CN-121980448-A - Intelligent detection method and system for railway sleeper through multi-source data fusion

CN121980448ACN 121980448 ACN121980448 ACN 121980448ACN-121980448-A

Abstract

The invention provides a method and a system for intelligently detecting railway sleeper by multi-source data fusion, which relate to the technical field of railway track detection and comprise the steps of obtaining multi-source original data of a railway track; the method comprises the steps of preprocessing data according to multi-source original data, eliminating inherent noise and environmental interference of a sensor through denoising operation to obtain standardized data, extracting features according to the standardized data to obtain a feature vector set of sleeper arrangement, carrying out data fusion according to the feature vector set to obtain comprehensive state scores of sleepers, carrying out anomaly detection according to the comprehensive state scores to obtain an anomaly identification result, carrying out predictive analysis according to the anomaly identification result, predicting future changes of sleeper states through historical data trend analysis through time sequence modeling, and generating maintenance suggestions based on the prediction result. The invention obviously improves the automation level, the result accuracy and the operation and maintenance efficiency of the detection process.

Inventors

WANG JINNAN
ZHAO JUN
LIN YAO
LI CHUNYU
MA XUETAO
HE YONG

Assignees

中铁工程设计咨询集团有限公司

Dates

Publication Date: 20260505
Application Date: 20251230

Claims (10)

1. The intelligent detection method for the railway sleeper of the multi-source data fusion is characterized by comprising the following steps of: The method comprises the steps of obtaining multi-source original data of a railway track, wherein the multi-source original data comprise infrared temperature measurement data, dynamic mechanical data, sound wave spectrum data, image data and running state data of a train; carrying out data preprocessing according to the multisource original data, eliminating inherent noise and environmental interference of a sensor through denoising operation, and uniformly scaling data of different dimensions to the same numerical range by combining normalization processing to obtain standardized data; extracting features according to the standardized data, and extracting geometrical features, dynamic features and thermal features of sleeper arrangements from the multi-source data to obtain a feature vector set of sleeper arrangements; Carrying out data fusion according to the feature vector set, and obtaining the comprehensive state score of the sleeper through a comprehensive fusion process of multi-source feature integration and prior information combination; Performing anomaly detection according to the comprehensive state scores, comparing real-time data with a historical normal mode through mode identification, and identifying sleeper arrangement anomalies based on threshold comparison to obtain anomaly identification results; and carrying out prediction analysis according to the abnormal identification result, predicting the future change of the sleeper state through time series modeling analysis historical data trend, and generating maintenance advice based on the prediction result.
2. The intelligent detection method for the railway sleeper of the multi-source data fusion according to claim 1, wherein the data preprocessing is performed according to the multi-source original data, inherent noise and environmental interference of a sensor are eliminated through denoising operation, and data of different dimensions are uniformly scaled to the same numerical range by combining normalization processing, so that standardized data is obtained, and the method comprises the following steps: noise elimination processing is carried out according to the multi-source original data, and the special rail vibration interference and thermal noise in sleeper detection are eliminated by applying a sensor specific noise model, so that preliminary cleaning data are obtained; Extracting effective data according to the preliminary cleaning data, and removing invalid data points by a screening criterion based on a sleeper space distribution rule to obtain screened data; and carrying out dimension unified processing according to the screened data, and converting the multi-source data into a dimensionless form based on data distribution characteristics to obtain standardized data.
3. The intelligent detection method for railroad ties through multi-source data fusion according to claim 1, wherein feature extraction is performed according to the standardized data, and a feature vector set of the ties is obtained by extracting geometric features, dynamic features and thermal features of the ties from the multi-source data, comprising: Carrying out multisource feature extraction according to the standardized data, extracting sleeper surface temperature distribution features from infrared temperature measurement data, extracting vibration frequency features from dynamic mechanical data, extracting sleeper spacing and angle features from image data, and integrating sound wave spectrum data and train state data features to obtain a preliminary feature pool; feature screening is carried out according to the preliminary feature pool, and redundant features are removed based on importance analysis of historical data and sleeper physical meaning verification to obtain a key feature set; and performing matrix construction processing according to the key feature set, and obtaining a feature vector set of sleeper arrangement by organizing feature vectors according to sleeper numbers and introducing a space-time expansion mechanism.
4. The intelligent detection method for the railway sleeper of the multi-source data fusion according to claim 1, wherein the data fusion is carried out according to the feature vector set, and the comprehensive state score of the sleeper is obtained through a comprehensive fusion process of multi-source feature integration and prior information combination, and the method comprises the following steps: Performing feature weight distribution processing according to the feature vector set, and distributing differential weights for the features through a weight distribution mechanism based on the historical detection precision of the sensor to obtain a weighted feature set; Carrying out probability fusion according to the weighted feature set, and updating the confidence coefficient of the sleeper state based on the combination of the prior probability and the real-time data likelihood to obtain a preliminary state score; and carrying out comprehensive evaluation processing according to the preliminary state scores, and obtaining the comprehensive state scores of the sleepers through a multi-dimensional score fusion and threshold comparison mechanism.
5. The intelligent detection method for the railway sleeper based on the multi-source data fusion according to claim 1, wherein the method is characterized in that the method comprises the steps of performing anomaly detection according to the comprehensive state score, comparing real-time data with a historical normal mode through mode identification, identifying sleeper arrangement anomalies based on threshold comparison, and obtaining an anomaly identification result, and comprises the following steps: establishing a multidimensional reference pattern library according to historical normal data, and constructing a reference pattern library containing space-time characteristics by analyzing sleeper temperature distribution, vibration characteristics, acoustic characteristics and stable association patterns of train operation states in a normal state; performing multi-source feature collaborative analysis according to the comprehensive state scores, extracting a current multi-dimensional feature sequence through a time sequence sliding window, and performing feature layer similarity matching with the reference pattern library to obtain a multi-dimensional deviation evaluation result; And carrying out anomaly comprehensive judgment according to the multi-dimensional deviation evaluation result, and identifying specific anomaly types and spatial distribution thereof by combining a dynamic threshold comparison mechanism with an anomaly type feature library to obtain an anomaly identification result.
6. The intelligent detection system for the railway sleeper of the multi-source data fusion is characterized by comprising the following components: the acquisition module is used for acquiring multi-source original data of the railway track, wherein the multi-source original data comprise infrared temperature measurement data, dynamic mechanical data, sound wave spectrum data, image data and running state data of a train; the processing module is used for carrying out data preprocessing according to the multi-source original data, eliminating inherent noise and environmental interference of the sensor through denoising operation, and uniformly scaling data with different dimensions to the same numerical range by combining normalization processing to obtain standardized data; The extraction module is used for extracting the characteristics according to the standardized data, and extracting the geometric characteristics, the dynamic characteristics and the thermal characteristics of sleeper arrangements from the multi-source data to obtain a characteristic vector set of sleeper arrangements; the fusion module is used for carrying out data fusion according to the feature vector set, and obtaining the comprehensive state score of the sleeper through the comprehensive fusion process of multi-source feature integration and prior information combination; the detection module is used for carrying out abnormality detection according to the comprehensive state scores, comparing real-time data with a historical normal mode through mode identification, and identifying sleeper arrangement abnormality based on threshold comparison to obtain an abnormality identification result; And the output module is used for carrying out prediction analysis according to the abnormal identification result, predicting the future change of the sleeper state through time sequence modeling analysis of historical data trend, and generating maintenance advice based on the prediction result.
7. The multi-source data fusion railroad tie intelligent detection system of claim 6, wherein the processing module comprises: the first processing unit is used for carrying out noise elimination processing according to the multi-source original data, and eliminating the special rail vibration interference and thermal noise in sleeper detection by applying a sensor specific noise model to obtain primary cleaning data; The second processing unit is used for extracting and processing effective data according to the preliminary cleaning data, and removing invalid data points through screening criteria based on sleeper space distribution rules to obtain screened data; And the third processing unit is used for carrying out dimension unified processing according to the screened data, and converting the multi-source data into a dimensionless form based on the data distribution characteristics to obtain standardized data.
8. The intelligent railroad tie detection system of multi-source data fusion of claim 6, wherein the extraction module comprises: The first extraction unit is used for carrying out multi-source feature extraction according to the standardized data, extracting sleeper surface temperature distribution features from infrared temperature measurement data, extracting vibration frequency features from dynamic data, extracting sleeper spacing and angle features from image data, and integrating sound wave spectrum data and train state data features to obtain a preliminary feature pool; the second extraction unit is used for carrying out feature screening according to the preliminary feature pool, analyzing importance and verifying sleeper physical meaning based on historical data, and eliminating redundant features to obtain a key feature set; And the third extraction unit is used for carrying out matrix construction processing according to the key feature set, and obtaining a feature vector set of sleeper arrangement by organizing feature vectors according to sleeper numbers and introducing a space-time expansion mechanism.
9. The intelligent railroad tie detection system of multi-source data fusion of claim 6, wherein the fusion module comprises: The first fusion unit is used for carrying out characteristic weight distribution processing according to the characteristic vector set, and distributing differential weights for all the characteristics through a weight distribution mechanism based on the historical detection precision of the sensor to obtain a weighted characteristic set; The second fusion unit is used for carrying out probability fusion according to the weighted feature set, and updating the confidence coefficient of the sleeper state based on the combination of the prior probability and the real-time data likelihood to obtain a preliminary state score; And the third fusion unit is used for carrying out comprehensive evaluation processing according to the preliminary state scores, and obtaining the comprehensive state scores of the sleepers through a multidimensional score fusion and threshold comparison mechanism.
10. The multi-source data fusion railroad tie intelligent detection system of claim 6, wherein the detection module comprises: The first detection unit is used for establishing a multi-dimensional reference pattern library according to historical normal data, and constructing a reference pattern library containing space-time characteristics by analyzing sleeper temperature distribution, vibration characteristics, acoustic characteristics and stable association patterns of train running states in a normal state; The second detection unit is used for carrying out multi-source feature collaborative analysis according to the comprehensive state scores, extracting a current multi-dimensional feature sequence through a time sequence sliding window, and carrying out feature layer similarity matching with the reference pattern library to obtain a multi-dimensional deviation evaluation result; And the third detection unit is used for carrying out abnormal comprehensive judgment according to the multi-dimensional deviation degree evaluation result, and identifying specific abnormal types and spatial distribution thereof by combining an abnormal type feature library through a dynamic threshold comparison mechanism to obtain an abnormal identification result.

Description

Intelligent detection method and system for railway sleeper through multi-source data fusion Technical Field The invention relates to the technical field of railway track detection, in particular to an intelligent detection method and system for a railway sleeper with multi-source data fusion. Background In the technical field of railway track detection and operation and maintenance, a sleeper is used as a core bearing component of a track structure, and the arrangement state and the health condition of the sleeper directly determine the geometric stability and smoothness of the track and the running safety of a train. Along with the continuous development of sensing technology and data acquisition and processing technology, railway detection is gradually evolved from early stage to adopting various online monitoring systems such as an infrared shaft temperature detection system, a track dynamic detection system, an acoustic detection system and the like depending on manual hiking inspection and handheld tool measurement, so that automatic acquisition of state parameters of sleeper and track parts is realized. However, the prior art means still has significant limitations when dealing with the complex problem of sleeper arrangement state, on one hand, the traditional manual inspection method has low efficiency and high cost, the detection result is easily influenced by personnel experience and subjective state, and the standardized monitoring of a large-scale road network is difficult to realize, on the other hand, even if an automatic detection device is adopted, the prior art method is also limited to independent analysis and simple threshold alarm of single type sensor data, and the effective fusion and collaborative analysis of multi-source heterogeneous data are lacking, so that the description dimension of sleeper state is single, the real physical state and evolution trend of sleeper state cannot be comprehensively and accurately reflected, and especially complex anomalies such as looseness, cracks, sedimentation and the like are difficult to be found early. Based on the shortcomings of the prior art, a need exists for a method and a system for intelligent detection of a railway sleeper by multi-source data fusion. Disclosure of Invention The invention aims to provide a railway sleeper intelligent detection method and system for multi-source data fusion, so as to solve the problems. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: in a first aspect, the present application provides a method for intelligently detecting a railroad sleeper with multi-source data fusion, including: The method comprises the steps of obtaining multi-source original data of a railway track, wherein the multi-source original data comprise infrared temperature measurement data, dynamic mechanical data, sound wave spectrum data, image data and running state data of a train; carrying out data preprocessing according to the multisource original data, eliminating inherent noise and environmental interference of a sensor through denoising operation, and uniformly scaling data of different dimensions to the same numerical range by combining normalization processing to obtain standardized data; extracting features according to the standardized data, and extracting geometrical features, dynamic features and thermal features of sleeper arrangements from the multi-source data to obtain a feature vector set of sleeper arrangements; Carrying out data fusion according to the feature vector set, and obtaining the comprehensive state score of the sleeper through a comprehensive fusion process of multi-source feature integration and prior information combination; Performing anomaly detection according to the comprehensive state scores, comparing real-time data with a historical normal mode through mode identification, and identifying sleeper arrangement anomalies based on threshold comparison to obtain anomaly identification results; and carrying out prediction analysis according to the abnormal identification result, predicting the future change of the sleeper state through time series modeling analysis historical data trend, and generating maintenance advice based on the prediction result. In a second aspect, the present application also provides a railroad sleeper intelligent detection system for multi-source data fusion, including: the acquisition module is used for acquiring multi-source original data of the railway track, wherein the multi-source original data comprise infrared temperature measurement data, dynamic mechanical data, sound wave spectrum data, image data and running state data of a train; the processing module is used for carrying out data preprocessing according to the multi-source original data, eliminating inherent noise and environmental interference of the sensor through denoising operation, and uniformly scaling data with different dimensions to the same numerical r