Train brake performance evaluation method and system based on artificial intelligence

Abstract

The invention provides a train brake performance evaluation method and system based on artificial intelligence, and the method comprises the steps: firstly obtaining a train brake full-cycle operation associated data set which covers operation parameters, environmental influences, historical performance records and other data; then, extracting a dynamic association relationship among brake components, and constructing an operation association chain which takes a time axis as a clue and of which the association strength evolves along with the operation stage; brake performance dynamic reference features including component association and cross-stage co-evolution features are generated based on the association chain evolution trajectory; calling a pre-trained artificial intelligence model, carrying out cross-dimension interactive mapping on the real-time operation features and the dynamic reference features, and generating a performance evolution evaluation result; and finally, generating a performance dynamic adjustment scheme including staged parameter adjustment and cross-stage collaborative optimization suggestions according to an evaluation result and an association chain trend. According to the method, accurate evaluation and dynamic optimization of the train brake performance are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a train brake performance evaluation method and system based on artificial intelligence. BACKGROUND

[0002] In the field of railway transportation, the train brake is a key component to ensure the safe operation of the train, and its stability and reliability are crucial. The traditional train brake performance evaluation method mainly relies on periodic manual detection and static data analysis. Manual detection requires professional technicians to check each component of the brake one by one, which is not only inefficient, but also difficult to cover the running state of the brake throughout the cycle, and is prone to miss some potential hidden troubles.

[0003] In terms of static data analysis, the existing technology usually only analyzes the running parameters of the brake at a specific time or under specific operating conditions, lacking comprehensive consideration of the full-cycle running data of the brake. At the same time, the existing method does not fully consider the dynamic correlation between the components of the brake and the characteristics of the above correlation changing with the running stage. For example, under different climate conditions, running speeds and line conditions, the performance of each component of the brake and their mutual influence will change, but the traditional method cannot accurately capture these changes, resulting in inaccurate and incomplete evaluation of the performance of the brake, making it difficult to predict the evolution trend of the brake performance in advance, and unable to take effective maintenance and optimization measures in time, thereby increasing the safety risk and maintenance cost of train operation. SUMMARY

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide a train brake performance evaluation method based on artificial intelligence, which comprises: obtaining a train brake full-cycle running correlation data set, the train brake full-cycle running correlation data set containing running parameter data, environmental influence data and historical performance evolution record data generated by each component of the brake at different running stages; extracting the dynamic correlation between the components of the brake from the train brake full-cycle running correlation data set, and constructing a brake running correlation chain, the brake running correlation chain connecting the correlation between each component with time axis as the clue, and presenting the evolution characteristics of the correlation strength with the change of the running stage; generating a brake performance dynamic benchmark feature based on the evolution track of the brake running correlation chain, the brake performance dynamic benchmark feature containing component correlation features at each running stage and cross-stage collaborative evolution features; The pre-trained brake performance dynamic evaluation artificial intelligence model is called to perform cross-dimension interaction mapping processing on the real-time collected brake operation characteristics and the brake performance dynamic benchmark characteristics, and a brake performance evolution evaluation result is generated. According to the brake performance evolution evaluation result, a brake performance dynamic adjustment scheme is generated in combination with an associated chain evolution trend, and the brake performance dynamic adjustment scheme includes phased component parameter adjustment directions and cross-stage collaborative optimization suggestions.

[0005] In another aspect, the embodiment of the present application also provides a train brake performance evaluation system based on artificial intelligence, which comprises a processor and a machine readable storage medium, the machine readable storage medium is connected with the processor, the machine readable storage medium is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the machine readable storage medium to realize the above-mentioned method.

[0006] Based on the above aspects, by obtaining a full-cycle operation associated data set of a train brake, a dynamic association relationship between brake components is extracted from the full-cycle operation associated data set and a brake operation associated chain is constructed, and the evolution characteristics of the association strength of the association relationship between components with the running stage are clearly presented with the time axis as a clue, so that the complex operation mechanism inside the brake can be deeply understood. The brake performance dynamic benchmark characteristics are generated based on the evolution track of the operation associated chain, which includes the component association characteristics of each running stage and the cross-stage collaborative evolution characteristics, the pre-trained artificial intelligence model is called to perform cross-dimension interaction mapping processing on the real-time collected brake operation characteristics and the dynamic benchmark characteristics, and an accurate brake performance evolution evaluation result is generated, so that the change trend of the brake performance can be found in time. Finally, the performance dynamic adjustment scheme including the phased component parameter adjustment directions and the cross-stage collaborative optimization suggestions is generated according to the evaluation result in combination with the associated chain evolution trend, the precise evaluation and dynamic optimization of the train brake performance are realized, the safety and reliability of the train operation are effectively improved, and the maintenance cost is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0007] Figure 1 is the execution flow diagram of the train brake performance evaluation method based on artificial intelligence provided by the embodiment of the present application.

[0008] Figure 2 is the schematic diagram of exemplary hardware and software components of the train brake performance evaluation system based on artificial intelligence provided by the embodiment of the present application. DETAILED DESCRIPTION

[0009] The present application will be specifically described below in combination with the drawings of the specification, Figure 1is a flowchart of a train brake performance evaluation method based on artificial intelligence provided by an embodiment of the present application. The train brake performance evaluation method based on artificial intelligence will be described in detail below.

[0010] Step S110: Obtain a train brake full-cycle operation associated data set, which contains operation parameter data, environmental influence data and historical performance evolution record data generated by each component of the brake at different operation stages.

[0011] In this embodiment, the air brake of a passenger train is taken as an object to obtain the full-cycle operation associated data set. The brake components include air compressors, air reservoirs, brake valves, brake cylinders, release valves, brake hoses, etc. Different operation stages include the starting stage, the stable operation stage, the braking stage and the stopping stage, and the time span of each stage is determined according to the train operation plan. The operation parameter data involves indicators such as the pressure, flow, temperature, action frequency and response time of each component; the environmental influence data includes external temperature, humidity, atmospheric pressure and vibration intensity during train operation; and the historical performance evolution record data contains performance state rating, maintenance time, component replacement type and performance decay trend of each component in the past multiple operation cycles.

[0012] For example, step S111: deploy full-cycle data acquisition devices covering each component of the brake, which include a parameter acquisition module, a stage identification module and a timestamp marking module. The parameter acquisition module acquires the operation parameter data of each component in real time, the stage identification module identifies the operation stage according to the brake working condition change, and the timestamp marking module adds a timestamp to the collected data.

[0013] A pressure sensor and a flow sensor are installed at an air compressor output end of the brake machine, a temperature sensor and a pressure sensor are installed on an outer wall of a wind storage cylinder, a displacement sensor and an action frequency counter are installed on the brake valve and the relief valve, a pressure sensor and a response time timer are installed in the brake cylinder, and a vibration sensor is installed at a brake hose connection. The above sensors collectively constitute a parameter acquisition module, which acquires running parameter data of corresponding components in real time. The stage identification module identifies the running stage by receiving signals such as a traction instruction, a brake instruction, and a parking instruction of the train, and combining parameter changes of components of the brake machine. For example, when the traction instruction is received and the air compressor pressure starts to rise and the brake valve is in a closed state, it is determined that the starting stage is reached; when the train speed is stable within a set range and the component parameters do not fluctuate sharply, it is determined that the stable running stage is reached; when the brake instruction is received, the brake valve is opened, and the brake cylinder pressure rises, it is determined that the braking stage is reached; and when the train speed is reduced to zero and all brake components are in a stationary state, it is determined that the stopping stage is reached. The timestamp marking module uses a clock signal synchronized with the train dispatching system to add a timestamp accurate to milliseconds to each acquired running parameter data, so as to ensure the time correlation of the data.

[0014] Step S112: connecting a train environment monitoring system, obtaining environment influence data in a brake machine running process from the train environment monitoring system, the environment influence data including temperature data, humidity data, vibration data, and air pressure data of each running stage, and adding corresponding stage identification and a timestamp to the environment influence data.

[0015] The full-cycle data acquisition device is connected with the train environment monitoring system through an Ethernet interface, and the environment influence data is obtained from the train environment monitoring system according to a preset communication protocol (such as a TCP / IP protocol). The train environment monitoring system is provided with environment monitoring sensors at the train head, tail, and middle part, and can collect environment data at different positions. For each stage of the brake machine running, temperature data, humidity data, vibration data, and air pressure data in the corresponding time period are filtered from the environment monitoring system. For example, in the starting stage, the environment data from the starting time to the ending time of the stage is obtained, and the environment data is added with the identification of the “starting stage” and the timestamp consistent with the format of the running parameter data, so as to ensure the time correspondence between the environment influence data and the running parameter data.

[0016] Step S113: accessing a train brake machine historical performance management database, and calling historical performance evolution record data of components of the brake machine from the train brake machine historical performance management database, the historical performance evolution record data including performance state records, maintenance records, and performance change trend descriptions of each component in each running stage in different historical cycles.

[0017] The train brake history performance management database is connected through a database access interface (such as an ODBC interface), the train brake history performance management database adopts a relational database structure, and historical data is classified and stored according to dimensions such as component type, running cycle, running stage, etc. When data is called, according to the component model and number of the current brake, the performance state records (such as the evaluation results of good, good, medium, and poor) of each component in the starting stage, stable running stage, braking stage, and stopping stage in the past multiple running cycles, maintenance records (such as maintenance date, maintenance content, and maintenance personnel), and performance change trend description (such as pressure decay rate and response time extension trend) are filtered out. For example, the pressure output stability rating of the air compressor in each stage in the past 12 running cycles, the monthly maintenance record, and the textual description of the pressure decay trend are called.

[0018] Step S114: Perform stage alignment processing on the collected running parameter data, environmental influence data, and historical performance evolution record data, match the data from different sources to the corresponding running stage according to the stage identifier and time stamp, and form a stage data combination.

[0019] The running stage identifier and time stamp in the running parameter data and environmental influence data are extracted, and the data is matched in units of running stages. For example, for the starting stage, all air compressor pressure data, air cylinder temperature data, and other running parameter data with time stamps in the starting time to the ending time of the stage are filtered out, as well as environmental temperature and humidity data in the corresponding time period, and then the historical performance evolution record data of the air compressor, air cylinder, and other components in the stage are associated to form a data combination of the starting stage. Similarly, data combinations of the stable running stage, braking stage, and stopping stage are formed. In the matching process, if the time stamp of a certain data is at the junction time of two stages, it is attributed according to the working condition change time point determined by the stage identification module, and it is ensured that each data is accurately matched to the corresponding running stage.

[0020] Step S115: The stage data combination is summarized to form a train brake full-cycle running associated data set containing running parameter data, environmental influence data, and historical performance evolution record data of each component of the train brake in different running stages.

[0021] The data of the starting stage, stable running stage, braking stage and stopping stage are combined in time sequence to summarize a structured data set. The data set is stored in JSON format, each running stage is a first-level node, each component is a second-level node, and running parameter data, environmental impact data and historical performance evolution record data are third-level nodes. For example, the starting stage node includes air compressors and air storage cylinders as second-level nodes, and the air compressor node includes pressure, flow and other running parameter data nodes, corresponding environmental impact data nodes and historical performance evolution record data nodes. After summarizing, the data set is checked for integrity. If some data of a component in a stage is missing, the missing position and type are marked for subsequent processing.

[0022] Step S120: Extracting dynamic correlation relationship between brake components from the train brake full-cycle running correlation data set to construct a brake running correlation chain. The brake running correlation chain serializes the correlation relationship of each component with time axis as the clue, and presents the correlation strength evolution characteristics with the change of running stage.

[0023] Based on the train brake full-cycle running correlation data set constructed above, the mutual influence relationship between the running parameter data of each component is analyzed to extract the dynamic correlation relationship. For example, the change of air compressor pressure data will affect the pressure data of air storage cylinder, and the pressure data of air storage cylinder will affect the action state of brake valve, and the action state of brake valve will further affect the pressure data of brake cylinder. These mutual influence relationships are dynamic correlation relationships. When constructing the brake running correlation chain, the correlation relationship of components in different running stages is arranged in time sequence with time axis as the horizontal main line, and the strength change of correlation relationship in different stages is also reflected. For example, in the braking stage, the correlation strength between brake valve and brake cylinder is higher than that in the stable running stage.

[0024] Step S121: Dividing the train brake full-cycle running correlation data set into running stage, and dividing the data into stage data units corresponding to each stage according to the starting stage, stable running stage, braking stage and stopping stage of the brake. Each stage data unit contains the running parameter data, environmental impact data and performance record data of each component in the stage.

[0025] According to the start time and end time of each operation phase determined by the phase identification module in step S111, the train brake full-cycle operation associated data set is divided. All data within the time range of the starting phase are extracted to form a starting phase data unit. Similarly, the data of the stable operation phase, braking phase and stopping phase are extracted to form corresponding phase data units. In each phase data unit, the operation parameter data is classified and arranged according to the components, and the environmental influence data and the performance record data of each component in the phase (such as performance state rating, whether abnormal, etc.) are also included. For example, in the starting phase data unit, the pressure change curve data and flow data of the air compressor in the starting phase, the temperature data and pressure data of the air cylinder, and the environmental temperature and humidity data in the phase are included, as well as the performance record data of the air compressor and the air cylinder in the phase.

[0026] Step S122: For each phase data unit, the parameter association relationship between any two components in the corresponding phase is extracted, the occurrence frequency, duration and association performance of the association relationship are recorded, and a phase-in associated relationship set is formed.

[0027] Taking the starting phase data unit as an example, the air compressor and the air cylinder are selected, and the relationship between the pressure parameter of the air compressor and the pressure parameter of the air cylinder is analyzed. When the air compressor pressure rises, the air cylinder pressure rises, that is, it is determined that there is a parameter association relationship between the two. The number of occurrences of the above association relationship in the starting phase is counted, which is the occurrence frequency; the time interval from the beginning to the end of the association relationship is recorded, which is the duration; and the association performance is described as the corresponding relationship between the air compressor pressure change rate and the air cylinder pressure change rate. In this way, the parameter association relationship between all pairs of components in the starting phase is analyzed to form the starting phase associated relationship set. Using the same method, the associated relationship sets in the stable operation phase, the braking phase and the stopping phase are formed.

[0028] Step S123: Compare the phase-in associated relationship sets of adjacent phase data units to identify the addition, disappearance and intensity change of the associated relationship, form phase-to-phase associated evolution information, and the phase-to-phase associated evolution information includes the associated relationship change type and change trigger condition.

[0029] The stage-in correlation set of the starting stage and the stable running stage is compared. If a parameter correlation between the brake valve and the release valve that does not exist in the starting stage appears in the stable running stage, it is determined to be a new correlation; if the high-frequency correlation between the air compressor and the air storage cylinder that exists in the starting stage has a significant decrease in frequency in the stable running stage, it is determined to be a change in intensity; if a certain correlation exists in the starting stage but does not exist in the stable running stage at all, it is determined to be a disappearing correlation. The change triggering condition is determined according to the working condition change and the parameter change of the adjacent stage, for example, the triggering condition of the newly added correlation between the brake valve and the release valve is that the action frequency of the brake valve is stable within a certain range after the train enters the stable running stage. In the above manner, the stage-in correlation sets of the stable running stage and the braking stage, and the braking stage and the stopping stage are compared in turn to identify various changes and form complete inter-stage correlation evolution information.

[0030] Step S124: Arrange the stage-in correlation sets in chronological order with the time axis as the horizontal clue, and connect the correlation relationships of adjacent stages with the inter-stage correlation evolution information as the vertical connection to form an initial brake machine running correlation chain.

[0031] The stage-in correlation sets of the starting stage, the stable running stage, the braking stage, and the stopping stage are arranged in sequence below the time axis from left to right representing the chronological order. According to the inter-stage correlation evolution information, the correlation relationships between adjacent stages that have a succession relationship are connected by lines. For newly added correlation relationships, a line is drawn from the starting position of the corresponding stage to indicate their appearance. For disappearing correlation relationships, the line is terminated at the end position of the corresponding stage. For correlation relationships with a change in intensity, the change in intensity is represented by the change in thickness of the line. For example, the correlation line between the air compressor and the air storage cylinder in the starting stage is continued to the stable running stage, and the line becomes thinner to indicate a decrease in intensity. The correlation line between the newly added brake valve and the release valve in the stable running stage is drawn from the starting position of the stable running stage. In the above manner, the initial brake machine running correlation chain is formed.

[0032] Step S125: Analyze the intensity change curve of each correlation relationship in the initial brake machine running correlation chain in the whole cycle, mark the peak points, valley points, and inflection points in the intensity change curve, determine the correlation intensity evolution characteristics, optimize the structure of the initial brake machine running correlation chain based on the characteristics, and generate a brake machine running correlation chain.

[0033] For each of the initial brake operation associated chain, such as the association between the air compressor and the air reservoir, the intensity value in the starting stage, the stable running stage, the braking stage and the stopping stage is extracted, and the intensity change curve is drawn with time as the horizontal axis and intensity value as the vertical axis. On the curve, the point with the highest intensity value is found as the peak point, the point with the lowest intensity value is found as the valley point, and the point where the direction of intensity change changes is found as the inflection point. For example, in the early stage of the starting stage, the intensity of the association rises rapidly to the peak point, then gradually decreases to the valley point in the stable running stage, and then slightly rises in the braking stage. The rising starting point is the inflection point. According to the distribution of these peak points, valley points and inflection points, the association intensity evolution characteristics are determined, such as the intensity of the association fluctuates greatly in the starting stage, and the intensity is stable in the stable running stage. Based on the characteristics, the initial brake operation associated chain is optimized, for example, the association relationship with intense change is marked with special color, and the drawing precision of the line is adjusted to highlight the key nodes, so as to generate the final brake operation associated chain.

[0034] Step S130: Based on the evolution trajectory of the brake operation associated chain, a brake performance dynamic benchmark feature is generated, which includes component association features of each running stage and cross-stage collaborative evolution features.

[0035] The evolution trajectory of the brake operation associated chain reflects the change process of each association relationship in different running stages. By feature extraction and analysis of the trajectory, a dynamic benchmark feature reflecting the normal performance state of the brake is generated. The component association features of each running stage are the core attribute set of the association relationship in this stage, and the cross-stage collaborative evolution features reflect the collaborative change law of the association relationship between different stages. Both of them constitute the brake performance dynamic benchmark feature.

[0036] Step S131: Traverse the time axis of the brake operation associated chain, extract the association relationship set corresponding to each running stage and the association strength evolution feature, and form a stage association data package.

[0037] According to the order of the time axis, the starting stage, the stable running stage, the braking stage and the stopping stage are traversed in turn. For each running stage, all the association relationships in this stage are extracted from the brake operation associated chain to form an association relationship set, and the intensity change curve, peak point, valley point, inflection point and other association strength evolution features of each association relationship in this stage are extracted. The above association relationship set and association strength evolution feature are packaged and stored to form a stage association data package corresponding to each running stage. For example, the starting stage association data package contains all the component association relationships in the starting stage and the intensity evolution feature data of each association relationship.

[0038] Step S132: Feature extraction is performed on the association relationship in each stage associated data packet to generate component association features of the running stage, which reflect the core performance of the association relationship between components in the running stage and the adaptability to the environmental impact data of the stage.

[0039] Taking the start-up stage associated data packet as an example, feature extraction is performed on each association relationship therein. For the association relationship between the air compressor and the air storage cylinder, the average strength value, the strength fluctuation amplitude, the peak occurrence time and other core performance features are extracted. At the same time, the relationship between the strength change of the association relationship and the environmental temperature and humidity in the start-up stage is analyzed. If the strength fluctuation amplitude of the association relationship increases when the environmental temperature rises, this adaptability feature is also included. In the same way, feature extraction is performed on all the association relationships in the start-up stage associated data packet, and the component association features of the start-up stage are formed after being summarized. The same process is used to generate the component association features of the stable running stage, the braking stage and the stopping stage.

[0040] Step S133: The component association features of adjacent running stages are compared and analyzed to identify the inheritance relationship, the variation relationship and the complementary relationship between the features, and the inter-stage feature association information is formed.

[0041] The component association features of the start-up stage and the stable running stage are compared. If the core performance of a component association feature of the stable running stage is basically the same as that of a component association feature of the start-up stage, only with a slight difference in strength value, it is determined that there is an inheritance relationship between the two. If the core performance of a component association feature of the stable running stage is obviously different from that of the corresponding feature of the start-up stage, such as the change of the strength change mode of the association relationship, it is determined that there is a variation relationship. If the function of a component association feature of the start-up stage is complementary to that of a component association feature of the stable running stage, which together guarantees the performance of the brake during the stage transition, it is determined that there is a complementary relationship. The specific performance of these relationships and the feature types involved are recorded to form the inter-stage feature association information between the start-up stage and the stable running stage. In the above manner, the component association features of the stable running stage and the braking stage, and the braking stage and the stopping stage are continuously compared to form a complete inter-stage feature association information.

[0042] Step S134: Based on the inter-stage feature association information, a cross-stage collaborative evolution model is constructed, which takes the component association features of each stage as input and outputs cross-stage collaborative evolution features reflecting the continuous change rule of the features, the cross-stage collaborative evolution features including the trigger factor and the evolution direction of the feature change.

[0043] The cross-stage synergistic evolution model adopts a multi-layer perceptron structure. The input layer includes all dimensional data of the associated features of the components in each stage, such as the average strength value of each associated relationship, the strength fluctuation amplitude, the adaptability parameter to the environment, etc. The hidden layer is set to three layers. The first layer is used to perform a preliminary nonlinear transformation on the input feature data. The second layer performs feature fusion on the transformed data. The third layer further refines the fused features and mines the synergistic relationship between the features. The output layer outputs the cross-stage synergistic evolution features, which include the trigger factors of feature changes (such as changes in environmental temperature, changes in component action frequency, etc.) and the evolution direction (such as strength increase, strength decrease, and fluctuation amplitude increase, etc.). In the training process of the model, historical inter-stage feature association information and corresponding cross-stage evolution results are used as training data. The weights of each layer are adjusted through the back propagation algorithm until the error between the cross-stage synergistic evolution features output by the model and the actual evolution results is within the preset range.

[0044] Step S135: Integrate the component association features of all operating stages and the cross-stage synergistic evolution features in chronological order, label the corresponding operating stage identifier and evolution node of each feature, and form the dynamic benchmark features of brake performance.

[0045] According to the chronological order from the start-up stage to the stop stage, the component association features of each stage are arranged in sequence, and then the cross-stage synergistic evolution features are inserted into the corresponding stage transition position. For example, the cross-stage synergistic evolution features from the start-up stage to the stable running stage are inserted after the component association features of the start-up stage and before the component association features of the stable running stage. Label the corresponding operating stage identifier (such as “start-up stage”, “stable running stage”, etc.) for each component association feature, and label the evolution node for the cross-stage synergistic evolution feature. The evolution node is the starting time point and key turning point of feature change. After integration, the dynamic benchmark features of brake performance are formed. The dynamic benchmark features of brake performance are stored in sequence and can fully reflect the normal performance evolution law of the brake in the whole cycle.

[0046] Step S140: Call the pre-trained artificial intelligence model for dynamic evaluation of brake performance, perform cross-dimensional interactive mapping processing on the real-time collected brake operating features and the dynamic benchmark features of brake performance, and generate the performance evolution evaluation results of the brake.

[0047] In this embodiment, the pre-trained brake performance dynamic evaluation artificial intelligence model is a sequence prediction model based on a Transformer architecture, which has been trained through a large amount of historical brake full-cycle operation correlation data, corresponding dynamic benchmark features and performance evaluation result labels. Real-time collected brake operation features are obtained through the same full-cycle data collection device as step S111, covering component parameter correlation data, environment adaptation data and stage transition data of the current operation stage. After inputting the real-time operation features and the brake performance dynamic benchmark features into the model, the model constructs a cross-dimension mapping relationship between the two through the feature interaction layer, mines the matching and deviation of the real-time features and the benchmark features, and then integrates and analyzes the results through the result generation layer to output an evaluation result containing the performance status of each component and the cross-stage coordination status.

[0048] Step S141: Obtain real-time collected brake operation features, which include component correlation features, real-time environment adaptation features and real-time stage transition features of the real-time operation stage.

[0049] Real-time parameter data in the current operation state is collected through sensors deployed on each component of the brake, such as current pressure output data of the air compressor, action frequency data of the brake valve, response time data of the brake cylinder, etc. Based on these parameter data, the parameter correlation between any two components in the current operation stage is analyzed, and the component correlation features of the real-time operation stage are extracted, which include correlation component pairs, correlation strength and correlation stability, etc. Real-time environment adaptation features are obtained through the train environment monitoring system to obtain current environmental data such as temperature, humidity, air pressure, etc., and the adaptation degree of the current component correlation features and the environmental data is generated, such as the change adaptation of the correlation strength between the brake hose and the brake cylinder when the environmental temperature rises. Real-time stage transition features are generated for the transition process between the current operation stage and the previous operation stage, analyzing the change rule, change rate and transition triggering condition of the component correlation features during the transition, such as the change of the correlation features between the brake valve and the air reservoir during the transition from the stable operation stage to the braking stage.

[0050] Step S142: Input the brake performance dynamic benchmark features and the real-time collected brake operation features into the feature interaction layer of the pre-trained brake performance dynamic evaluation artificial intelligence model.

[0051] The feature interaction layer of the brake performance dynamic evaluation artificial intelligence model comprises a feature standardization module, a feature alignment module and an interaction mapping module. The dynamic benchmark features of the brake performance and the real-time collected brake operation features are first input into the feature standardization module. The feature standardization module adopts the min-max standardization method to convert the dimensional data of the two types of features to the same data interval, eliminating the influence of dimensional differences. The standardized features enter the feature alignment module. The feature alignment module aligns the real-time operation features and the corresponding stage features in the dynamic benchmark features in the time dimension according to the feature operation stage identifier and the timestamp, ensuring that the features at the same time node can be effectively compared. The aligned features are then input into the interaction mapping module. The interaction mapping module calculates the correlation weight between the real-time features and the benchmark features in each dimension through the attention mechanism.

[0052] Step S143: constructing a cross-dimension mapping rule of the dynamic benchmark features and the real-time operation features through the feature interaction layer, the cross-dimension mapping rule comprising a stage matching rule, a feature type corresponding rule and an evolution trend comparison rule.

[0053] The interaction mapping module of the feature interaction layer first constructs a stage matching rule based on the aligned feature data, then constructs a feature type corresponding rule according to the feature attributes, and finally constructs an evolution trend comparison rule in combination with the time sequence change of the features, forming a complete cross-dimension mapping rule system. The above rules will serve as the basis for the interactive comparison of the dynamic benchmark features and the real-time operation features, ensuring the normativity and accuracy of the comparison process and avoiding evaluation errors caused by mismatched feature dimensions or inconsistent comparison logic.

[0054] Step S1431: analyzing the operation stage identifier in the dynamic benchmark features of the brake performance, determining the time range, environmental features and core correlation relationship types of each stage, and constructing a stage feature library.

[0055] The operation stage identifiers of the component correlation features and the cross-stage coordinated evolution features in the dynamic benchmark features of the brake performance are extracted, such as the “start-up stage” and the “stable operation stage”. The starting time and the ending time of each stage are determined according to the timestamp corresponding to the stage identifier, forming the time range of each stage. The environmental influence data corresponding to each stage is selected from the dynamic benchmark features, and the environmental features of each stage are extracted, such as the environmental temperature range of the start-up stage and the vibration intensity range of the stable operation stage. At the same time, the occurrence frequency and the strength of the correlation relationship in each stage are analyzed to determine the core correlation relationship type of each stage, such as the pressure correlation between the air compressor and the air storage cylinder in the start-up stage, and the action correlation between the brake valve and the brake cylinder in the braking stage. The time range, environmental features and core correlation relationship types of each stage are stored by stage, and a stage feature library is constructed.

[0056] Step S1432: extracting real-time running phase information in the real-time collected brake machine running features, comparing the real-time running phase information with the phase features in the phase feature library, determining the reference phase corresponding to the real-time running features, forming a phase matching rule, and the phase matching rule includes phase time overlap degree judgment, environment feature similarity judgment and core correlation type matching judgment.

[0057] The real-time running phase information is extracted from the real-time collected brake machine running features, including the starting time of the current running phase, the current environment data and the current core correlation type. The overlap ratio of the time range of the current running phase and the time range of each phase in the phase feature library is calculated, that is, the phase time overlap degree. The higher the overlap ratio is, the higher the time matching degree is. The cosine similarity algorithm is used to calculate the similarity of the current environment data and the environment features of each phase in the phase feature library, and the environment feature similarity is obtained. The consistency of the current core correlation type and the core correlation type of each phase in the phase feature library is compared, and the core correlation type matching judgment is performed. When the phase time overlap degree, the environment feature similarity and the core correlation type matching degree all reach the preset threshold, it is determined that the phase is the reference phase corresponding to the real-time running features, and thus the phase matching rule is formed.

[0058] Step S1433: type labeling is performed on the component correlation features and the cross-phase collaborative evolution features in the dynamic reference features, and the feature types are divided into parameter correlation type, environment adaptation type and evolution trend type.

[0059] The attributes of the component correlation features in the dynamic reference features are analyzed. If a certain feature mainly reflects the parameter mutual influence relationship between components, such as the corresponding relationship between the air compressor pressure and the air cylinder pressure, it is labeled as the parameter correlation type. If a certain feature mainly reflects the adaptation of the component correlation relationship to the environmental factors, such as the adaptation relationship between the brake valve action frequency and the environmental temperature, it is labeled as the environment adaptation type. For the cross-phase collaborative evolution features, if they mainly reflect the change law of the correlation features between different phases, such as the change trend of the brake cylinder pressure correlation feature from the stable running phase to the braking phase, they are labeled as the evolution trend type. According to the above standard, the types of all features in the dynamic reference features are labeled to ensure that each feature corresponds to a unique feature type.

[0060] Step S1434: type labeling is performed on the component correlation features, real-time environment adaptation features and real-time phase transition features in the real-time running features, so as to be consistent with the feature type division standard of the dynamic reference features, forming a feature type corresponding rule, and the feature type corresponding rule includes feature type definition, type identification basis and feature conversion mode when the types are not matched.

[0061] The feature type classification standard of the dynamic reference feature is referred to, and the real-time running feature is labeled with a type. The component correlation feature reflecting the mutual influence of component parameters in the real-time running feature is labeled as a parameter correlation type, the real-time environment adaptation feature is labeled as an environment adaptation type, and the real-time stage transition feature reflecting the feature change law during the stage transition period is labeled as an evolution trend type. Based on this, the feature type definition is determined, that is, the parameter correlation type feature is a feature reflecting the interaction relationship of component parameters, the environment adaptation type feature is a feature reflecting the correlation relationship and environment adaptation condition, and the evolution trend type feature is a feature reflecting the time evolution law of the feature. At the same time, the type identification basis is determined, that is, the type is judged according to the core attribute and description content of the feature. If a certain feature in the real-time running feature cannot be directly matched with the above type, the feature is processed through a feature conversion mode, such as a feature containing both parameter correlation and environment adaptation attributes is disassembled into two independent features, respectively corresponding to the parameter correlation type and the environment adaptation type, thereby forming a feature type corresponding rule.

[0062] Step S1435: Analyzing the evolution curve of the cross-stage collaborative evolution feature in the dynamic reference feature, extracting the change direction, change rate and change period of the evolution curve, and constructing an evolution trend template.

[0063] For each cross-stage collaborative evolution feature in the dynamic reference feature, the corresponding evolution curve data is extracted, and the evolution curve data takes time as the horizontal axis and feature intensity as the vertical axis. The change direction is determined by analyzing the slope change of the evolution curve. If the slope is positive, the change direction is upward, if the slope is negative, the change direction is downward, and if the slope is zero, the change direction is stable. The change amount of the feature intensity in a unit of time is calculated to determine the change rate. The greater the change amount, the faster the change rate. The change period is determined by identifying the repeated feature intensity fluctuation mode in the evolution curve. The time interval between two same fluctuation modes is the change period. The extracted change direction, change rate and change period are classified and arranged according to the type of the cross-stage collaborative evolution feature, and an evolution trend template is constructed. Each template corresponds to a typical evolution law of a type of cross-stage collaborative evolution feature.

[0064] Step S1436: Determining the change law of the real-time stage transition feature in the real-time running feature, forming a real-time evolution trend description, and constructing an evolution trend comparison rule. The evolution trend comparison rule includes evolution direction consistency judgment, change rate deviation range setting and change period coincidence degree judgment.

[0065] Analyze the parameter change data of the real-time stage transition characteristics in the real-time operation characteristics, determine its change direction, change rate and change period, and form a real-time evolution trend description. Compare the real-time evolution trend description with the evolution trend template to determine whether the change directions of the two are consistent. If the real-time change direction is the same as the template change direction, the evolution direction consistency is judged to be compliant; if different, it is judged to be non-compliant. Set the change rate deviation range. When the difference between the real-time change rate and the template change rate is within this range, it is judged to be compliant; if it exceeds the range, it is judged to be non-compliant. Calculate the overlap ratio of the real-time change period and the template change period, that is, the change period overlap. When the overlap ratio reaches the preset threshold, it is judged to be compliant. Combining the above three judgment criteria, construct the evolution trend comparison rules.

[0066] Step S144: Based on the cross-dimensional mapping rules, interactively compare the component association features of each stage in the dynamic baseline features, the cross-stage collaborative evolution features and the corresponding features in the real-time operation features to explore the evolutionary matching relationship and deviation association relationship between the features.

[0067] After determining the baseline phase corresponding to the real-time operational feature based on the phase matching rules within the cross-dimensional mapping rules, the dynamic baseline feature is then matched with similar real-time operational features according to the feature type matching rules. Evolutionary trend-based features are then specifically compared using the evolutionary trend comparison rules. During this comparison, the matching of feature attributes is recorded. For features with consistent attributes or differences within the allowable range, an evolutionary matching relationship is determined. For features with attribute differences outside the allowable range, the causes and impact of the deviation are analyzed and determined to be a deviation association relationship, thus completing a comprehensive interactive comparison of the two types of features.

[0068] Step S1441: According to the stage matching rules, the real-time operation features are grouped with the dynamic baseline features of the corresponding baseline stage to form feature comparison groups. Each feature comparison group contains the component association features of the baseline stage, the cross-stage co-evolution features and the corresponding features in the real-time operation features.

[0069] Based on the stage matching rules formed in step S1432, if the benchmark stage corresponding to the real-time operation feature is determined to be the stable operation stage, the component association features of the stable operation stage and the cross-stage co-evolution features between the stable operation stage and adjacent stages are extracted from the dynamic benchmark features. These features are combined with the stable operation stage component association features, real-time environment adaptation features, and real-time stage transition features in the real-time operation features to form a stable operation stage feature comparison group. Using the same method, if the real-time operation feature corresponds to other benchmark stages, feature comparison groups for the corresponding stages are formed separately, ensuring that the benchmark features and real-time features within each feature comparison group belong to the same operation stage or stage transition process.

[0070] Step S1442: Based on the feature type corresponding rule, in each feature comparison group, the parameter association type features of the dynamic reference features are corresponded with the parameter association type features of the real-time running features, the environment adaptation type features are corresponded with the environment adaptation type features, and the evolution trend type features are corresponded with the evolution trend type features.

[0071] In the stable running phase feature comparison group, according to the feature type corresponding rule, the parameter association type component association features (such as the parameter interaction features of the brake valve and the relief valve) in the stable running phase of the dynamic reference features are corresponded with the parameter association type component association features in the stable running phase of the real-time running features; the environment adaptation type component association features (such as the adaptation features of the brake hose association strength and the humidity) in the stable running phase of the dynamic reference features are corresponded with the real-time environment adaptation features in the real-time running features; the evolution trend type cross-stage collaborative evolution features between the stable running phase and the braking phase of the dynamic reference features are corresponded with the real-time stage transition features (the features of the stable running phase to the braking phase) in the real-time running features. The other feature comparison groups also complete the corresponding matching of the same type of features in the above manner.

[0072] Step S1443: Attribute comparison is performed on the corresponding type features. The parameter association type features compare the associated component pairs and the associated performances, the environment adaptation type features compare the environment factors and the adaptation performances of the associated relationships, and the evolution trend type features compare the change directions and the change nodes.

[0073] For the corresponding group of the parameter association type features, whether the associated component pairs in the dynamic reference features and the real-time running features are consistent is compared, such as whether they are both the associated brake valve and relief valve, and then whether the associated performances are compared, such as whether the response modes of the relief valve when the brake valve acts are the same. For the corresponding group of the environment adaptation type features, whether the adaptation performances of the environment factors (such as temperature and humidity) and the associated relationships are compared, such as whether the change amplitudes of the brake hose association strength are consistent under the same humidity condition. For the corresponding group of the evolution trend type features, whether the change directions are the same is compared, such as whether the association strength rises when transitioning from the stable running phase to the braking phase; and then whether the change nodes are consistent is compared, such as whether the time points at which the association strength starts to rise are consistent.

[0074] Step S1444: If the attribute comparison results of the corresponding type features meet the preset matching threshold range, the feature relationship of the feature comparison group is marked as an evolution matching relationship, and the matched feature attributes and the matching degree description are recorded.

[0075] The matching threshold of the preset parameter correlation type feature is that the correlation components are completely consistent and the similarity of the correlation performance reaches a preset proportion. The matching threshold of the environment adaptation type feature is that the deviation of the environment factor adaptation performance is within a preset range. The matching threshold of the evolution trend type feature is that the change direction is consistent and the time difference of the change node is within a preset interval. If the attribute comparison result of the parameter correlation type feature of the brake valve and the release valve in the stable running stage feature comparison group meets the matching threshold of the parameter correlation type feature, the feature relationship of the feature comparison group is marked as an evolution matching relationship, and the matched attribute is recorded as "consistent correlation component pair, correlation performance similarity meets the requirement", and the matching degree is described as "high matching".

[0076] Step S1445: If the attribute comparison result of the corresponding type feature exceeds the preset matching threshold range, analyze the stage position, feature type, and evolution node of the deviation, determine the influence range and possible trigger factors of the deviation, mark the feature relationship of the feature comparison group as a deviation correlation relationship, and record the deviation attribute, deviation degree, and deviation trigger factor description.

[0077] If the attribute comparison result of the brake hose correlation strength and the humidity environment adaptation type feature in the stable running stage feature comparison group shows that the change amplitude of the brake hose correlation strength in the real-time feature exceeds the deviation range of the dynamic reference feature, it is determined that the stage position of the deviation is the stable running stage, and the feature type is the environment adaptation type. Analyze the evolution node of the deviation correlation as the middle stage of the stable running stage, and the influence range involves the correlation stability of the brake hose and the brake cylinder. Combined with the current environment data and component parameters, it is determined that the possible trigger factor is that the actual humidity exceeds the humidity range in the reference environment and the brake hose is slightly aged. The feature relationship of the feature comparison group is marked as a deviation correlation relationship, and the deviation attribute is recorded as "deviation of brake hose correlation strength and humidity adaptation", the deviation degree is "moderate deviation", and the deviation trigger factor description is "actual humidity exceeds the standard and brake hose is slightly aged".

[0078] Step S1446: Traverse all feature comparison groups, summarize the evolution matching relationship and the deviation correlation relationship, and form a complete feature interaction comparison result set.

[0079] The above attribute comparison, relationship marking, and recording operations are performed on the feature comparison groups corresponding to the starting stage, stable running stage, braking stage, and stopping stage in sequence. All feature groups marked as evolution matching relationship and deviation correlation relationship and their detailed information are summarized to form a complete result set containing the feature interaction of each stage.

[0080] Step S145: Through the result generation layer of the brake performance dynamic evaluation artificial intelligence model, the evolution matching relationship and the deviation correlation relationship are integrated, the operating stages and evolution nodes corresponding to the deviations are marked, and the brake performance evolution evaluation results including component performance evolution status evaluation and cross-stage collaborative status evaluation are generated.

[0081] The result generation layer of the AI ​​model for dynamic brake performance evaluation includes a feature relationship integration module, a deviation location module, and an evaluation report generation module. The feature relationship integration module receives a set of feature interaction comparison results and classifies and integrates the evolutionary matching relationships by operating stage to form a feature matching summary for each stage. It also classifies and integrates the deviation association relationships by deviation type and impact range to form a deviation summary. The deviation location module, based on the records of deviation association relationships, annotates the specific operating stage (e.g., the mid-stage of the stable operation stage) and evolutionary nodes (e.g., key time points when the association strength changes) corresponding to each deviation. The evaluation report generation module generates a component performance evolution status assessment and a cross-stage collaborative status assessment based on the feature matching summary and deviation location results. The component performance evolution status assessment describes whether the performance evolution of each component conforms to the baseline trend based on the evolutionary matching and deviation of its associated features. The cross-stage collaborative status assessment describes whether the cross-stage collaboration is normal based on the stage transition process based on the matching and deviation of evolutionary trend-type features, ultimately forming a complete brake performance evolution assessment result.

[0082] Step S150: Generate a dynamic brake performance adjustment plan based on the brake performance evolution evaluation result and the evolution trend of the associated chain. The dynamic brake performance adjustment plan includes phased component parameter adjustment directions and cross-stage collaborative optimization suggestions.

[0083] Based on the component performance deviations and cross-stage collaborative deviations identified in the brake performance evolution evaluation results, combined with the evolution trend of the brake operation association chain (such as the changing pattern of association strength and the evolution direction of the association relationship), adjustment directions for component parameters are formulated for different operation stages, and collaborative optimization suggestions are formulated for the stage transition process. The above adjustment directions and optimization suggestions need to take into account the association relationship between components and the influence of environmental factors to ensure that the solution can effectively correct performance deviations and restore the brake performance to the baseline evolution trend.

[0084] Step S151: parsing the component performance evolution status evaluation in the brake performance evolution evaluation result, identifying the components whose performance evolution deviates from the baseline trend and the corresponding deviation stages, and forming a staged list of components to be adjusted.

[0085] The performance evolution state evaluation in the brake machine performance evolution evaluation result is analyzed, and the content describing the performance deviation from the benchmark trend is extracted. If the evaluation result shows that the performance evolution of the brake hose deviates from the benchmark trend in the middle of the stable running stage, and the performance evolution of the brake cylinder deviates from the benchmark trend in the early stage of the braking stage, the brake hose and the brake cylinder are identified as deviation components, and the corresponding deviation stages are the middle of the stable running stage and the early stage of the braking stage respectively. The above information is classified and arranged according to the deviation stage to form a staged component list to be adjusted, such as the brake hose to be adjusted in the middle of the stable running stage and the brake cylinder to be adjusted in the early stage of the braking stage.

[0086] Step S152: For each component in the staged component list to be adjusted and the corresponding stage, analyze the reason for the performance evolution of the component deviating from the benchmark trend, which is determined based on the deviation association relationship and the evolution trend of the associated chain of the characteristics of the component in the stage.

[0087] For the brake hose to be adjusted in the middle of the stable running stage, the component association characteristics of the brake hose in the middle of the stable running stage are extracted, and the benchmark component association characteristics in the middle of the stable running stage in the dynamic benchmark characteristics of the brake machine are extracted. The component association characteristics of the brake hose are compared with the benchmark component association characteristics, and the deviation attribute list is formed by combining the deviation association relationship between them to determine the characteristics attributes with deviations. The deviation attribute list may include the association strength deviation between the brake hose and the brake valve, the pressure conduction delay deviation of the brake hose, etc.

[0088] Each deviation attribute in the deviation attribute list is analyzed, and the environment influence data in the middle of the stable running stage is combined to judge whether the environment factor causes the deviation attribute. If the humidity in the stage is higher than the benchmark environment humidity, and the pressure conduction delay deviation of the brake hose has a corresponding relationship with the humidity increase, the environment adaptation deviation is taken as one of the reasons. The evolution characteristics of the association relationship before and after the middle of the stable running stage in the brake machine running associated chain are called, and the influence of the association relationship in the previous stage (the early stage of the stable running stage) on the associated characteristics of the brake hose in the current stage is analyzed. If the association strength between the brake valve and the brake hose in the early stage of the stable running stage abnormally fluctuates and the fluctuation continues to affect the middle stage, the evolution transmission deviation of the previous stage is taken as one of the reasons. The association relationship between the brake hose and other components (such as the brake cylinder and the air reservoir) in the middle of the stable running stage is analyzed. If the pressure output of the brake cylinder is unstable in the stage, the load of the brake hose is abnormal, and then the associated characteristics deviation is caused, the associated component influence deviation is taken as one of the reasons. The environment adaptation deviation, the evolution transmission deviation of the previous stage, and the associated component influence deviation are combined, and the amplification effect of the characteristics of the continuous decline of the association strength of the brake hose in the evolution trend of the associated chain on the deviation is formed to form the complete reason description of the performance evolution deviation of the brake hose from the benchmark trend in the middle of the stable running stage.

[0089] For the brake cylinder to be adjusted at the initial stage of the braking phase, the same analysis process is adopted. The component correlation features of the brake cylinder at the initial stage of the braking phase and the corresponding reference component correlation features are extracted, and after comparison, the deviation attribute list is determined. Then, in combination with the environmental influence data of this stage, the correlation evolution features of the previous stage (the end of the stable running phase), and the correlation relationship with other components (such as brake valve and relief valve), the causes of the performance deviation of the brake cylinder are comprehensively determined, which may include environmental adaptation deviation caused by excessive vibration intensity at the initial stage of the braking phase, evolution conduction deviation caused by unstable pressure of the air storage cylinder at the end of the stable running phase, and correlation component influence deviation caused by delayed response of the brake valve, etc.

[0090] Step S1521: Extract the component correlation features of the component to be adjusted in the corresponding stage, and extract the reference component correlation features of the performance dynamic reference features of the brake machine in this stage.

[0091] For each component in the list of components to be adjusted in stages, according to its corresponding deviation stage, all correlation feature data of the component in this stage are located and extracted from the brake machine running correlation data, which covers the correlation strength, correlation response time, correlation parameter cooperative change mode, etc. between the component and all other related components. At the same time, from the performance dynamic reference features of the brake machine, according to the identification of the deviation stage, the reference component correlation features corresponding to the component to be adjusted in this stage are extracted, which include correlation strength reference values, correlation response time reference ranges, and correlation parameter cooperative change reference modes of the same dimension, etc., to ensure that the extracted correlation features of the component to be adjusted and the reference component correlation features are completely matched in feature dimension and stage range.

[0092] Step S1522: Compare the component correlation features of the component to be adjusted with the reference component correlation features, combine the deviation correlation relationship, determine the feature attributes with deviations, and form a deviation attribute list.

[0093] The component correlation features of the component to be adjusted and the reference component correlation features are compared one by one according to the feature dimension, and the deviation correlation relationship between the two is referred to, and all feature dimensions that do not meet the requirements of the reference features are screened out, which are the feature attributes with deviations. The above feature attributes with deviations are sorted according to the importance, and the deviation forms of each attribute are recorded in turn, such as correlation strength lower than the reference value, correlation response time exceeding the reference range, correlation parameter cooperative change mode inconsistent with the reference mode, etc., to finally form a structured deviation attribute list.

[0094] Step S1523: Analyze each deviation attribute in the deviation attribute list, and combine it with the environmental impact data of this stage to determine whether environmental factors cause the deviation attribute to be generated. If the environmental factors are related to the deviation attribute, the environmental adaptation deviation is regarded as one of the reasons.

[0095] Analyze each deviation attribute in the deviation attribute list one by one, and retrieve the complete environmental impact data for that deviation stage, including the change curves of specific environmental parameters such as temperature, humidity, vibration, and air pressure. By analyzing the correlation between the occurrence time and change trend of the deviation attribute and the change time and change trend of the environmental parameters, if the occurrence time of a deviation attribute coincides with the abnormal change time of the environmental parameters and the change trend remains consistent, it is determined that environmental factors caused the deviation attribute, and the environmental adaptation deviation is included in the scope of the cause of the component performance deviation.

[0096] Step S1524: Retrieve the evolution characteristics of the association relationship before and after this stage in the brake operation association chain, analyze the impact of the association relationship in the previous stage on the association characteristics of the components to be adjusted in the current stage, and determine whether the current deviation is caused by the abnormal evolution of the association in the previous stage. If there is an impact, the evolution conduction deviation in the previous stage will be regarded as one of the reasons.

[0097] The evolutionary characteristics of the association relationships of the component to be adjusted, both in the previous and subsequent stages corresponding to the deviation phase, are extracted from the brake operation association chain. Focus is placed on the intensity changes, stability, and abnormal fluctuations of all associations related to the component to be adjusted in the previous stage. The evolutionary characteristics of these previous stages are analyzed to determine if there are any anomalies, and whether these anomalies are transmitted through inter-component associations to the current stage, leading to deviations in the association characteristics of the component to be adjusted. If there is a clear transmission path and evidence of an impact, the deviation in the previous stage's evolutionary transmission is listed as one of the causes.

[0098] Step S1525: Analyze the association relationship between the component to be adjusted and other components at this stage to determine whether the deviation of the associated characteristics of the component to be adjusted is caused by abnormal performance of other components. If so, the deviation of the associated components is considered as one of the reasons.

[0099] The network of relationships between the component to be adjusted and all related components (such as upstream supply components, downstream execution components, and coordinated control components) during the deviation phase is analyzed, and the performance data and correlation characteristics of these related components during that phase are extracted. Each related component is individually inspected for performance anomalies, such as parameters outside the normal range, delayed action responses, unstable correlation strength, etc., and analyzed to determine whether such anomalies directly or indirectly affect the correlation characteristics of the component to be adjusted. If there is a clear causal relationship between the performance anomaly of a related component and the deviation of the correlation characteristics of the component to be adjusted, the deviation is considered a contributing factor.

[0100] Step S1526: Synthesize the environment adaptation deviation, the previous stage evolution conduction deviation, the associated component influence deviation and other deviation types, combine the amplification or inhibition effect of the associated chain evolution trend on the deviation, form the complete reason description of the performance evolution deviation of the component from the benchmark trend in the corresponding stage.

[0101] Summarize the various deviation reasons obtained by the above analysis, and identify whether there are other uncaptured deviation types (such as inherent deviation caused by component aging, historical deviation left by improper maintenance, etc.). Combine the influence of the associated feature deviation of the component to be adjusted in the braking machine running associated chain evolution trend, judge whether the evolution trend amplifies the deviation degree or suppresses the deviation development. Sort the various deviation reasons according to the influence degree, describe the action mechanism, the manifestation form and the correlation degree with the deviation of each reason in detail, and finally form a logically complete and clearly hierarchical reason description.

[0102] Step S153: According to the reasons, combine the running parameter adaptation range of the component in the corresponding stage and the intra-stage association relationship with other components, determine the parameter adjustment direction of the component in the stage, form the phased component parameter adjustment direction, and the phased component parameter adjustment direction contains the stage occasion of parameter adjustment, adjustment trend and adaptation requirements with stage environment.

[0103] For the brake hose in the middle stage of stable running stage, according to the reasons of performance deviation, obtain the running parameter adaptation range of the brake hose in the middle stage of stable running stage, which contains the working pressure allowed interval of the brake hose, the pressure conduction delay allowed range, the associated strength adaptation interval with the brake valve, etc. For the environment adaptation deviation (too high humidity causes pressure conduction delay), it is determined that the pressure compensation parameter of the brake hose needs to be adjusted, and the adjustment trend is to improve the pressure compensation coefficient; for the previous stage evolution conduction deviation (abnormal fluctuation of brake valve associated strength), the coupling parameter of brake hose and brake valve needs to be adjusted synchronously, and the adjustment trend is to optimize the coupling frequency; for the associated component influence deviation (abnormal brake cylinder load), the flow regulation parameter of the brake hose needs to be adjusted, and the adjustment trend is to increase the flow threshold.

[0104] The adjustment trend of these parameters is analyzed whether it is in line with the running parameter adaptation range of the brake hose in the middle of the stable running stage. If the pressure compensation coefficient is increased and it is still in the allowed interval, it is confirmed that the adjustment trend is feasible. At the same time, it is analyzed whether the parameter adjustment will affect the relationship between the brake hose and other components (such as brake cylinder, air reservoir). If the adjustment of the flow regulating parameter may break the pressure supply balance with the air reservoir, the pressure adaptation auxiliary parameter of the air reservoir needs to be set at the same time when the flow parameter is adjusted. The stage opportunity of parameter adjustment is the period when the environmental humidity decreases in the middle of the stable running stage, so as to reduce the interference of environmental factors on the adjustment effect, and to clearly adjust the environmental conditions such as temperature and humidity in this stage, and to form the parameter adjustment direction of the brake hose in the middle of the stable running stage.

[0105] For the brake cylinder in the initial stage of braking, the running parameter adaptation range of the brake cylinder in the initial stage of braking is obtained according to its deviation reason, including the brake pressure output range, the response time range, and the cooperative action threshold with the brake valve. For the environmental adaptation deviation caused by excessive vibration intensity, the damping buffer parameter of the brake cylinder is adjusted, and the adjustment trend is to increase the buffer coefficient. For the evolution conduction deviation of the previous stage caused by unstable air pressure of the air reservoir, the pressure receiving sensitivity parameter of the brake cylinder is adjusted, and the adjustment trend is to reduce the sensitivity. For the related component influence deviation caused by the response delay of the brake valve, the action trigger parameter of the brake cylinder is adjusted, and the adjustment trend is to trigger the time in advance. It is verified whether these adjustment trends are within the adaptation range, and the influence on the relationship with other components is evaluated, and the adjustment opportunity is determined as after the vibration peak in the initial stage of braking, which meets the air pressure condition in this stage, and the phased parameter adjustment direction of the brake cylinder is formed.

[0106] Step S1531: Obtain the running parameter adaptation range of the component in the corresponding stage, which contains the allowed fluctuation interval of each running parameter of the component in the stage and the adaptation interval with the environmental characteristics.

[0107] By querying the brake machine component design manual, historical running data statistics results and technical specifications provided by the manufacturer, the running parameter adaptation range of the component to be adjusted in the corresponding deviation stage is obtained. The running parameter adaptation range is for each running parameter of the component (such as pressure, flow, temperature, response time, action frequency, etc.), and clearly defines the allowed fluctuation interval of its normal work, that is, the upper limit and lower limit. At the same time, combined with the typical environmental characteristics (such as temperature range, humidity range, vibration level, etc.) in this stage, the adaptation interval of each running parameter corresponding to different environmental characteristics is determined to ensure that the parameter can still maintain normal function under certain environmental conditions.

[0108] Step S1532: For each reason of the component performance evolution deviating from the benchmark trend, determine the corresponding adjustment parameter type, which includes environment adaptation type parameter, association synergy type parameter, and stage transition type parameter.

[0109] According to each specific reason of the component performance deviation, analyze the operating parameter category affected by the reason, and then determine the corresponding adjustment parameter type. If the reason is related to environmental factors (such as deviation caused by environmental humidity and temperature), the adjustment parameter type is the environment adaptation type parameter, which is used to optimize the adaptation ability of the component to environmental changes; if the reason involves abnormal association relationship with other components (such as insufficient association strength and synergy action disorder), the adjustment parameter type is the association synergy type parameter, which is used to improve the synergy working effect between components; if the reason is related to improper parameter connection in the stage transition process, the adjustment parameter type is the stage transition type parameter, which is used to ensure smooth transition of parameters in stage switching.

[0110] Step S1533: Analyze the difference between the current parameter performance of the adjustment parameter type and the benchmark parameter performance, and combine the operating parameter adaptation range of the stage to preliminarily determine the parameter adjustment trend. If the current parameter performance exceeds the upper limit of the adaptation interval, the preliminary adjustment trend is to reduce the parameter; if it is lower than the lower limit of the adaptation interval, the preliminary adjustment trend is to increase the parameter.

[0111] For each determined adjustment parameter type, extract its current parameter performance data (such as current pressure value, current response time, and current association strength value), and compare it with the benchmark parameter performance (such as benchmark pressure value, benchmark response time, and benchmark association strength value) to calculate the difference between them. Combine the operating parameter adaptation range of the corresponding stage to determine whether the current parameter performance exceeds the range. If the current parameter performance is higher than the upper limit of the adaptation interval, preliminarily determine the adjustment trend to reduce the parameter; if the current parameter performance is lower than the lower limit of the adaptation interval, preliminarily determine the adjustment trend to increase the parameter; if the current parameter performance is within the adaptation interval but deviates from the benchmark performance, determine the adjustment trend to approach the benchmark value according to the deviation direction.

[0112] Step S1534: Analyze the association relationship between the component and other components in the stage to determine whether the preliminarily determined parameter adjustment trend will affect the parameter performance of other components to exceed their adaptation range. If there is an effect, analyze the parameter sensitivity of the associated component to adjust the amplitude of the preliminary adjustment trend.

[0113] The connection relationship and parameter interaction logic of the to-be-adjusted component with all associated components at the corresponding stage are combed. After the simulation of the preliminary determined parameter adjustment trend is implemented, the possible changes in the parameter performance of the associated components are determined. If the simulation result shows that the parameter performance of a certain associated component may exceed its own operating parameter adaptation range, the sensitivity of the associated component to the adjustment parameter is further analyzed, i.e., the change amplitude of the parameter of the associated component when the adjustment parameter changes by one unit. According to the sensitivity analysis result, the amplitude of the preliminary adjustment trend is appropriately reduced or enlarged, so as to ensure that the parameter of the to-be-adjusted component after adjustment meets the requirements, and at the same time, the parameter of the associated component can also be maintained within the adaptation range.

[0114] Step S1535: In combination with the environmental characteristics of the stage, it is determined whether the adjusted parameter trend is adapted to the environmental characteristics. If not, the parameter adjustment trend is further corrected, so that the parameter after adjustment can adapt to the environment of the stage.

[0115] The environmental characteristic data of the corresponding stage are called, including the specific values, change frequency and fluctuation amplitude of the environmental parameters. The parameter value corresponding to the adjusted parameter trend is substituted into the component operation simulation model under the environmental characteristics, to determine whether the parameter can normally function under the environmental conditions, or whether a new deviation caused by insufficient environmental adaptability will occur. If the simulation result shows that the parameter adjustment trend is not adapted to the environmental characteristics, for example, the adjusted pressure parameter still has a conduction delay in a high-humidity environment, the parameter adjustment trend is further corrected according to the specific influence of the environmental characteristics, such as increasing the amplitude of pressure compensation or adjusting the dynamic response speed of the parameter, until the parameter after adjustment can adapt to the environmental characteristics of the stage.

[0116] Step S1536: The stage timing of parameter adjustment, the gradual change requirement of adjustment amplitude and the cooperative adjustment sequence with associated components are marked, to form a phased component parameter adjustment direction.

[0117] According to the operating condition characteristics and environmental change law of the corresponding stage, the best stage timing of parameter adjustment is determined, for example, the period with relatively stable environmental parameters, the period with medium load of the component or the non-critical operating period within the stage is selected for adjustment. For the adjustment amplitude, the gradual change requirement is set to avoid the impact of parameter mutation on the components and associated systems, for example, it is specified that the target value is gradually adjusted in multiple times within a certain time, and the amplitude of each adjustment does not exceed the preset threshold. The dependency relationship of the components associated with the to-be-adjusted component to the adjustment is analyzed, and the cooperative adjustment sequence is determined, such as adjusting the parameters of the upstream supply components first, then adjusting the parameters of the to-be-adjusted components, and finally adjusting the parameters of the downstream execution components, to ensure the coordination and stability of the parameter adjustment of the entire associated system. The stage timing, gradual change requirement and cooperative adjustment sequence are integrated with the parameter adjustment trend to form a complete phased component parameter adjustment direction.

[0118] Step S154: Analyze the cross-stage coordination state evaluation in the brake performance evolution evaluation result, identify the coordination link deviating from the benchmark trend and the corresponding transition stage, and form a cross-stage optimization coordination link list.

[0119] The cross-stage coordination state evaluation in the brake performance evolution evaluation result is analyzed in detail, and the evaluation description of the coordination relationship in the transition process of different stages is extracted. If the evaluation result shows that, in the process of transition from the stable running stage to the braking stage, the coordination link between the air storage cylinder and the brake valve deviates from the benchmark trend, specifically, the supply pressure speed of the air storage cylinder to the brake valve in the transition stage is lower than the benchmark supply pressure speed; in the transition from the braking stage to the stopping stage, the coordination link between the brake cylinder and the relief valve has a coordination action delay, deviating from the benchmark coordination sequence, then the coordination link between the air storage cylinder and the brake valve, and the coordination link between the brake cylinder and the relief valve are identified as the coordination link deviating from the benchmark trend, and the corresponding transition stages are the stable running-braking transition stage and the braking-stopping transition stage respectively. The above coordination links and corresponding transition stages are sorted according to the transition order to form a cross-stage optimization coordination link list.

[0120] Step S155: For each coordination link and corresponding transition stage in the cross-stage optimization coordination link list, analyze the reason for the deviation of the coordination evolution from the benchmark trend, which is determined based on the deviation association relationship and the evolution trend of the associated chain of the cross-stage coordination evolution characteristics and the benchmark characteristics.

[0121] For the air storage cylinder and brake valve coordination link in the stable running-braking transition stage, the cross-stage coordination evolution characteristics of the coordination link in the transition stage are extracted, and the benchmark cross-stage coordination evolution characteristics of the corresponding transition stage in the dynamic benchmark characteristics of the brake performance are extracted. After comparing the two, the deviation attributes such as supply pressure response time deviation and pressure coordination precision deviation are determined. Combined with the environmental influence data of the transition stage (such as vibration data and temperature data of the transition stage), it is judged whether the environmental factors cause the deviation. If the vibration intensity of the transition stage is higher than the benchmark value, and the supply pressure response time deviation increases with the increase of vibration intensity, then the environmental adaptation deviation is one of the reasons. The associated evolution characteristics before and after the transition stage in the brake running associated chain are called, if the pressure stability deviation of the air storage cylinder at the end of the stable running stage is transmitted to the transition stage, then the evolution transmission deviation of the previous stage is one of the reasons. Analyze the association relationship between the air storage cylinder and the brake valve in the coordination link, if the abnormal valve opening regulation of the brake valve leads to the increase of supply pressure resistance, then the associated component influence deviation is one of the reasons. Comprehensive these reasons, and combined with the amplification effect of the continuous weakening of the associated strength of the coordination link in the associated chain evolution trend, the reason description of the deviation of the coordination link is formed.

[0122] For the brake cylinder and relief valve coordination link in the brake-stop transition phase, the same flow is analyzed. The cross-phase coordination evolution characteristics and benchmark characteristics of this link are extracted, and the deviation attributes (such as action coordination timing deviation, pressure release coordination deviation) are determined. Combined with the transition phase environment data, the associated evolution characteristics of the previous phase (the end of the braking phase), and the internal component association, the possible causes include environmental adaptation deviation caused by sudden pressure drop in the transition phase, previous phase conduction deviation caused by brake cylinder pressure residue at the end of the braking phase, and associated component influence deviation caused by insufficient opening speed of the relief valve.

[0123] Step S156: According to the reasons, combined with the phased component parameter adjustment direction and cross-phase association, the cross-phase coordination optimization suggestion for this coordination link is made, which includes parameter coordination adjustment mode in the transition phase, associated component cooperation adjustment requirement and evolution trend correction direction.

[0124] For the air storage cylinder and brake valve coordination link in the stable operation-brake transition phase, according to its deviation reasons and the air storage cylinder and brake valve respective phased component parameter adjustment direction, the cross-phase coordination optimization suggestion is made. In terms of parameter coordination adjustment mode, considering the existence of supply pressure response time deviation and pressure coordination precision deviation, the coordination adjustment mode of "air storage cylinder pre-pressurization + brake valve early opening" is adopted, that is, the air storage cylinder increases the pre-pressurization parameter at the end of the stable operation phase according to the phased adjustment direction, and the brake valve opens the valve port in the early stage of the transition phase, and the adjustment time difference of the two is set to one fifth of the transition phase time. In terms of associated component cooperation adjustment requirement, the upstream air compressor is required to maintain stable pressure supply in the transition phase, and the downstream brake cylinder is prepared for pressure reception, and its parameter adjustment needs to be synchronized with the coordination adjustment of the air storage cylinder and the brake valve, for example, the pressure reception sensitivity adjustment of the brake cylinder needs to be started after the pre-pressurization of the air storage cylinder is completed. In terms of evolution trend correction direction, aiming at the evolution trend of continuously weakening association strength of this coordination link, by setting the phased target threshold of the association strength of the air storage cylinder and the brake valve, combined with the actual association strength monitoring result after parameter coordination adjustment, the pre-pressurization parameter and the valve opening amplitude are dynamically fine-tuned, gradually making the association strength return to the benchmark evolution trend, and ensuring that the supply pressure speed and pressure precision in the transition phase meet the benchmark requirements.

[0125] For the brake cylinder and relief valve coordination link in the brake-stop transition stage, the corresponding cross-stage coordination optimization suggestions are formulated according to the deviation causes and the adjustment direction of the two-stage parameters. In terms of parameter coordination adjustment mode, due to the existence of action coordination time sequence deviation and pressure release coordination deviation, the "brake cylinder pressure pre-decay + relief valve gradient opening" coordination adjustment mode is adopted, that is, the brake cylinder starts the pressure pre-decay program in advance according to the two-stage adjustment direction at the end of the braking stage, and the relief valve gradually increases the opening degree according to the preset gradient in the transition stage, and the gradient interval is set to one sixth of the transition stage time, so as to realize the precise matching of pressure release and relief action. In terms of related component cooperation adjustment requirements, it is clear that the air storage cylinder needs to maintain low pressure stable output in the transition stage, the brake valve needs to send a signal to trigger the relief valve action when the brake cylinder pressure pre-decay reaches the preset value, and the parameter adjustment time sequence of the two needs to be consistent with the coordination adjustment of the brake cylinder and the relief valve, for example, the low pressure stable parameter setting of the air storage cylinder needs to be completed after the brake cylinder pressure pre-decay parameter is determined. In terms of evolution trend correction direction, in view of the trend that the action coordination of the coordination link is gradually delayed, a coordination time sequence deviation warning mechanism is established to monitor the matching degree of the brake cylinder pressure decay rate and the relief valve opening progress in real time, and the pressure pre-decay rate and the relief valve opening gradient are dynamically adjusted according to the warning signal to promote the evolution trajectory of the coordination time sequence to return to the benchmark.

[0126] Step S157: Integrate the two-stage component parameter adjustment direction and the cross-stage coordination optimization suggestion according to the time axis and the evolution node of the related chain, mark the priority and implementation conditions of the adjustment, and generate the dynamic adjustment scheme of the brake performance.

[0127] Taking the time axis as the core clue, arrange all the two-stage component parameter adjustment directions such as the parameter adjustment direction of the brake hose in the middle of the stable running stage and the parameter adjustment direction of the brake cylinder in the early stage of the braking stage on the time axis in the order of corresponding running stages. At the same time, insert the cross-stage coordination optimization suggestions such as the coordination optimization suggestions of the air storage cylinder and the brake valve in the stable running-braking transition stage and the coordination optimization suggestions of the brake cylinder and the relief valve in the brake-stop transition stage into the corresponding transition stage positions on the time axis, to ensure the continuity of the two-stage adjustment and cross-stage optimization in time.

[0128] According to the evolution nodes of the brake machine running related chain, associate and mark each adjustment direction and optimization suggestion with the corresponding evolution node (such as the associated intensity peak point, the inflection point, and the stage transition starting point), and clearly indicate the specific evolution node targeted by the adjustment measures. For example, the parameter adjustment direction of the brake hose is marked with the associated intensity inflection point in the middle of the stable running stage, and the coordination optimization suggestions of the air storage cylinder and the brake valve are marked with the starting node of the stable running-braking transition stage.

[0129] According to the influence degree and urgency of the adjustment measures on the performance of the brake, the adjustment priority is divided. The adjustment measures that directly affect the brake safety (such as the parameter adjustment of the brake cylinder in the early stage of the braking phase) are set as the highest priority, the adjustment measures that improve the stability of the performance (such as the parameter adjustment of the brake hose) are set as the medium priority, and the adjustment measures that optimize the synergy efficiency (such as the synergy optimization suggestion in the transition phase) are set as the next high priority.

[0130] For each adjustment measure, the implementation conditions are specified. The implementation conditions include environmental conditions (such as the environmental temperature, humidity in the preset range), working condition conditions (such as the train running speed, load in a specific interval), and pre-adjustment completion conditions (such as the pre-sequencing adjustment of the associated components has been executed and the expected effect has been achieved). For example, the implementation conditions of the parameter adjustment of the brake cylinder in the early stage of the braking phase are set as the environmental vibration intensity being lower than the preset threshold, the train load being between sixty percent and ninety percent of the rated load, and the pre-pressurization adjustment of the air reservoir has been completed.

[0131] The integrated phased component parameter adjustment direction, cross-phase synergy optimization suggestion, and corresponding priority, implementation conditions are arranged in a structured form to form a complete brake performance dynamic adjustment scheme. The brake performance dynamic adjustment scheme is presented in the form of a document, including adjustment measure list, time schedule table, priority order table, implementation condition description table, etc. for subsequent execution and monitoring.

[0132] Figure 2 A schematic diagram of exemplary hardware and software components of the train brake performance evaluation system 100 based on artificial intelligence provided by some embodiments of the present application, which can implement the idea of the present application, is shown. For example, the processor 120 can be used on the train brake performance evaluation system 100 based on artificial intelligence, and is used to execute the functions in the present application.

[0133] The train brake performance evaluation system 100 based on artificial intelligence can be a general server or a special-purpose server, both of which can be used to implement the train brake performance evaluation method based on artificial intelligence of the present application. Although only one server is shown in the present application, for the sake of convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0134] For example, the artificial intelligence based train brake performance evaluation system 100 can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, a ROM, or a RAM, or any combination thereof. Illustratively, the artificial intelligence based train brake performance evaluation system 100 can also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The methods of the present application can be implemented according to these program instructions. The artificial intelligence based train brake performance evaluation system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0135] For ease of illustration, only one processor is described in the artificial intelligence based train brake performance evaluation system 100. However, it should be noted that the artificial intelligence based train brake performance evaluation system 100 in the present application can also include multiple processors, so the steps performed by one processor described in the present application can also be jointly performed or separately performed by multiple processors. For example, if the processor of the artificial intelligence based train brake performance evaluation system 100 performs steps A and B, it should be understood that steps A and B can also be jointly performed by two different processors or separately performed in one processor. For example, a first processor performs step A, a second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0136] In addition, the present application also provides a readable storage medium, wherein computer executable instructions are pre-set in the readable storage medium, and when a processor executes the computer executable instructions, the artificial intelligence based train brake performance evaluation method is realized.

[0137] It should be noted that, in order to simplify the expression of the present application and to help the understanding of one or more embodiments of the present application, in the foregoing description of the embodiments of the present application, various features are sometimes combined into one embodiment, figure or description thereof.

Claims

1. A train brake performance evaluation method based on artificial intelligence, characterized in that: The method comprises: Acquire a full-cycle operation-related data set of a train brake, wherein the full-cycle operation-related data set of a train brake includes operation parameter data, environmental impact data, and historical performance evolution record data generated by various components of the brake at different operation stages; Extracting dynamic association relationships between brake components from the train brake full-cycle operation association data set, and constructing a brake operation association chain. The brake operation association chain connects the association relationships of various components in series using a time axis as a clue, and presents association strength evolution characteristics as the operation stage changes; Generate a dynamic benchmark feature of brake performance based on the evolution trajectory of the brake operation association chain, wherein the dynamic benchmark feature of brake performance includes component association features of each operation stage and cross-stage collaborative evolution features; Invoking a pre-trained artificial intelligence model for dynamic brake performance evaluation, performing cross-dimensional interactive mapping processing on the real-time collected brake operation characteristics and the dynamic brake performance benchmark characteristics, and generating brake performance evolution evaluation results; Based on the brake performance evolution evaluation results and combined with the evolution trend of the associated chain, a dynamic adjustment plan for brake performance is generated. The dynamic adjustment plan for brake performance includes phased component parameter adjustment directions and cross-stage collaborative optimization suggestions.

2. The train brake performance evaluation method based on artificial intelligence according to claim 1 is characterized in that: The step of extracting dynamic association relationships between brake components from the train brake full cycle operation association data set and constructing a brake operation association chain includes: The train brake full cycle operation associated data set is divided into operation stages, and the data is divided into stage data units corresponding to each stage according to the brake start-up stage, stable operation stage, braking stage, and stop stage. Each stage data unit contains the operating parameter data, environmental impact data, and performance record data of each component in that stage; For each stage data unit, extract the parameter association relationship between any two components in the corresponding stage, record the frequency, duration and performance of the association relationship, and form a set of association relationships within the stage; Comparing the intra-stage association relationship sets of data units in adjacent stages, identifying the addition, disappearance and strength changes of association relationships, and forming inter-stage association evolution information, wherein the inter-stage association evolution information includes the association relationship change type and the change triggering condition; Using the time axis as the horizontal clue, the association relationship sets within each stage are arranged in chronological order. At the same time, the association evolution information between stages is used as the vertical connection to connect the association relationships of adjacent stages in series to form the initial brake operation association chain. Analyze the strength change curves of each association relationship in the initial brake operation association chain within the entire cycle, mark the peak points, valley points and inflection points in the strength change curve, determine the association strength evolution characteristics, optimize the structure of the initial brake operation association chain based on the characteristics, and generate the brake operation association chain.

3. The train brake performance evaluation method based on artificial intelligence according to claim 1 is characterized in that: Generating a dynamic benchmark characteristic of brake performance based on the evolution trajectory of the brake operation association chain includes: Traversing the time axis of the brake operation association chain, extracting the association relationship set and association strength evolution characteristics corresponding to each operation stage, and forming a stage association data packet; Extract the features of the association relationships in the association data packets of each stage to generate the component association features of the operation stage. The component association features reflect the core performance of the association relationships of the components in the operation stage and their adaptability to the environmental impact data of the stage. Compare and analyze the component-related features of adjacent operation stages, identify the inheritance relationship, variation relationship, and complementary relationship between features, and form inter-stage feature association information; Based on the feature correlation information between stages, a cross-stage co-evolution model is constructed. This cross-stage co-evolution model takes the component correlation features of each stage as input and outputs cross-stage co-evolution features that reflect the continuous change law of features. The cross-stage co-evolution features include the triggering factors and evolution direction of feature changes; The component-related features of all operating stages and the cross-stage co-evolution features are integrated in chronological order, and the operating stage identifiers and evolution nodes corresponding to each feature are marked to form the dynamic benchmark features of the brake performance.

4. The train brake performance evaluation method based on artificial intelligence according to claim 3 is characterized in that: The pre-trained brake performance dynamic evaluation artificial intelligence model is called to perform cross-dimensional interactive mapping processing on the real-time collected brake operation characteristics and the brake performance dynamic benchmark characteristics to generate a brake performance evolution evaluation result, including: Acquiring real-time collected brake operation characteristics, wherein the real-time collected brake operation characteristics include component association characteristics, real-time environment adaptation characteristics, and real-time stage transition characteristics in the real-time operation stage; Inputting the brake performance dynamic benchmark features and the real-time collected brake operation features into the feature interaction layer of a pre-trained brake performance dynamic evaluation artificial intelligence model; Constructing cross-dimensional mapping rules between dynamic baseline features and real-time operation features through the feature interaction layer, wherein the cross-dimensional mapping rules include stage matching rules, feature type corresponding rules, and evolution trend comparison rules; Based on the cross-dimensional mapping rules, interactive comparison is performed between the component association features of each stage in the dynamic baseline features, the cross-stage co-evolution features and the corresponding features in the real-time operation features to explore the evolutionary matching relationship and deviation association relationship between the features; Through the result generation layer of the brake performance dynamic evaluation artificial intelligence model, the evolution matching relationship and the deviation correlation relationship are integrated, the operating stages and evolution nodes corresponding to the deviations are marked, and the brake performance evolution evaluation results including component performance evolution status evaluation and cross-stage collaborative status evaluation are generated.

5. The train brake performance evaluation method based on artificial intelligence according to claim 4 is characterized in that: The cross-dimensional mapping rules between dynamic baseline features and real-time operation features are constructed through the feature interaction layer, including: Analyze the operating stage identifiers in the dynamic benchmark characteristics of the brake performance, determine the time range, environmental characteristics and core correlation relationship types of each stage, and build a stage characteristic library; Extracting real-time operation phase information from the real-time collected brake operation characteristics, comparing the real-time operation phase information with the phase characteristics in the phase characteristic library, determining a reference phase corresponding to the real-time operation characteristics, and forming a phase matching rule, wherein the phase matching rule includes determining a phase time overlap, determining an environmental feature similarity, and determining a core association relationship type matching rule; The component association features and cross-stage collaborative evolution features in the dynamic benchmark features are labeled, and the feature types are divided into parameter association type, environment adaptation type, and evolution trend type; The component association features, real-time environment adaptation features, and real-time stage transition features in the real-time operation features are labeled with types to make them consistent with the feature type classification standards of the dynamic baseline features, forming feature type correspondence rules. The feature type correspondence rules include feature type definitions, type identification basis, and feature conversion methods when types do not match; Analyze the evolution curve of the cross-stage co-evolution feature in the dynamic baseline feature, extract the change direction, change rate and change period of the evolution curve, and construct an evolution trend template; Determine the change rules of real-time stage transition characteristics in real-time operation characteristics, form a real-time evolution trend description, and construct evolution trend comparison rules. The evolution trend comparison rules include evolution direction consistency judgment, change rate deviation range setting, and change cycle coincidence judgment.

6. The train brake performance evaluation method based on artificial intelligence according to claim 4 is characterized in that: Based on the cross-dimensional mapping rules, interactive comparison is performed on the component association features of each stage in the dynamic baseline features, the cross-stage co-evolution features and the corresponding features in the real-time operation features to mine the evolutionary matching relationship and deviation association relationship between the features, including: According to the stage matching rules, the real-time operation features are grouped with the dynamic baseline features of the corresponding baseline stage to form feature comparison groups. Each feature comparison group contains the component association features of the baseline stage, the cross-stage co-evolution features, and the corresponding features in the real-time operation features. Based on the feature type correspondence rules, within each feature comparison group, the parameter-related features of the dynamic baseline features are matched with the parameter-related features of the real-time operation features, the environment adaptation features are matched with the environment adaptation features, and the evolution trend features are matched with the evolution trend features; Attribute comparison is performed on features of corresponding types. Parameter-related features compare related component pairs and related performances. Environment-adaptive features compare the adaptation performance of environmental factors and related relationships. Evolution-trend features compare change directions and change nodes. If the attribute comparison result of the corresponding type feature meets the preset matching threshold range, the feature relationship of the feature comparison group is marked as an evolutionary matching relationship, and the matching feature attributes and matching degree description are recorded; If the attribute comparison result of the corresponding type of feature exceeds the preset matching threshold range, the stage position, feature type and associated evolution node where the deviation occurs are analyzed to determine the impact range and possible triggering factors of the deviation. The feature relationship of the feature comparison group is marked as a deviation association relationship, and the deviation attribute, deviation degree and deviation triggering factor description are recorded; Traverse all feature comparison groups, summarize the evolutionary matching relationships and deviation correlation relationships, and form a complete feature interaction comparison result set.

7. The train brake performance evaluation method based on artificial intelligence according to claim 4 is characterized in that: The generating of a brake performance dynamic adjustment plan based on the brake performance evolution evaluation result and the associated chain evolution trend includes: Analyzing the component performance evolution status evaluation in the brake performance evolution evaluation result, identifying the components whose performance evolution deviates from the baseline trend and the corresponding deviation stages, and forming a staged list of components to be adjusted; For each component and corresponding stage in the staged list of components to be adjusted, analyzing the reason why the performance evolution of the component deviates from the baseline trend, the reason being determined based on the deviation correlation relationship between the component's associated features and the baseline features at that stage and the evolution trend of the associated chain; Based on the reasons, combined with the operating parameter adaptation range of the component in the corresponding stage and its intra-stage association with other components, the parameter adjustment direction of the component in the corresponding stage is determined, forming a phased component parameter adjustment direction, which includes the phase timing of parameter adjustment, adjustment trend, and adaptation requirements with the phase environment; Analyzing the cross-stage collaborative state evaluation in the brake performance evolution evaluation results, identifying the collaborative links and corresponding transition stages where the cross-stage collaborative evolution deviates from the baseline trend, and forming a list of cross-stage collaborative links to be optimized; For each collaborative link and corresponding transition stage in the list of cross-stage collaborative links to be optimized, analyzing the reason why the collaborative evolution deviates from the baseline trend, wherein the reason is determined based on the deviation correlation relationship between the cross-stage collaborative evolution characteristics and the baseline characteristics and the evolution trend of the correlation chain; Based on the reasons, combined with the phased component parameter adjustment direction and cross-phase association relationships, formulate cross-phase collaborative optimization suggestions for this collaborative link. The cross-phase collaborative optimization suggestions include the parameter collaborative adjustment method in the transition phase, the coordination adjustment requirements of the associated components, and the direction of the evolution trend correction; The phased component parameter adjustment directions and cross-phase collaborative optimization suggestions are integrated according to the timeline and associated chain evolution nodes, the adjustment priority and implementation conditions are marked, and a dynamic adjustment plan for brake performance is generated.

8. The train brake performance evaluation method based on artificial intelligence according to claim 7 is characterized in that: The analyzing, for each component and corresponding stage in the list of components to be adjusted in stages, the reason why the performance evolution of the component deviates from the baseline trend includes: Extracting component association features of the components to be adjusted in the corresponding stages, and simultaneously extracting reference component association features of the stage in the dynamic reference features of brake performance; Compare the component-related features of the component to be adjusted with the reference component-related features, combine the deviation association relationship, determine the feature attributes with deviations, and form a deviation attribute list; Analyze each deviation attribute in the deviation attribute list, and combine it with the environmental impact data of this stage to determine whether the environmental factor causes the deviation attribute. If the environmental factor is related to the deviation attribute, then consider the environmental adaptation deviation as one of the causes; Retrieve the evolution characteristics of the association relationships before and after the current stage in the brake operation association chain, analyze the impact of the association relationships in the previous stage on the association characteristics of the components to be adjusted in the current stage, and determine whether the current deviation is caused by abnormal association evolution in the previous stage. If there is an impact, the deviation caused by the evolutionary conduction in the previous stage is considered as one of the causes; Analyze the relationship between the component to be adjusted and other components at this stage to determine whether the deviation of the associated characteristics of the component to be adjusted is caused by abnormal performance of other components. If so, the deviation of the associated components is considered as one of the reasons; By comprehensively considering the environmental adaptation deviation, the evolutionary conduction deviation in the previous stage, the deviation of the impact of related components and other deviation types, and combining the amplification or suppression effect of the evolutionary trend of the associated chain on the deviation, a complete description of the reasons why the performance evolution of the component in the corresponding stage deviates from the baseline trend is formed.

9. The train brake performance evaluation method based on artificial intelligence according to claim 7 is characterized in that: The step of determining the parameter adjustment direction of the component in the corresponding stage based on the reasons and combining the operating parameter adaptation range of the component in the corresponding stage and the correlation relationship with other components in the stage, thereby forming a phased component parameter adjustment direction, includes: Obtaining an operating parameter adaptation range of the component in the corresponding stage, wherein the operating parameter adaptation range includes an allowable fluctuation range of each operating parameter of the component in the stage and an adaptation range with environmental characteristics; For each reason why the component performance evolution deviates from the baseline trend, determine the corresponding parameter type that needs to be adjusted, where the parameter types that need to be adjusted include environment adaptation parameters, correlation and coordination parameters, and stage transition parameters; Analyze the difference between the current parameter performance of the parameter type to be adjusted and the baseline parameter performance. Combined with the operating parameter adaptation range of this stage, preliminarily determine the parameter adjustment trend. If the current parameter performance exceeds the upper limit of the adaptation range, the preliminary adjustment trend is to reduce the parameter. If it is below the lower limit of the adaptation range, the preliminary adjustment trend is to increase the parameter. Analyze the relationship between this component and other components at this stage to determine whether the initially determined parameter adjustment trend will affect the parameter performance of other components beyond their adaptation range. If there is an impact, analyze the parameter sensitivity of the associated components and adjust the amplitude of the initial adjustment trend. Combined with the environmental characteristics of this stage, determine whether the adjusted parameter trend is compatible with the environmental characteristics. If not, further revise the parameter adjustment trend so that the adjusted parameters can adapt to the environment of this stage. Mark the stage timing of parameter adjustment, the gradual requirements of the adjustment range, and the coordinated adjustment sequence with related components to form the phased component parameter adjustment direction.

10. A train brake performance evaluation system based on artificial intelligence, characterized in that: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the artificial intelligence-based train brake performance evaluation method described in any one of claims 1 to 9.

Images