Fault prediction method and system for rail braking equipment

Abstract

The invention relates to the technical field of traffic fault prediction, in particular to a rail braking equipment fault prediction method and system, and the method comprises the steps: obtaining the operation state data of rail braking equipment; a dynamic fault association diagram is constructed, and a fault transmission relation between brake equipment components is obtained; dynamically adjusting the edge weight of the dynamic fault association diagram according to the running state data of the track braking equipment, calculating the PageRank value of the node of the dynamic fault association diagram to identify a key component, and searching a target fault propagation path by adopting a Dijkstra algorithm; constructing a multi-task predictive maintenance decision-making tree, inputting the running state data of the track braking equipment, the key component and the target fault propagation path into the multi-task predictive maintenance decision-making tree, and outputting a multi-dimensional prediction result through multi-layer decision-making; and generating a priority maintenance suggestion according to the multi-dimensional prediction result. Therefore, the problems that dynamic changes of the components are ignored, the relation between the components is cut, and maintenance decision support is lacked in the prior art are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traffic failure prediction, and in particular to a rail brake equipment failure prediction method and system. BACKGROUND

[0002] At present, the maintenance mode of rail brake equipment is still dominated by regular maintenance and after-maintenance. Regular maintenance is a comprehensive inspection and maintenance of the equipment according to a fixed cycle, which ignores the differences between individual equipment in actual operation, and is prone to problems such as over-maintenance causing resource waste, or insufficient maintenance causing safety hazards. After-maintenance is a remedial measure after the occurrence of a fault, which can solve the existing fault, but the delay of the train and the interruption of the operation are inevitable when the fault occurs, and even serious safety accidents such as rear-end collision and derailment may occur, causing huge losses to the society and the economy.

[0003] With the development of information technology and data science, data-driven failure prediction methods have begun to emerge in the field of rail transportation, including expert systems and machine learning technologies. However, some methods only focus on the detection of a single component or a certain type of fault, which breaks the complex fault propagation chain between brake equipment components and cannot accurately deduce the development trend of the fault. Some methods use static models, which are difficult to adapt to the dynamic process of frequent changes in working conditions and continuous degradation of performance during equipment operation. Moreover, most of the existing prediction results only provide fault warning information, lack of deep support for maintenance decision-making, and cannot accurately guide the reasonable allocation of maintenance resources and the scientific determination of maintenance priority. SUMMARY

[0004] The present application provides a rail brake equipment failure prediction method and system to solve the problems of ignoring the dynamic changes of components, breaking the relationship between components, and lacking support for maintenance decision-making in the prior art.

[0005] The first aspect of the present application provides a method for predicting faults of rail brake equipment, comprising the following steps: obtaining operating status data of the rail brake equipment; constructing a dynamic fault association graph to obtain fault transmission relationships between brake equipment components, wherein constructing the dynamic fault association graph includes abstracting the physical components of the rail brake equipment into graph nodes, constructing directed edges based on physical connection relationships, energy transfer paths, and control signal flows, and setting initial weights according to historical fault statistical data; dynamically adjusting the edge weights of the dynamic fault association graph according to the operating status data of the rail brake equipment, calculating the PageRank values ​​of the nodes of the dynamic fault association graph to identify key components, and using the Dijkstra algorithm to find target fault propagation paths; constructing a multi-task predictive maintenance decision tree, inputting the operating status data of the rail brake equipment, the key components, and the target fault propagation paths into the multi-task predictive maintenance decision tree, and outputting multi-dimensional prediction results through multi-layer decision-making, wherein the multi-dimensional prediction results include a comprehensive risk index, fault components, fault types and corresponding probabilities, expected occurrence times, and affected ranges; and generating priority maintenance recommendations based on the multi-dimensional prediction results.

[0006] Optionally, the identifying of key components by calculating the PageRank value of the dynamic fault association graph node includes: in the dynamic fault association graph, the PageRank value represents the influence of the component in the fault propagation network, calculating the PageRank value of each component in turn, calculating the mean and standard deviation based on the PageRank value of each calculated component, if the PageRank value of the component is greater than the sum of the mean and 1.5 times the standard deviation, then the component is determined to be a core component; otherwise, if the PageRank value of the component is greater than the mean, then the component is determined to be an important component; otherwise, it is a general component; obtain the core components and important components as key components.

[0007] Optionally, the PageRank value calculation formula is:

[0008] in, Indicates the component node to be calculated, Indicates pointing to a node The node set of Represents a slave node To Node The edge weight of , d is the damping coefficient; Optionally, the using the Dijkstra algorithm to find the target fault propagation path includes: using the Dijkstra algorithm, starting from the key component node, and searching for a fault propagation path with the largest sum of weights as the target fault propagation path, wherein the specific steps are as follows: S1, remove edges whose weight is lower than the target threshold; S2. Initialize the distance array, the visited set, and the predecessor node array, wherein the distance array is used to store the cumulative weight of the maximum probability path from the starting node S to each node. Initially, the dist value of the starting node is set to 1, and the dist values ​​of the remaining nodes are set to 0. The visited set is used to mark the nodes with the determined maximum probability path, and the predecessor node array is used to store the predecessor node of each node on the maximum probability path; S4. Select the node U with the largest dist value from all unvisited nodes and add it to the visited set; S5. For all adjacent nodes V of node U, calculate the cumulative propagation probability from S to V via U. The specific calculation formula is:

[0009] in, is the edge weight from node U to V; S6. If the cumulative propagation probability is greater than the dist value of the adjacent node V, the dist value of the adjacent node V is updated to the cumulative propagation probability, and the predecessor node of V is recorded as U in the predecessor node array; S7, loop through steps S4-S6 until all reachable nodes are processed; S8. Starting from the last node, trace back to the starting point along the predecessor node array, and obtain the target fault propagation path after reversal.

[0010] Optionally, the multi-task predictive maintenance decision tree includes: an input layer, a shared feature layer, task-specific branches, and an output fusion layer, wherein the input layer is used to receive rail brake equipment operating status data, key component information, and fault propagation paths; the shared feature layer is used to extract common feature representations of the input data and screen key features; the task-specific branches are used to construct multiple branches to perform prediction tasks separately; and the output fusion layer is used to integrate the prediction results of each branch and apply decision rules to generate a final multidimensional prediction result.

[0011] Optionally, the task-specific branches include: a fault component prediction branch, a fault classification branch, a fault time prediction branch, and an impact range assessment branch.

[0012] Optionally, inputting the rail brake equipment operating status data, the key components, and the target fault propagation path into the multi-task predictive maintenance decision tree, and outputting a multi-dimensional prediction result through multi-layer decision-making includes: Faulty component prediction branch: Detect abnormal parameters based on thresholds, calculate the abnormality score of the component, and output a list of key abnormal components. The abnormality score calculation formula is:

[0013] wherein v is a component, n is the number of parameters of the component, is the importance weight of the parameter, is the i-th parameter deviation degree of the component v, is the value of the component v; Failure classification branch: output the failure type with the maximum probability and its occurrence probability using the random forest classifier; Failure time prediction branch: extract parameter degradation features, use an LSTM network to predict the parameter evolution trend, and predict the expected failure occurrence time; Impact range evaluation branch: calculate the impact score based on the failure propagation path, and output the list of affected components and their impact degree; The second aspect embodiment of the present application provides a rail brake equipment failure prediction system, comprising: a data acquisition module, a failure correlation module, an association range calculation module, a multi-task prediction module, and a maintenance suggestion generation module, wherein the data acquisition module is configured to acquire rail brake equipment operating state data; the failure correlation module is configured to construct a dynamic failure correlation graph, obtain the failure transmission relationship between brake equipment components, and dynamically adjust the edge weight of the dynamic failure correlation graph according to the rail brake equipment operating state data; the association range calculation module is configured to calculate the PageRank value of the nodes of the dynamic failure correlation graph to identify key components, and use the Dijkstra algorithm to find a target failure propagation path; the multi-task prediction module is configured to construct a multi-task predictive maintenance decision tree, input the rail brake equipment operating state data, the key components, and the target failure propagation path into the multi-task predictive maintenance decision tree, and output multi-dimensional prediction results through multi-layer decision, wherein the multi-dimensional prediction results include a comprehensive risk index, failure components, failure types and corresponding probabilities, expected occurrence times, and affected ranges; and the maintenance suggestion generation module is configured to generate priority maintenance suggestions according to the multi-dimensional prediction results.

[0014] The third aspect embodiment of the present application provides an electronic device, comprising: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are configured to execute a rail brake equipment failure prediction method as described in the above embodiments.

[0015] The fourth aspect embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, which is executed by a processor to implement a rail brake equipment failure prediction method as described in the above embodiments.

[0016] The beneficial effects achieved by the above-mentioned application are as follows: The present embodiment comprehensively collects equipment operating status data to construct a dynamic fault association graph based on physical connections, energy transfer, and control signal flows. This graph accurately depicts the complex fault transmission relationships between components and dynamically adjusts edge weights based on real-time data, effectively reflecting dynamic changes in equipment operation. It uses PageRank values ​​to quickly locate key components and leverages the Dijkstra algorithm to efficiently identify fault propagation paths, significantly improving the accuracy and efficiency of fault prediction. A multi-task predictive maintenance decision tree deeply integrates equipment operating data, key component information, and fault path characteristics. Through multi-layered decision-making, it outputs multi-dimensional results covering fault components, types, probabilities, expected occurrence times, and affected areas, providing comprehensive, three-dimensional insights into potential faults. Prioritized maintenance recommendations generated based on multi-dimensional predictions comprehensively consider the degree of fault risk, urgency of occurrence, and impact range, scientifically allocate maintenance resources, and accurately plan maintenance tasks. This avoids the waste of resources from scheduled maintenance and eliminates operational interruptions and safety hazards caused by post-maintenance repairs, thereby improving equipment reliability, effectively ensuring the safety and efficiency of rail transit operations, and promoting the intelligent and refined operation and maintenance of rail braking equipment. This solves the problems in the existing technology such as ignoring the dynamic changes of components, severing the relationship between components, and lacking maintenance decision support.

[0017] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a flow chart of a rail brake equipment fault prediction method provided according to an embodiment of the present application; Figure 2 Schematic diagram of the structure of a rail brake equipment fault prediction system provided according to an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0020] A rail brake equipment fault prediction method and system according to an embodiment of the present application will be described below with reference to the accompanying drawings. In view of the problems of neglecting dynamic changes of components, breaking the relationship between components, and lacking maintenance decision support in the prior art mentioned in the background art, the present application provides a rail brake equipment fault prediction method. In the method, a dynamic fault correlation graph based on physical connection, energy transmission and control signal flow is constructed by comprehensively collecting equipment operation state data, the complex fault transmission relationship between components is accurately described, and the edge weight is dynamically adjusted in combination with real-time data to effectively reflect the dynamic changes in equipment operation. The PageRank value is used to quickly locate the key components, and the Dijkstra algorithm is used to efficiently lock the fault propagation path, greatly improving the accuracy and efficiency of fault prediction. The multi-task predictive maintenance decision tree deeply integrates equipment operation data, key component information and fault path features, and outputs multi-dimensional results including fault components, types, probabilities, predicted occurrence times and affected ranges through multi-layer decision-making, realizing all-round and stereoscopic insight into potential faults. The priority maintenance recommendations generated based on multi-dimensional prediction can comprehensively consider the fault risk degree, occurrence urgency and impact range, scientifically allocate maintenance resources, and accurately plan maintenance tasks, avoiding both resource waste of regular maintenance and operation interruption and safety hazards caused by after-maintenance, improving equipment reliability, effectively ensuring the safety and efficiency of rail transit operation, and promoting the intelligent and refined operation and maintenance of rail brake equipment. Thus, the problems of neglecting dynamic changes of components, breaking the relationship between components, and lacking maintenance decision support in the prior art are solved.

[0021] Specifically, Figure 1 A flowchart of a rail brake equipment fault prediction method provided by an embodiment of the present application is shown in FIG. 1.

[0022] As Figure 1 shown, the rail brake equipment fault prediction method includes the following steps: In step S101, rail brake equipment operation state data is acquired.

[0023] The rail brake equipment includes various brake valves, air compressors, air cylinders, brake cylinders, basic brake devices (brake levers, connecting rods, brake shoes), traction motors, brake resistance cabinets, converters, control contactors, brake control units, etc.

[0024] The rail brake equipment operation state data includes pressure data, temperature data, displacement data, and vibration data, which are acquired by corresponding sensors.

[0025] It can be understood that the embodiments of the application construct a three-dimensional monitoring network of the running state of the brake device by acquiring the running state data of the track brake device through the sensor. Specifically, the pressure data can master the brake power transmission condition in real time, timely find abnormal pressure fluctuation, and avoid brake failure caused by insufficient or excessive brake pressure; the temperature data can sensitively capture the temperature change of the component, quickly locate the abnormal heating point, and prevent performance degradation or safety hazards caused by excessive wear and poor heat dissipation of the component; the displacement data accurately presents the working position state of the component, effectively avoiding abnormal brake force output caused by component position deviation; the vibration data can identify potential faults such as component loosening and bearing wear in the early stage by analyzing the subtle changes of frequency and amplitude.

[0026] In step S102, a dynamic fault correlation graph is constructed to obtain the fault transmission relationship between the components of the brake device, wherein the dynamic fault correlation graph is constructed by abstracting the physical components of the track brake device as graph nodes, constructing directed edges based on physical connection relationship, energy transmission path and control signal flow, and setting initial weights according to historical fault statistical data.

[0027] Specifically, the physical components of the track brake device are abstracted as graph nodes, wherein the node attributes include: component identifier, component type, current running state, real-time detection parameter; Based on the physical connection relationship, the energy transmission path and the control signal flow, the directed edges are constructed, wherein the physical connection relationship represents the direct mechanical connection, the energy transmission path represents the energy flow path including hydraulic energy, electrical energy and mechanical energy, and the control signal flow represents the transmission path of the control instruction or feedback signal. The edge attributes of the directed edges include: edge type and weight, wherein the edge type includes three types of physical connection relationship, energy transmission path and control signal flow; According to the historical fault statistical data, an initial weight is given to each edge, and the weight reflects the fault transmission probability; For example, the components such as brake valve, air compressor and brake cylinder are abstracted as nodes, such as the brake valve node records its identifier “BV-01”, component type “control type”, current state “on” and real-time pressure detection parameter, etc. Based on the physical connection relationship, the directed edges are constructed between the brake cylinder and the foundation brake device, representing mechanical force transmission; the edges are constructed between the air compressor and the total air cylinder according to the energy transmission path, embodying the flow of compressed air; the edges are constructed between the brake valve and the brake control unit according to the control signal flow, transmitting the brake instruction. According to the historical statistics, there have been 10 cases of brake cylinder pressure abnormality caused by brake valve failure, and a higher initial weight 0.8 is given to the edges between them, reflecting the fault transmission probability.

[0028] It can be understood that, by constructing a dynamic fault correlation graph, the embodiments of the application can abstract track braking equipment components as nodes and give rich attributes, construct directed edges based on multi-dimensional relationships, set initial weights based on historical data, and systematically and visually present complex fault transmission relationships between equipment components. The node attributes can be used to track the running state of the components in real time and quickly locate abnormalities. Based on the three types of edge relationships of physics, energy and signals, the fault propagation path can be deeply mined, and the limitations of traditional methods of only focusing on single connections can be broken through. By giving initial weights based on historical data, a quantitative basis for fault transmission probability is provided, making the prediction more scientific.

[0029] In step S103, the edge weight of the dynamic fault correlation graph is dynamically adjusted according to the track braking equipment running state data, the PageRank value of the nodes of the dynamic fault correlation graph is calculated to identify key components, and the Dijkstra algorithm is used to find the target fault propagation path.

[0030] Among them, the track braking equipment running state data is threshold detected, the real-time running state data is compared with the normal range of the preset parameter, the current abnormal score is calculated, and the edge weight is dynamically adjusted according to the abnormal score.

[0031] It should be noted that the formula for calculating the abnormal score is:

[0032]

[0033] Wherein, n is the number of running state parameters of the track braking equipment components, is the parameter weight, is the abnormal score of the i-th parameter, and f is a nonlinear mapping function, is the real-time monitoring value of the i-th running state parameter of the component, is the normal mean value of the i-th running state parameter of the component, is the normal standard deviation of the i-th running state parameter of the component.

[0034] The edge weight dynamic adjustment formula is:

[0035] Wherein, is the edge weight from component i to component j at time t, is the comprehensive abnormal score of component i, is an adjustment coefficient, is an edge type correction factor, and different edge types take different values.

[0036] It can be understood that, by means of the dynamic weight adjustment mechanism, the embodiments of the present application enable the fault correlation graph to reflect the device running state in real time, accurately capture the dynamic characteristics of fault propagation by converting sensor data into edge weight changes, quickly locate core components such as brake valves, control units, etc. by analyzing the importance of nodes in the graph through the PageRank algorithm, and make maintenance resources focus on high-risk areas first; the Dijkstra algorithm can efficiently find the most likely fault propagation path, providing data support for subsequent determination of fault impression range.

[0037] In the embodiments of the present application, identifying the key components by calculating the PageRank values of the nodes of the dynamic fault correlation graph comprises: In the dynamic fault correlation graph, the PageRank value represents the influence of the component in the fault propagation network, and the PageRank value of each component is calculated in turn, the mean and standard deviation are calculated according to the PageRank values of the components, if the PageRank value of the component is greater than the sum of the mean and 1.5 times the standard deviation, the component is determined to be a core component, otherwise, if the PageRank value of the component is greater than the mean, the component is determined to be an important component, otherwise, it is a general component. The core components and important components are obtained as the key components.

[0038] For example, a certain dynamic fault correlation graph contains 20 component nodes including brake valves, air compressors, brake cylinders, brake control units, and basic brake devices. By calculating the PageRank values of each node, the mean is 0.05 and the standard deviation is 0.03.

[0039] The PageRank value of the brake control unit is 0.12, which is significantly greater than the mean + 1.5 times the standard deviation (0.05 + 1.5 x 0.03 = 0.095), so it is determined to be a core component. The brake control unit is the control hub of the air brake and electric signal, connecting multiple components such as brake valves and anti-skid devices, so its fault propagation influence is the largest. The PageRank value of the brake valve is 0.07, which is between the mean (0.05) and 0.095, so it is determined to be an important component. The brake valve is a core component of the driver's control brake, and its fault will directly affect the transmission of the brake command. The PageRank value of the basic brake device is 0.04, which is lower than the mean, so it is determined to be a general component. The fault of the basic brake device is usually local wear, which has less impact on the whole system.

[0040] It can be understood that, by means of the PageRank algorithm, the influence score of the component in the dynamic fault correlation graph is calculated, and the statistical screening strategy of the mean and the standard deviation is combined, so that the rail braking equipment components can be systematically divided into core, important and general components. The limitation of the traditional experience-based judgment of the key components is broken, the topological importance of the components in the fault propagation network is objectively reflected by the graph theory model, the PageRank value of the core component is significantly higher than the mean and the standard deviation threshold because the core component is located at the hub position of the fault propagation, and the chain fault is easily caused once the core component fails; the important component plays a key role in the local fault propagation path, and the core component and the important component are taken as the key components, so that the high-risk components can be focused, and more resources can be invested in the link which has a great influence on the system safety.

[0041] In the embodiment of the application, the PageRank value calculation formula is:

[0042] wherein, represents the component node to be calculated, represents the node set pointing to the node , represents the edge weight from the node to the node , and d is the damping coefficient. In the embodiment of the application, the Dijkstra algorithm is used to find the target fault propagation path, including: using the Dijkstra algorithm, taking the key component node as the starting point, and searching for the fault propagation path with the maximum weight sum as the target fault propagation path, and the specific steps are as follows: S1, remove the edges with a weight lower than the target threshold; S2, initialize the distance array, the visited set and the predecessor node array, wherein the distance array is used to save the cumulative weight of the maximum probability path from the starting node S to each node, the dist value of the starting node is set to 1 at the beginning, and the dist values of the remaining nodes are set to 0, the visited set is used to mark the nodes whose maximum probability path has been determined, and the predecessor node array is used to save the predecessor node of each node in the maximum probability path; S4, select the node U with the maximum dist value from all unvisited nodes and add it to the visited set; S5, calculate the cumulative propagation probability from S to V through U for all adjacent nodes V of the node U, and the specific calculation formula is:

[0043] wherein, is the edge weight from the node U to the node V; ​S6. If the cumulative propagation probability is greater than the dist value of the adjacent node V, then update the dist value of the adjacent node V to the cumulative propagation probability, and record the predecessor node of V as U in the predecessor node array; S7, loop through steps S4-S6 until all reachable nodes are processed; S8. Starting from the last node, trace back to the starting point along the predecessor node array, and obtain the target fault propagation path after reversal.

[0044] For example, a dynamic fault association graph contains node A (brake valve), node B (brake cylinder), node C (brake control unit), and node D (basic brake device). The edge weight from node A to node B is 0.8, the edge weight from node A to node C is 0.5, and the edge weight from node B to node D is 0.7.

[0045] Taking node A as the key component, the steps to calculate the target path are as follows: Remove edges with weights below the target threshold of 0.3 and keep the edges 、 、 D.

[0046] Initialize the distance array, set the dist value of node A to 1, and the dist values ​​of the remaining nodes to 0.

[0047] Initialize the visited set and predecessor node array and set them to empty.

[0048] Select node A with the largest dist value, add it to the visited set, and calculate the cumulative propagation probability from node A to adjacent nodes B and C in turn. The cumulative propagation probability from node A to adjacent node B is , the cumulative propagation probability from node A to adjacent node C is , update the dist values ​​of node B and node C respectively and 0.5, record in the predecessor node array that the predecessor node of B is A, and the predecessor node of C is A.

[0049] Among the unvisited nodes, node B has the largest dist value, so node B is added to the visited set.

[0050] Calculate the cumulative propagation probability from node B to adjacent node D as follows: , because 0.56 is greater than 0, the dist value of node D is updated to 0.56, and the predecessor node of D is recorded as B in the predecessor node array.

[0051] Tracing back from node D to node A, we obtain the target path A→B→D, with a cumulative propagation probability of 0.56.

[0052] It can be understood that the embodiments of the present application take the Dijkstra algorithm as the starting point of the key components, search the fault propagation path with the maximum weight sum as the target path, can make full use of the quantitative expression of the edge weight of the dynamic fault correlation graph to the fault conduction probability, and accurately locate the path most likely to occur fault propagation.

[0053] In step S104, a multi-task predictive maintenance decision tree is constructed, and the track braking equipment operation state data and the key components and the target fault propagation path are input into the multi-task predictive maintenance decision tree, and a multi-dimensional prediction result is output through multi-layer decision.

[0054] The multi-dimensional prediction result includes a comprehensive risk index, a fault component, a fault type and a corresponding probability, a predicted occurrence time, and an affected range.

[0055] It can be understood that the embodiments of the present application deeply integrate the real-time operation state data of the equipment, the key component information and the fault propagation path, through the multi-layer decision structure, not only can accurately identify the components and types that may occur faults, but also can quantify the fault occurrence probability, estimate the occurrence time and define the affected range, can master the overall situation of the fault in advance, and systematically analyze from the root cause to the affected range, so as to formulate more targeted maintenance strategy and optimize maintenance resource scheduling.

[0056] In the embodiments of the present application, the multi-task predictive maintenance decision tree includes an input layer, a shared feature layer, a task-specific branch, and an output fusion layer, wherein the input layer is used to receive the track braking equipment operation state data, the key component information and the fault propagation path, the shared feature layer is used to extract the public feature representation of the input data, and the key features are filtered, the task-specific branch is used to construct a plurality of branches to respectively execute the prediction task, and the output fusion layer is used to integrate the prediction results of each branch, and apply a decision rule to generate a final multi-dimensional prediction result.

[0057] The output fusion layer integrates the prediction results of each branch, calculates a comprehensive risk index, sorts according to the risk priority, and outputs a priority maintenance prediction list, and each maintenance prediction list content includes a comprehensive risk index, a fault component, a fault type and a corresponding probability, a predicted occurrence time, and an affected range.

[0058] The urgency is determined by the predicted occurrence time. The urgency is determined by the predicted occurrence time.

[0059] For example, the operating state data such as pressure, temperature, etc., the key component information of the brake cylinder and the air compressor, and the two fault propagation paths of “brake cylinder → brake pipeline → brake shoe” and “air compressor → air cylinder → brake valve” are input into the multi-task predictive maintenance decision tree. After the input layer receives the data, the shared feature layer extracts common features such as pressure fluctuation and temperature change trend, and filters out key features such as pressure abnormal amplitude and temperature rising rate. In the task-specific branch, the fault component prediction branch determines that the abnormal score of the brake cylinder is 15 and the abnormal score of the air compressor is 8, which are added to the key abnormal component list and output; the fault classification branch calculates that the “leakage” probability of the brake cylinder is 80% and the “wear beyond limit” probability of the air compressor is 75%; the fault time prediction branch shows that the brake cylinder has a remaining predicted occurrence time of 36 hours and the air compressor has a predicted occurrence time of 60 hours; the influence range evaluation branch concludes that the brake cylinder fault will moderately affect the brake pipeline and slightly affect the brake shoe, and the air compressor fault will slightly affect the air cylinder and the brake valve. The output fusion layer integrates the results, and the comprehensive risk index formula is calculated. The comprehensive risk index of the brake cylinder is 0.384, and the comprehensive risk index of the air compressor is 0.135. The maintenance prediction list is output according to the risk priority. The brake cylinder is ranked first due to higher risk, and the maintenance prediction list includes the comprehensive risk index 0.384, the fault component brake cylinder, the fault type leakage probability 80%, the predicted occurrence time 36 hours, and the affected range brake pipeline (moderate) and brake shoe (slight); the air compressor is ranked second, and the corresponding prediction results are displayed.

[0060] It can be understood that, in the embodiments of the present application, the input layer realizes centralized reception of multi-source data, providing comprehensive information for subsequent analysis; the shared feature layer extracts common features and filters key elements, effectively reducing data redundancy and enhancing the model's ability to capture core information; the task-specific branch builds independent and parallel analysis paths for different prediction targets such as fault components, fault types, probabilities, times, and influence ranges, achieving specialized prediction; the output fusion layer integrates the results of each branch, and uses decision rules to weigh comprehensively, outputting complete and accurate multi-dimensional predictions. The division of labor among the layers not only avoids repeated calculations to reduce computational complexity, but also improves the timeliness of prediction through multi-task parallelism, while the fusion mechanism ensures the accuracy and reliability of the results, providing solid technical support for the intelligent operation and maintenance of rail brake equipment.

[0061] In the embodiments of the present application, the task-specific branch includes: a fault component prediction branch, a fault classification branch, a fault time prediction branch, and an influence range evaluation branch.

[0062] It can be understood that, by dividing the task-specific branch into multiple branches, each branch has a clear division of labor and handles different dimensions of prediction tasks, which can deal with the complexity of rail brake equipment fault prediction in a refined manner, significantly improving the accuracy of prediction and the effectiveness of decision-making.

[0063] In the embodiments of the present application, the track braking device operating state data and the key components and target failure propagation path are input into the multi-task predictive maintenance decision tree, and a multi-dimensional prediction result is output through multiple layers of decision-making, including: Fault component prediction branch: based on threshold detection of abnormal parameters, calculate the abnormal score of the component, output the list of key abnormal components, wherein the abnormal score calculation formula is:

[0064] Wherein v is the component, n is the number of parameters of the component, is the importance weight of the parameter, is the i-th parameter deviation degree of component v, is the value of component v. Fault classification branch: use the random forest classifier to output the fault type with the maximum probability and its occurrence probability; Fault time prediction branch: extract parameter degradation features, use LSTM network to predict parameter evolution trend, and predict fault expected occurrence time; Impact range evaluation branch: based on the fault propagation path, calculate the impact score, and output the list of affected components and their impact degree; Specifically, in the random forest classifier, the average value of the prediction probability of each fault type by all decision trees is calculated, and the fault type with the maximum probability and its probability are output, wherein the fault types include wear limit, deformation limit, loosening, leakage, aging, failure, and the prediction probability formula of the random forest classifier is:

[0065] Wherein T represents the number of decision trees, represents the prediction probability of the t-th tree for the fault type , and is the fault type set after integration.

[0066] Extracting parameter degradation features means calculating the degradation rate of the operating parameter, and the degradation rate calculation formula is:

[0067] Wherein, is the operating parameter value at the current time, the reference value is the parameter value in the previous time window, is the time of the time window.

[0068] Based on the fault propagation path, the impact score is calculated, specifically, the path weight of the fault propagation path node, the corresponding PageRank value and the cumulative propagation probability of the fault propagation path are multiplied to obtain.

[0069] It can be understood that the embodiment of the present application integrates data and algorithms of different dimensions through a multi-task decision tree, uses real-time monitoring data to capture equipment anomalies, and improves prediction accuracy by combining historical data patterns. It can detect potential faults and the scope of fault impact in advance, and can optimize resource allocation and improve equipment reliability and operational efficiency based on maintenance sorting based on a comprehensive risk index.

[0070] In step S105 , priority maintenance suggestions are generated based on the multi-dimensional prediction results.

[0071] Specifically, the multi-dimensional prediction result is a priority maintenance prediction list, where each maintenance prediction list includes a comprehensive risk index, faulty components, fault type and corresponding probability, expected occurrence time, and affected scope.

[0072] Maintenance levels are divided according to the comprehensive risk index. If the comprehensive risk index is greater than 0.5, it is emergency maintenance and requires immediate shutdown and maintenance; if the comprehensive risk index is between 0.35-0.5, it is high-priority maintenance and needs to be arranged within 12 hours; if the comprehensive risk index is between 0.2-0.35, it is medium-priority maintenance and needs to be arranged within 24 hours; if the comprehensive risk index is less than 0.2, it is low-priority maintenance and needs to be arranged within 72 hours.

[0073] According to the set template, the comprehensive risk index, fault component, fault type and corresponding probability, expected occurrence time, and affected scope are filled into the template to generate a maintenance reference report to provide maintenance personnel with a maintenance basis.

[0074] It can be understood that the embodiment of the present application divides maintenance levels according to a quantitative comprehensive risk index, which can clearly define the priority of handling different faults; the standardized report template integrates complex fault information into structured content, and maintenance personnel can quickly obtain key information without repeated sorting, and can accurately prepare spare parts and deploy professional personnel in advance, greatly shortening maintenance preparation time.

[0075] A method for predicting faults of rail brake equipment will be described in detail below through a specific embodiment, as follows: The subway operation and maintenance center collects operating status data of the train's brake equipment components when braking occurs on a certain line of trains. The operating status data includes brake cylinder pressure, air cylinder pressure, hydraulic line pressure, brake disc temperature, motor temperature, hydraulic oil temperature, brake cylinder piston displacement, brake shoe clearance, air compressor vibration, and brake caliper vibration data.

[0076] The physical components of the train braking equipment are abstracted as graph nodes, including brake valves, air compressors, air cylinders, brake cylinders, foundation brake devices (brake levers, connecting rods, brake shoes), traction motors, brake resistance cabinets, converters, control contactors, brake control units, etc. Based on physical connection relationship, energy transmission path, control signal flow, directed edges are constructed, such as air compressor supplying air to air cylinder, establishing “air compressor→air cylinder” edge; brake control unit controls brake valve, establishing “brake control unit→brake valve” edge. According to historical fault statistics, the initial weight of the edge is set, for example, “brake valve→brake cylinder” has many historical associated faults, and the initial weight is set to 0.6.

[0077] Based on real-time data, the edge weight of the fault association graph is dynamically adjusted. Real-time data shows that the brake cylinder pressure abnormally drops, the weight of “brake valve→brake cylinder” is increased to 0.7, and the weight of “brake cylinder→brake pipeline” is increased from 0.5 to 0.8. The PageRank value of the node is calculated to determine that the brake valve and the brake cylinder are key components. Using Dijkstra algorithm, the target fault propagation path “brake valve→brake cylinder→brake pipeline” is obtained.

[0078] The running data, key components and target fault propagation path are input into the multi-task predictive maintenance decision tree. The fault component prediction branch calculates that the brake cylinder abnormal score exceeds the threshold value, and is listed as a key abnormal component; the fault classification branch determines that the brake cylinder “leakage” fault probability reaches 80% through random forest; the fault time prediction branch predicts that the fault is expected to occur in about 18 hours using LSTM network; the influence range evaluation branch determines that the fault will have a serious impact on the brake pipeline and a moderate impact on the brake shoe.

[0079] The output fusion layer calculates the comprehensive risk index of the brake cylinder as 0.52. Since there is only one fault component, the priority maintenance prediction list only has the multi-dimensional prediction result of the brake cylinder component, which includes the comprehensive risk index 0.52, the fault component is the brake cylinder, the fault type is the leakage probability of 80%, the expected occurrence time is 18 hours, and the affected range is the brake pipeline (serious) and the brake shoe (moderate).

[0080] According to the comprehensive risk index 0.52, it is determined that it is an emergency repair level. According to the multi-dimensional prediction result, a maintenance reference report is generated, and the maintenance reference report is sent to the maintenance personnel and the maintenance personnel is notified to stop operation and maintain immediately.

[0081] In summary, the embodiment of the present application can accurately depict the complex fault transmission relationship between components by collecting equipment running state data, constructing a dynamic fault correlation graph based on physical connection, energy transmission and control signal flow, and dynamically adjusting the edge weight combined with real-time data, which can effectively reflect the dynamic changes in equipment operation; the key components are quickly located by using the PageRank value, the fault propagation path is efficiently locked by using the Dijkstra algorithm, and the multi-dimensional prediction results are deeply fused according to the multi-task predictive maintenance decision tree to realize the all-round and stereoscopic insight into the potential fault; the priority maintenance suggestion generated based on the multi-dimensional prediction can comprehensively consider the fault risk degree, the occurrence urgency and the influence range, scientifically allocate maintenance resources, and accurately plan maintenance tasks.

[0082] Next, an orbit braking equipment fault prediction system according to an embodiment of the present application is described with reference to the accompanying drawings.

[0083] Figure 2 is a structural schematic diagram of an orbit braking equipment fault prediction system according to an embodiment of the present application.

[0084] As shown in Figure 2 , the orbit braking equipment fault prediction system 10 includes a data acquisition module 100, a fault correlation module 200, an association range calculation module 300, a multi-task prediction module 400, and a maintenance suggestion generation module 500.

[0085] The data acquisition module 100 is configured to acquire orbit braking equipment running state data; the fault correlation module 200 is configured to construct a dynamic fault correlation graph to obtain the fault transmission relationship between the components of the braking equipment, and dynamically adjust the edge weight of the dynamic fault correlation graph according to the orbit braking equipment running state data; the association range calculation module 300 is configured to calculate the PageRank value of the nodes of the dynamic fault correlation graph to identify key components, and find the target fault propagation path by using the Dijkstra algorithm; the multi-task prediction module 400 is configured to construct a multi-task predictive maintenance decision tree, input the orbit braking equipment running state data, the key components and the target fault propagation path into the multi-task predictive maintenance decision tree, and output multi-dimensional prediction results through multi-layer decision, wherein the multi-dimensional prediction results include a comprehensive risk index, fault components, fault types and corresponding probabilities, predicted occurrence time, and affected range; and the maintenance suggestion generation module 500 is configured to generate priority maintenance suggestions according to the multi-dimensional prediction results.

[0086] According to the track brake equipment fault prediction system provided by the embodiment of the application, the dynamic fault correlation graph based on physical connection, energy transmission and control signal flow is constructed by comprehensively collecting equipment operation state data, the complex fault transmission relationship between components is accurately described, the edge weight is dynamically adjusted in combination with real-time data, and the dynamic changes in the equipment operation are effectively reflected; the key components are quickly located by using the PageRank value, the fault propagation path is efficiently locked by using the Dijkstra algorithm, and the accuracy and efficiency of fault prediction are greatly improved; the multi-task predictive maintenance decision tree deeply integrates equipment operation data, key component information and fault path features, outputs multi-dimensional results covering fault components, types, probabilities, predicted occurrence time and affected range through multi-layer decision, realizes all-around and stereoscopic insight into potential faults, and generates priority maintenance suggestions based on multi-dimensional prediction. The fault risk degree, occurrence urgency and influence range are comprehensively considered, maintenance resources are scientifically allocated, maintenance tasks are accurately planned, resource waste of regular maintenance is avoided, operation interruption and safety hazards caused by after-maintenance are eliminated, equipment reliability is improved, the safety and efficiency of rail transit operation are effectively ensured, and the rail brake equipment operation and maintenance are promoted to the intelligent and refined direction. Thus, the problems of ignoring component dynamic changes, separating component relationships and lacking maintenance decision support in the prior art are solved.

[0087] Figure 3 The structure schematic diagram of the electronic device provided by the embodiment of the application is provided. The electronic device can include: The memory 301, the processor 302, and the computer program stored in the memory 301 and executable on the processor 302.

[0088] The processor 302 implements the track brake equipment fault prediction method provided in the above embodiment when executing the program.

[0089] Further, the electronic device further includes: The communication interface 303 is used for communication between the memory 301 and the processor 302.

[0090] The memory 301 is used to store the computer program executable on the processor 302.

[0091] The memory 301 can include a high-speed RAM memory, and can also include a non-volatile memory such as at least one disk memory.

[0092] If the memory 301, the processor 302 and the communication interface 303 are implemented independently, the communication interface 303, the memory 301 and the processor 302 can be connected with each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 3 Only one thick line is used to represent the bus in the figure, but it does not mean that there is only one bus or only one type of bus.

[0093] Optionally, in a specific implementation, if the memory 301, the processor 302 and the communication interface 303 are integrated on a chip, the memory 301, the processor 302 and the communication interface 303 can complete communication between each other through an internal interface.

[0094] The processor 302 can be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application.

[0095] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the track brake device fault prediction method.

[0096] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples without contradiction.

[0097] Furthermore, the terms "first", "second", etc. are used herein only to describe different steps or features and do not imply a relative importance or a specific order of steps or features. Thus, features defined with "first", "second" etc. can include at least one of the features implicitly or explicitly. In the description of the application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise expressly specified.

[0098] Any process or method descriptions or blocks in flow charts described herein and elsewhere can be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps in the process. Alternate implementations are included within the scope of the preferred embodiments of the present application in which the functions performed by the various processes described herein and elsewhere are allocated differently among the components of the preferred embodiments, such as according to the functions performed by the various components, in a substantially simultaneous manner, or according to a different order.

[0099] It should be understood that aspects of the present application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the various steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As such, if implemented in hardware and in another embodiment, the hardware can include any or a combination of the following technologies, which are all well known in the art: a discrete logic circuit having logic gates for implementing logic functions upon an application of data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0100] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. Further, those of skill in the art would understand that the preferred embodiments of the present application can be implemented by a variety of means, including as a process, an apparatus, a machine, or a combination thereof. In another embodiment, the information, instructions, or commands can be downloaded over the Internet or over another network by way of data signals embodied in carrier waves.

Claims

1. A method for predicting rail brake equipment failure, characterized in that: The following steps are involved: Obtain operating status data of rail brake equipment; Construct a dynamic fault association graph to obtain the fault transmission relationship between the brake equipment components. Constructing the dynamic fault association graph involves abstracting the physical components of the rail brake equipment into graph nodes, constructing directed edges based on physical connection relationships, energy transfer paths, and control signal flows, and setting initial weights based on historical fault statistics. Dynamically adjust the edge weights of the dynamic fault association graph according to the operating status data of the rail brake equipment, calculate the PageRank values ​​of the nodes of the dynamic fault association graph to identify key components, and use the Dijkstra algorithm to find the target fault propagation path; Constructing a multi-task predictive maintenance decision tree, inputting the rail brake equipment operating status data, the key components, and the target fault propagation path into the multi-task predictive maintenance decision tree, and outputting a multi-dimensional prediction result through multi-layer decision-making, wherein the multi-dimensional prediction result includes a comprehensive risk index, fault component, fault type and corresponding probability, expected occurrence time, and affected range; Generate priority maintenance recommendations based on the multi-dimensional prediction results.

2. A rail brake equipment fault prediction method according to claim 1, characterized in that: The step of identifying key components by calculating the PageRank values ​​of the nodes in the dynamic fault correlation graph includes: In the dynamic fault association graph, the PageRank value represents the influence of the component in the fault propagation network. The PageRank value of each component is calculated in turn. The mean and standard deviation are calculated based on the PageRank value of each component. If the PageRank value of the component is greater than the sum of the mean and 1.5 times the standard deviation, the component is determined to be a core component. Otherwise, if the PageRank value of the component is greater than the mean, the component is determined to be an important component. Otherwise, it is a general component. Get core components and important components as key components.

3. A rail brake equipment fault prediction method according to claim 2, characterized in that: The PageRank value calculation formula is: in, Indicates the component node to be calculated, Indicates pointing to a node The node set of Represents a slave node To Node is the edge weight, and d is the damping coefficient.

4. A rail brake equipment fault prediction method according to claim 1, characterized in that: The method of using the Dijkstra algorithm to find the target fault propagation path includes: using the Dijkstra algorithm, starting from the key component node, searching for the fault propagation path with the largest sum of weights as the target fault propagation path, wherein the specific steps are as follows: S1, remove edges whose weight is lower than the target threshold; S2. Initialize the distance array, the visited set, and the predecessor node array, wherein the distance array is used to store the cumulative weight of the maximum probability path from the starting node S to each node. Initially, the dist value of the starting node is set to 1, and the dist values ​​of the remaining nodes are set to 0. The visited set is used to mark the nodes with the determined maximum probability path, and the predecessor node array is used to store the predecessor node of each node on the maximum probability path; S4. Select the node U with the largest dist value from all unvisited nodes and add it to the visited set; S5. For all adjacent nodes V of node U, calculate the cumulative propagation probability from S to V via U. The specific calculation formula is: in, is the edge weight from node U to V; S6. If the cumulative propagation probability is greater than the dist value of the adjacent node V, the dist value of the adjacent node V is updated to the cumulative propagation probability, and the predecessor node of V is recorded as U in the predecessor node array; S7, loop through steps S4-S6 until all reachable nodes are processed; S8. Starting from the last node, trace back to the starting point along the predecessor node array, and obtain the target fault propagation path after reversal.

5. A rail brake equipment fault prediction method according to claim 1, characterized in that: The multi-task predictive maintenance decision tree includes: an input layer, a shared feature layer, task-specific branches, and an output fusion layer. The input layer is used to receive rail brake equipment operating status data, key component information, and fault propagation paths. The shared feature layer is used to extract common feature representations of the input data and screen key features. The task-specific branches are used to construct multiple branches to perform prediction tasks separately. The output fusion layer is used to integrate the prediction results of each branch and apply decision rules to generate the final multi-dimensional prediction results.

6. A rail brake equipment fault prediction method according to claim 5, characterized in that: The task-specific branches include: a fault component prediction branch, a fault classification branch, a fault time prediction branch, and an impact range assessment branch.

7. A rail brake equipment fault prediction method according to claim 1 or 6, characterized in that: Inputting the rail brake equipment operating status data, the key components, and the target fault propagation path into the multi-task predictive maintenance decision tree, and outputting a multi-dimensional prediction result through multi-layer decision-making includes: Faulty component prediction branch: Detect abnormal parameters based on thresholds, calculate the abnormality score of the component, and output a list of key abnormal components. The abnormality score calculation formula is: Where v is the component, n is the number of parameters of the component, is the importance weight of the parameter, is the deviation of the i-th parameter of component v, Is the component v value; Fault classification branch: Use the random forest classifier to output the fault type with the highest probability and its probability of occurrence; Failure time prediction branch: Extracts parameter degradation features, uses LSTM networks to predict parameter evolution trends, and predicts the expected time of failure. Impact scope assessment branch: Calculates the impact score based on the fault propagation path and outputs a list of affected components and their impact levels.

8. A rail brake equipment fault prediction system according to claim 1, characterized in that: include: Data acquisition module, fault association module, association range calculation module, multi-task prediction module, maintenance suggestion generation module, among which, The data acquisition module is used to obtain the operating status data of the track brake equipment; The fault association module is used to construct a dynamic fault association graph, obtain the fault transmission relationship between the brake equipment components, and dynamically adjust the edge weights of the dynamic fault association graph according to the operating status data of the rail brake equipment; The association range calculation module is used to calculate the PageRank value of the nodes in the dynamic fault association graph to identify key components and use the Dijkstra algorithm to find the target fault propagation path; The multi-task prediction module is used to construct a multi-task predictive maintenance decision tree, input the rail brake equipment operating status data, the key components, and the target fault propagation path into the multi-task predictive maintenance decision tree, and output a multi-dimensional prediction result through multi-layer decision making, wherein the multi-dimensional prediction result includes a comprehensive risk index, fault component, fault type and corresponding probability, expected occurrence time, and affected range; The maintenance suggestion generating module is used to generate priority maintenance suggestions according to the multi-dimensional prediction results.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a rail brake equipment fault prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement a rail brake equipment fault prediction method according to any one of claims 1 to 7.

Images