Rail train sanding system health state prediction and management system and method based on multi-source information fusion

Abstract

The invention relates to a rail train sanding system health state prediction and management system and method based on multi-source information fusion, and relates to the technical field of rail transit vehicle operation and maintenance and state monitoring. The technical problem that the known, evaluable and predictable refined operation and maintenance requirements of modern intelligent rail transit on a key system cannot be met in the prior art is solved. The invention discloses a health state prediction and management system for a rail train sanding system. The health state prediction and management system comprises a data acquisition and sensing layer, a data fusion and processing layer and an intelligent analysis and application layer which are connected in sequence. According to the system and the method disclosed by the invention, the technical spanning from'no monitoring 'to'full perception' is realized; revolutionary transformation of a maintenance mode is promoted; an innovative health state evaluation index system is constructed; the depth data value is mined, and multi-department collaboration is enabled; the system is high in compatibility and low in implementation cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rail transit vehicle operation and state monitoring, in particular to a health state prediction and management system and method for a rail train sanding system based on multi-source information fusion. BACKGROUND

[0002] The sanding system is a key auxiliary system of the rail train, which effectively improves the adhesion coefficient between the wheel and the rail by spraying quartz sand on the rail surface, prevents the wheel from idling or sliding during traction or braking, and is directly related to the running safety and efficiency of the train. However, the management and maintenance of the sanding system is in a state of extreme extensive and passive, and there are the following significant technical gaps and defects:

[0003] (1) State unknown: the existing system only has the simplest low sand level alarm function, and cannot perform any quantitative monitoring and recording on the sanding amount, sanding frequency, and sanding behavior under different modes (manual / automatic). The operation department has no knowledge of sand consumption, and cannot perform fine management and cost accounting.

[0004] (2) Outdated maintenance strategy: maintenance work is completely based on fixed cycles (such as replacing sand every 30,000 kilometers) or post-fault processing (such as dredging after sand pipe blockage). This "time-based maintenance" mode is extremely uneconomical, which may lead to premature replacement of sand before it is exhausted, or safety hazards due to unexpected depletion of sand and potential blockage of the pipeline.

[0005] (3) Information isolation and lack of intelligent analysis: the sanding system operates as an independent unit, and its data is completely isolated from other systems of the train (such as traction, braking, positioning, and fault diagnosis systems). It is impossible to correlate and analyze the sanding behavior with specific operating scenarios (such as line sections, slopes, weather conditions, and driver operation habits), so it is impossible to evaluate the adhesion utilization efficiency and provide data support for optimizing driving strategies or line maintenance.

[0006] (4) Relies on human experience: sand level inspection, sand quality judgment (whether it is damp and hardened), and other work highly depend on visual inspection and experience judgment by on-board mechanics or ground maintenance personnel, lack objective and quantitative evaluation methods, and are inefficient and unreliable.

[0007] Therefore, the existing technology cannot meet the fine operation and maintenance requirements of modern intelligent rail transit for key systems, and an innovative management method is urgently needed. SUMMARY

[0008] The present application aims to solve the technical problem that the prior art cannot meet the fine operation and maintenance requirements of modern intelligent rail transit on the key system knowable, evaluable and predictable, and provides a health state prediction and management system and method for a rail train sanding system based on multi-source information fusion.

[0009] To solve the above technical problems, the technical scheme of the present application is as follows:

[0010] A health state prediction and management system for a rail train sanding system based on multi-source information fusion, comprising, in sequence, a data acquisition and perception layer, a data fusion and processing layer, and an intelligent analysis and application layer.

[0011] The data acquisition and perception layer is provided with a data acquisition and analysis module; the data acquisition and analysis module is used to collect raw data from the vehicle MVB / CAN bus and identify the source of the sanding instruction;

[0012] The data fusion and processing layer is provided with a multi-source information fusion module and a space-time database; the multi-source information fusion module is used to associate and synchronize the sanding instruction with the mileage, GPS positioning and time information; and the space-time database is used to store the associated raw data and the calculated index data.

[0013] The intelligent analysis and application layer is provided with a health state prediction engine, a maintenance decision and work order management production module, and a visual display and decision support platform; the health state prediction engine is built-in with a prediction algorithm model and is used to perform sand consumption prediction and residual life calculation; the maintenance decision and work order management production module is used to automatically generate maintenance suggestions and work orders according to the prediction results; and the visual display and decision support platform is used to graphically display the system state, analysis results and alarm information to the user.

[0014] A health state prediction and management method for a rail train sanding system suitable for the health state prediction and management system for a rail train sanding system based on multi-source information fusion, comprising the following steps:

[0015] Step S1: multi-source data acquisition and instruction identification;

[0016] The data acquisition and analysis module is used to collect vehicle network bus data in real time, analyze and accurately distinguish the source type of the sanding instruction, and synchronously collect the real-time mileage information, GPS positioning information and time information of the train;

[0017] Step S2: data association and fusion;

[0018] The multi-source information fusion module is applied to label each sanding record with a source tag, and synchronize and associate the sanding record with corresponding mileage, GPS positioning and time information to form a raw data set with rich context information; and the raw data set is stored in a space-time database.

[0019] Step S3: health index calculation and feature extraction;

[0020] The health state prediction engine is applied to calculate a series of health and performance evaluation indexes based on the raw data set; the health and performance evaluation indexes support multi-dimensional filtering and aggregation analysis according to line, season and weather condition;

[0021] Step S4: health state prediction and decision generation;

[0022] The health state prediction engine is applied to build a time series prediction model based on a historical sand consumption data sequence, to predict sand consumption and residual sand life in a future specific period; and when the prediction value is lower than a preset threshold, a maintenance decision and work order management production module is applied to automatically generate a maintenance work order.

[0023] Step S5: data visualization and decision support;

[0024] The visualization display and decision support platform is applied to provide a man-machine interactive interface, display health indexes, prediction results, alarm information and multi-dimensional analysis and comparison results, and provide support for operation and maintenance decision.

[0025] In the above technical solution, the health and performance evaluation indexes in step S3 include: sand consumption per 10,000 kilometers.

[0026] In the above technical solution, step S1 is specifically: a data acquisition and analysis module continuously monitors a vehicle bus; when a sanding electromagnetic valve driving signal is detected, it is accurately judged according to the signal characteristics whether the sanding is from driver manual operation, anti-slip system triggering or anti-slip system triggering; at the same time, accurate GPS coordinates, mileage information and time stamp at this moment are obtained from the train TCMS system.

[0027] In the above technical solution, step S2 is specifically: the multi-source information fusion module binds each sanding event with the context information at the time of occurrence; the system periodically aggregates data to calculate core indexes of each sand box or each train; all data are stored in a space-time database.

[0028] In the above technical solution, step S3 is specifically: the health state prediction engine reads historical sand consumption rate time series data of a specific vehicle; an autoregressive integrated moving average model or a long short-term memory network is used for modeling to predict sand consumption in a future period or a future mileage.

[0029] In the above technical solution, step S4 is specifically: the health state prediction engine combines the current sand level or the last sanding amount to calculate the remaining service life; when the predicted value is lower than the safety threshold, the maintenance decision and work order management production module automatically generates a work order suggesting sanding in the maintenance management system and pushes it to the visual display and decision support platform.

[0030] In the above technical solution, step S5 is specifically: through the visual display and decision support platform, it is determined whether the triggering frequency and sand consumption of the anti-slip sanding of a certain vehicle model in a specific long and steep slope section are significantly higher than those in other sections, prompting the maintenance department to check the track surface condition of the section or prompting the maintenance department to optimize the operation guide of the section to reduce sand consumption and wheel-rail wear from the source.

[0031] The present application has the following beneficial effects:

[0032] The track train sanding system health state prediction and management system and method based on multi-source information fusion of the present application have the following advantages:

[0033] (1) It realizes the technical leap from "no monitoring" to "full perception": for the first time, it realizes the fine, quantitative and traceable full perception of the working state of the sanding system, fills the data gap in this field, and lays a solid foundation for intelligent management.

[0034] (2) It promotes a revolutionary change in the maintenance mode: through the data-driven prediction model, it realizes the change from "periodic maintenance" to "predictive maintenance" of the maintenance work, realizes accurate prediction and on-demand triggering, and significantly improves the economy and safety.

[0035] (3) It builds an innovative health state evaluation index system: the first "sand consumption per million kilometers" and a series of other indicators provide a unified, standardized quantitative scale and scientific basis for evaluating the performance of the sanding system and even the adhesion utilization efficiency of the whole train.

[0036] (4) It excavates the value of deep data and empowers multi-department cooperation: the data generated by the present application has a value far beyond the sanding system itself. It provides unprecedented data insight and decision support for the maintenance department to optimize driving strategies, the maintenance department to evaluate line conditions, and the material department to accurately purchase sand, realizing cross-professional collaborative value.

[0037] (5) The system has strong compatibility and low implementation cost: the core data comes from the existing network of the vehicle, only software functions and data analysis logic need to be added, and the existing vehicle is friendly to modification, easy to deploy on a large scale, and has a high return on investment. BRIEF DESCRIPTION OF DRAWINGS

[0038] The present application will be further described in detail below in combination with the drawings and specific embodiments.

[0039] Figure 1 The logical architecture and data flow diagram of the track train sanding system health state prediction and management system based on multi-source information fusion of the application. DETAILED DESCRIPTION

[0040] The application idea of the application is:

[0041] The track train sanding system health state prediction and management method based on multi-source information fusion of the application realizes predictive health management (PHM) of the sanding system based on multi-source information fusion and big data analysis technology.

[0042] The track train sanding system health state prediction and management method based on multi-source information fusion of the application is used to realize intelligent management of the track train (including high-speed motor train unit, subway vehicle, locomotive, etc.) sanding system from perception, evaluation to prediction in the whole life cycle.

[0043] The application will be described in detail below with reference to the accompanying drawings.

[0044] As shown in the figure, the track train sanding system health state prediction and management system based on multi-source information fusion of the application comprises, which are connected in sequence: a data acquisition and perception layer, a data fusion and processing layer, and an intelligent analysis and application layer. Figure 1

[0045] The data acquisition and perception layer is provided with a data acquisition and analysis module; the data acquisition and analysis module is used to collect original data from the vehicle MVB / CAN bus and analyze and identify the source of the sanding instruction.

[0046] The data fusion and processing layer is provided with a multi-source information fusion module and a space-time database; the multi-source information fusion module is used to associate and synchronize the sanding instruction with the mileage, GPS positioning and time information; the space-time database is used to store the associated original data and the calculated index data; the multi-source information fusion module is provided with a feature extraction and index calculation engine, which is used to calculate the sand consumption per 10,000 kilometers, the proportion of the instruction source, and the section sand consumption intensity.

[0047] The intelligent analysis and application layer is provided with a health state prediction engine, a maintenance decision and work order management production module, and a visual display and decision support platform; the health state prediction engine is built-in with a prediction algorithm model, which is used to perform sand consumption prediction and residual life calculation; the maintenance decision and work order management production module is used to automatically generate maintenance suggestions and work orders according to the prediction results; the visual display and decision support platform is used to graphically display the system state, analysis results and alarm information to the user.

[0048] ​The track train sanding system health state prediction and management method suitable for the track train sanding system health state prediction and management system based on multi-source information fusion of the application comprises the following steps:

[0049] Step S1: multi-source data acquisition and instruction recognition;

[0050] The application data acquisition and analysis module acquires vehicle network bus data in real time, analyzes and accurately distinguishes the source type of the sanding instruction, and synchronously acquires real-time mileage information, GPS positioning information and time information of the train;

[0051] Step S2: data correlation and fusion;

[0052] The multi-source information fusion module applies a source label to each sanding record, and synchronously correlates and associates the source label with corresponding mileage, position and time information with high-precision time stamp, to form an original data set with rich context information; the original data set is stored in a space-time database;

[0053] Step S3: health index calculation and feature extraction;

[0054] The health state prediction engine calculates a series of health and performance evaluation indexes based on the original data set, wherein the core health and performance evaluation index is "ton-kilometer sand consumption"; the health and performance evaluation indexes support multi-dimensional filtering and aggregation analysis according to lines, seasons and weather conditions;

[0055] Step S4: health state prediction and decision generation;

[0056] The health state prediction engine constructs a time series prediction model based on a historical sand consumption data sequence, predicts the sand consumption and remaining sand life in a future specific period, and automatically generates a maintenance work order when the prediction value is lower than a preset threshold.

[0057] Step S5: data visualization and decision support;

[0058] The visualization display and decision support platform provides a man-machine interactive interface for displaying health indexes, prediction results, alarm information and multi-dimensional analysis and comparison results, and provides support for operation and maintenance decisions.

[0059] The track train sanding system health state prediction and management method suitable for the track train sanding system health state prediction and management system based on multi-source information fusion of the application will be described in detail below with reference to the accompanying drawings, such as Figure 1 The track train sanding system health state prediction and management method comprises the following steps:

[0060] Step S1: multi-source data acquisition and instruction recognition, data acquisition;

[0061] The data acquisition and analysis module collects vehicle network bus data in real time, analyzes and accurately distinguishes the source type of the sanding instruction; synchronously collects real-time mileage information, GPS positioning information and time information of the train. The data acquisition and analysis module continuously monitors the vehicle bus; when detecting the sanding electromagnetic valve driving signal, according to the signal characteristics (such as instruction address, message ID), it accurately judges whether this sanding is from the driver's manual operation, the anti-slip system trigger or the anti-slip system trigger; at the same time, the accurate GPS coordinates, mileage information and time stamp at this moment are obtained from the train TCMS system.

[0062] Step S2: data association and fusion, data fusion and index calculation;

[0063] The multi-source information fusion module labels each sanding record with a source label, and synchronously associates it with the corresponding mileage, location and time information with high-precision time stamp, forming a raw data set with rich context information. The multi-source information fusion module binds each sanding event with the context information (location, mileage, time) at the time of occurrence. The system aggregates data regularly (such as daily) and calculates the core index of each sand box or each train - "sand consumption per million kilometers" (total sand consumption / total mileage*10000). At the same time, other indexes such as "automatic sanding proportion", "sand consumption of certain line interval" can be calculated. All data is stored in a space-time database.

[0064] Steps S3-S4: health index calculation and feature extraction, health state prediction and decision generation;

[0065] Step S3 is: the health state prediction engine calculates a series of health and performance evaluation indexes based on the raw data set, and the core health and performance evaluation index is "sand consumption per million kilometers"; the health and performance evaluation index supports multi-dimensional filtering and aggregation analysis according to line, season and weather conditions;

[0066] Step S4 is: the health state prediction engine reads the historical sand consumption data sequence data of a specific vehicle, builds a time series prediction model based on the historical sand consumption data sequence, predicts the sand consumption and remaining sand life in a specific future period; when the predicted value is lower than the preset threshold, the maintenance decision and work order management production module automatically generates a maintenance work order.

[0067] Step S3 and S4 are specifically: using ARIMA (autoregressive integrated moving average model) or LSTM (long short-term memory network) algorithm for modeling, predicting the sand consumption in a certain period of time (such as 7 days) or a certain mileage (such as 3000 kilometers). Combined with the current sand level (if any) or the last sanding amount, the remaining service life is calculated. When the predicted value is lower than the safety threshold, the maintenance decision and work order management production module automatically generates a "suggestion to add sand" work order in the maintenance management system, and pushes it to the relevant personnel's visual display and decision support platform through the interface.

[0068] Step S5: data visualization and decision support, advanced analysis application;

[0069] The application visual display and decision support platform provides a human-computer interaction interface for displaying health indicators, prediction results, alarm information and multi-dimensional analysis comparison results, and provides support for operation and maintenance decision-making.

[0070] For example, through the visual display and decision support platform, the management personnel can find that the triggering frequency and sand consumption of the anti-slip sanding of a certain vehicle model in a certain long and steep slope section are significantly higher than those in other sections. This insight can prompt the maintenance department to check the track surface condition of this section, or prompt the maintenance department to optimize the operation guide of this section, to reduce sand consumption and wheel-rail wear from the source.

[0071] Obviously, the above embodiments are only examples for clear illustration, and are not limitations on the embodiments. Based on the above description, other different forms of changes or variations can be made by those of ordinary skill in the art. Here, it is not necessary and impossible to exhaust all the embodiments. The obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. A health state prediction and management system for a rail train sanding system based on multi-source information fusion, characterized in that, Comprise sequentially connected: data acquisition and perception layer, data fusion and processing layer, intelligent analysis and application layer; The data acquisition and perception layer is provided with a data acquisition and analysis module; the data acquisition and analysis module is used for collecting original data from the vehicle MVB / CAN bus and identifying the source of the sanding instruction; The data fusion and processing layer is provided with a multi-source information fusion module and a space-time database; the multi-source information fusion module is used for associating and synchronizing the sanding instruction with mileage, GPS positioning and time information; the space-time database is used for storing the associated original data and calculated index data; The intelligent analysis and application layer is provided with a health state prediction engine, a maintenance decision and work order management production module and a visual display and decision support platform; the health state prediction engine is internally provided with a prediction algorithm model, which is used for performing sand consumption prediction and residual life calculation; the maintenance decision and work order management production module is used for automatically generating maintenance suggestions and work orders according to the prediction results; the visual display and decision support platform is used for graphically displaying system state, analysis results and alarm information to the user.

2. A track train sanding system health state prediction and management method suitable for the track train sanding system health state prediction and management system based on multi-source information fusion of claim 1, characterized in that, Comprise the following steps: Step S1: multi-source data acquisition and instruction identification; Apply the data acquisition and analysis module to collect vehicle network bus data in real time, analyze and accurately distinguish the source type of the sanding instruction; simultaneously collect real-time mileage information, GPS positioning information and time information of the train; Step S2: data association and fusion; Apply the multi-source information fusion module to label each sanding record with a source label, and synchronize and associate it with the corresponding mileage, GPS positioning and time information with high-precision timestamp, forming an original data set with rich context information; the original data set is stored in the space-time database; Step S3: health index calculation and feature extraction; Apply the health state prediction engine to calculate a series of health and performance evaluation indexes based on the original data set; the health and performance evaluation indexes support multi-dimensional filtering and aggregation analysis according to line, season and weather conditions; Step S4: health state prediction and decision generation; Apply the health state prediction engine to build a time series prediction model based on historical sand consumption data sequences, predict the sand consumption and residual sand life in a specific future period; When the prediction value is lower than the preset threshold, apply the maintenance decision and work order management production module to automatically generate a maintenance work order; Step S5: data visualization and decision support; Apply the visual display and decision support platform to provide a human-computer interaction interface, display health indicators, prediction results, alarm information and multi-dimensional analysis comparison results, and provide support for operation and maintenance decision.

3. The railcar health state prediction and management method of claim 2, wherein, The health and performance evaluation indexes in step S3 include: sand consumption per 10,000 kilometers.

4. The railcar health state prediction and management method of claim 2, wherein, Step S1 is specifically: the data acquisition and analysis module continuously monitors the vehicle bus; when a sanding electromagnetic valve driving signal is detected, it is accurately judged whether this sanding is from driver manual operation, anti-idling system triggering or anti-skid system triggering according to the signal characteristics; at the same time, the accurate GPS coordinates, mileage information and timestamp at this moment are obtained from the train TCMS system.

5. The railcar health prognosis and management method of claim 2, wherein, Step S2 is specifically: the multi-source information fusion module binds each sanding event with the context information at the time of occurrence; the system periodically aggregates data and calculates the core indicators of each sand box or each train; all data are stored in the space-time database.

6. The railcar health prognosis and management method of claim 2, wherein, Step S3 is specifically: the health state prediction engine reads the historical sand consumption rate time series data of a specific vehicle; an autoregressive integrated moving average model or a long short-term memory network is used for modeling to predict the sand consumption in a future period of time or a certain mileage.

7. The railcar health state prediction and management method of sanding system of claim 2, wherein, Step S4 is specifically: the health state prediction engine calculates the remaining service life in combination with the current sand level or the last sanding amount. When the predicted value is lower than the safety threshold, the maintenance decision and work order management production module automatically generates a work order for sanding in the maintenance management system and pushes it to the visual display and decision support platform.

8. The railcar health state prediction and management method of sanding system of claim 2, wherein, Step S5 is specifically: through the visual display and decision support platform, it is determined whether the triggering frequency and sand consumption of anti-spin sanding of a certain vehicle model in a specific long and steep slope section are significantly higher than those in other sections, prompting the road department to check the track surface condition of the section or prompting the railway department to optimize the operation guide of the section to reduce sand consumption and wheel-rail wear from the source.

Images