Method and platform for predicting service life of running mechanism of rail grinding wagon based on multi-index fusion

Abstract

The invention discloses a multi-index fusion steel rail grinding wagon running mechanism service life prediction method, which comprises the following working steps: S1, information acquisition: acquiring running state data of key parts of a running mechanism, including a lubrication state, an abrasion condition and use time; s2, constructing a model: constructing a dichotomy model by adopting a deep neural network, predicting the health state of the key part, and outputting a health state classification result; and S3, information analysis: designing a signal fusion algorithm, training a deep learning model based on the health state information of the plurality of parts, evaluating the overall health state of the walking mechanism, and classifying the health state as one of'fault 'or'non-fault'. According to the method, the deep neural network is used for modeling the part data, so that the prediction precision is improved; intelligent prediction and life stage division effectively reduce unnecessary maintenance and replacement, prolong the service life of equipment and reduce the total maintenance cost.

Description

Technical Field

[0001] The present invention relates to the technical field of rail grinding vehicle running mechanisms, and in particular to a multi-index integrated life prediction method and platform for rail grinding vehicle running mechanisms. Background Art

[0002] The running mechanism of a rail grinding vehicle is a critical system for vehicle operation, and its health directly impacts the vehicle's safety and operating efficiency. Currently, running mechanism maintenance relies primarily on regular inspections, a method that is inefficient and costly, and also lacks accurate prediction and maintenance capabilities for critical components.

[0003] With increasing railway traffic and the long-term, high-intensity operation of equipment, running mechanisms face a greater risk of wear and damage. Therefore, accurately assessing their health status and predicting their remaining lifespan have become critical to ensuring the safe operation of railway equipment and improving equipment utilization. However, existing running mechanism health monitoring methods rely primarily on manual inspections or single-sensor data, failing to obtain comprehensive and real-time information on equipment health, making it difficult to accurately predict equipment failures and remaining service life.

[0004] As mentioned above, we have designed a multi-index fusion rail grinding vehicle running mechanism life prediction method and platform to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a multi-index fusion rail grinding vehicle running mechanism life prediction method and platform.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A multi-index integrated life prediction method for the running mechanism of a rail grinding vehicle includes the following steps:

[0008] Step S1, information collection: collecting operating status data of key components of the running mechanism, including lubrication status, wear condition, and usage time;

[0009] Step S2: Model building: A binary classification model is constructed using a deep neural network to predict the health status of key components and output the health status classification results;

[0010] Step S3, information analysis: Design a signal fusion algorithm, train a deep learning model based on the health status information of multiple components, evaluate the overall health status of the running mechanism, and classify it as "faulty" or "not faulty";

[0011] Step S4, information processing: For the case where the overall health status is "fault" in step S3, a fault alarm message is output; for the case where the overall health status is "not faulty", the running mechanism is divided into multiple life stages based on historical data and fuzzy rule sets, and the current life stage is determined based on real-time data.

[0012] Preferably, the data collection uses multiple sensors, including vibration sensors, temperature sensors and displacement sensors, to monitor the status of key components.

[0013] Preferably, the deep neural network uses a convolutional neural network (CNN) or a long short-term memory network (LSTM) to model component data to improve prediction accuracy.

[0014] Preferably, the signal fusion algorithm adopts a weighted fusion method or an adaptive fusion method to integrate the status information of different components to optimize the overall health status assessment.

[0015] Preferably, the life stage prediction is based on a fuzzy rule set, which divides the state of the running mechanism into multiple life stages and performs dynamic adjustments based on real-time collected data.

[0016] A platform for predicting the life of the running mechanism of a rail grinding vehicle by integrating multiple indicators, comprising:

[0017] Data acquisition module, used to collect running status data of the running mechanism;

[0018] Data preprocessing module for data normalization and feature extraction;

[0019] Component failure prediction module, used to predict the health status of key components;

[0020] Overall health status assessment module, used to assess the overall health status of the running mechanism;

[0021] Life prediction module, used to estimate the remaining service life based on fuzzy rules and historical data;

[0022] Visualization and alarm module, used to provide data display, real-time monitoring and fault warning.

[0023] Preferably, the data preprocessing module performs signal denoising using one of wavelet transform or Fourier transform, and performs feature dimensionality reduction using principal component analysis (PCA) or autoencoder.

[0024] Preferably, the life prediction module dynamically adjusts the life stage division based on fuzzy rules and full life cycle data to improve the accuracy of life prediction; the visualization and alarm module provides real-time status monitoring, trend analysis, anomaly detection and automatic alarm functions.

[0025] Compared with the existing technology, the present invention has the following advantages: by collecting multi-dimensional operating data of the running mechanism in real time, combined with deep learning algorithms and fuzzy rules, it can achieve high-precision health status prediction and lifespan estimation. Compared with the existing technology, the present invention has the following advantages:

[0026] High accuracy: Utilizing deep neural networks to model component data improves prediction accuracy.

[0027] Real-time: Real-time monitoring and dynamic adjustment of life stages to provide timely feedback on equipment status.

[0028] Enhanced safety: Through timely fault warning and life prediction, sudden equipment failures are avoided and the safety and reliability of the equipment are improved.

[0029] Reduce maintenance costs: Intelligent prediction and life stage division effectively reduce unnecessary maintenance and replacement, extend the service life of equipment, and reduce overall maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a platform framework diagram of a multi-index fusion rail grinding vehicle running mechanism life prediction method and platform proposed by the present invention. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0032] Reference Figure 1 A multi-index integrated life prediction method for the running mechanism of a rail grinding vehicle includes the following steps:

[0033] Step S1, information collection: collecting operating status data of key components of the running mechanism, including lubrication status, wear status, and usage time. Multiple sensors, including vibration sensors, temperature sensors, and displacement sensors, are used to monitor the status of key components.

[0034] Step S2: Model building: A binary classification model is constructed using a deep neural network to predict the health status of key components and output the health status classification results. The deep neural network uses a convolutional neural network (CNN) or a long short-term memory network (LSTM) to model component data to improve prediction accuracy.

[0035] Step S3: Information Analysis: Design a signal fusion algorithm, train a deep learning model based on the health status information of multiple components, and evaluate the overall health status of the running mechanism, classifying it as "faulty" or "not faulty." The signal fusion algorithm uses a weighted fusion method or an adaptive fusion method to integrate the status information of different components to optimize the overall health status assessment.

[0036] Step S4, information processing: For the case where the overall health status is "fault" in step S3, a fault alarm message is output; for the case where the overall health status is "not faulty", the running mechanism is divided into multiple life stages based on historical data and a fuzzy rule set, and the current life stage is determined based on real-time data. The life stage prediction is based on a fuzzy rule set, which divides the state of the running mechanism into multiple life stages and dynamically adjusts it based on real-time collected data.

[0037] This embodiment also relates to a platform for a multi-index integrated rail grinding vehicle running mechanism life prediction method, which includes:

[0038] Data acquisition module, used to collect running status data of the running mechanism;

[0039] Data preprocessing module for data normalization and feature extraction;

[0040] Component failure prediction module, used to predict the health status of key components;

[0041] Overall health status assessment module, used to assess the overall health status of the running mechanism;

[0042] Life prediction module, used to estimate the remaining service life based on fuzzy rules and historical data;

[0043] Visualization and alarm module, used to provide data display, real-time monitoring and fault warning.

[0044] More specifically, the data preprocessing module uses wavelet transform or Fourier transform to perform signal denoising, and performs feature dimensionality reduction through principal component analysis (PCA) or autoencoder; the component failure prediction module uses convolutional neural network (CNN) or long short-term memory network (LSTM) to model time series data to improve prediction accuracy; the signal fusion algorithm uses weighted fusion method or adaptive fusion method to integrate the status information of different components to optimize the overall health status assessment; the life prediction module dynamically adjusts the life stage division based on fuzzy rules and full life cycle data to improve life prediction accuracy; the visualization and alarm module provides real-time status monitoring, trend analysis, anomaly detection and automatic alarm functions.

[0045] Combined with the above-mentioned multi-index fusion rail grinding vehicle running mechanism life prediction method and platform, the following steps are implemented:

[0046] Step S1: Information Collection

[0047] Multiple sensors (including vibration sensors, temperature sensors, acceleration sensors, and displacement sensors) collect real-time operating data from key components of the running mechanism. This data includes, but is not limited to, component lubrication status, wear, temperature changes, vibration frequency, and displacement deviation. This data reflects the running status of the running mechanism and provides a foundation for subsequent health status prediction and lifespan assessment.

[0048] Step S2: Build a health status prediction model

[0049] A binary classification model is constructed using a deep neural network (DNN), specifically a convolutional neural network (CNN) or a long short-term memory (LSTM) network. The trained model is fed with collected data on key components, predicting the health status of each component and outputting a "healthy" or "faulty" classification. This model effectively identifies key components in different states and provides data support for overall health assessment.

[0050] Step S3: Information analysis and health status assessment

[0051] Based on the component health status information obtained in step S2, a signal fusion algorithm is used for comprehensive analysis. This signal fusion algorithm can be a weighted or adaptive fusion method, fusing the health status data of different components to assess the overall health status of the running mechanism. Ultimately, the model outputs the overall health status as "faulty" or "not faulty."

[0052] Step S4: Life cycle stage prediction and dynamic adjustment

[0053] When the overall health status is "faulty," the system immediately outputs a fault alarm, indicating the need for maintenance or repair. If the overall health status is "not faulty," the running mechanism's lifespan is divided into multiple stages (e.g., early, mid, and late) using fuzzy rules and historical data. Based on real-time data and historical trends, the running mechanism's current lifespan stage is dynamically determined and the lifespan divisions are adjusted.

[0054] Data collection and processing

[0055] The data acquisition module includes vibration sensors, temperature sensors, acceleration sensors, and displacement sensors, enabling real-time monitoring of the operating status of key components of the running mechanism. The collected data is normalized, feature extracted, and denoised (e.g., using wavelet transforms). Principal component analysis (PCA) or an autoencoder is then used to reduce feature dimensionality, thereby reducing redundant data and improving computational efficiency.

[0056] Signal fusion and health status assessment

[0057] A comprehensive analysis of the health status data of different components is performed using either a weighted fusion method or an adaptive fusion method. The weighted fusion method weights the health status data of each component based on its importance, while the adaptive fusion method adjusts the weight of each component's status information based on real-time monitoring data to optimize the accuracy of the overall health status assessment.

[0058] Lifespan Prediction and Fuzzy Rules

[0059] The life prediction module dynamically predicts the life of the running mechanism based on a fuzzy rule set. The fuzzy rule set models the characteristics of each running mechanism life stage and dynamically adjusts the system based on real-time and historical data to accurately predict the remaining useful life.

[0060] Preferred embodiments

[0061] Data collection: Through vibration sensors, temperature sensors, acceleration sensors and displacement sensors, the running data of the running mechanism is comprehensively collected.

[0062] Deep Neural Networks: Use convolutional neural networks (CNNs) or long short-term memory networks (LSTMs) to model time series data and accurately predict the health status of each component.

[0063] Signal fusion algorithm: weighted fusion method or adaptive fusion method is used to integrate the health status information of different components and optimize the overall health status assessment.

[0064] Life stage prediction: Based on fuzzy rule sets and historical data, the life of the running mechanism is predicted in stages and dynamically adjusted according to real-time data.

[0065] Example 1

[0066] The running mechanism of a rail grinding vehicle includes multiple key components (such as wheels, bearings, and gears). In this embodiment, a combination of vibration sensors, temperature sensors, acceleration sensors, and displacement sensors is used to collect real-time operating status data on these components. The data preprocessing module normalizes the raw data, removes noise using wavelet transform, and reduces the data dimensionality using principal component analysis (PCA).

[0067] A convolutional neural network (CNN) model is trained and fed with a dataset to predict health status. The model outputs a health status classification for each component ("healthy" or "faulty"). A weighted fusion method is then used to combine the health status of each component to determine the overall health status of the running mechanism.

[0068] When the overall health status is "faulty," the system outputs a fault alarm message through the visualization and alarm module, prompting maintenance personnel to address the issue. If the overall health status is "not faulty," the system enters the life stage prediction process. Based on a set of fuzzy rules, the system divides the running mechanism life into early, mid, and late stages. Real-time monitoring data helps the system dynamically adjust the life stages.

[0069] Example 2

[0070] Another rail grinding vehicle uses acceleration, vibration, and temperature sensors to collect real-time data on the running mechanism's status. The data preprocessing module processes the collected data, removing noise and extracting key features. A long short-term memory (LSTM) network is used to model the time series data and combine historical and real-time monitoring data to assess the overall health status.

[0071] The signal fusion algorithm uses an adaptive fusion approach, weighting data based on the health and importance of different components. The life prediction module divides the running mechanism into life stages based on a set of fuzzy rules and makes real-time adjustments.

[0072] The multi-index fusion method of the present invention achieves high-precision health status prediction and lifespan estimation by collecting multi-dimensional operating data of the running mechanism in real time and combining deep learning algorithms and fuzzy rules. Compared with existing technologies, the present invention has the following advantages:

[0073] High accuracy: Utilizing deep neural networks to model component data improves prediction accuracy.

[0074] Real-time: Real-time monitoring and dynamic adjustment of life stages to provide timely feedback on equipment status.

[0075] Enhanced safety: Through timely fault warning and life prediction, sudden equipment failures are avoided and the safety and reliability of the equipment are improved.

[0076] Reduce maintenance costs: Intelligent prediction and life stage division effectively reduce unnecessary maintenance and replacement, extend the service life of equipment, and reduce overall maintenance costs.

[0077] The present invention can be widely used in health monitoring and life prediction of various equipment such as rail grinding vehicles and rail engineering vehicles, providing strong technical support for improving the safety and efficiency of railway transportation systems.

[0078] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art will appreciate that the present invention is not limited to the foregoing embodiments and that various modifications and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such modifications and improvements are intended to fall within the scope of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-index fusion method for predicting the life of a rail grinding vehicle running mechanism, characterized in that: The following steps are included: Step S1, information collection: collecting operating status data of key components of the running mechanism, including lubrication status, wear condition, and usage time; Step S2: Model building: A binary classification model is constructed using a deep neural network to predict the health status of key components and output the health status classification results; Step S3, information analysis: Design a signal fusion algorithm, train a deep learning model based on the health status information of multiple components, evaluate the overall health status of the running mechanism, and classify it as "faulty" or "not faulty"; Step S4, information processing: If the overall health status is "fault" in step S3, a fault alarm message is output. If the overall health status is "not faulty", the running mechanism is divided into multiple life stages based on historical data and fuzzy rule sets, and the current life stage is determined based on real-time data.

2. The multi-index fusion rail grinding vehicle running mechanism life prediction method according to claim 1 is characterized in that: The data acquisition uses multiple sensors, including vibration sensors, temperature sensors and displacement sensors, to monitor the status of key components.

3. The multi-index fusion life prediction method for the running mechanism of a rail grinding vehicle according to claim 1 is characterized in that: The deep neural network uses a convolutional neural network (CNN) or a long short-term memory network (LSTM) to model component data to improve prediction accuracy.

4. The multi-index fusion life prediction method for the running mechanism of a rail grinding vehicle according to claim 1 is characterized in that: The signal fusion algorithm adopts a weighted fusion method or an adaptive fusion method to integrate the status information of different components to optimize the overall health status assessment.

5. The multi-index fusion life prediction method for the running mechanism of a rail grinding vehicle according to claim 1 is characterized in that: The life stage prediction is based on a fuzzy rule set, which divides the state of the running mechanism into multiple life stages and performs dynamic adjustments based on real-time collected data.

6. A platform for predicting the life of the running mechanism of a rail grinding vehicle based on the multi-index fusion method according to claims 1-5, characterized in that: The platform includes: Data acquisition module, used to collect running status data of the running mechanism; Data preprocessing module for data normalization and feature extraction; Component failure prediction module, used to predict the health status of key components; Overall health status assessment module, used to assess the overall health status of the running mechanism; Life prediction module, used to estimate the remaining service life based on fuzzy rules and historical data; Visualization and alarm module, used to provide data display, real-time monitoring and fault warning.

7. The platform of the multi-index fusion rail grinding vehicle running mechanism life prediction method according to claim 6, characterized in that: The data preprocessing module performs signal denoising using one of wavelet transform or Fourier transform, and performs feature dimensionality reduction using principal component analysis (PCA) or autoencoder.

8. The platform of the multi-index fusion rail grinding vehicle running mechanism life prediction method according to claim 6, characterized in that: The life prediction module dynamically adjusts the life stage division based on fuzzy rules and full life cycle data to improve the accuracy of life prediction; the visualization and alarm module provides real-time status monitoring, trend analysis, anomaly detection and automatic alarm functions.

Images