A truck operation safety monitoring method and system based on TFDS

Abstract

The present invention provides a method and system for monitoring the operation safety of a freight car based on TFDS, which relates to the technical field of railway freight car fault detection, including: obtaining first information, second information, third information, fourth information and fifth information; performing matrix construction processing according to the third information to obtain initial load distribution information; performing dynamic analysis according to the initial load distribution information and the fourth information to obtain dynamic load distribution information; performing correlation analysis according to the dynamic load analysis information and the second information to obtain the axle load distribution state; performing model construction processing according to the axle load distribution state and the first information to obtain a chain graph model; dynamically adjusting and evaluating the stress state and coupling relationship of each node in the chain graph model according to the fifth information to obtain fault warning information. The present invention can reflect the load distribution of the carriage and the stress transfer between axles in real time by fully integrating multi-source data.

Description

Technical Field

[0001] The present invention relates to the technical field of railway freight car fault detection, and in particular to a freight car operation safety monitoring method and system based on TFDS. Background Art

[0002] With the continuous growth of railway freight demand, the operation safety of freight cars has become a vital part of the railway transportation system. The current freight car safety monitoring technology mainly relies on traditional single-point sensor monitoring systems and static analysis methods. These technologies usually judge the operation status of freight cars through simple load measurement and axle stress monitoring, but it is difficult to capture the coupling effect between axles under complex dynamic conditions in real time. In recent years, the railway freight car operation fault dynamic image detection system (TFDS) has begun to play an increasingly important role in the railway field. TFDS is an intelligent system that integrates high-precision speed measurement system, high-speed digital image acquisition, large-capacity image data real-time processing and precise positioning pattern recognition technology, which can effectively monitor and identify various faults in freight car operation. However, in the existing technology, the application of TFDS is still limited, and it is usually only used for single data source analysis, such as axle image monitoring or basic judgment of load distribution, which is difficult to fully reflect the complex load distribution and coupling effect between axles. In addition, the existing fault prediction is mostly based on simple threshold judgment, which cannot fully consider the complex coupling effect under dynamic conditions, resulting in high false alarm rate and missed alarm rate, and it is difficult to achieve accurate safety monitoring.

[0003] Based on the defects of the existing technology, there is an urgent need for a truck operation safety monitoring method and system based on TFDS. Summary of the invention

[0004] The purpose of the present invention is to provide a truck operation safety monitoring method and system based on TFDS to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for monitoring truck operation safety based on TFDS, comprising:

[0006] Acquire first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes load data of each carriage and distribution of goods in the carriage, the fourth information includes train running data, and the fifth information is real-time monitoring data;

[0007] Performing matrix construction processing according to the third information, and obtaining initial load distribution information for reflecting the static load of each carriage by calculating and generating a load distribution matrix of each carriage;

[0008] Performing dynamic analysis based on the initial load distribution information and the fourth information, generating a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtaining dynamic load distribution information of each carriage under different running conditions;

[0009] Performing correlation analysis based on the dynamic load analysis information and the second information, extracting features from the axle deformation image data, and performing correlation analysis in combination with the dynamic load distribution information, calculating the deformation degree of each axle and its corresponding load stress, and obtaining the axle load distribution state;

[0010] Performing model building processing according to the axle load distribution state and the first information, building a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles;

[0011] The stress state and coupling relationship of each node in the chain diagram model are dynamically adjusted and evaluated according to the fifth information, high-risk axles are identified, chain failures are predicted, and fault warning information is obtained.

[0012] In the second aspect, the present application also provides a truck operation safety monitoring system based on TFDS, including:

[0013] an acquisition module, used to acquire first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes the load data of each carriage and the distribution of goods in the carriage, the fourth information includes the running data of the train, and the fifth information is real-time monitoring data;

[0014] A calculation module, configured to perform matrix construction processing according to the third information, and obtain initial load distribution information reflecting the static load of each carriage by calculating and generating a load distribution matrix of each carriage;

[0015] an analysis module, configured to perform dynamic analysis according to the initial load distribution information and the fourth information, generate a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtain dynamic load distribution information of each carriage under different running conditions;

[0016] a correlation module, configured to perform correlation analysis based on the dynamic load analysis information and the second information, extract features from the axle deformation image data, and perform correlation analysis in combination with the dynamic load distribution information, thereby calculating the deformation degree of each axle and its corresponding load stress, and obtaining the axle load distribution state;

[0017] A construction module, configured to perform model construction processing according to the axle load distribution state and the first information, and to construct a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles;

[0018] An identification module is used to dynamically adjust and evaluate the stress state and coupling relationship of each node in the chain diagram model according to the fifth information, identify high-risk axles and predict chain failures, and obtain fault warning information.

[0019] The beneficial effects of the present invention are:

[0020] The present invention uses the TFDS system to efficiently acquire axle deformation image data, combines axle sensor data and load distribution information, and comprehensively integrates multi-source data, which can reflect the load distribution of the carriage and the stress transfer between axles in real time, significantly improving the accuracy and comprehensiveness of data analysis; the present invention establishes an inter-axle coupling relationship model based on stress transfer and coupling effects, dynamically analyzes the chain effect between axles, and can predict the occurrence of chain failures by accurately calculating the coupling relationship between each axle, thereby improving the early warning capability of systemic failures.

[0021] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 It is a flow chart of a truck operation safety monitoring method based on TFDS described in an embodiment of the present invention;

[0024] Figure 2 Schematic diagram of the chain graph model described in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0027] Embodiment 1:

[0028] This embodiment provides a truck operation safety monitoring method based on TFDS.

[0029] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300, step S400, step S500 and step S600.

[0030] Step S100, obtaining first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes the load data of each carriage and the distribution of goods in the carriage, the fourth information includes the running data of the train, and the fifth information is real-time monitoring data;

[0031] It should be noted that the first information is the axle sensor data, which mainly includes basic physical quantities such as temperature, vibration, and stress, and can directly reflect the working status of the axle during operation. The second information is collected through the railway freight car operation fault dynamic image detection system (TFDS). This system is based on high-precision speed measurement, digital image acquisition and pattern recognition technology, and captures the deformation of the axle in real time, providing accurate deformation image data. The third information covers the load data and cargo distribution of each carriage, which is the basis for load analysis and dynamic center of gravity calculation. The fourth information includes the train's operating data, such as speed, acceleration, and route characteristics (such as curves, ramps, etc.). The fifth information is real-time monitoring data, which integrates the instant feedback of all the aforementioned information for real-time adjustment and prediction. By integrating these five types of information, the system can comprehensively and dynamically grasp the status of the train during operation.

[0032] Step S200: performing matrix construction processing according to the third information, and obtaining initial load distribution information for reflecting the static load of each carriage by calculating and generating the load distribution matrix of each carriage;

[0033] It should be noted that this step first divides each car into several partitions and accurately calculates the load of each partition. These load data include the weight of the cargo and its specific position in the car. Next, the mass center method is used to combine the load of each partition with its position coordinates in the car coordinate system, and the load of the entire car is vectorized through mass distribution. Subsequently, the system uses a matrix transformation method to integrate these load distribution vectors into a matrix structure that reflects the overall static load of the car, namely the load distribution matrix. This matrix not only describes the load distribution of each partition in the car, but also reflects the offset of the center of gravity of the car under the load of each partition. In this way, the system can accurately capture the load state of each car under static conditions.

[0034] Step S300, performing dynamic analysis based on the initial load distribution information and the fourth information, generating a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtaining dynamic load distribution information of each carriage under different running conditions;

[0035] It can be understood that this step can predict the offset of the center of gravity of the carriage and its impact on the axle load in real time by accurately simulating the dynamic load changes of the train under different working conditions.

[0036] Step S400, performing correlation analysis based on the dynamic load analysis information and the second information, extracting features from the axle deformation image data, and performing correlation analysis in combination with the dynamic load distribution information, calculating the deformation degree of each axle and its corresponding load stress, and obtaining the axle load distribution state;

[0037] It should be noted that this step can not only identify the real-time stress state of the axle, but also predict possible stress failure areas by deeply combining image features with dynamic load information, thereby achieving more accurate safety monitoring and fault prediction.

[0038] Step S500: performing model building processing according to the axle load distribution state and the first information, and building a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles;

[0039] It can be understood that this step, by constructing an accurate chain diagram model, can intuitively understand and analyze the coupling relationship between axles, thereby providing a scientific basis for predicting potential chain failures and optimizing safety monitoring.

[0040] Step S600: dynamically adjust and evaluate the stress state and coupling relationship of each node in the chain graph model according to the fifth information, identify high-risk axles and predict chain failures, and obtain fault warning information.

[0041] It should be noted that this step can accurately identify and predict potential risks in truck operation, especially cascading failures between multiple axles, through dynamic fusion and evaluation of real-time data. Compared with traditional static monitoring methods, this dynamic adjustment and prediction mechanism based on the chain graph model significantly improves the accuracy and timeliness of fault warnings, providing more reliable protection for the safe operation of trucks.

[0042] Further, step S200 includes step S210 to step S240.

[0043] Step S210: Perform preliminary load data processing according to the third information, calculate the load distribution of each partition in the carriage, and convert the load of each partition into a position vector in the coordinate system using the mass center method to obtain the load distribution vector of each carriage;

[0044] Specifically, this step applies the mass center method, which combines the load of each partition and its position coordinates in the car coordinate system by weighted averaging to obtain the load distribution vector of the entire car. During the calculation process, the mass center position of the entire car is obtained by accumulating the load and position of each partition, and expressed in the form of a load distribution vector. This processing method can not only accurately describe the distribution of cargo in the car, but also reveal the impact of the load of each partition on the overall center of gravity of the car. Through this step, a high-precision load distribution vector is obtained, which reflects the center of gravity and load status of the car under static conditions. The calculation formula is:

[0045]

[0046] Where i is the number of the carriage; j is the number of the partition; n is the number of partitions in the i-th carriage; V i represents the load distribution vector of the i-th carriage; m ij represents the load of the jth partition in the i-th carriage; r ij Represents the position vector of the jth partition in the i-th carriage in the carriage coordinate system.

[0047] Step S220, calculate the center of gravity according to the load distribution vector, convert the load distribution vector of each carriage into the center of gravity coordinates of the carriage by means of a matrix transformation method, and obtain the center of gravity position matrix of the carriage;

[0048] In the process of center of gravity calculation, the basic transformation matrix is ​​first applied to perform linear transformation on the load distribution vector to obtain the preliminary center of gravity coordinates. However, considering the complex nonlinear factors that may exist in the load distribution of the carriage in actual applications (such as uneven load distribution, the influence of the shape of the carriage, etc.), a nonlinear correction matrix and corresponding nonlinear functions are further introduced. These nonlinear correction terms ensure that the final calculated center of gravity position matrix is ​​more accurate by adjusting and compensating for subtle effects that cannot be captured by the basic linear transformation, especially when dealing with complex load distribution or carriage deformation. The calculation formula is:

[0049]

[0050] Where i represents the serial number of the carriage; C i represents the center of gravity position matrix of the i-th carriage; T i Represents the basic transformation matrix; V i represents the load distribution vector of the i-th car; k represents the serial number of the nonlinear correction matrix; p represents the total number of nonlinear correction items; N ik represents the kth nonlinear correction matrix of the ith carriage; f k (V i ) represents the load distribution vector V i The kth nonlinear function of .

[0051] Step S230, performing static load calculation according to the center of gravity position matrix, analyzing the moment balance between the center of gravity position and the axle, and calculating the static load of each axle to obtain a static load vector reflecting the stress state of the axle;

[0052] Specifically, this step first determines the spatial position of the center of gravity of each car relative to the axle based on the center of gravity position matrix obtained in the previous step. Since the center of gravity position determines the proportion of the load borne by each axle, the system establishes a torque balance equation on this basis. Through the torque balance analysis, the system considers the mechanical environment of each axle, including the position of the center of gravity of the car relative to the axle and the geometric distribution of each axle. After establishing the torque balance equation, by solving these equations, the load distribution of each axle under the current static conditions is obtained, that is, the static load vector. This vector quantifies the force borne by each axle, reflecting the impact of the car center of gravity offset on the force on each axle. The form of the torque balance equation is as follows:

[0053]

[0054] Where s represents the serial number of the axle; q represents the total number of axles in the i-th carriage; L is represents the arm length of the sth axle in the ith carriage; F is represents the force on the sth axle in the ith carriage; W i represents the total load of the i-th carriage; C i represents the center of gravity position matrix of the i-th carriage; P is represents the position of the sth axle in the ith carriage.

[0055] Step S240, performing matrix construction processing according to the static load vector, integrating the static load data of each car into a matrix structure reflecting the axle load distribution of the entire train by constructing a load influence matrix, and obtaining an initial load distribution matrix.

[0056] Specifically, this step first uses the static load vector of each car as the basic data to form the local load matrix of each car. Next, according to the structural characteristics of the whole train and the mutual influence between cars, a load influence matrix is ​​constructed, which is used to describe the load transfer relationship and interaction between cars. The load influence matrix not only takes into account the static load distribution of each car, but also integrates the mutual influence between cars due to structural coupling, uneven load and other factors. The design of this matrix structure fully reflects the overall mechanical environment of the train. Preferably, a multidimensional load influence matrix is ​​generated by analyzing the load distribution ratio between adjacent cars and the stress transfer path of the connecting parts between cars. Finally, through matrix operations, the static load vectors of each car are combined with the load influence matrix to integrate into an initial load distribution matrix reflecting the load state of the axles of the entire train. The initial load distribution matrix contains the load distribution information of each car on different axles, which is an accurate expression of the comprehensive load state of the entire train under static conditions.

[0057] Further, step S300 includes step S310 to step S340.

[0058] Step S310, calculating the inertial force of each carriage under different road sections and speed conditions according to the initial load distribution information and the fourth information, and obtaining an inertial force distribution matrix of each carriage;

[0059] It can be understood that this step generates the inertia force distribution matrix of each car by calculating the inertia force of different partitions of each car point by point. The matrix describes in detail the distribution of inertia force at different positions in the car under various operating conditions. In particular, under complex road conditions (such as curves and ramps), the influence of the load of each partition on the overall inertia force distribution is considered, and the dynamic load changes of each car are accurately expressed in matrix form.

[0060] Step S320, performing a gravity center offset analysis according to the inertial force distribution matrix, calculating the gravity center offset of each car under the action of the inertial force by applying a moment balance analysis based on the gravity center offset, and obtaining a dynamic gravity center offset vector of each car;

[0061] Specifically, first, the inertial force distribution of each car under different road sections and speed conditions is determined based on the inertial force distribution matrix. These inertial forces act on each partition of the car, causing the overall center of gravity of the car to shift. Next, the moment balance analysis method is applied to consider the inertial force of each partition and its lever arm relative to the geometric center of the car, and calculate the influence of these moments on the overall center of gravity position of the car. The core of the moment balance analysis is to calculate the degree of center of mass shift according to the effect of inertial force on each part of the car. By establishing the balance equation between the center of gravity of the car and the inertial force of each partition, the center of gravity shift of each car can be solved. Finally, the calculation results are integrated into a dynamic center of gravity shift vector, which reflects the center of gravity shift of the car under different operating conditions. This vector not only shows the change of the center of gravity relative to the initial position of the car, but also captures the complex dynamic behavior caused by the inertial force, especially in complex sections such as curves and ramps. The center of gravity shift vector can accurately describe the dynamic change of the center of gravity of the car.

[0062] Step S330, performing load distribution processing according to the dynamic center of gravity offset vector, calculating the load change of each axle under different operating conditions through a load distribution adjustment algorithm based on center of gravity movement, and obtaining a dynamic load distribution matrix;

[0063] It should be noted that this step uses the dynamic center of gravity offset vector as input to identify the changes in the center of gravity position of each car under different operating conditions. These changes directly affect the stress state of each axle, because the movement of the center of gravity will cause the load in the car to be redistributed, which in turn affects the vertical load of the axle. Preferably, a load distribution adjustment algorithm based on the movement of the center of gravity is adopted, which takes into account the offset of the center of gravity of the car relative to the position of each axle and recalculates the load of each axle under dynamic conditions. Specifically, when the center of gravity shifts in a certain direction, the axle close to the direction will bear a greater load, while the axle away from the direction will bear a smaller load. By applying the relationship between mechanical balance and center of gravity offset, the load changes of each axle are accurately calculated. The algorithm not only takes into account linear load changes, but also handles the complexity of nonlinear load distribution, especially in extreme conditions such as sharp turns or ramp driving, which can reflect the true distribution of the car load. Finally, these calculation results are integrated into a dynamic load distribution matrix, which describes in detail the load changes on each axle under various dynamic operating conditions.

[0064] Step S340: Perform dynamic analysis and processing according to the dynamic load distribution matrix, combine the time series analysis of the operation data, and accumulate the load changes of each car during the entire operation process to obtain the dynamic load distribution information of each car under different operating conditions.

[0065] It should be noted that the dynamic load distribution matrix records in detail the load status of each axle at a specific time and operating conditions. However, train operation is a continuous dynamic process, and the carriage load will continue to adjust with time and changes in the operating environment. Therefore, the system must perform time series analysis to capture the trend and cumulative effect of load changes over time.

[0066] This step combines the instantaneous data in the dynamic load distribution matrix with the historical data of train operation by applying the time series analysis method to analyze the evolution trajectory of the load in different time periods. Time series analysis not only takes into account the immediate changes in the load, but also analyzes the cumulative impact of the load throughout the operation process, especially the stress concentration or stress fatigue caused by long-term effects. In addition, by accumulating the changes in the load, the system can identify significant changes in the load in specific sections (such as curves, ramps) or specific working conditions (such as acceleration and deceleration). Finally, these analysis results are integrated into dynamic load distribution information, which comprehensively reflects the load distribution status of each car under different operating conditions. This information not only shows the distribution of the load at different time points, but also reveals how the load is gradually accumulated or transferred throughout the operation process.

[0067] Further, step S400 includes step S410 to step S440.

[0068] Step S410, performing preliminary processing on axle deformation according to dynamic load distribution information, and obtaining theoretical deformation distribution of each axle by simulating and calculating stress and strain of the axle under different dynamic load conditions;

[0069] Specifically, the simulation process simulates the complex stress conditions that the axle is subjected to in actual operation, including static loads, dynamic inertia forces, bending moments, and possible impact forces. The superposition of these forces will cause stress concentration and corresponding strain changes in different parts of the axle. Through simulation, the stress distribution diagram and corresponding deformation mode of the axle under different operating conditions can be calculated, providing a theoretical deformation distribution, that is, the deformation characteristics that the axle should exhibit under ideal conditions. Furthermore, factors such as material properties, temperature changes, and fatigue effects need to be considered to make the simulation results closer to the actual situation. The theoretical deformation distribution not only provides a standard reference for subsequent deformation monitoring and data analysis, but also identifies possible stress concentration areas and potential weak points, providing a scientific basis for the safety assessment of the axle.

[0070] Step S420, performing feature extraction processing on the axle deformation image data according to the second information, performing multi-scale feature extraction on the axle deformation image data through a convolutional neural network, and obtaining an image feature vector reflecting the actual deformation of the axle;

[0071] Specifically, this step first normalizes the original image data through the preprocessing step, removes noise and enhances the key deformation features of the image to ensure the quality of the input data. Then the preprocessed image data is input into the convolutional neural network. With its unique convolutional layer structure, the convolutional neural network performs convolution operations on the image through multiple convolution kernels, and extracts deformation features such as edges, corners, and curvatures in the image layer by layer. By identifying features of different scales and different levels, the system can capture the deformation details of the axle at the micro and macro levels. Through the dimensionality reduction operation of the pooling layer, the system retains the key features in the image, reduces the data dimension, and improves the computational efficiency. Finally, through the fully connected layer, the system combines multi-scale features into a high-dimensional image feature vector. The image feature vector accurately reflects the deformation characteristics of the axle in actual operation, including but not limited to bending, twisting, wear and other deformation modes. Unlike single-scale image processing, the multi-scale feature extraction method of the convolutional neural network can comprehensively analyze the subtle differences in the deformation image, making the system more sensitive and accurate when dealing with complex deformation situations.

[0072] Step S430, performing deformation stress correlation processing according to the theoretical deformation distribution and the image feature vector, calculating the difference value by matching the image feature vector with the theoretical deformation distribution point by point, and analyzing the difference pattern to obtain the deformation deviation vector;

[0073] It can be understood that the theoretical deformation distribution represents the deformation state that the axle should exhibit under ideal load conditions, while the image feature vector reflects the deformation exhibited by the axle in actual operation. This step calculates the difference value between each corresponding point through an accurate point-by-point matching algorithm. This difference value represents the deviation between the theoretical deformation and the actual deformation, which may be caused by various factors, such as local concentration of load, fatigue effect of materials, or unexpected mechanical impact. In order to understand these deviations more comprehensively, it is necessary to further analyze the spatial distribution and pattern of the difference values. For example, identifying that there are large deviation values ​​in certain areas and showing a consistent trend may indicate some kind of stress concentration or potential structural failure. Finally, the deformation deviation vector is obtained, which not only quantifies the difference between the theoretical deformation and the actual deformation, but also provides key information for in-depth understanding of the stress and deformation state of the axle.

[0074] Step S440, calculate and process the stress state according to the deformation deviation vector, use the deformation deviation data based on the inverse stress derivation algorithm to inversely calculate the load actually applied to the axle, and calculate the actual stress distribution of each axle through the stress-strain relationship, calculate the actual load stress of the axle, and obtain the final axle load distribution state.

[0075] It can be understood that the reverse stress derivation algorithm is based on the principles of mechanical equilibrium and material mechanics. Through known deformation and material properties, the external force and load applied to the axle are calculated in reverse. Specifically, this step first combines the deformation deviation vector with the geometric structure and material properties of the axle to establish a stress-strain relationship model. Through this model, the deviation can be converted into a stress value, and combined with the geometric distribution of the axle, the actual load applied at a specific position can be inferred. Next, the stress-strain relationship is used to further calculate the stress distribution of each axle in actual operation. The stress-strain relationship describes the deformation behavior of the material under stress, which can help the system accurately calculate various stress concentration areas and possible fatigue areas. Finally, these stress distribution information are integrated to generate the actual load stress of the axle, and construct a complete axle load distribution state. The stress-strain relationship is expressed as:

[0076]

[0077] Among them, σ x , σ y and σ z Respectively represent the normal stress in the x, y, and z directions; τ xy , τ yz and τ zx Respectively represent the shear stress on the xy, yz, and zx planes; E represents Young's modulus; v represents Poisson's ratio; ∈ x ,∈ y and ∈ zRespectively represent the normal strain in the x, y, and z directions; γ xy , γ yz and γ zx They represent the shear strains on the xy, yz, and zx planes, respectively.

[0078] Further, step S500 includes step S510 to step S540.

[0079] Step S510: Perform preliminary analysis and processing of coupling effects according to the axle load distribution state and the first information, quantify the coupling relationship between the axles using a multivariate regression analysis algorithm, determine the influence of stress changes of any axle on adjacent axles by analyzing the stress transfer path of each axle, and generate a coupling coefficient matrix;

[0080] It can be understood that this step applies a multivariate regression analysis algorithm to quantify the relationship between the axles. Specifically, the algorithm evaluates how the stress changes of one axle are transmitted to the adjacent axles through mechanical connections and structural coupling by analyzing the stress transfer path between the axles. In order to more accurately describe this coupling relationship, multiple variables are incorporated into the regression model, including the stress level, geometric position, load history, etc. of each axle. The coupling strength between different axles is determined by regression analysis, that is, the degree of influence of the stress change of a certain axle on other axles. Finally, a coupling coefficient matrix is ​​generated based on the results of the regression analysis. Each element of the matrix represents the coupling coefficient between the two axles. The higher the value, the stronger the stress transfer. Through this matrix, the system can clearly understand the stress transfer path and coupling effect between the axles.

[0081] Step S520, performing stress transfer analysis and processing according to the coupling coefficient matrix, converting the preliminary coupling coefficient matrix into edge weights in the axle network based on the stress propagation algorithm of graph theory, simulating the transmission and attenuation of stress in the axle network, and calculating the coupling strength between the axles to obtain a coupling strength matrix;

[0082] It should be noted that this step regards the coupling coefficient matrix as an adjacency matrix in a graph theory problem, where each axle corresponds to a node in the graph, the connecting edges between the nodes correspond to the coupling relationship between the axles, and the edge weights are determined by the coupling coefficients. The higher the coupling coefficient, the larger the edge weight, indicating that the stress transfer is stronger. Next, the stress propagation algorithm in graph theory is used to analyze this axle network. The algorithm takes into account the transmission path, path length, and stress attenuation effect of stress in the network. During the stress transfer process, stress not only propagates along the direct path in the axle network, but may also be transmitted through multiple intermediary nodes (other axles). Therefore, it is necessary to calculate the stress transfer path between the axles and consider the attenuation effect of stress during the transfer process. This attenuation can be linear or nonlinear, depending on the mechanical connection between the axles, the material properties, and the complexity of the load transfer path. By simulating the transmission of stress in the entire axle network, the comprehensive coupling strength between the axles can be calculated.

[0083] Step S530, constructing chain graph nodes and chains according to the coupling strength matrix, treating the axles as nodes in the graph through a network analysis algorithm, using the coupling strength as the weight of the chain, constructing a chain graph model that reflects the physical coupling relationship between the axles, and obtaining a preliminary chain graph structure;

[0084] It is understandable that in this embodiment, information is first extracted from the coupling strength matrix, and each axle is defined as a node in the graph. Each node represents the state of an axle, such as stress level, deformation degree, etc. Next, the coupling strength between the axles (i.e., the element value in the coupling strength matrix) is used as the weight of the chain. The larger the weight of the chain, the stronger the stress transfer relationship between the two axles and the more significant the coupling effect. Then, a chain graph model is constructed by a network analysis algorithm. The algorithm not only considers the directly connected axle pairs, but also identifies which nodes play a key role in the stress transfer and coupling effect of the entire system by analyzing the overall structure in the axle network. By identifying high-weight chains and key nodes, it is determined which axles play a core role in the overall network and which axles may become the path of fault propagation. In the process of constructing the chain graph, the topological features in the axle network are also considered, such as the degree of the node, that is, the number of chains connected to the node, and the connectivity and clustering coefficient of the network. These topological features help to reveal the complex coupling relationship between axles, especially in the interaction between multiple axles, and key nodes and chains with potential chain effects can be identified.

[0085] like Figure 2 As shown, Figure 2The figure is a schematic diagram of the chain graph model. The nodes in the figure represent axles. The node names such as "Axle1" to "Axle 6" identify the specific axle numbers. The circular frame of the node symbolizes the physical existence of the axle and displays its status information, such as stress level or deformation degree. The nodes are of equal size to ensure equal importance in the figure. The edges represent the coupling relationship or physical connection between the axles. The thickness of the edge is proportional to the weight, indicating the coupling strength. Thicker edges indicate stronger stress transfer between axles. The numbers on the edges (weight labels) specify the numerical value of the coupling strength. The edge weight label is located in the middle of the edge, slightly offset from the node, and is used to identify the coupling strength between axles, providing quantitative information to help understand the interaction between axles. Key nodes in the figure, such as nodes connecting multiple high coupling strength edges (such as "Axle 4"), play an important role in fault propagation or stress transfer and need to be monitored. Edges with high coupling strength, such as "Axle 4" to "Axle 5", indicate stronger coupling and are areas of stress concentration, requiring special attention. In addition, the cross connections in the figure (such as "Axle 1" to "Axle 3") show the complex coupling relationship between axles, which is not limited to the direct connection between adjacent axles and may lead to complex stress distribution, which needs to be considered in stress analysis and failure prediction.

[0086] Step S540, performing coupling weight optimization processing according to the preliminary chain graph structure, adjusting the coupling weight of the chain by comprehensively considering the dynamic load changes of the axles and the stress history data, optimizing the interaction model between the axles, and obtaining the final chain graph model.

[0087] Specifically, this step considers the following factors: whether the dramatic fluctuations in dynamic loads have caused changes in the stress transfer paths between axles, and whether the stress history data indicates that the coupling relationship between some axles has been strengthened or weakened due to long-term stress accumulation. Through these analyses, the system can identify which chains need to increase their weights to reflect stronger coupling relationships, and which chains need to reduce their weights to reflect weakened coupling effects.

[0088] Further, step S600 includes step S610 to step S640.

[0089] Step S610: dynamically monitor the stress state according to the fifth information and the chain graph model, dynamically estimate the current stress state of each axle node by combining the historical stress data and the real-time monitoring data, and obtain a real-time stress state vector;

[0090] It can be understood that the real-time stress state vector represents the actual stress state of each axle node under the current operating conditions, and reflects the stress transfer and diffusion caused by dynamic load changes or coupling effects. The real-time stress state vector provides basic data support for subsequent high-risk axle identification and fault prediction.

[0091] Step S620: dynamically adjust the coupling relationship according to the real-time stress state vector, update the coupling relationship between nodes in the chain graph model by combining the current stress state and the historical coupling weight through the Bayesian network-based relationship inference algorithm, and obtain an adjusted coupling relationship matrix;

[0092] It should be noted that the Bayesian network is a probabilistic model that can use conditional probabilities to infer the relationship between variables under uncertain conditions. In this embodiment, the Bayesian network is used to learn the dependency between the real-time stress state vector and the historical coupling weights to infer the latest coupling strength between the axles under current conditions. Specifically, the Bayesian network recalculates the coupling weights between each pair of axles by updating the conditional probabilities between the nodes (axles) in the chain graph model. This process not only takes into account the impact of the current stress state on the coupling relationship, but also combines the long-term coupling trends observed in historical data.

[0093] Step S630: perform high-risk axle identification processing according to the coupling relationship matrix, use the stress state of the axle node and the adjusted coupling relationship as input data, perform cluster analysis on the nodes, divide axles into different risk categories by identifying abnormal patterns of stress and coupling relationship, and identify nodes with excessive stress or abnormal coupling strength to obtain a list of high-risk axles;

[0094] Specifically, this step combines the adjusted coupling relationship matrix with the real-time stress state vector, which comprehensively reflects the physical state of each axle node and its coupling with other axles. The stress state data provides the instantaneous stress level of each axle, while the coupling relationship matrix reveals the interaction strength between the axles. By integrating these data, the complex interactions existing in the axle network and their impact on the overall stress distribution can be captured. Next, the axle nodes are classified using a cluster analysis algorithm. Cluster analysis is an unsupervised learning method that divides data points (i.e., axle nodes in this embodiment) into different groups or clusters by analyzing the similarities between them. Abnormal patterns of stress states and coupling relationships are identified by clustering algorithms, such as stress values ​​of some axles being significantly higher than other axles, or abnormal increases or decreases in coupling strength between some nodes. These anomalies indicate potential risk areas or fault points. Specifically, by analyzing the stress distribution, coupling strength distribution, and their positions in the network, axle nodes that exhibit excessive stress or abnormal coupling relationships are identified. Through this clustering process, axle nodes of different risk categories can be divided. Preferably, axle nodes are divided into low-risk, medium-risk and high-risk categories, and axles with excessive stress or significantly abnormal coupling strength are specially marked. These high-risk axles are included in the high-risk axle list and become the focus of subsequent chain failure prediction and early warning.

[0095] Step S640: Perform chain failure prediction processing according to the high-risk axle list, simulate the chain reaction caused by the high-risk axle failure by using the Markov chain Monte Carlo method, predict the system-level failure risk, and obtain failure warning information.

[0096] It should be noted that the Markov Chain Monte Carlo method is a numerical simulation method based on random sampling, which is used to estimate the probability distribution in complex systems that is difficult to calculate directly.

[0097] Further, step S640 includes step S641 to step S644.

[0098] Step S641: Perform preliminary fault propagation path processing according to the high-risk axle list, and obtain a potential fault propagation path set by constructing a fault propagation path model between high-risk axles;

[0099] Specifically, all axle nodes marked as high risk are first extracted from the list of high-risk axles. These nodes are considered to be potential fault sources because of their characteristics of excessive stress or abnormal coupling strength. Next, the coupling relationship matrix in the chain graph model is used to analyze the coupling strength between high-risk nodes and their mutual connections in the axle network. The higher the coupling strength between node pairs, the greater the possibility of fault propagation. Then, based on these analysis results, a fault propagation path model between high-risk axles is constructed. During the construction process, direct and indirect connection paths between high-risk nodes are identified, and the feasibility of fault propagation on each path is evaluated. A direct path refers to the situation where two high-risk nodes are connected through a single chain, while an indirect path includes the situation where they are connected through multiple intermediate nodes. Furthermore, not only the propagation of faults on direct paths, but also complex paths that are indirectly propagated through multiple nodes are considered. Finally, the fault propagation paths are integrated to generate a potential fault propagation path set. This path set contains all possible fault propagation paths and identifies the connection nodes, coupling strength, and potential propagation direction of each path.

[0100] Step S642: performing chain reaction probability calculation processing according to the potential fault propagation path set, calculating the state transition probability of each propagation path by using a state transition matrix construction algorithm based on a Markov chain, and obtaining a fault propagation probability matrix;

[0101] It can be understood that this step first extracts all possible fault propagation paths from the potential fault propagation path set. Each path represents a route where a high-risk axle fault may propagate to other axles. Next, a state transfer matrix is ​​established based on the principle of Markov chain, which is used to describe the probability of a fault propagating from one node (axle) to the next node. The state transition probability reflects the possibility of a fault propagating from one axle to the next axle on a given path. These state transition probabilities are determined by analyzing the coupling strength between each pair of connected nodes in the coupling relationship matrix, as well as historical fault data (such as the frequency and path of fault occurrence). Specifically, the state transition probability of fault propagation is higher between axle pairs with higher coupling strength. These probability values ​​are then filled in the corresponding positions of the state transfer matrix to form a complete state transfer matrix.

[0102] In order to further improve the matrix, a variety of possible external factors are also considered, preferably such as operating conditions, dynamic load changes, and abnormal patterns in historical data. These factors may affect the propagation path and speed of the fault, so in the process of constructing the state transfer matrix, the system integrates these data to adjust and optimize the transition probability. Finally, by calculating the state transition probabilities of all potential fault propagation paths, a fault propagation probability matrix is ​​generated. Each element of the matrix represents the probability of the fault propagating from one node to the next node between different axles. This matrix not only describes the possible propagation direction of the fault, but also quantifies the uncertainty in the propagation process.

[0103] Step S643: perform system-level fault risk assessment processing according to the fault propagation probability matrix, perform multiple random samplings using the Markov chain Monte Carlo simulation algorithm to simulate the propagation process of high-risk axle faults in the system, perform statistical analysis on the simulation results, evaluate the intensity and scope of the chain reaction, and calculate the overall fault risk of the system to obtain a system-level fault risk assessment result;

[0104] Preferably, this step uses the Markov chain Monte Carlo algorithm to perform a large number of random samplings, and each sampling simulates a possible fault propagation path. In each simulation, one or more high-risk axles are randomly selected as the initial fault point, and then the process of fault propagation from one axle to another is gradually simulated according to the state transition probability. Due to the randomness of the Markov chain Monte Carlo algorithm, the propagation path of each simulation is different, so all possible fault propagation scenarios can be covered by multiple iterative sampling. In each simulation, the path of fault propagation, the number of axles affected, the propagation time, and the change of coupling relationship during the propagation process are recorded. By performing statistical analysis on all simulation results, common patterns of fault propagation can be identified, and the intensity of the chain reaction (such as the number of axles affected, the propagation speed, etc.) and the scope of the impact (such as the axle groups involved and possible system-level impacts) can be evaluated. Next, the overall fault risk of the system is calculated based on the statistical analysis results. This fault risk depends not only on the results of a single simulation, but also takes into account the similarity and cumulative effect between different simulation results. By summarizing all simulation results, a comprehensive system-level fault risk assessment is generated, which quantifies the probability and severity of the overall system-level chain reaction that may be caused by the fault. The final system-level failure risk assessment results not only identify the axle nodes and paths that are most likely to cause serious chain failures, but also provide specific risk levels (such as low, medium, and high risks) and recommended preventive measures.

[0105] Step S644: perform fault warning information generation processing according to the system-level fault risk assessment result, and obtain final fault warning information by setting a risk threshold and generating corresponding warning information.

[0106] It is understandable that this step first extracts key data from the system-level fault risk assessment results, including the possibility of each high-risk axle and its chain failure, the fault propagation path, the number of axles affected, and the intensity and scope of fault propagation. Then set risk thresholds, which are usually determined based on industry standards, historical data, operating conditions, and management strategies. Preferably, different thresholds are set for different levels of risk, such as minor fault warning, moderate risk warning, and severe risk warning. Next, the system-level fault risk assessment results are compared with the preset thresholds. For risk results that exceed the set thresholds, corresponding warning information is generated, including the cause of the fault, the location of the high-risk axle, the axles and transmission paths that may be affected, the expected fault propagation time, and the recommended countermeasures. Furthermore, the warning information can be further classified according to the urgency of fault propagation and the possible scope of impact, so that managers can prioritize the most serious risks. Finally, the system integrates all warning information to generate a final fault warning information report. The report not only covers a detailed analysis of various risks, but also provides operational recommendations based on risk assessment, such as slowing down, increasing monitoring frequency, or performing axle maintenance. Through this early warning information, managers can understand potential failure risks in advance and take corresponding preventive measures to reduce or avoid the occurrence of system-level failures.

[0107] Embodiment 2:

[0108] like Figure 2 As shown, this embodiment provides a truck operation safety monitoring system based on TFDS, and the system includes:

[0109] an acquisition module, used to acquire first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes the load data of each carriage and the distribution of goods in the carriage, the fourth information includes the running data of the train, and the fifth information is real-time monitoring data;

[0110] A calculation module, used for performing matrix construction processing according to the third information, and obtaining initial load distribution information reflecting the static load of each carriage by calculating and generating a load distribution matrix of each carriage;

[0111] An analysis module is used to perform dynamic analysis based on the initial load distribution information and the fourth information, generate a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtain dynamic load distribution information of each carriage under different running conditions;

[0112] A correlation module is used to perform correlation analysis based on the dynamic load analysis information and the second information, extract features from the axle deformation image data, and perform correlation analysis in combination with the dynamic load distribution information to calculate the deformation degree of each axle and its corresponding load stress, thereby obtaining the axle load distribution state;

[0113] A construction module is used to perform model construction processing according to the axle load distribution state and the first information, and to construct a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles;

[0114] The identification module is used to dynamically adjust and evaluate the stress state and coupling relationship of each node in the chain diagram model according to the fifth information, identify high-risk axles and predict chain failures, and obtain fault warning information.

[0115] In a specific implementation disclosed in the present application, the calculation module includes:

[0116] a first calculation unit, configured to perform preliminary load data processing according to the third information, calculate the load distribution of each partition in the carriage, and convert the load of each partition into a position vector in a coordinate system by using a mass center method, so as to obtain a load distribution vector of each carriage;

[0117] A second calculation unit is used to calculate the center of gravity according to the load distribution vector, and convert the load distribution vector of each carriage into the center of gravity coordinates of the carriage by a matrix transformation method to obtain the center of gravity position matrix of the carriage;

[0118] The third calculation unit is used to calculate the static load according to the center of gravity position matrix, analyze the moment balance between the center of gravity position and the axle, and calculate the static load of each axle to obtain a static load vector reflecting the stress state of the axle;

[0119] The first construction unit is used to perform matrix construction processing according to the static load vector, and integrate the static load data of each car into a matrix structure reflecting the axle load distribution of the entire train by constructing a load influence matrix to obtain an initial load distribution matrix.

[0120] In a specific embodiment disclosed in the present application, the analysis module includes:

[0121] a fourth calculation unit, for calculating the inertia force of each carriage under different road sections and speed conditions according to the initial load distribution information and the fourth information, and obtaining an inertia force distribution matrix of each carriage;

[0122] A first analysis unit is used to perform a gravity center offset analysis according to the inertial force distribution matrix, calculate the gravity center offset of each car under the inertial force by applying a moment balance analysis based on the gravity center offset, and obtain a dynamic gravity center offset vector of each car;

[0123] A first distribution unit is used to perform load distribution processing according to the dynamic center of gravity offset vector, calculate the load change of each axle under different operating conditions through a load distribution adjustment algorithm based on center of gravity movement, and obtain a dynamic load distribution matrix;

[0124] The second analysis unit is used to perform dynamic analysis processing according to the dynamic load distribution matrix. By combining the time series analysis of the operation data and accumulating the load changes of each car during the entire operation process, the dynamic load distribution information of each car under different operating conditions is obtained.

[0125] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A truck operation safety monitoring method based on TFDS, characterized in that: include: Acquire first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes load data of each carriage and distribution of goods in the carriage, the fourth information includes train running data, and the fifth information is real-time monitoring data; Performing matrix construction processing according to the third information, and obtaining initial load distribution information for reflecting the static load of each carriage by calculating and generating a load distribution matrix of each carriage; Performing dynamic analysis based on the initial load distribution information and the fourth information, generating a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtaining dynamic load distribution information of each carriage under different running conditions; Performing correlation analysis based on the dynamic load analysis information and the second information, extracting features from the axle deformation image data, and performing correlation analysis in combination with the dynamic load distribution information, calculating the deformation degree of each axle and its corresponding load stress, and obtaining the axle load distribution state; Performing model building processing according to the axle load distribution state and the first information, building a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles; The stress state and coupling relationship of each node in the chain diagram model are dynamically adjusted and evaluated according to the fifth information, high-risk axles are identified, chain failures are predicted, and fault warning information is obtained.

2. The truck operation safety monitoring method based on TFDS according to claim 1 is characterized in that: Performing matrix construction processing according to the third information, calculating and generating the load distribution matrix of each carriage to obtain initial load distribution information for reflecting the static load of each carriage, including: Performing preliminary load data processing according to the third information, calculating the load distribution of each partition in the carriage, and converting the load of each partition into a position vector in a coordinate system using a mass center method, thereby obtaining a load distribution vector of each carriage; The center of gravity is calculated according to the load distribution vector, and the load distribution vector of each carriage is converted into the center of gravity coordinates of the carriage by a matrix transformation method to obtain the center of gravity position matrix of the carriage; Static load calculation is performed according to the center of gravity position matrix, and the static load of each axle is calculated by moment balance analysis between the center of gravity position and the axle to obtain a static load vector reflecting the stress state of the axle; A matrix construction process is performed according to the static load vector, and the static load data of each carriage is integrated into a matrix structure reflecting the axle load distribution of the entire train by constructing a load influence matrix, thereby obtaining an initial load distribution matrix.

3. The truck operation safety monitoring method based on TFDS according to claim 1 is characterized in that: Dynamic analysis is performed according to the initial load distribution information and the fourth information, and the dynamic center of gravity offset vector of each carriage is generated by analyzing and calculating the running characteristics of the train on different sections, so as to obtain the dynamic load distribution information of each carriage under different running conditions, including: Calculating the inertia force of each carriage under different road sections and speed conditions according to the initial load distribution information and the fourth information to obtain an inertia force distribution matrix of each carriage; Performing a gravity center offset analysis according to the inertial force distribution matrix, calculating the gravity center offset of each car under the action of the inertial force by applying a moment balance analysis based on the gravity center offset, and obtaining a dynamic gravity center offset vector of each car; Performing load distribution processing according to the dynamic center of gravity offset vector, calculating the load change of each axle under different operating conditions through a load distribution adjustment algorithm based on center of gravity movement, and obtaining a dynamic load distribution matrix; Dynamic analysis and processing is performed according to the dynamic load distribution matrix, and the dynamic load distribution information of each car under different operating conditions is obtained by combining the time series analysis of the operating data and accumulating the load changes of each car during the entire operation process.

4. The truck operation safety monitoring method based on TFDS according to claim 1 is characterized in that: According to the dynamic load analysis information and the second information, correlation analysis is performed, by extracting features from the axle deformation image data, and combining the dynamic load distribution information to perform correlation analysis, the deformation degree of each axle and its corresponding load stress are calculated, and the axle load distribution state is obtained, including: Performing preliminary processing on axle deformation according to the dynamic load distribution information, and obtaining theoretical deformation distribution of each axle by simulating and calculating stress and strain of the axle under different dynamic load conditions; Performing feature extraction processing on the axle deformation image data according to the second information, performing multi-scale feature extraction on the axle deformation image data through a convolutional neural network, and obtaining an image feature vector reflecting the actual deformation of the axle; Performing deformation stress correlation processing according to the theoretical deformation distribution and the image feature vector, calculating difference values ​​by matching the image feature vector with the theoretical deformation distribution point by point, and analyzing the difference pattern to obtain a deformation deviation vector; The stress state is calculated and processed according to the deformation deviation vector, and the load actually applied to the axle is inversely calculated using the deformation deviation data based on the inverse stress deduction algorithm. The actual stress distribution of each axle is calculated through the stress-strain relationship, and the actual load stress of the axle is calculated to obtain the final axle load distribution state.

5. The truck operation safety monitoring method based on TFDS according to claim 1 is characterized in that: A model is constructed according to the axle load distribution state and the first information, and a chain diagram model for reflecting the coupling relationship between the axles is constructed based on the stress transfer and coupling effect mechanism between the axles, including: Performing preliminary analysis and processing of coupling effects according to the axle load distribution state and the first information, quantifying the coupling relationship between the axles using a multivariate regression analysis algorithm, determining the influence of stress changes of any axle on adjacent axles by analyzing the stress transfer path of each axle, and generating a coupling coefficient matrix; Perform stress transfer analysis and processing according to the coupling coefficient matrix, convert the preliminary coupling coefficient matrix into edge weights in the axle network based on the stress propagation algorithm of graph theory, simulate the transmission and attenuation of stress in the axle network, and calculate the coupling strength between each axle to obtain a coupling strength matrix; According to the coupling strength matrix, chain graph nodes and chains are constructed, the axles are regarded as nodes in the graph through a network analysis algorithm, the coupling strength is used as the weight of the chain, a chain graph model reflecting the physical coupling relationship between the axles is constructed, and a preliminary chain graph structure is obtained; The coupling weight optimization process is performed according to the preliminary chain graph structure, the coupling weight of the chain is adjusted by comprehensively considering the dynamic load change of the axle and the stress history data, the interaction model between the axles is optimized, and the final chain graph model is obtained.

6. The truck operation safety monitoring method based on TFDS according to claim 1 is characterized in that: According to the fifth information, the stress state and coupling relationship of each node in the chain graph model are dynamically adjusted and evaluated, high-risk axles are identified and chain failures are predicted, and fault warning information is obtained, including: Performing dynamic monitoring processing of the stress state according to the fifth information and the chain diagram model, dynamically estimating the current stress state of each axle node by combining historical stress data and real-time monitoring data, and obtaining a real-time stress state vector; Dynamically adjust the coupling relationship according to the real-time stress state vector, and update the coupling relationship between nodes in the chain graph model by combining the current stress state and the historical coupling weight through a Bayesian network-based relationship inference algorithm to obtain an adjusted coupling relationship matrix; According to the coupling relationship matrix, high-risk axles are identified, the stress state of the axle nodes and the adjusted coupling relationship are used as input data, the nodes are clustered, and axles of different risk categories are divided by identifying abnormal patterns of stress and coupling relationship, and nodes with excessive stress or abnormal coupling strength are identified to obtain a list of high-risk axles; A chain failure prediction process is performed according to the high-risk axle list, and the chain reaction caused by the high-risk axle failure is simulated by the Markov chain Monte Carlo method to predict the system-level failure risk and obtain failure warning information.

7. The truck operation safety monitoring method based on TFDS according to claim 6 is characterized in that: According to the high-risk axle list, chain failure prediction processing is performed, and the chain reaction caused by high-risk axle failure is simulated by Markov chain Monte Carlo method to predict system-level failure risk and obtain failure warning information, including: Performing preliminary fault propagation path processing according to the high-risk axle list, and obtaining a potential fault propagation path set by constructing a fault propagation path model between high-risk axles; Performing chain reaction probability calculation processing according to the potential fault propagation path set, calculating the state transition probability of each propagation path by a state transition matrix construction algorithm based on a Markov chain, and obtaining a fault propagation probability matrix; Performing system-level fault risk assessment processing according to the fault propagation probability matrix, performing multiple random samplings using a Markov chain Monte Carlo simulation algorithm to simulate the propagation process of high-risk axle faults in the system, statistically analyzing the simulation results, assessing the intensity and scope of the chain reaction, and calculating the overall system fault risk to obtain a system-level fault risk assessment result; Fault warning information generation processing is performed according to the system-level fault risk assessment result, and the final fault warning information is obtained by setting a risk threshold and generating corresponding warning information.

8. A truck operation safety monitoring system based on TFDS, characterized in that: include: an acquisition module, used to acquire first information, second information, third information, fourth information and fifth information, wherein the first information is axle sensor data, the second information is axle deformation image data collected by a railway freight car running fault dynamic image detection system, the third information includes the load data of each carriage and the distribution of goods in the carriage, the fourth information includes the running data of the train, and the fifth information is real-time monitoring data; A calculation module, configured to perform matrix construction processing according to the third information, and obtain initial load distribution information reflecting the static load of each carriage by calculating and generating a load distribution matrix of each carriage; an analysis module, configured to perform dynamic analysis according to the initial load distribution information and the fourth information, generate a dynamic center of gravity offset vector of each carriage by analyzing and calculating the running characteristics of the train on different sections, and obtain dynamic load distribution information of each carriage under different running conditions; a correlation module, configured to perform correlation analysis based on the dynamic load analysis information and the second information, extract features from the axle deformation image data, and perform correlation analysis in combination with the dynamic load distribution information, thereby calculating the deformation degree of each axle and its corresponding load stress, and obtaining the axle load distribution state; A construction module, configured to perform model construction processing according to the axle load distribution state and the first information, and to construct a chain graph model for reflecting the coupling relationship between the axles based on the stress transfer and coupling effect mechanism between the axles, wherein the nodes in the chain graph model represent the states of the axles, the chains represent the physical coupling relationship between the axles, and the coupling weight represents the interaction strength between the axles; An identification module is used to dynamically adjust and evaluate the stress state and coupling relationship of each node in the chain diagram model according to the fifth information, identify high-risk axles and predict chain failures, and obtain fault warning information.

9. The truck operation safety monitoring system based on TFDS according to claim 8 is characterized in that: The calculation module comprises: a first calculation unit, configured to perform preliminary load data processing according to the third information, calculate the load distribution of each partition in the carriage, and convert the load of each partition into a position vector in a coordinate system by using a mass center method, so as to obtain a load distribution vector of each carriage; A second calculation unit is used to calculate the center of gravity according to the load distribution vector, and convert the load distribution vector of each carriage into the center of gravity coordinates of the carriage by a matrix transformation method to obtain the center of gravity position matrix of the carriage; a third calculation unit, configured to calculate the static load according to the center of gravity position matrix, analyze the moment balance between the center of gravity position and the axle, and calculate the static load of each axle to obtain a static load vector reflecting the stress state of the axle; The first construction unit is used to perform matrix construction processing according to the static load vector, and integrate the static load data of each car into a matrix structure reflecting the axle load distribution of the entire train by constructing a load influence matrix to obtain an initial load distribution matrix.

10. The truck operation safety monitoring system based on TFDS according to claim 8 is characterized in that: The analysis module comprises: a fourth calculation unit, configured to calculate the inertia force of each carriage under different road sections and speed conditions according to the initial load distribution information and the fourth information, and obtain an inertia force distribution matrix of each carriage; A first analysis unit is used to perform a gravity center offset analysis according to the inertial force distribution matrix, calculate the gravity center offset of each car under the inertial force by applying a moment balance analysis based on the gravity center offset, and obtain a dynamic gravity center offset vector of each car; a first distribution unit, configured to perform load distribution processing according to the dynamic center of gravity offset vector, calculate the load variation of each axle under different operating conditions through a load distribution adjustment algorithm based on center of gravity movement, and obtain a dynamic load distribution matrix; The second analysis unit is used to perform dynamic analysis processing according to the dynamic load distribution matrix, and obtain the dynamic load distribution information of each car under different operating conditions by combining the time series analysis of the operating data and accumulating the load changes of each car during the entire operation process.

Images