Railway carriage monitoring method and device and electronic equipment

Abstract

The invention provides a railway carriage monitoring method and device and electronic equipment, and relates to the field of data processing. The method comprises the following steps: acquiring electromagnetic induction data which is sent by sensor equipment and aims at a target carriage; performing finite element analysis and inverse problem solution on the electromagnetic induction data to generate a carriage electromagnetic image; acquiring deployment position data of the sensor equipment on the target carriage; determining abnormal data of the target carriage according to the deployment position data and the carriage electromagnetic image; and inputting the abnormal data into a convolutional neural network, and determining an abnormal mode of the target carriage. By implementing the technical scheme provided by the invention, the monitoring accuracy is improved.

Description

Technical Field

[0001] The present application relates to the technical field of data processing, and in particular to a railway carriage monitoring method, device and electronic equipment. Background Art

[0002] Railway transportation is an important mode of logistics and passenger transportation worldwide. With its large capacity, high efficiency and low energy consumption, it has become a core component of the modern transportation system. As a key carrier of railway transportation, the operating status of railway carriages is directly related to the safety and efficiency of the overall transportation system.

[0003] At present, the condition monitoring of railway carriages usually relies on manual inspections. However, manual inspections have problems such as reliance on experience, low efficiency, and limited coverage. In particular, in areas that are difficult to directly observe, such as the bottom structure and suspension system of the carriage, potential fault risks are difficult to detect in time, resulting in low monitoring accuracy.

[0004] Therefore, a railway carriage monitoring method, device and electronic equipment are urgently needed. Summary of the invention

[0005] The present application provides a railway carriage monitoring method, device and electronic equipment to improve monitoring accuracy.

[0006] In a first aspect of the present application, a railway carriage monitoring method is provided, the method comprising: obtaining electromagnetic induction data for a target carriage sent by a sensor device; performing finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the carriage; obtaining deployment position data of the sensor device on the target carriage; determining abnormal data of the target carriage based on the deployment position data and the carriage electromagnetic image; and inputting the abnormal data into a convolutional neural network to determine an abnormal pattern of the target carriage.

[0007] By adopting the above technical solutions, the electromagnetic induction data in the operation of the carriage is collected in real time to capture the dynamic changes of the carriage state and avoid the lag of traditional manual detection. Electromagnetic induction data can reflect information in multiple dimensions such as carriage structure, electrical state and dynamic interaction, covering a wider monitoring range, including hidden problems that are difficult to detect with the naked eye. Finite element analysis can restore the complex distribution of the electromagnetic field around the carriage from the collected electromagnetic induction data through precise mathematical modeling, eliminating the error of pure statistical methods. Even in a complex operating environment, the calculation model can be dynamically adjusted to ensure the accuracy of data processing. Inverse problem solving combined with electromagnetic image generation technology can intuitively display the electromagnetic state of the carriage in the form of an image, which is convenient for technicians to understand and diagnose abnormalities. The spatial resolution and dynamic changes of electromagnetic images can accurately locate the problem area and provide a direct basis for subsequent maintenance. Combining the deployment location of the sensor with the electromagnetic image, the specific carriage part corresponding to the abnormal data can be clarified to improve the positioning accuracy. By inputting abnormal data into the convolutional neural network, intelligent mapping from data to pattern is realized, and abnormal patterns can be quickly identified. Abnormal pattern recognition can integrate multi-dimensional data, greatly reduce the problem of false alarms or missed alarms in traditional monitoring, and improve the reliability of carriage status detection. Therefore, it is easy to improve monitoring accuracy.

[0008] Optionally, the acquiring of electromagnetic induction data for the target car sent by the sensor device specifically includes: receiving first data for the body of the target car sent by an electromagnetic induction antenna, the electromagnetic induction antenna being arranged circumferentially around the body; receiving second data for the bottom of the target car sent by an electromagnetic sensor, the electromagnetic sensor being arranged on a railway track corresponding to the target car; performing data processing on the first data and the second data to obtain the electromagnetic induction data, the data processing including denoising, filtering and normalization processing.

[0009] By adopting the above technical solution, the joint collection of electromagnetic induction data of the vehicle body and the electromagnetic induction data of the bottom of the vehicle can cover the key structural areas of the car body and ensure the comprehensiveness of monitoring. A single collection device may have blind spots, and the sensors on the track may not be able to monitor the data of the high-altitude vehicle body. These blind spots are eliminated through the upper and lower linkage arrangement. By verifying and supplementing the data at different positions, potential faults or abnormalities can be identified more accurately. The electromagnetic induction antenna is set around the circumference of the vehicle body, which can achieve 360-degree all-round monitoring. No matter which side of the car the abnormality occurs, the corresponding electromagnetic signal changes can be captured. The electromagnetic sensor on the track specializes in collecting data from the bottom of the car, which can accurately monitor the car body. The dynamic state of the bottom of the car, such as looseness of the suspension system, wear of the connector, fatigue cracking of the bottom plate and other problems. Since the contact between the car and the track is dynamic during operation, the sensors on the track can capture dynamic characteristics such as vibration and uneven load distribution during operation, providing more accurate basic data for further analysis. The sensors on the track do not need to be installed directly on the car, avoiding the impact of equipment maintenance and installation on the normal operation of the car, and have the reliability of long-term monitoring. Data processing of the first data and the second data can improve the stability and robustness of the data, reduce the probability of false alarms and missed alarms, and establish a comprehensive electromagnetic distribution model of the entire car to provide a more comprehensive evaluation of the operating status.

[0010] Optionally, the electromagnetic induction data is subjected to finite element analysis and inverse problem solving to generate an electromagnetic image of a vehicle compartment, specifically comprising: acquiring structural data of the target vehicle compartment; determining a geometric model based on the structural data; discretizing control equations corresponding to the geometric model using a preset solver to generate a sparse linear equation group; and processing the sparse linear equation group by a conjugate gradient method to obtain an original electromagnetic field distribution.

[0011] By adopting the above technical solutions, after obtaining the structural data of the target car, the geometric model can be accurately generated according to the size and shape of the actual car. This modeling method ensures that the basis of finite element analysis is based on the real physical characteristics of the car, rather than idealized or simplified assumptions. By discretizing the continuous control equations, complex physical problems can be converted into discrete equations that can be processed by computers. This method provides highly accurate numerical solutions that can capture subtle electromagnetic changes inside and outside the car structure. The sparse linear equations represent the discretization results of the control equations. The conjugate gradient method is used to solve sparse matrix problems, which has faster calculation speed and less memory usage. It is suitable for handling large-scale problems and can be solved efficiently with lower computing resources, reducing memory pressure and computing time during the calculation process. Through finite element analysis and inverse problem solving, the original electromagnetic field distribution map finally obtained not only shows the spatial distribution of electromagnetic induction data, but also reflects the electromagnetic characteristics of different areas inside and outside the car. This is crucial for the health monitoring of the car and helps to find potential faults or problem areas. The original electromagnetic field distribution map can show the changes of the electromagnetic field in different parts of the car, helping engineers to locate problem areas and find structural defects or abnormalities in time. The electromagnetic field distribution map can reveal the relationship between the change of electromagnetic signals and the structure of the car, and help analyze whether the change of electromagnetic signals is related to the car fault. With these data, the abnormal parts of the car can be accurately located. Inverse problem solving can infer the electromagnetic field distribution and potential defects inside the car through the signals collected by external sensors without directly contacting the car or disassembling the structure. This method avoids the destructive inspection of the car structure by traditional detection methods, and is non-invasive and efficient. Inverse problem solving can infer the complex electromagnetic field distribution inside the car from the actual measurement data, providing more accurate fault analysis and abnormality identification capabilities than conventional methods.

[0012] Optionally, the performing finite element analysis and inverse problem solving on the electromagnetic induction data to generate the electromagnetic image of the vehicle compartment specifically also includes: determining the actual electromagnetic field distribution according to the electromagnetic induction data; constructing an objective function according to the actual electromagnetic field distribution and the original electromagnetic field distribution; solving the objective function to obtain a solution result; and mapping the solution result into a three-dimensional image to obtain the electromagnetic image of the vehicle compartment.

[0013] By adopting the above technical solution, by analyzing the electromagnetic induction data, the actual electromagnetic field distribution inside and outside the car can be accurately calculated, revealing the true electromagnetic characteristics of the car. This helps to capture potential problems that are difficult to detect with traditional monitoring methods. The actual electromagnetic field distribution is closely related to the operating state of the car, so this analysis can reflect the changes in the electromagnetic response of the car under different working conditions in real time, helping to detect anomalies and faults in a timely manner. By accurately obtaining the actual electromagnetic field distribution, the deviation that may be caused by predicting the electromagnetic field only through theoretical models is avoided, thereby improving the reliability of the analysis results. By constructing the objective function, the difference between the actual electromagnetic field distribution and the original electromagnetic field distribution can be quantitatively analyzed. This process can further optimize the model and parameters and improve the analysis accuracy. By solving the objective function, the optimal solution between the actual electromagnetic field distribution and the theoretical model can be obtained. This process is based on the solution of the inverse problem, which is suitable for processing complex electromagnetic induction data and can provide an accurate electromagnetic field distribution solution. The solution of the objective function helps to accurately describe the details of the electromagnetic environment inside and outside the car, and helps to analyze the source and impact of electromagnetic anomalies, especially when dealing with cars with complex geometric shapes and structures. By mapping the solution results into a three-dimensional image, the electromagnetic field distribution of the carriage can be presented clearly and intuitively, making it easier for technicians to visually diagnose electromagnetic anomalies. This visualization not only improves the efficiency of analysis, but also enhances the understanding and intuitive perception of complex data.

[0014] Optionally, determining the abnormal data of the target compartment according to the deployment position data and the compartment electromagnetic image specifically includes: determining a first position and a second position according to the deployment position data, the first position and the second position being any two mutually symmetrical positions among a plurality of deployment positions; performing feature extraction on the compartment electromagnetic image to obtain a plurality of abnormal feature areas; acquiring a first area and a second area from the plurality of abnormal feature areas, the first area being an area corresponding to the first position, and the second area being an area corresponding to the second position; determining a first electromagnetic field distribution corresponding to the first area, and a second electromagnetic field distribution corresponding to the second area; if it is determined that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric distributions, determining the data corresponding to each of the first area and the second area as the abnormal data.

[0015] By adopting the above technical solution, by utilizing symmetry, the abnormal situation of electromagnetic field distribution can be judged more efficiently, reducing the complexity of analysis. Comparison between symmetrical areas can eliminate local changes in the electromagnetic field caused by environmental interference or random noise, thereby more accurately locating the real abnormality. By determining any two symmetrical positions, the method does not rely on a specific deployment position, can be applied to various car structures and sensor layout methods, and improves the versatility and flexibility of the analysis model. By extracting the features of electromagnetic images, areas where abnormalities may exist can be quickly extracted from complex image data, significantly improving the detection efficiency. The feature extraction process simplifies large-scale electromagnetic image data into several abnormal feature areas, greatly reducing the complexity of subsequent calculations and the burden of data processing. Feature extraction technology is usually combined with image processing algorithms or deep learning models to achieve automated operations, thereby avoiding the uncertainty of manual screening of data. By focusing the analysis on the first area and the second area, the interference of non-related data can be reduced, further improving the accuracy of the analysis. Analyzing the various characteristics of the electromagnetic field in the symmetrical area can fully reflect the changes in the electromagnetic field of the car, providing a multi-dimensional basis for abnormal judgment. This method can quickly locate the distribution characteristics of the symmetrical area and provide a reliable analysis method for real-time status monitoring of the car.

[0016] Optionally, determining the abnormal data of the target compartment based on the deployment position data and the compartment electromagnetic image specifically also includes: if it is determined that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then determining that the data corresponding to the abnormal characteristic area is the abnormal data.

[0017] By adopting the above technical solution and using magnetic flux density as a judgment indicator, a clear, standardized and quantitative judgment standard is provided for the abnormal detection of carriages, avoiding the interference of subjective factors. Magnetic flux density can intuitively reflect the degree of abnormality of electromagnetic field distribution. This judgment method can effectively distinguish normal fluctuations from real abnormalities. By simply comparing with the preset threshold, potential abnormal areas can be quickly screened out. This method has low computational complexity and can significantly improve the data processing speed. Compared with complex abnormality judgment models, it is only necessary to compare the magnetic flux density with the threshold to determine whether it is abnormal, thereby achieving a higher degree of automation and reducing the cost of manual analysis. Changes in magnetic flux density are highly sensitive to material defects and reduced electromagnetic shielding effectiveness, and can capture subtle anomalies that are difficult to detect with other indicators. When electromagnetic field abnormalities suddenly occur during the operation of the carriage, these abnormal data can be captured and marked in real time through threshold judgment.

[0018] Optionally, the abnormal data is input into a convolutional neural network to determine the abnormal mode of the target car, specifically including: if it is determined through the convolutional neural network that the abnormal data indicates that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric distributions, then the abnormal mode is determined to be car tilt and / or overload; if it is determined through the convolutional neural network that the abnormal data indicates that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then the abnormal mode is determined to be loose car parts.

[0019] By adopting the above technical solutions, the convolutional neural network can automatically extract the features of the input data, reduce human intervention, and realize the rapid judgment of abnormal patterns. Abnormal data may contain a large amount of multi-dimensional information. CNN can efficiently process these complex data through feature learning, avoiding the inefficiency caused by manual analysis in traditional methods. This method clearly distinguishes different abnormal patterns, has high adaptability, and can expand the recognition of more patterns. The asymmetry of the electromagnetic field distribution is closely related to the mechanical state of the car tilt and overload. Through this judgment logic, these problems can be quickly identified to avoid greater safety hazards caused by abnormal conditions. Asymmetric distribution judgment reduces the impact of environmental interference and random noise on the detection results and improves the robustness of abnormal pattern recognition. Car tilt and overload will directly lead to uneven distribution of electromagnetic fields. This feature is highly correlated with electromagnetic induction data, providing a clear physical basis for identification. Loose components usually lead to abnormal local electromagnetic characteristics, which are manifested as magnetic flux density exceeding the normal range. By judging the threshold of magnetic flux density, these abnormal areas can be accurately located. The loose problem may not be obvious in the early stage, but the change of magnetic flux density can be used as an early indication to facilitate timely warning and maintenance. The magnetic flux density threshold is combined with CNN to decompose the complex pattern recognition problem into simple threshold comparison and deep learning analysis, which improves the transparency and explainability of recognition.

[0020] In a second aspect of the present application, a railway carriage monitoring device is provided, which includes an acquisition module and a processing module, wherein the acquisition module is used to acquire electromagnetic induction data for a target carriage sent by a sensor device; the processing module is used to perform finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the carriage; the acquisition module is also used to acquire deployment position data of the sensor device on the target carriage; the processing module is also used to determine abnormal data of the target carriage based on the deployment position data and the carriage electromagnetic image; the processing module is also used to input the abnormal data into a convolutional neural network to determine the abnormal mode of the target carriage.

[0021] In the third aspect of the present application, an electronic device is provided, which includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs the method described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, the method described above is executed.

[0023] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: By collecting electromagnetic induction data in the operation of the carriage in real time, the dynamic changes of the carriage state can be captured to avoid the lag of traditional manual detection. Electromagnetic induction data can reflect information in multiple dimensions such as carriage structure, electrical state and dynamic interaction, covering a wider monitoring range, including hidden problems that are difficult to detect with the naked eye. Finite element analysis can restore the complex distribution of the electromagnetic field around the carriage from the collected electromagnetic induction data through precise mathematical modeling, eliminating the errors of pure statistical methods. Even in a complex operating environment, the calculation model can be dynamically adjusted to ensure the accuracy of data processing. Inverse problem solving combined with electromagnetic image generation technology can intuitively display the electromagnetic state of the carriage in the form of an image, which is convenient for technicians to understand and diagnose abnormalities. The spatial resolution and dynamic changes of electromagnetic images can accurately locate the problem area and provide a direct basis for subsequent maintenance. Combining the deployment location of the sensor with the electromagnetic image, the specific carriage part corresponding to the abnormal data can be clarified to improve the positioning accuracy. By inputting abnormal data into the convolutional neural network, intelligent mapping from data to pattern is realized, and abnormal patterns can be quickly identified. Abnormal pattern recognition can integrate multi-dimensional data, greatly reduce the problem of false alarms or missed alarms in traditional monitoring, and improve the reliability of carriage status detection. Therefore, it is easy to improve monitoring accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of a flow chart of a railway carriage monitoring method provided in an embodiment of the present application; Figure 2 Another schematic diagram of a flow chart of a railway carriage monitoring method provided in an embodiment of the present application; Figure 3 A schematic diagram of a module of a railway carriage monitoring device provided in an embodiment of the present application; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0025] Explanation of the reference numerals: 31, acquisition module; 32, processing module; 41, processor; 42, communication bus; 43, user interface; 44, network interface; 45, memory. DETAILED DESCRIPTION

[0026] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0027] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0028] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0029] Railway transportation is an important logistics and passenger transportation method widely used around the world. With its significant advantages such as large transportation capacity, high operating efficiency and relatively low energy consumption, it has become a core component of the modern transportation system. Whether it is cross-border freight, intercity railway transportation, or urban rail transit, the railway system plays an indispensable role in the global transportation network. In this context, railway carriages are the key carriers of railway transportation. The safety and stability of their operating status are directly related to the operating efficiency of the overall transportation system and the safety of passengers and goods.

[0030] Monitoring the operating status of railway carriages is an important part of ensuring transportation safety, extending the service life of equipment, and improving transportation efficiency. However, in actual applications, the monitoring of railway carriage status is still mainly based on manual inspections. However, the coverage of manual inspections is limited. Especially for the bottom structure and suspension system of the carriage, since these areas are relatively hidden and the inspection conditions are poor, it is difficult to achieve full coverage by manpower, and there is often a risk that potential faults cannot be discovered in time, resulting in low monitoring accuracy.

[0031] In order to solve the above technical problems, the present application provides a railway carriage monitoring method, referring to Figure 1 , Figure 1 A flow chart of a railway carriage monitoring method provided in an embodiment of the present application. The method is applied to a server and specifically includes steps S110 to S150, which are as follows: S110, acquiring electromagnetic induction data for the target compartment sent by the sensor device.

[0032] Specifically, the server, as a central processing unit, is responsible for collecting data from various sensors distributed on railway carriages and tracks. These data are transmitted to the server through wireless communications, such as Wi-Fi, LoRa, 5G, or wired communications, such as Ethernet. Sensor devices include but are not limited to the following two: electromagnetic induction antennas are arranged around the carriage, mainly used to capture the electromagnetic characteristics of the surface of the carriage, such as the conductive state of the metal shell of the carriage, the magnetic field distribution, etc. The electromagnetic sensor is installed on the track or the bottom of the carriage to monitor the changes in the electromagnetic field caused by the bottom of the carriage and the operating status, such as local electromagnetic anomalies caused by wear, cracks or loose components of the steel structure at the bottom of the car. Electromagnetic induction data refers to the electromagnetic field information detected by the sensor, including magnetic flux density, magnetic field strength, eddy current distribution, etc. These data can reflect changes in the material properties, structural health status and operating status of the carriage. After receiving the data, the server performs further processing and storage.

[0033] In a possible implementation, obtaining electromagnetic induction data for a target car sent by a sensor device specifically includes: receiving first data for a body of the target car sent by an electromagnetic induction antenna, the electromagnetic induction antenna being arranged circumferentially around the body; receiving second data for a bottom of the target car sent by an electromagnetic sensor, the electromagnetic sensor being arranged on a railway track corresponding to the target car; performing data processing on the first data and the second data to obtain electromagnetic induction data, the data processing including denoising, filtering and normalization processing.

[0034] Specifically, the electromagnetic induction antenna is installed around the car body and distributed along the circumference of the car body. It is responsible for detecting the electromagnetic response characteristics of the car body shell, including the conductivity of the body material, the eddy current distribution on the shell, etc. Its focus is to reflect the structural condition of the car body surface or near the surface. The electromagnetic sensor is located on the railway track where the car body runs, focusing on monitoring the changes in the electromagnetic field distribution at the bottom of the car body. These sensors analyze the status of the metal components at the bottom of the car body, such as the suspension system, the bottom plate, etc., especially detecting electromagnetic anomalies caused by wear, looseness or cracks. The two types of sensors transmit the collected data to the central server in real time through wireless communication or wired connection. The original signal may be mixed with external noise, such as environmental magnetic field fluctuations and interference from other electronic devices. Denoising algorithms, such as wavelet denoising or mean filtering, can clean up these irrelevant signals and retain the core electromagnetic characteristics. Filtering further eliminates interference signals in a specific frequency range, such as high-frequency noise, and retains effective low-frequency or medium-frequency signals. The signals collected by different sensors may have amplitude differences due to different device sensitivities. Normalization can standardize the data to a unified scale, which is convenient for subsequent analysis of the overall electromagnetic field distribution of the car body.

[0035] S120, performing finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the vehicle compartment.

[0036] Specifically, finite element analysis is a numerical calculation method used to solve the problem of physical field distribution of complex structures. In the embodiment of the present application, the electromagnetic field distribution of the car and its surroundings is simulated by finite element analysis. A geometric model is established based on the structural data of the target car, such as size, material properties, etc., which is the basis of finite element analysis. The car body geometric model includes the car body, the bottom of the car and key components, such as the suspension system and the floor structure. The geometric model is then discretized into multiple small units, such as triangles or tetrahedron units, for mathematical calculations. The electromagnetic field distribution follows Maxwell's equations. These equations describe the electric field, the magnetic field and their interactions, and the finite element method is used to convert the partial differential equations into a computable sparse linear equation system. These equations are solved by numerical algorithms to obtain the electromagnetic field distribution of each grid unit in the car. The purpose of solving the inverse problem is to infer the structural state that causes this distribution, such as material defects, damage locations, etc., based on the electromagnetic induction data collected by the sensor. The server maps the results of the inverse problem solution to the car body geometric model to generate a three-dimensional image. In addition, different colors or brightness can be used to represent the electromagnetic field intensity or the location of the abnormal area. The electromagnetic image of the car can clearly show the structural status and fault location.

[0037] In one possible implementation, finite element analysis and inverse problem solving are performed on electromagnetic induction data to generate an electromagnetic image of the vehicle compartment, specifically including: acquiring structural data of the target vehicle compartment; determining a geometric model based on the structural data; discretizing the control equations corresponding to the geometric model using a preset solver to generate a sparse linear equation group; and processing the sparse linear equation group using a conjugate gradient method to obtain the original electromagnetic field distribution.

[0038] Specifically, structural data refers to information describing the geometric shape and physical properties of the target car body, including the size, shape, material properties, such as conductivity, magnetic permeability, and related physical parameters of the car body. For example, the structural data of the car body may include information such as the length of the car body (15 meters), the height of the car body (4 meters), and the thickness of the car body (2 mm), and the electromagnetic properties of the car body materials, such as the conductivity and magnetic permeability of the car body, are also required. The geometric model is to create a three-dimensional representation of the car body by modeling the structural data, usually including the body, chassis, and key components inside the car body. Through the structural data of the car body, the car body model is converted into a geometric shape suitable for numerical calculation using CAD software or finite element modeling tools. For this car body, the model may include the rectangular shape of the car body, the configuration of the bottom suspension system, etc. The finite element method converts continuous physical problems into discrete mathematical problems by discretizing the original partial differential equations. The control equations mentioned here refer to equations that describe electromagnetic fields, such as Maxwell's equations. Through the discretization process, these equations are converted into matrix form to form a sparse linear equation system. The behavior of electromagnetic fields is usually described by Maxwell's equations, which describe the distribution of electric and magnetic fields inside and outside the cabin in electromagnetic detection of the cabin. Through the finite element method, these continuous governing equations are transformed into a finite number of algebraic equations that describe the values ​​of the electromagnetic fields at different locations.

[0039] The conjugate gradient method is an iterative algorithm specifically used to solve sparse linear equations. Since the matrix of the electromagnetic field problem is sparse, the conjugate gradient method can effectively calculate the value of the electromagnetic field in each calculation unit. Through multiple iterations, the conjugate gradient method can gradually approach the solution of the equation until the predetermined accuracy is met. This method is suitable for processing large sparse matrices and has high computational efficiency. For example, by using the conjugate gradient method to solve the above sparse linear equations, the values ​​of the electromagnetic field in each unit can be obtained. For example, at different positions on the top and bottom of the car, the magnetic field intensity, magnetic flux, etc. are calculated. Through the above process, the electromagnetic field distribution inside and outside the car can be finally obtained. These electromagnetic field data usually include electric field intensity, magnetic field intensity, and magnetic flux density. The distribution of the electromagnetic field can reveal the electromagnetic characteristics of the car structure, such as whether there are abnormalities, such as cracks, corrosion, looseness, etc.

[0040] In a possible implementation, finite element analysis and inverse problem solving are performed on electromagnetic induction data to generate an electromagnetic image of the vehicle compartment, which specifically includes: determining the actual electromagnetic field distribution based on the electromagnetic induction data; constructing an objective function based on the actual electromagnetic field distribution and the original electromagnetic field distribution; solving the objective function to obtain a solution result; and mapping the solution result into a three-dimensional image to obtain an electromagnetic image of the vehicle compartment.

[0041] Specifically, first, the server needs to calculate the actual distribution of the electromagnetic field based on the electromagnetic induction data obtained from the sensor. The electromagnetic field is composed of an electric field and a magnetic field. The electromagnetic induction data usually provides the data of the magnetic field change around the car. Through these data, the electromagnetic field distribution inside and outside the car can be inferred. For example, assuming that the sensors are located at different positions around the car, the collected electromagnetic induction data may be the magnetic field strength values ​​at different positions. By processing these data, the actual magnetic field distribution can be obtained, such as the magnetic field strength at different positions on the top, bottom, and side of the car. The objective function is a mathematical expression used for optimization and solution. It is constructed by comparing the difference between the actual electromagnetic field distribution and the original electromagnetic field distribution. In this way, the objective function can reveal the deviation between the actual electromagnetic field distribution and the original electromagnetic field distribution, which reflects the abnormality or fault of the car structure. Assuming that the original electromagnetic field distribution model shows that there should be a uniform magnetic field distribution at the bottom of the car, but the actual electromagnetic field data shows that the magnetic field strength in some areas is low, this difference will be reflected through the objective function, thereby helping to identify abnormalities.

[0042] After obtaining the objective function, the server needs to solve this function through numerical optimization methods to obtain the corrected or updated value of the electromagnetic field. This process is actually to find the best match between the actual electromagnetic field distribution and the original electromagnetic field distribution by minimizing the objective function. Commonly used solution methods include gradient descent method, Newton method, etc. Through these methods, the parameters of the electromagnetic field distribution are continuously adjusted until the value of the objective function reaches the minimum, which means that the difference between the actual electromagnetic field distribution and the original electromagnetic field distribution has been minimized. For example, suppose that a new electromagnetic field distribution is obtained after minimizing the objective function, in which the magnetic field distribution at the bottom of the car is corrected to be closer to the theoretical value, which may mean that there is some structural damage or fault at the bottom of the car. Finally, the electromagnetic field distribution results obtained by solving can be mapped into an electromagnetic image of the car through three-dimensional visualization technology. This three-dimensional electromagnetic image can clearly show the electromagnetic field distribution in each area of ​​the car, helping to further analyze and judge whether there is anomaly or fault. In three-dimensional space, the electromagnetic field intensity of each point can be expressed by color, density, etc., so as to generate an intuitive electromagnetic image. These images will show the distribution of electromagnetic fields in various parts of the car, and abnormal areas will appear as image features that are significantly different from normal areas. For example, through mapping, a three-dimensional image of the electromagnetic image of the car cabin may show that the magnetic flux density in a certain area of ​​the bottom of the car cabin is significantly higher than other areas, indicating that there may be structural problems in that area, such as looseness or cracks in the suspension system.

[0043] S130: Acquire deployment position data of the sensor device on the target carriage.

[0044] Specifically, sensor devices will be pre-installed in different locations in the car to fully monitor the status of the car. These sensors may be installed on the top, bottom, side or other key areas of the car. Each sensor has a specific installation location, and the server needs to record the data of these installation locations. These deployment location data include the physical location coordinates of the sensor, such as X, Y, and Z coordinates, as well as the relative position relationship between the sensor and other parts of the car, such as the body, chassis, suspension system, etc. These data are crucial for subsequent data analysis because the signals collected by the sensor are closely related to its location.

[0045] For example, suppose a freight train named "Carriage A1" is being monitored. The car is equipped with multiple sensors to monitor different parts of the car. For example, the electromagnetic sensor is installed at the bottom of the car to detect changes in the electromagnetic field at the bottom of the car to determine whether there are potential suspension system problems. The acceleration sensor is installed on the side of the car to detect vibrations in the car and analyze whether there are loose or damaged parts. The temperature sensor is installed inside the car to monitor temperature changes inside the car and identify possible electrical faults. The deployment location data includes the specific coordinates of each sensor. For example, the electromagnetic sensor may be located at the center of the bottom of the car with coordinates (0, -5, -1). The acceleration sensor is located on the right side of the car, 10 meters away from the front of the car, with coordinates (10, 5, 0). The temperature sensor is located in the upper left corner of the car, 3 meters away from the rear of the car, with coordinates (-3, -2, 1). These deployment location data are sent to the server via wireless communication or wired network, and the server uses these data to analyze the specific information collected by each sensor.

[0046] When there is an abnormality in the electromagnetic image or other detection data of the vehicle cabin, the server can combine the location data to accurately locate the location of the problem. For example, if the electromagnetic sensor detects an abnormal electromagnetic field distribution, the location data can help determine whether there is a problem with the suspension system under the vehicle or a damaged component. By understanding the location distribution of sensors, the server can optimize the data collection strategy and increase the monitoring density of key components or possible fault areas.

[0047] S140: Determine abnormal data of the target compartment based on the deployment position data and the compartment electromagnetic image.

[0048] Specifically, the server uses the deployment location data to understand the spatial location of each data point in the electromagnetic image. For example, the electromagnetic image may show an abnormal electromagnetic field distribution at a certain location on the bottom of the car, and the deployment location data tells the server that this location corresponds to a key part of the bottom of the car. Combining the two, the server can determine the problem area and further analyze its cause. Abnormal data refers to data that deviates from normal working conditions or predetermined thresholds. These abnormal data may include asymmetric electromagnetic field distribution, magnetic flux density beyond the normal range, or inconsistent electromagnetic field distribution with other parts of the car. By analyzing these abnormal data, the server can determine whether the car is at risk of failure or damage.

[0049] In a possible implementation, abnormal data of a target car is determined based on deployment position data and an electromagnetic image of the car, specifically including: determining a first position and a second position based on the deployment position data, the first position and the second position being any two mutually symmetrical positions among a plurality of deployment positions; performing feature extraction on the electromagnetic image of the car to obtain a plurality of abnormal feature areas; obtaining a first area and a second area from the plurality of abnormal feature areas, the first area being an area corresponding to the first position, and the second area being an area corresponding to the second position; determining a first electromagnetic field distribution corresponding to the first area, and a second electromagnetic field distribution corresponding to the second area; if it is determined that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric distributions, determining the data corresponding to each of the first area and the second area as abnormal data.

[0050] Specifically, the first position and the second position are selected through the deployment position data. The two positions should be symmetrical, that is, they are equidistant and located in two relative areas of the carriage. For example, the left and right sides, front and rear ends of the carriage, etc. The server extracts the characteristics of electromagnetic field changes from the electromagnetic image, that is, abnormal fluctuations or changes in the electromagnetic field distribution in the area. The server then selects areas where abnormalities may exist from the extracted features. These areas may be places where the electromagnetic field distribution is significantly different from that in normal conditions, indicating that a certain component of the carriage may have problems, such as looseness, damage, etc. According to the deployment position data, the abnormal areas of the first position and the second position correspond to two symmetrical areas of the carriage respectively. For example, if the first position is located on the left side of the carriage, the first area is the electromagnetic field distribution on the left side; similarly, the second position corresponds to the electromagnetic field distribution on the right side of the carriage. The electromagnetic field distribution of these two areas is analyzed separately to determine whether they are consistent. If the electromagnetic field distribution of the first position and the second position is asymmetric, for example, the electromagnetic field distribution on the left side is abnormal and the right side is normal, it may indicate that there are different degrees of faults or damage in these positions. If an asymmetric electromagnetic field distribution is found, it is considered that the area is abnormal. The abnormal data is determined by comparing the electromagnetic field distribution of the first area and the second area. If there is an asymmetry or obvious difference in the electromagnetic field distribution of the two areas, the electromagnetic data corresponding to these areas are determined to be abnormal data. These abnormal data can help identify the fault area in the car and further conduct detailed inspection or repair.

[0051] For example, suppose that the target carriage of a train is undergoing electromagnetic monitoring. The specific steps are as follows: Two electromagnetic sensors of carriage A1 are installed at the bottom of the left and right sides of the carriage, respectively, to record the electromagnetic induction data of the left side (first position) and the right side (second position). The installation positions of these two positions are symmetrical to each other and are located on both sides of the center of the bottom of the car. The electromagnetic image of carriage A1 is generated through the data collected by the sensors, showing the electromagnetic field distribution at the bottom of the carriage. During the analysis process, it was found that the electromagnetic field density in the left (first position) area was higher than the normal level, while the electromagnetic field density in the right (second position) area was within the normal range. The first area on the left and the second area on the right correspond to the first position and the second position, respectively. After detailed analysis, the electromagnetic field distribution on the left side showed abnormality, while the electromagnetic field distribution on the right side was normal. Comparing the electromagnetic field distribution of the first position (left side) with the second position (right side), it was found that the two were asymmetric. The abnormal electromagnetic field on the left side indicates the existence of uneven magnetic flux density, which may be due to problems with the suspension system or wheels at the bottom of the carriage. Since the abnormal distribution of the electromagnetic field on the left side is different from the symmetrical electromagnetic field distribution on the right side, the data at these two positions are determined to be abnormal data. Based on these abnormal data, it can be inferred that there may be a fault at the bottom left side of the car.

[0052] Therefore, by combining the deployment location data and the electromagnetic image of the carriage, the server can use symmetry analysis to determine abnormalities in the carriage. If asymmetry in the electromagnetic field distribution is found at a symmetrical position, the server can confirm that the electromagnetic data in that area is abnormal data. This method not only improves the accuracy of fault detection, but also can quickly locate the fault area, help carry out targeted maintenance and repairs, and ensure the safety of railway transportation.

[0053] In a possible implementation, abnormal data of the target compartment is determined based on the deployment position data and the electromagnetic image of the compartment, which specifically includes: if it is determined that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then the data corresponding to the abnormal characteristic area is determined to be abnormal data.

[0054] Specifically, in the previous steps, the server extracts the characteristic areas with abnormalities by analyzing the electromagnetic induction data of the carriage and generating electromagnetic images. These abnormal areas are manifested as abnormalities in the electromagnetic field distribution, which may be caused by damage, looseness or other faults in the carriage components. These abnormal areas may include the bottom of the car, the body, connecting parts, etc., and the electromagnetic image can show the magnetic field density and changes in these areas. Magnetic flux density refers to the magnetic flux passing through a unit area, which is a parameter that describes the strength of the electromagnetic field. In the electromagnetic induction of the carriage, the magnetic flux density is used to measure the strength of the electromagnetic field. The higher the magnetic flux density, the greater the change in the electromagnetic field, which may reflect the abnormality of certain components, such as loose components, wear, etc. The preset threshold is a standard value set in advance by the server for comparison with the actual measured magnetic flux density. If the magnetic flux density of an abnormal area exceeds this preset threshold, it can be considered that there is a fault or abnormality in the area. This threshold is usually determined by a large amount of experimental data or historical data to ensure that the system can accurately distinguish between normal and abnormal magnetic field distributions during detection. For example, if the magnetic flux density of a normal car compartment area is usually between 100 microteslas and 300 microteslas, any area above 300 microteslas can be considered abnormal. When the magnetic flux density of an abnormal area is greater than or equal to the preset threshold, the system will determine the data in that area as abnormal data. This means that there may be a fault or damage in the area and further inspection or repair is required. If the magnetic flux density is lower than the preset threshold, the electromagnetic field in the area can be considered normal and will not be considered to be faulty.

[0055] S150, inputting the abnormal data into a convolutional neural network to determine the abnormal pattern of the target compartment.

[0056] Specifically, a convolutional neural network is a deep learning model that is particularly good at processing image data. Through multiple layers of convolutional layers, pooling layers, fully connected layers and other structures, the convolutional neural network automatically extracts features from the raw data and is able to perform pattern recognition and classification. The goal is to input abnormal data into the convolutional neural network, learn the characteristics of the abnormal data through the model, and finally identify the specific abnormal pattern of the carriage. The output of the convolutional neural network is the classification result of the abnormal data, which can be used to determine the abnormal pattern of the carriage. Through the training of the convolutional neural network, the model can identify different types of abnormalities, such as tilting of the carriage, eccentric loading, loose components, etc. These abnormal patterns are caused by factors such as magnetic flux density, structural changes in the carriage, and component location. The trained convolutional neural network model can match the input abnormal data with the known abnormal pattern, and finally output the result to determine the type of abnormality in the target carriage.

[0057] In one possible implementation, refer to Figure 2 , Figure 2Another flow chart of a railway carriage monitoring method provided by an embodiment of the present application. The abnormal data is input into a convolutional neural network to determine the abnormal mode of the target carriage, specifically including steps S210 to S220, the above steps are as follows: S210, through the convolutional neural network, if it is determined that the abnormal data indicates that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetrically distributed, then the abnormal mode is determined to be a tilt and / or eccentric load of the carriage; S220, through the convolutional neural network, if it is determined that the abnormal data indicates that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then the abnormal mode is determined to be a loose carriage component.

[0058] Specifically, through the reasoning process of the convolutional neural network, if a specific electromagnetic field distribution pattern is detected from the input abnormal data, the abnormal mode of the car can be determined. Specifically, there are two main abnormal modes in the embodiment of the present application. First, when the electromagnetic field distribution of the car is asymmetric, that is, the electromagnetic field is greatly different on the left and right sides or the front and back parts, this may indicate that the load of the car is uneven, resulting in tilting or eccentric loading of the car.

[0059] Secondly, if the magnetic flux density on the left side of the car is much higher than on the right side, this may be a sign of car eccentric loading, indicating that the car is loaded more heavily on one side, resulting in an asymmetric electromagnetic field distribution. The convolutional neural network will identify this asymmetric electromagnetic field distribution and classify it as a tilted or eccentrically loaded car.

[0060] Furthermore, if certain areas in the electromagnetic induction data show a higher magnetic flux density and exceed the preset threshold, this may indicate that there is some abnormality in the area, such as a loose or damaged part in the car. If the magnetic flux density in a certain area at the bottom of the car suddenly increases and exceeds the normal value, this may indicate that the parts in the area are loose or damaged, resulting in abnormal changes in the electromagnetic field. The convolutional neural network will use this data to determine the abnormality of the area and output the abnormal pattern of loose car parts.

[0061] The present application also provides a railway carriage monitoring device, referring to Figure 3 , Figure 3 A module schematic diagram of a railway carriage monitoring device provided in an embodiment of the present application. The railway carriage monitoring device is a server, and the server includes an acquisition module 31 and a processing module 32, wherein the acquisition module 31 acquires electromagnetic induction data for a target carriage sent by a sensor device; the processing module 32 performs finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the carriage; the acquisition module 31 acquires the deployment position data of the sensor device on the target carriage; the processing module 32 determines the abnormal data of the target carriage based on the deployment position data and the electromagnetic image of the carriage; the processing module 32 inputs the abnormal data into a convolutional neural network to determine the abnormal pattern of the target carriage.

[0062] In a possible implementation, the acquisition module 31 acquires electromagnetic induction data for the target car sent by the sensor device, specifically including: the acquisition module 31 receives first data for the body of the target car sent by the electromagnetic induction antenna, and the electromagnetic induction antenna is arranged around the body; the acquisition module 31 receives second data for the bottom of the target car sent by the electromagnetic sensor, and the electromagnetic sensor is arranged on the railway track corresponding to the target car; the processing module 32 processes the first data and the second data to obtain electromagnetic induction data, and the data processing includes denoising, filtering and normalization processing.

[0063] In one possible implementation, the processing module 32 performs finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the vehicle compartment, specifically including: the acquisition module 31 acquires the structural data of the target vehicle compartment; the processing module 32 determines the geometric model based on the structural data; the processing module 32 uses a preset solver to discretize the control equations corresponding to the geometric model to generate a sparse linear equation group; the processing module 32 processes the sparse linear equation group by the conjugate gradient method to obtain the original electromagnetic field distribution.

[0064] In a possible implementation, the processing module 32 performs finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the vehicle compartment, which specifically includes: the processing module 32 determines the actual electromagnetic field distribution based on the electromagnetic induction data; the processing module 32 constructs an objective function based on the actual electromagnetic field distribution and the original electromagnetic field distribution; the processing module 32 solves the objective function to obtain a solution result; the processing module 32 maps the solution result into a three-dimensional image to obtain an electromagnetic image of the vehicle compartment.

[0065] In a possible implementation, the processing module 32 determines abnormal data of the target compartment based on the deployment position data and the electromagnetic image of the compartment, specifically including: the processing module 32 determines a first position and a second position based on the deployment position data, the first position and the second position being any two mutually symmetrical positions among a plurality of deployment positions; the processing module 32 performs feature extraction on the electromagnetic image of the compartment to obtain a plurality of abnormal feature areas; the processing module 32 obtains a first area and a second area from the plurality of abnormal feature areas, the first area being an area corresponding to the first position, and the second area being an area corresponding to the second position; the processing module 32 determines a first electromagnetic field distribution corresponding to the first area, and a second electromagnetic field distribution corresponding to the second area; if the processing module 32 determines that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric distributions, then the data corresponding to each of the first area and the second area are determined to be abnormal data.

[0066] In a possible implementation, the processing module 32 determines the abnormal data of the target compartment based on the deployment position data and the electromagnetic image of the compartment, and specifically includes: if the processing module 32 determines that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then the data corresponding to the abnormal characteristic area is determined to be abnormal data.

[0067] In one possible implementation, the processing module 32 inputs the abnormal data into a convolutional neural network to determine the abnormal mode of the target car, specifically including: if the processing module 32 determines through the convolutional neural network that the abnormal data indicates that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric, then the abnormal mode is determined to be car tilt and / or overload; if the processing module 32 determines through the convolutional neural network that the abnormal data indicates that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, then the abnormal mode is determined to be loose car parts.

[0068] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0069] The present application also provides an electronic device, referring to Figure 4 , Figure 4 The electronic device may include: at least one processor 41 , at least one network interface 44 , a user interface 43 , a memory 45 , and at least one communication bus 42 .

[0070] The communication bus 42 is used to realize the connection and communication between these components.

[0071] The user interface 43 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 43 may also include a standard wired interface and a wireless interface.

[0072] The network interface 44 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0073] Among them, the processor 41 may include one or more processing cores. The processor 41 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 45, and calling data stored in the memory 45. Optionally, the processor 41 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 41 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 41, and it can be implemented separately through a chip.

[0074] Among them, the memory 45 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 45 includes a non-transitory computer-readable storage medium. The memory 45 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 45 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store data involved in the above-mentioned method embodiments, etc. The memory 45 may also be optionally at least one storage device located away from the aforementioned processor 41. As Figure 4 As shown, the memory 45 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program of a railway carriage monitoring method.

[0075] exist Figure 4In the electronic device shown, the user interface 43 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 41 can be used to call an application program storing a railway carriage monitoring method in the memory 45, and when executed by one or more processors, the electronic device executes one or more methods in the above-mentioned embodiments.

[0076] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0077] The present application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors, the electronic device executes one or more of the methods described in the above embodiments.

[0078] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0079] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0080] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0081] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0082] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0083] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, it will be easy for those skilled in the art to think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A railway carriage monitoring method, characterized in that: The method comprises: Acquire electromagnetic induction data of the target compartment sent by the sensor device; Performing finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the vehicle compartment; Acquiring deployment position data of the sensor device on the target carriage; Determining abnormal data of the target carriage according to the deployment position data and the carriage electromagnetic image; The abnormal data is input into a convolutional neural network to determine the abnormal pattern of the target compartment.

2. The railway carriage monitoring method according to claim 1, characterized in that: The obtaining of electromagnetic induction data for the target compartment sent by the sensor device specifically includes: receiving first data of a vehicle body of the target vehicle compartment sent by an electromagnetic induction antenna, wherein the electromagnetic induction antenna is arranged circumferentially around the vehicle body; Receiving second data for the bottom of the target carriage sent by an electromagnetic sensor, wherein the electromagnetic sensor is arranged on a railway track corresponding to the target carriage; The first data and the second data are processed to obtain the electromagnetic induction data, wherein the data processing includes denoising, filtering and normalization processing.

3. The railway carriage monitoring method according to claim 1, characterized in that: The performing of finite element analysis and inverse problem solving on the electromagnetic induction data to generate the electromagnetic image of the carriage specifically includes: Acquiring structural data of the target carriage; Determining a geometric model according to the structural data; Using a preset solver to discretize the control equations corresponding to the geometric model to generate a sparse linear equation group; The sparse linear equations are processed by the conjugate gradient method to obtain the original electromagnetic field distribution.

4. The railway carriage monitoring method according to claim 3, characterized in that: The performing finite element analysis and inverse problem solving on the electromagnetic induction data to generate the electromagnetic image of the carriage specifically includes: determining actual electromagnetic field distribution according to the electromagnetic induction data; constructing an objective function according to the actual electromagnetic field distribution and the original electromagnetic field distribution; Solving the objective function to obtain a solution result; The solution result is mapped into a three-dimensional image to obtain the electromagnetic image of the vehicle compartment.

5. The railway carriage monitoring method according to claim 1, characterized in that: The determining, according to the deployment position data and the carriage electromagnetic image, abnormal data of the target carriage specifically includes: Determine a first position and a second position according to the deployment position data, wherein the first position and the second position are any two mutually symmetrical positions among the plurality of deployment positions; Extracting features from the electromagnetic image of the carriage to obtain multiple abnormal feature areas; Acquire a first region and a second region from the plurality of abnormal feature regions, wherein the first region is a region corresponding to the first position, and the second region is a region corresponding to the second position; Determine a first electromagnetic field distribution corresponding to the first area, and a second electromagnetic field distribution corresponding to the second area; If it is determined that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetric distributions, then the data corresponding to the first area and the data corresponding to the second area are determined to be the abnormal data.

6. The railway carriage monitoring method according to claim 5, characterized in that: The determining the abnormal data of the target carriage according to the deployment position data and the carriage electromagnetic image specifically includes: If it is determined that the magnetic flux density corresponding to the abnormal characteristic region is greater than or equal to a preset threshold, the data corresponding to the abnormal characteristic region is determined to be the abnormal data.

7. The railway carriage monitoring method according to claim 6, characterized in that: The step of inputting the abnormal data into a convolutional neural network to determine the abnormal mode of the target compartment specifically includes: If it is determined by the convolutional neural network that the abnormal data indicates that the first electromagnetic field distribution and the second electromagnetic field distribution are asymmetrically distributed, then the abnormal mode is determined to be a car tilt and / or eccentric loading; Through the convolutional neural network, if it is determined that the abnormal data indicates that the magnetic flux density corresponding to the abnormal characteristic area is greater than or equal to a preset threshold, the abnormal mode is determined to be loose car body components.

8. A railway carriage monitoring device, characterized in that: The railway carriage monitoring device comprises an acquisition module (31) and a processing module (32), wherein: The acquisition module (31) is used to acquire electromagnetic induction data for the target compartment sent by the sensor device; The processing module (32) is used to perform finite element analysis and inverse problem solving on the electromagnetic induction data to generate an electromagnetic image of the vehicle compartment; The acquisition module (31) is also used to acquire the deployment position data of the sensor device on the target carriage; The processing module (32) is further used to determine abnormal data of the target carriage based on the deployment position data and the carriage electromagnetic image; The processing module (32) is further used to input the abnormal data into a convolutional neural network to determine the abnormal mode of the target compartment.

9. An electronic device, characterized in that: The electronic device comprises a processor (41), a memory (45), a user interface (43) and a network interface (44), wherein the memory (45) is used to store instructions, the user interface (43) and the network interface (44) are both used to communicate with other devices, and the processor (41) is used to execute the instructions stored in the memory (45) so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.

Images