Measuring method and measuring system for parameters of contact rail system

Abstract

The invention relates to the technical field of rail transit equipment measurement and monitoring, in particular to a method and a system for measuring parameters of a contact rail system, which are used for acquiring data by using a magnetic field inductive sensor and an optical interference sensor and generating high-precision contact rail parameters in combination with a multi-level adaptive Bayesian reasoning and physical modeling technology. And then, anomaly detection and verification are performed on the parameter data by using a graph neural network and a dynamic resonance analysis technology, and finally verified anomaly feature data and optimized parameter data are integrated to generate a comprehensive metro contact rail system measurement report and maintenance suggestion. According to the invention, the measurement precision and the anomaly detection capability of the metro contact rail system are obviously improved, and scientific support is provided for safe operation of the metro system.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit equipment measurement and monitoring, and in particular to a method and system for measuring contact rail system parameters. Background Art

[0002] The measurement of contact rail system parameters is of great significance in rail transit systems. Whether it is a maglev track or a subway track, the physical parameters of the contact rail (such as deformation, stress concentration, etc.) are directly related to the safety and stability of the train operation. However, there are significant differences between the maglev contact rail and the subway contact rail in terms of structure, operating environment and measurement requirements. The measurement of the maglev contact rail mainly focuses on the precise alignment and relative position parameters between the track and the maglev train, which are crucial to maintaining the suspension stability of the train. The prior art (Chinese invention patent, publication number: CN115930793A, name: A maglev contact rail parameter measurement system and device) mainly uses laser sensors and embedded computing systems to automatically measure the three-dimensional data and track gauge value of the maglev contact rail. However, due to the high precision requirements and relatively stable physical conditions of the maglev system, this method faces certain limitations when dealing with more complex subway contact rail systems.

[0003] The main difference between the subway contact rail system and the maglev contact rail system is that the subway contact rail needs to withstand greater physical stress and frequent changes in train loads, which makes it more susceptible to structural deformation, wear and stress concentration. In addition, the operating environment of the subway contact rail is more complex and is affected by multiple factors such as temperature, humidity, and vibration, resulting in more significant nonlinear dynamic behavior and resonance phenomena. Therefore, the measurement of subway contact rails requires not only high-precision data acquisition, but also more complex and sophisticated anomaly detection and dynamic analysis technologies.

[0004] Whether it is a maglev contact rail or a subway contact rail, the challenge is how to accurately measure the physical parameters of the contact rail under complex operating conditions and promptly detect potential structural anomalies. Existing measurement methods have problems such as insufficient accuracy, high missed detection rate, and inability to fully capture the nonlinear dynamic characteristics of the system when dealing with these challenges.

[0005] Due to the complexity of its operating environment, subway contact rails are more susceptible to external environmental influences, resulting in more variable structural states, especially in terms of stress concentration, resonance, and nonlinear dynamic behavior. Existing measurement methods are difficult to effectively address these challenges. Therefore, a new measurement method is needed that can not only solve these common problems, but also provide a more comprehensive and accurate measurement and detection solution for the unique needs of subway contact rails. Summary of the invention

[0006] In view of the many problems existing in the above-mentioned prior art, the present invention provides a method and system for measuring contact rail system parameters. The present invention combines multi-level adaptive Bayesian reasoning with physical modeling technology to achieve high-precision calculation of complex parameters of subway contact rail systems, and effectively identifies and verifies potential anomalies in the system through the combination of dynamic resonance analysis and graph neural networks. Finally, by integrating the generated optimized parameters and abnormal feature data, a comprehensive health assessment and maintenance recommendations for the contact rail system are provided, which significantly improves the reliability and safety of the system.

[0007] A method for measuring contact rail system parameters comprises the following steps:

[0008] Using a magnetic field induction sensor and an optical interference sensor to collect initial data of the subway contact rail, and synchronously processing and pre-screening the initial data to generate pre-screening data;

[0009] The pre-screened data is subjected to adaptive filtering, multi-resolution analysis and normalization processing, and then feature selection and dimensionality reduction processing are performed to generate key feature data;

[0010] By using a method combining multi-level adaptive Bayesian reasoning with physical modeling, multi-mode fusion and complex parameter calculation are performed on the key feature data, and the calculation results are optimized through an adversarial learning mechanism to generate optimized contact rail parameter data;

[0011] Performing multi-dimensional nonlinear analysis on the optimized contact rail parameter data, combining chaos coupling analysis with graph neural network to perform complex anomaly detection, and generating verified abnormal feature data;

[0012] The verified abnormal characteristic data and the optimized contact rail parameter data are integrated to generate a contact rail system measurement report.

[0013] Preferably, the data synchronization processing includes performing time alignment processing on the collected initial magnetic field induction data and initial optical interference data respectively, and performing spatial coordinate standardization processing on the aligned data.

[0014] Preferably, the multi-resolution analysis adopts wavelet transform technology, which specifically includes performing wavelet decomposition on the pre-screened data, extracting characteristic signals of the contact rail of different frequency components, wherein the high-frequency components are used to identify subtle surface defects and stress concentrations, and the low-frequency components are used to capture the overall deformation trend of the contact rail, and the decomposed components are reconstructed to generate multi-resolution feature data containing multi-scale feature information.

[0015] Preferably, the complex parameter calculation includes a method combining multi-level adaptive Bayesian reasoning with physical modeling, specifically including the following steps:

[0016] Perform Bayesian reasoning on key feature data, set prior probability distribution, and perform initial reasoning based on observed data to generate first reasoning parameter data;

[0017] Inputting the first inference parameter data into a physical modeling module, performing mechanical modeling on the contact rail system, and generating physical modeling data through mechanical equations and physical constraints;

[0018] Feeding the physical modeling data back to the Bayesian reasoning model, adjusting the prior parameters in the reasoning model, and performing multi-level adaptive reasoning until a final reasoning result that meets the physical conditions is generated to form final reasoning parameter data;

[0019] The final contact rail parameter data is calculated according to the following calculation expression:

[0020]

[0021] Among them, P(θ|D) represents the posterior probability of parameter θ when the observed data D is known; P(D|θ) represents the likelihood function of the observed data D when the parameter is assumed to be θ; P(θ) represents the prior probability of parameter θ; P(D) represents the marginal likelihood of the observed data D.

[0022] Preferably, the calculation results are further optimized through an adversarial learning mechanism, including setting up a generation network and a discriminant network, the generation network generates simulated contact rail parameter data based on the final inference parameter data, the discriminant network distinguishes between the simulated data and the real data, and in each iteration, adjusts the parameters of the generation network to make the generated data closer to the real data, and at the same time adjusts the discriminant ability of the discriminant network until the output data of the generation network is indistinguishable from the real data, thereby generating optimized contact rail parameter data.

[0023] Preferably, the multi-dimensional nonlinear analysis includes:

[0024] The chaotic characteristics of the contact rail system are quantitatively analyzed by calculating the Lyapunov index to generate chaotic characteristic data. The calculation expression of the Lyapunov index is as follows:

[0025]

[0026] Where λ represents the Lyapunov exponent; t represents time; δX i (t) represents the small disturbance of the contact rail system characteristics at time t; δX i (0) represents a small disturbance at the initial moment;

[0027] Combined with chaotic coupling analysis, the nonlinear dynamic behavior between various characteristics in the contact rail system is identified and nonlinear coupling characteristic data is generated.

[0028] Preferably, the nonlinear coupling characteristic data is combined with the structural modeling of the graph neural network, including using the graph neural network to model the spatial structure of the contact rail system, mapping the nonlinear coupling characteristic data into graph structure data, wherein the nodes represent the key positions of the contact rails and the edges represent the connection relationships of the contact rails, and analyzing the nodes and edges in the graph structure data, identifying local structural anomalies in the contact rail system through the graph neural network model, and generating structural anomaly data.

[0029] Preferably, the structural anomaly data is combined with dynamic resonance analysis to perform complex anomaly detection, including resonance analysis of the stress response of the contact rail system at different frequencies, identifying structural anomalies caused by the resonance effect, combining the resonance analysis results with the structural anomaly data, generating verified abnormal feature data, and feeding back to the optimized contact rail parameter data generation process.

[0030] Preferably, the generation of the contact rail system measurement report includes generating a health assessment report and maintenance recommendations for the subway contact rail system based on the verified abnormal feature data and the optimized contact rail parameter data, and applying the health assessment report and maintenance recommendations to the real-time monitoring system of the subway contact rail system.

[0031] A measuring system for implementing the method for measuring the parameters of a contact rail system, comprising the following components:

[0032] Magnetic field induction sensor module, used to collect initial magnetic field induction data of subway contact rail;

[0033] Optical interference sensor module, used to collect initial optical interference data of subway contact rails;

[0034] A data synchronization and pre-screening module, used for synchronously processing and pre-screening the initial magnetic field sensing data and the initial optical interference data to generate pre-screening data;

[0035] A data preprocessing module, including an adaptive filtering unit, a multi-resolution analysis unit and a normalization processing unit, for processing the pre-screening data and generating key feature data;

[0036] A multi-mode fusion and complex parameter calculation module, including a multi-level adaptive Bayesian reasoning unit, a physical modeling unit and an adversarial learning unit, is used to perform multi-mode fusion, complex parameter calculation and optimization on the key feature data to generate optimized contact rail parameter data;

[0037] A multi-dimensional nonlinear analysis and anomaly detection module, including a chaotic coupling analysis unit and a graph neural network unit, is used to perform nonlinear analysis and complex anomaly detection on the optimized contact rail parameter data to generate verified abnormal feature data;

[0038] The data integration and report generation module is used to integrate the verified abnormal feature data with the optimized contact rail parameter data and generate a contact rail system measurement report.

[0039] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0040] The present invention realizes the accurate calculation of complex parameters of subway contact rail system by combining multi-level adaptive Bayesian reasoning with physical modeling, and effectively solves the problem of insufficient accuracy of existing magnetic levitation contact rail measurement technology when processing subway contact rails.

[0041] The present invention realizes efficient identification of local abnormal structures of subway contact rail systems by combining dynamic resonance analysis with graph neural networks, overcoming the defect that the prior art cannot cope with the complex operating environment of subway contact rails.

[0042] The present invention generates a comprehensive subway contact rail system measurement report by integrating the verified abnormal feature data with the optimized contact rail parameter data, which not only improves the measurement accuracy but also provides a scientific basis for the preventive maintenance of the subway system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the process of the present invention;

[0044] Figure 2 A flowchart of multi-level adaptive Bayesian reasoning and physical modeling in the present invention;

[0045] Figure 3 It is a schematic diagram of the structural analysis of the graph neural network in the present invention;

[0046] Figure 4 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0047] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0048] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0049] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0050] like Figure 1 As shown, a method for measuring contact rail system parameters comprises the following steps:

[0051] Using a magnetic field induction sensor and an optical interference sensor to collect initial data of the subway contact rail, and synchronously processing and pre-screening the initial data to generate pre-screening data;

[0052] Preferably, the data synchronization processing includes performing time alignment processing on the collected initial magnetic field induction data and initial optical interference data respectively, and performing spatial coordinate standardization processing on the aligned data.

[0053] The magnetic field induction sensor is mainly used to detect the stress distribution and material properties inside the contact rail, while the optical interferometer sensor is used to accurately measure the tiny deformation and geometric deviation of the contact rail surface. By combining these two sensors, the key parameters of the contact rail system can be fully obtained to ensure the comprehensiveness and accuracy of the data.

[0054] After data collection, in order to ensure the comparability of multi-source data and the accuracy of subsequent processing, the initial magnetic field induction data and the initial optical interference data must be synchronized. The core of data synchronization processing lies in time alignment and spatial coordinate standardization. Time alignment processing ensures that the data collected at the same time point are consistent and avoids data deviation caused by collection at different times. In the present invention, time alignment processing is specifically implemented by adding an accurate timestamp to each data point, which ensures that during subsequent analysis, data from different data sources can be compared and fused under the same time reference, thereby improving the accuracy of the analysis.

[0055] At the same time, spatial coordinate standardization is crucial for unifying multidimensional data collected by different sensors under the same spatial reference. Since the magnetic field induction sensor and the optical interference sensor collect data at different spatial positions respectively, in order to ensure the spatial consistency of the data, the coordinate system needs to be standardized. By performing spatial coordinate standardization on the aligned data, these data can be mapped to a unified coordinate system, so that in the subsequent data fusion and analysis process, data from different data sources can be accurately spatially compared and calculated. This process is achieved in the present invention by accurately calibrating the sensor installation position and mathematically mapping the data to ensure that the spatial position of each data point remains consistent, thereby laying the foundation for generating high-quality pre-screening data.

[0056] After completing the data synchronization processing, pre-screening is performed to further remove noise and invalid information in the data and improve the data quality. Pre-screening is mainly achieved through threshold filtering and noise suppression algorithms, with the aim of retaining key information related to the health status and performance of the contact rail system and excluding invalid data points that may be caused by external interference or sensor errors. After pre-screening processing, the generated pre-screening data will be used for subsequent in-depth analysis and calculation to ensure the efficiency and accuracy of the entire measurement system.

[0057] The pre-screened data is subjected to adaptive filtering, multi-resolution analysis and normalization processing, and then feature selection and dimensionality reduction processing are performed to generate key feature data;

[0058] Preferably, the multi-resolution analysis adopts wavelet transform technology, which specifically includes performing wavelet decomposition on the pre-screened data, extracting characteristic signals of the contact rail of different frequency components, wherein the high-frequency components are used to identify subtle surface defects and stress concentrations, and the low-frequency components are used to capture the overall deformation trend of the contact rail, and the decomposed components are reconstructed to generate multi-resolution feature data containing multi-scale feature information.

[0059] Adaptive filtering is a technology that dynamically adjusts filter parameters and can adjust the filter response in real time according to the characteristics of the input data to maximize the retention of useful signals and suppress noise. In the present invention, adaptive filtering is used to remove residual noise in pre-screened data while retaining signal components related to the actual state of the contact rail system. This processing method ensures the clarity and authenticity of the data, laying a good foundation for subsequent multi-resolution analysis.

[0060] Multi-resolution analysis is an indispensable step in the measurement of contact rail system parameters. It captures various characteristics of the contact rail by analyzing signals at different scales. It is particularly suitable to use wavelet transform technology for multi-resolution analysis because wavelet transform can decompose signals into different frequency components, thereby providing a localized description of the signal in both time and frequency domains. In the present invention, wavelet transform is specifically applied to wavelet decomposition of pre-screened data to extract contact rail characteristic signals of different frequency components. High-frequency components are mainly used to identify subtle defects and stress concentration areas on the surface of the contact rail. These high-frequency information can reflect small-scale wear or cracks on the surface of the track and are important indicators of the health of the track. Low-frequency components are used to capture the overall deformation trend of the contact rail. These low-frequency information can show the macro deformation of the track system, such as overall bending or settlement. By decomposing and reconstructing the data of these different frequency components, multi-resolution feature data containing multi-scale feature information is finally generated. This process ensures the comprehensiveness of data analysis, so that the contact rail characteristics from micro to macro are effectively captured and expressed.

[0061] Normalization is an important step to ensure that different feature data have a unified scale. Since different feature data of the contact rail system may have large differences in numerical range and unit, directly comparing these data may lead to deviations in the analysis results. Through normalization, all feature data can be scaled to a unified range, making the subsequent feature selection and dimensionality reduction process more accurate and reliable. In the present invention, normalization not only improves the consistency of data, but also provides more suitable input data for subsequent algorithm processing, further improving the robustness of the entire measurement system.

[0062] Feature selection and dimensionality reduction are key steps to extract the most representative information from a large amount of feature data. During the feature selection process, the system will evaluate the contribution of each feature to the contact rail state and select the features that best reflect the health of the system. Dimensionality reduction processing reduces the number of features and the dimension of the data while retaining information that is critical to the state of the contact rail system. This step is particularly important in the present invention because it not only reduces the complexity of data processing, but also improves the accuracy of the analysis by excluding redundant and irrelevant information.

[0063] like Figure 2 As shown, a method combining multi-level adaptive Bayesian reasoning and physical modeling is used to perform multi-mode fusion and complex parameter calculation on the key feature data, and the calculation results are optimized through an adversarial learning mechanism to generate optimized contact rail parameter data;

[0064] Bayesian reasoning is a reasoning method based on probability theory, which derives the posterior probability distribution by combining prior knowledge and observed data. In the parameter measurement of the contact rail system, multi-level adaptive Bayesian reasoning can handle the complexity and uncertainty of multi-source data. Specifically, this method first uses prior knowledge (such as historical data or design parameters of the contact rail) to set the initial probability model, and then combines the key feature data extracted from the previous data processing steps to perform reasoning and generate preliminary parameter estimates. In the reasoning process at each level, the Bayesian model can adaptively adjust its prior distribution to better adapt to the new data input, which enables the model to continuously correct its prediction results and gradually approach the actual state of the contact rail.

[0065] However, relying solely on Bayesian reasoning may not be enough to capture the complex physical phenomena in the contact rail system, so the present invention further introduces physical modeling as a supplement. Physical modeling provides additional constraints for Bayesian reasoning by simulating the actual physical properties of the contact rail system, such as stress distribution, material properties, geometric deformation, etc. Specifically, physical modeling uses known physical laws and material mechanics models to generate physical parameters related to the state of the contact rail. Combining these physical parameters with the results of Bayesian reasoning can significantly improve the prediction accuracy of the model. For example, in stress analysis, physical modeling can provide the stress distribution of the contact rail under different load conditions, while Bayesian reasoning can be corrected according to the actual measured stress data, thereby generating comprehensive stress parameters that are more in line with the actual situation.

[0066] In order to further optimize these complex parameter calculations, the present invention introduces an adversarial learning mechanism, in particular, a generative adversarial network (GAN). GAN consists of a generative network and a discriminative network. The generative network attempts to generate simulated data of contact rail parameters, while the discriminative network attempts to distinguish the differences between these simulated data and the actual measured data. In the process of adversarial learning, the generative network continuously adjusts its parameters so that the generated data is closer and closer to the real data, while the discriminative network gradually improves its recognition ability through iterative training. Ultimately, the optimization results output by the generative network are almost indistinguishable from the actual data, which not only optimizes the accuracy of the parameter calculation results, but also improves the robustness and generalization ability of the model. For example, when predicting the geometric deformation of the contact rail, the generative network can generate highly realistic deformation simulation results by learning the actual deformation data, while the discriminative network promotes the continuous improvement of the generative network by identifying the subtle differences between these simulated data and the actual data, until the simulation results are highly consistent with the actual data.

[0067] Through this method of combining multi-level adaptive Bayesian reasoning with physical modeling, and with the optimization of the adversarial learning mechanism, the present invention can achieve accurate calculation and prediction of complex parameters of the contact rail system. This method can not only effectively handle the fusion of multi-source data, but also provide stable and reliable results in the face of uncertainty and complexity, thereby ensuring the safe operation and effective maintenance of the contact rail system. This multi-mode fusion calculation method not only retains the flexibility of statistical reasoning, but also combines the authenticity of physical modeling. Ultimately, through the optimization of the adversarial learning mechanism, the contact rail parameter measurement results are more accurate and robust, and can meet the needs of various complex working conditions.

[0068] Preferably, the complex parameter calculation includes a method combining multi-level adaptive Bayesian reasoning with physical modeling, specifically including the following steps:

[0069] Perform Bayesian reasoning on key feature data, set prior probability distribution, and perform initial reasoning based on observed data to generate first reasoning parameter data;

[0070] Inputting the first inference parameter data into a physical modeling module, performing mechanical modeling on the contact rail system, and generating physical modeling data through mechanical equations and physical constraints;

[0071] Feeding the physical modeling data back to the Bayesian reasoning model, adjusting the prior parameters in the reasoning model, and performing multi-level adaptive reasoning until a final reasoning result that meets the physical conditions is generated to form final reasoning parameter data;

[0072] The final contact rail parameter data is calculated according to the following calculation expression:

[0073]

[0074] Among them, P(θ|D) represents the posterior probability of parameter θ when the observed data D is known; P(D|θ) represents the likelihood function of the observed data D when the parameter is assumed to be θ; P(θ) represents the prior probability of parameter θ; P(D) represents the marginal likelihood of the observed data D.

[0075] In the present invention, complex parameter calculation is a key step, especially when the system needs to process multi-dimensional and multi-variable feature data. In order to ensure the accuracy of the calculation results and the robustness of the system, the present invention adopts a method combining multi-level adaptive Bayesian reasoning with physical modeling. This method combines the advantages of statistical reasoning and physical modeling, not only can accurately predict the state of the contact rail system, but also can handle the uncertainty and complexity in the observed data, thereby generating optimized contact rail parameter data.

[0076] Multi-level adaptive Bayesian reasoning is the core of this method. Bayesian reasoning derives the posterior probability distribution by combining prior knowledge and observation data, providing a powerful tool for parameter estimation of the contact rail system. In the present invention, the initial stage of Bayesian reasoning first processes the key feature data, and generates the first level of reasoning parameter data by setting the prior probability distribution and combining the observation data. This step uses prior knowledge (such as historical data or design parameters of the contact rail) to guide the reasoning process, and at the same time combines the current observation data to adjust the model's prediction to make it more in line with the actual situation. Through this adaptive approach, the model can gradually approach the actual state of the system under the continuously updated observation data.

[0077] However, Bayesian reasoning itself may be insufficient when faced with complex physical phenomena. Therefore, the present invention introduces physical modeling as a supplement. Physical modeling simulates the actual behavior of the contact rail system through mechanical equations and physical constraints, which is particularly important when dealing with problems such as stress distribution and geometric deformation of the contact rail. In this step, the first reasoning parameter data is input into the physical modeling module to generate physical modeling data that reflects the real physical properties of the contact rail system. Physical modeling provides additional constraints for Bayesian reasoning by using known material mechanics and geometric deformation models, so that the reasoning results are more accurate and in line with actual physical conditions.

[0078] The physical modeling data is fed back to the Bayesian reasoning model to further adjust the prior parameters in the reasoning model. This feedback mechanism enables Bayesian reasoning to make adaptive adjustments based on the physical modeling data, approaching the optimal parameter estimates layer by layer. Through a multi-level adaptive reasoning process, the model ultimately generates a reasoning result that meets the physical conditions, forming the final reasoning parameter data.

[0079] In order to ensure that the final calculation result has a high degree of accuracy in the practical application of the contact rail system, the present invention uses the Bayesian formula to calculate the final contact rail parameter data. The specific calculation formula is:

[0080]

[0081] Among them, P(θ|D) represents the posterior probability of parameter θ when the observed data D is known, which represents the optimal estimate of the contact rail system parameters after the given observed data; P(D|θ) represents the likelihood function of the observed data D under the assumption that the parameter is θ, which reflects the degree of fit between the parameter and the observed data; P(θ) represents the prior probability of parameter θ, which is the preliminary estimate of the parameter before the observed data; and P(D) represents the marginal likelihood of the observed data D, which ensures the normalization of the calculation results.

[0082] Through this calculation method, the contact rail parameter data finally generated not only contains the direct information of the observed data, but also combines the constraints provided by physical modeling to ensure the accuracy and stability of parameter estimation. For example, when measuring the stress distribution of the contact rail, Bayesian reasoning can calculate the most likely stress distribution based on the observed stress data combined with the material mechanical properties in physical modeling. The application of this method not only improves the accuracy of the contact rail system parameter measurement, but also enhances the system's ability to cope with complex physical phenomena.

[0083] Through the combination of multi-level adaptive Bayesian reasoning and physical modeling, the present invention successfully achieves accurate calculation of complex parameters of the contact rail system. This method effectively integrates the advantages of statistical reasoning and physical modeling, and ensures high accuracy and reliability of contact rail parameter data through adaptive adjustment and feedback mechanism, providing a solid technical foundation for the monitoring and maintenance of the contact rail system.

[0084] Preferably, the calculation results are further optimized through an adversarial learning mechanism, including setting up a generation network and a discriminant network, the generation network generates simulated contact rail parameter data based on the final inference parameter data, the discriminant network distinguishes between the simulated data and the real data, and in each iteration, adjusts the parameters of the generation network to make the generated data closer to the real data, and at the same time adjusts the discriminant ability of the discriminant network until the output data of the generation network is indistinguishable from the real data, thereby generating optimized contact rail parameter data.

[0085] In the measurement of contact rail system parameters, in order to further improve the accuracy and reliability of parameter calculation results, the present invention introduces an adversarial learning mechanism. The adversarial learning mechanism, especially the generative adversarial network (GAN), enables the model to generate contact rail parameter data that is closer to reality through the game relationship between the generative network and the discriminative network, thereby optimizing the calculation results.

[0086] The core idea of ​​the generative adversarial network is to improve the performance of the model through the competition between the two networks. Specifically, in the present invention, the task of the generative network is to generate simulated contact rail parameter data based on the final inference parameter data, while the discriminant network attempts to distinguish the difference between these simulated data and the actual observed data. The generative network and the discriminant network are continuously trained in this process. The generative network attempts to generate contact rail parameter data as close to the real one as possible, while the discriminant network continuously improves its recognition ability to distinguish as accurately as possible which data is generated and which is actually observed.

[0087] In each iteration, the parameters of the generator network are adjusted according to the feedback from the discriminator network, so that the generated data gradually approaches the real data. This adjustment process can be understood as the generator network constantly trying to learn the real characteristics of the contact rail system and gradually reduce the gap with the actual observed data. The discriminator network plays the role of a "referee" in this process. Its feedback helps the generator network to continuously correct its output until the generated data and the real data are almost the same in the eyes of the discriminator network.

[0088] For example, in the stress distribution measurement of the contact rail system, the generation network can generate a simulation diagram of the stress distribution based on the inference parameter data obtained by preliminary calculation. The discriminant network compares these simulation diagrams with the actual measured stress distribution diagrams, identifies the differences and feeds them back to the generation network. The generation network will adjust its strategy for generating stress distribution diagrams based on the feedback from the discriminant network, so that the generated simulation diagrams gradually approach the actual measurement results. As the iteration proceeds, the generated stress distribution diagram will become closer and closer to the real image until the discriminant network cannot distinguish between the two. At this time, the output of the generation network is the optimized contact rail parameter data, which can truly reflect the actual state of the contact rail system.

[0089] Through this adversarial learning mechanism, the model can not only generate highly realistic contact rail parameter data, but also show strong adaptability and robustness when facing complex and changeable contact rail systems. This optimization process greatly improves the accuracy of parameter measurement, so that the final generated data is not only highly accurate in theory, but also shows excellent performance in practical applications. Through the introduction of adversarial learning mechanism, the parameter measurement of the contact rail system no longer relies solely on traditional physical modeling and statistical reasoning, but through deep learning, it fully utilizes the powerful ability of generative adversarial networks to achieve accurate prediction and optimization of complex system parameters, providing a more reliable technical means for the monitoring and maintenance of the contact rail system.

[0090] Performing multi-dimensional nonlinear analysis on the optimized contact rail parameter data, combining chaos coupling analysis with graph neural network to perform complex anomaly detection, and generating verified abnormal feature data;

[0091] In the present invention, multi-dimensional nonlinear analysis of optimized contact rail parameter data and complex anomaly detection combined with chaotic coupling analysis and graph neural network are important steps in identifying and evaluating potential system failures. This process, through the combination of advanced nonlinear analysis methods and graph neural networks, can deeply reveal the complex dynamic behavior in the contact rail system and accurately locate abnormal features.

[0092] Multidimensional nonlinear analysis is a key technology for dealing with complex systems. During operation, the contact rail system is affected by many factors, such as dynamic loads of trains, temperature changes, and material aging. These factors often cause the contact rail system to exhibit nonlinear characteristics, and traditional linear analysis methods are difficult to capture these complex dynamic behaviors. Multidimensional nonlinear analysis can effectively reveal nonlinear phenomena in the contact rail system, such as periodic oscillations, chaos, and multi-stable states, by analyzing the dynamic changes of the system in multiple dimensions. This analysis can not only help us understand the operating status of the contact rail system, but also identify potential unstable factors in the system.

[0093] As an advanced nonlinear analysis method, chaos coupling analysis focuses on identifying and analyzing complex dynamic behaviors caused by chaotic phenomena in the system. In the contact rail system, chaos coupling analysis can be used to detect chaotic behaviors caused by external disturbances or internal faults. For example, when the contact rail is subjected to continuous dynamic loads, its response may show chaotic characteristics, which are often related to potential failures of the system. Through chaos coupling analysis, the chaotic behavior in the system can be identified and its impact on the stability of the contact rail system can be evaluated, so as to find possible fault points in advance.

[0094] like Figure 3 As shown, in order to further improve the accuracy of anomaly detection, the present invention combines a graph neural network (GNN). A graph neural network is a deep learning model specifically used to process graph structure data. In the contact rail system, it can be used to model and analyze the spatial structure of the system. The contact rail system can be abstracted as a graph structure consisting of multiple nodes (such as key positions of the track) and edges (such as the connection relationship of the track). In this graph structure, the state of each node depends not only on its own characteristics, but also on the influence of adjacent nodes. The graph neural network can capture the complex dependencies between nodes by recursively calculating the nodes and edges of the graph structure, thereby identifying local anomalies in the system. For example, if the stress distribution of a certain section of track is significantly different from that of its adjacent track sections, the graph neural network can identify this anomaly and mark it as a potential fault point by analyzing the state of the node and its surrounding nodes.

[0095] In practical applications, the optimized contact rail parameter data is first subjected to multi-dimensional nonlinear analysis to identify possible chaotic behaviors and nonlinear dynamic characteristics in the system. Then, these nonlinear characteristic data are input into the graph neural network, which models and analyzes the spatial structure of the entire contact rail system, and finally generates verified abnormal characteristic data by identifying abnormal connections and state differences between nodes. These abnormal characteristic data can not only accurately locate the fault location in the system, but also provide a scientific basis for system maintenance and optimization.

[0096] In general, through the combination of multi-dimensional nonlinear analysis and chaotic coupling analysis, the present invention can deeply reveal the complex dynamic behavior in the contact rail system. The introduction of graph neural network provides a powerful tool for anomaly detection, which can accurately locate anomalies in the spatial structure of the system. This multi-level analysis and detection method not only improves the accuracy of contact rail system parameter measurement, but also significantly enhances the system's fault detection capability, providing a strong guarantee for ensuring the safe and stable operation of the contact rail system.

[0097] Preferably, the multi-dimensional nonlinear analysis includes:

[0098] The chaotic characteristics of the contact rail system are quantitatively analyzed by calculating the Lyapunov index to generate chaotic characteristic data. The calculation expression of the Lyapunov index is as follows:

[0099]

[0100] Where λ represents the Lyapunov exponent; t represents time; δX i (t) represents the small disturbance of the contact rail system characteristics at time t; δX i (0) represents a small disturbance at the initial moment;

[0101] Combined with chaotic coupling analysis, the nonlinear dynamic behavior between various characteristics in the contact rail system is identified and nonlinear coupling characteristic data is generated.

[0102] In the present invention, multi-dimensional nonlinear analysis is an important method to effectively identify and quantify complex dynamic behaviors in the system, especially showing significant advantages in detecting chaotic characteristics and nonlinear coupling phenomena of the system. Through Lyapunov index calculation and chaotic coupling analysis, the potential dynamic instability and nonlinear correlation characteristics of the contact rail system in actual operation can be deeply revealed, thereby generating key data that is helpful for fault detection and prevention.

[0103] The Lyapunov index is an important indicator for measuring the chaotic characteristics of a system. In the present invention, the chaotic characteristics of the contact rail system are quantitatively analyzed by calculating the Lyapunov index, and the sensitivity of the system to the initial conditions can be evaluated. Specifically, the Lyapunov index represents how the small difference between the states of two adjacent track systems changes over time.

[0104] If the Lyapunov exponent is positive, it indicates that the system is highly sensitive to small changes in initial conditions, the system exhibits chaotic characteristics, and even very small disturbances may cause drastic changes in the system state. In the contact rail system, this chaotic characteristic may reflect the unstable state of the track under complex stress conditions or potential material fatigue problems. Through quantitative analysis of the Lyapunov exponent, these potential chaotic behaviors can be identified and provide early warning signals for system maintenance. For example, in stress monitoring, areas with high Lyapunov exponents may indicate that the track in this area is experiencing highly complex stress changes and requires special attention.

[0105] In addition, combined with chaos coupling analysis, the nonlinear dynamic behavior between various features in the contact rail system can be further identified. Chaos coupling analysis is a method that can detect and quantify the nonlinear relationship between different system variables. In the contact rail system, there are often complex coupling relationships between various stress, deformation, and vibration characteristics. These coupling relationships are not simple linear relationships, but nonlinear interactions affected by multiple factors. Through chaos coupling analysis, these nonlinear coupling behaviors can be identified, revealing which features have the closest relationship and which may be potential sources of failure. For example, if the analysis finds that there is a strong nonlinear coupling relationship between the lateral vibration and longitudinal deformation of the track, this may indicate that the structure of the track in this area has begun to become abnormal and may require further inspection and repair.

[0106] By combining Lyapunov exponent calculation and chaotic coupling analysis, the present invention can deeply analyze the dynamic behavior of the contact rail system and generate nonlinear coupling characteristic data that are helpful for fault prediction and prevention. These characteristic data not only provide a scientific basis for system maintenance and optimization, but also can significantly improve the operational safety and stability of the contact rail system by early detection and identification of potential system instability. Through this series of analyses, the present invention can effectively deal with various problems that may arise in the contact rail system under complex working conditions and ensure the long-term stability and safe operation of the track system.

[0107] Preferably, the nonlinear coupling characteristic data is combined with the structural modeling of the graph neural network, including using the graph neural network to model the spatial structure of the contact rail system, mapping the nonlinear coupling characteristic data into graph structure data, wherein the nodes represent the key positions of the contact rails and the edges represent the connection relationships of the contact rails, and analyzing the nodes and edges in the graph structure data, identifying local structural anomalies in the contact rail system through the graph neural network model, and generating structural anomaly data.

[0108] In the present invention, the combination of nonlinear coupling feature data and structural modeling of graph neural networks is a key step in identifying and locating local structural anomalies in the system. This process combines the complex spatial structure of the contact rail system with the deep learning model, which can not only efficiently capture abnormal features in the system, but also provide accurate fault location, providing strong support for system maintenance and optimization.

[0109] Graph Neural Network (GNN) is a deep learning model suitable for processing graph structured data. The contact rail system is essentially a complex spatial network with multiple nodes and multiple connection relationships, which makes the graph neural network an ideal tool for analyzing the system. In the present invention, the nonlinear coupling characteristic data is first mapped to graph structured data through preprocessing. In this graph structure, the nodes represent the key positions of the contact rail, such as track joints, bending points or stress concentration areas; the edges represent the connection relationships between these nodes, such as the physical connection between track segments or the stress transfer path.

[0110] Using graph neural networks to model the spatial structure of the contact rail system means capturing the complex dependencies between nodes and edges in the graph by recursively processing them. Specifically, the graph neural network transmits the information of each node to its adjacent nodes through a hierarchical propagation mechanism, thereby updating the state of the node. This information transmission and update process enables each node to reflect not only its own characteristics, but also the relationship between it and its adjacent nodes. For example, the stress state of a track segment depends not only on its own characteristics, but also on the stress state of the adjacent track segments. Through multi-layer propagation and updating, the graph neural network is able to capture the complex dependencies and nonlinear characteristics of the system on a global scale.

[0111] In practical applications, this process can be explained in the following way: suppose that an abnormal stress concentration occurs at the bend of a certain section of contact rail. This abnormal stress may not only be caused by the characteristics of the node itself, but also by the state of the surrounding nodes. In this context, the graph neural network can identify the abnormality of the node by analyzing the relationship between the node and the surrounding nodes, and evaluate whether this abnormality will be transmitted to other nodes, thereby causing larger system problems. In this way, the graph neural network can effectively identify local structural abnormalities in the contact rail system and generate corresponding structural abnormality data.

[0112] The generated structural anomaly data is extremely valuable because it not only reveals potential problems currently existing in the contact rail system, but also provides information on the path and scope of possible fault spread. This information is of great significance for system maintenance and fault prevention. For example, if the graph neural network analyzes that a structural anomaly at a track joint may affect multiple adjacent track sections, maintenance personnel can prioritize the inspection and repair of these critical areas to avoid systemic failures.

[0113] By combining nonlinear coupling feature data with the structural modeling of graph neural networks, the present invention successfully achieves accurate anomaly detection and fault location of the contact rail system. This method can not only cope with the complex spatial structure of the contact rail system, but also efficiently process and analyze data in a large-scale network, provide accurate fault warnings and maintenance recommendations, and ultimately ensure the long-term stable operation of the contact rail system.

[0114] Preferably, the structural anomaly data is combined with dynamic resonance analysis to perform complex anomaly detection, including resonance analysis of the stress response of the contact rail system at different frequencies, identifying structural anomalies caused by the resonance effect, combining the resonance analysis results with the structural anomaly data, generating verified abnormal feature data, and feeding back to the optimized contact rail parameter data generation process.

[0115] In the present invention, complex anomaly detection by combining structural anomaly data with dynamic resonance analysis is an important step to ensure safe and stable operation of the system. This process identifies the stress response characteristics of the contact rail system at different frequencies, analyzes the structural anomalies caused by the resonance effect, and combines these analysis results with the existing structural anomaly data to generate more accurate anomaly feature data. These data are not only used to verify and optimize the contact rail parameters, but also provide an important basis for system maintenance and preventive repair.

[0116] Dynamic resonance analysis is a technique used to identify the response characteristics of a structure under different frequency excitations. In a contact rail system, due to the periodic loads generated by train operation, the track structure may experience excitations of different frequencies, which may induce resonance under certain conditions. Resonance means that when the natural frequency of the system is close to the external excitation frequency, the vibration response of the system will be significantly amplified, which may cause fatigue damage or failure of the structure. In the present invention, the goal of dynamic resonance analysis is to identify and evaluate the stress response of the contact rail system at these critical frequencies to detect potential structural abnormalities in advance.

[0117] Based on the structural anomaly data generated in the previous steps, the present invention performs dynamic resonance analysis on the contact rail system at multiple frequencies. By analyzing the stress response of the track at these frequencies, it is possible to identify which frequencies of excitation are most detrimental to the track system. For example, a section of track may produce significant vibration responses when a train passes by. If these responses are close to the natural frequency of the track, the system will experience resonance effects, leading to stress concentration and structural fatigue. By performing detailed resonance analysis on these frequency points, it is possible to identify structural anomalies caused by resonance effects and determine the potential impact of these anomalies on the system.

[0118] In order to better understand this process, a practical example can be given: suppose that the natural frequency of a section of contact rail is around 15Hz, and the wheel-rail contact frequency of the train during operation fluctuates between 12Hz and 18Hz. In this case, the track will experience excitation with a frequency close to its natural frequency under certain operating conditions, which may cause significant resonance. Through dynamic resonance analysis, the present invention can accurately identify abnormal stress responses within this frequency range and determine whether these responses will have an adverse effect on the track structure. If the analysis results show that the section of track has obvious stress concentration at 16Hz, this may mean that the track is at risk of fatigue damage during long-term operation and requires intensive monitoring and maintenance.

[0119] After completing the dynamic resonance analysis, the system combines these resonance analysis results with the existing structural anomaly data to generate verified anomaly feature data. The purpose of this step is to further verify the existence and severity of anomalies by combining the quantitative data of the resonance analysis with the previously identified structural anomaly data. Through this combination process, the system can more comprehensively evaluate the health status of the contact rail system and provide a reliable basis for subsequent optimization and decision-making.

[0120] These verified abnormal characteristic data are not only fed back to the contact rail parameter data generation process to further optimize the parameter calculation model, but can also be directly used to guide the system maintenance strategy. For example, if the verified abnormal data of a certain section of track shows that it has a significant resonance effect at a specific frequency, the maintenance team can give priority to this section of track, reduce the resonance risk by slowing down the train speed or installing vibration reduction devices, and avoid serious failures.

[0121] In general, by combining structural anomaly data with dynamic resonance analysis, the present invention can effectively identify and verify complex anomalies in the contact rail system. This process not only improves the accuracy of anomaly detection, but also provides a scientific basis for the long-term operation and preventive maintenance of the system. Ultimately, this multi-level analysis method ensures the overall stability and safety of the contact rail system, allowing the track to maintain good working condition under complex working conditions.

[0122] The verified abnormal characteristic data and the optimized contact rail parameter data are integrated to generate a contact rail system measurement report.

[0123] Preferably, the generation of the contact rail system measurement report includes generating a health assessment report and maintenance recommendations for the subway contact rail system based on the verified abnormal feature data and the optimized contact rail parameter data, and applying the health assessment report and maintenance recommendations to the real-time monitoring system of the subway contact rail system.

[0124] In the present invention, the verified abnormal characteristic data is integrated with the optimized contact rail parameter data to generate a contact rail system measurement report, which is a key step to comprehensively evaluate the health status of the system and make maintenance recommendations. This process not only integrates the results of multi-level analysis and optimization, but also provides a scientific basis and real-time monitoring support for the safe operation of the subway system.

[0125] The verified abnormal feature data are identified through the previous multi-dimensional nonlinear analysis, dynamic resonance analysis and graph neural network model. These data accurately reflect the potential problems in the contact rail system, such as structural anomalies, resonance effects, etc., while the optimized contact rail parameter data are obtained through a combination of multi-level adaptive Bayesian reasoning and physical modeling, representing the optimal state of the system after comprehensive analysis and optimization calculation. The integration of these two types of data aims to combine the results of abnormal detection and optimization calculation to generate a comprehensive and accurate contact rail system measurement report.

[0126] During the integration process, the system first unifies the various characteristic data into the same analysis framework to ensure that data from different sources can complement and verify each other. For example, if a section of contact rail shows stress anomaly in the optimized parameter calculation, and structural instability is also found in the anomaly detection, this double verification can greatly improve the accuracy of anomaly judgment. The integrated data not only provides a more comprehensive perspective on the system status, but also makes the conclusions in the measurement report more credible.

[0127] The generation of the contact rail system measurement report is not just a simple data summary, but an in-depth analysis and assessment of the system health status. The report includes two main parts: health assessment report and maintenance recommendations.

[0128] The health assessment report provides a detailed description of the current status of the contact rail system by analyzing the integrated data. The report will include the overall health of the contact rail, potential risk areas, and predicted failure trends. For example, if a section of track has frequently experienced resonance anomalies in the past few months, the report will record this in detail and predict the risk of failure that may occur in the future. The generation of the health assessment report not only relies on current data, but also combines historical data and trend analysis to provide more forward-looking maintenance recommendations.

[0129] Maintenance recommendations are specific operational guidelines based on the health assessment report, designed to help the subway maintenance team take preventive measures to reduce the possibility of failures. These recommendations usually include track sections that are prioritized for inspection and maintenance, recommended maintenance methods, and possible optimization solutions. For example, if the report indicates that a section of track has stress concentration at the joint, maintenance recommendations may include increasing the monitoring frequency of the joint, slowing down the speed of trains passing through the section, or adding vibration reduction facilities at the joint. With these recommendations, subway system managers can take measures before failures occur to reduce system risks.

[0130] The generated contact rail system measurement report is not just a static document, but is combined with the real-time monitoring system of the subway contact rail system to form a dynamic and real-time updateable monitoring system. Once the health assessment report and maintenance recommendations are generated, they will be input into the real-time monitoring system and become the basis for system operation and maintenance. For example, when the real-time monitoring system detects that the operating parameters of a certain section of track are consistent with the predicted data in the measurement report, the system will automatically remind the maintenance team to take corresponding measures, thereby achieving a more accurate and timely maintenance response.

[0131] like Figure 4 As shown, a measuring system for implementing the method for measuring the parameters of the contact rail system comprises the following components:

[0132] Magnetic field induction sensor module, used to collect initial magnetic field induction data of subway contact rail;

[0133] Optical interference sensor module, used to collect initial optical interference data of subway contact rails;

[0134] The magnetic field induction sensor module and the optical interference sensor module are used to collect the initial magnetic field induction data and initial optical interference data of the subway contact rail respectively. These two sensor modules ensure the comprehensiveness and accuracy of the data and capture the important features of the contact rail in different dimensions.

[0135] The data synchronization and pre-screening module is used to synchronously process and pre-screen the initial magnetic field sensing data and the initial optical interference data to generate pre-screening data; the synchronous processing ensures the time consistency of data collection between different sensors, and the pre-screening can remove noise or preliminary data that does not meet the requirements.

[0136] The data preprocessing module includes an adaptive filtering unit, a multi-resolution analysis unit and a normalization processing unit, which are used to process the pre-screening data and generate key feature data; this process effectively extracts the core features of the contact rail through filtering and data analysis, and improves the accuracy of subsequent analysis.

[0137] The multi-mode fusion and complex parameter calculation module includes a multi-level adaptive Bayesian reasoning unit, a physical modeling unit and an adversarial learning unit, which are used to perform multi-mode fusion, complex parameter calculation and optimization on the key feature data to generate optimized contact rail parameter data; the multi-mode fusion and optimization technology of this module makes the calculation of contact rail parameters more accurate and can better adapt to complex working conditions.

[0138] The multi-dimensional nonlinear analysis and anomaly detection module includes a chaotic coupling analysis unit and a graph neural network unit, which are used to perform nonlinear analysis and complex anomaly detection on the optimized contact rail parameter data to generate verified abnormal feature data; through these advanced analysis technologies, potential anomalies in the contact rail system can be effectively detected, especially anomalies under complex dynamic behaviors.

[0139] The data integration and report generation module is used to integrate the verified abnormal characteristic data with the optimized contact rail parameter data and generate a contact rail system measurement report. This report not only records the health status of the system in detail, but also provides a scientific basis for maintenance, ensuring the safe operation of the subway system.

[0140] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0141] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for measuring contact rail system parameters, characterized in that: The following steps are involved: Using a magnetic field induction sensor and an optical interferometer sensor to collect initial data of the subway contact rail, and synchronously processing and pre-screening the initial data to generate pre-screening data; Performing adaptive filtering, multi-resolution analysis and normalization on the pre-screened data, and then performing feature selection and dimensionality reduction to generate key feature data; Using a method combining multi-level adaptive Bayesian reasoning with physical modeling, multi-mode fusion and complex parameter calculation are performed on the key feature data, and the calculation results are optimized through an adversarial learning mechanism to generate optimized contact rail parameter data; Performing multi-dimensional nonlinear analysis on the optimized contact rail parameter data, combining chaotic coupling analysis with graph neural network to perform complex anomaly detection, and generating verified anomaly feature data; The verified abnormal characteristic data is integrated with the optimized contact rail parameter data to generate a contact rail system measurement report.

2. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The data synchronization processing includes performing time alignment processing on the collected initial magnetic field induction data and initial optical interference data, and performing spatial coordinate standardization processing on the aligned data.

3. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The multi-resolution analysis uses wavelet transform technology, specifically including wavelet decomposition of pre-screened data to extract characteristic signals of the contact rail at different frequency components. The high-frequency components are used to identify subtle surface defects and stress concentrations, while the low-frequency components are used to capture the overall deformation trend of the contact rail. The decomposed components are then reconstructed to generate multi-resolution feature data containing multi-scale feature information.

4. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The complex parameter calculation includes a method combining multi-level adaptive Bayesian reasoning with physical modeling, specifically including the following steps: Perform Bayesian inference on key feature data, set prior probability distribution, and perform initial inference based on observed data to generate first inference parameter data; Inputting the first inference parameter data into a physical modeling module, performing mechanical modeling on the contact rail system, and generating physical modeling data through mechanical equations and physical constraints; Feeding the physical modeling data back to the Bayesian reasoning model, adjusting the prior parameters in the reasoning model, and performing multi-level adaptive reasoning until a final reasoning result that meets the physical conditions is generated to form final reasoning parameter data; The final contact rail parameter data is calculated according to the following calculation expression: Among them, P(D|θ) represents the posterior probability of parameter θ when the observation data D is known; P(D|θ) represents the likelihood function of the observation data D when the parameter is assumed to be θ; P(θ) represents the prior probability of parameter θ; P(D) represents the marginal likelihood of the observation data D.

5. The method for measuring subway contact rail system parameters according to claim 4, characterized in that: The calculation results are further optimized through an adversarial learning mechanism, including setting up a generative network and a discriminative network. The generative network generates simulated contact rail parameter data based on the final inference parameter data, and the discriminative network distinguishes between the simulated data and the real data. In each iteration, the parameters of the generative network are adjusted to make the generated data closer to the real data. At the same time, the discriminative ability of the discriminative network is adjusted until the output data of the generative network is indistinguishable from the real data, thereby generating optimized contact rail parameter data.

6. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The multi-dimensional nonlinear analysis includes: The chaotic characteristics of the contact rail system are quantitatively analyzed by calculating the Lyapunov index to generate chaotic characteristic data. The calculation expression of the Lyapunov index is as follows: Where λ represents the Lyapunov exponent; t represents time; δX i (t) represents the small disturbance of the contact rail system characteristics at time t; δX i (0) represents a small perturbation at the initial moment; Combined with chaotic coupling analysis, the nonlinear dynamic behavior between various characteristics in the contact rail system is identified and nonlinear coupling characteristic data is generated.

7. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The nonlinear coupling feature data is combined with the structural modeling of the graph neural network, including using the graph neural network to model the spatial structure of the contact rail system, mapping the nonlinear coupling feature data into graph structure data, wherein the nodes represent the key positions of the contact rails and the edges represent the connection relationships of the contact rails, and analyzing the nodes and edges in the graph structure data, identifying local structural anomalies in the contact rail system through the graph neural network model, and generating structural anomaly data.

8. The method for measuring subway contact rail system parameters according to claim 7, characterized in that: The structural anomaly data is combined with dynamic resonance analysis to perform complex anomaly detection, including resonance analysis of the stress response of the contact rail system at different frequencies, identifying structural anomalies caused by the resonance effect, combining the resonance analysis results with the structural anomaly data, generating verified anomaly feature data, and feeding it back to the optimized contact rail parameter data generation process.

9. The method for measuring subway contact rail system parameters according to claim 1, characterized in that: The generation of the contact rail system measurement report includes generating a health assessment report and maintenance recommendations for the subway contact rail system based on the verified abnormal feature data and the optimized contact rail parameter data, and applying the health assessment report and maintenance recommendations to the real-time monitoring system of the subway contact rail system.

10. A measuring system for implementing the method for measuring contact rail system parameters according to any one of claims 1 to 9, characterized in that: Includes the following components: Magnetic field induction sensor module, used to collect initial magnetic field induction data of subway contact rail; Optical interferometry sensor module, used to collect initial optical interferometry data of subway contact rails; a data synchronization and pre-screening module, configured to synchronously process and pre-screen the initial magnetic field sensing data and the initial optical interference data to generate pre-screening data; A data preprocessing module, comprising an adaptive filtering unit, a multi-resolution analysis unit and a normalization processing unit, for processing the pre-screening data and generating key feature data; A multi-mode fusion and complex parameter calculation module, including a multi-level adaptive Bayesian reasoning unit, a physical modeling unit, and an adversarial learning unit, is used to perform multi-mode fusion, complex parameter calculation, and optimization on the key feature data to generate optimized contact rail parameter data; A multi-dimensional nonlinear analysis and anomaly detection module, including a chaotic coupling analysis unit and a graph neural network unit, is used to perform nonlinear analysis and complex anomaly detection on the optimized contact rail parameter data to generate verified anomaly feature data; The data integration and report generation module is used to integrate the verified abnormal feature data with the optimized contact rail parameter data and generate a contact rail system measurement report.

Images