme recognition. It shows that the BERT-LDA model can improve the accuracy and precision of topic recognition to some extent. Based on this, the characterisation of the expert group portrait in the field of highly skilled personnel is analysed in terms of personal attributes, research themes and other dimensions, with a view to providing support for the development of the field of highly skilled personnel.

To clarify the frequency components of the sound source of the newly developed locomotive and the contribution of each sound source to the overall noise, so as to implement effective noise reduction treatment and rectification measures, and ensure that the locomotive meets the noise test acceptance standards. The source, distribution and main frequency components of locomotive noise are deeply analyzed by using microphone array sound source identification algorithm. Through the sound intensity method, the contribution of different noise sources and frequency components to the overall noise is accurately evaluated, and based on this, a targeted noise reduction scheme for locomotives is proposed. In the process of the experiment, the noise data of the locomotive running is collected, and the algorithm is used for detailed analysis. The experimental results show that the method can accurately identify and quantify the contributions of different noise sources, which provides a strong theoretical support for traffic noise control. This research result not only has theoretical value, but also shows significant noise reduction effect in practical application, which is of great significance to traffic noise control.

Using Python software and pertinent knowledge of the Radon transform, the challenge of improving CT system parameter calibration and reconstructing high-quality images is solved in this paper. First, the correlations and rotation centers between the CT system parameters are established by reviewing and classifying pertinent data. Then, the affine transformation brought on by external causes is resolved using the picture that the iRadon transform has rebuilt. The 256×256 absorptivity matrix is once more obtained using the modified filtered back projection procedure, and the associated point's absorptivity is obtained using the Lagrange polynomial method for high precision. Finally, another new algorithm is chosen to recalculate using the obtained data, starting with the detector unit spacing calculation, and the difference between the two is compared to assess correctness and stability. Based on this, a better calibration template is suggested, one that eliminates the need for unique positioning techniques, makes calibration processes simpler, and also increases stability. Each of these advances imaging studies and CT system parameter calibration.