Railway Switch Power Curve Monitoring for Noise-Robust Fault Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current railway signal monitoring systems for switch equipment rely heavily on labor-intensive maintenance and human experience, leading to inefficiencies and inaccurate fault identification, particularly for new abnormal conditions, with limited model generalization and inadequate noise handling.
Innovation Solution
A method utilizing complementary ensemble empirical mode decomposition (CEEMD) to extract intrinsic mode function (IMF) components and fuzzy entropy features, combined with grey relational analysis and fuzzy clustering, to calculate similarities between real-time and historical switch action power curves for precise fault diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning and data mining technologies are used for abnormality diagnosis of electrical characteristic curves, then abnormality diagnosis accuracy is improved to a certain extent, but the model's generalization capability is limited for new abnormal conditions
Solution Approach 1:
The patent transforms the fixed machine learning model into a flexible similarity comparison framework by changing the diagnostic parameters from predefined features to dynamic similarity metrics. The grey relational analysis and fuzzy clustering methods enable the system to adapt to new abnormal conditions by comparing real-time curves with historical data without requiring model retraining, thus improving generalization capability while maintaining diagnosis accuracy
Solution Approach 2:
The patent creates a digital copy of historical switch action power curves and uses similarity comparison techniques to match real-time curves against these copies. This copying approach allows the system to diagnose new abnormal conditions by finding similar patterns in historical data, eliminating the need for continuous model training and improving adaptability to unseen faults
2Measurement precision
If machine learning methods are used for electrical characteristic curve diagnosis, then abnormality identification capability is enhanced, but handling of noise and interference remains inadequate
Solution Approach 1:
The patent extracts the essential diagnostic information from noisy electrical characteristic curves by using grey relational analysis and fuzzy clustering. These methods separate the meaningful pattern information from noise and interference by comparing overall curve shapes and key特征 points rather than relying on individual data points, thereby enhancing abnormality identification while filtering out harmful factors
Solution Approach 2:
The patent introduces similarity metrics as an intermediary between the noisy real-time curve and the clean historical reference curves. The grey relational analysis and fuzzy clustering algorithms act as mediators that compute similarity degrees without being directly affected by noise, thus protecting the diagnostic process from harmful factors while maintaining high identification capability
3Ease of operation
If traditional maintenance methods relying on worker experience are used, then operational simplicity is maintained, but fault identification accuracy deteriorates due to false and missed determination
Solution Approach 1:
The patent enables the monitoring system to perform self-diagnosis by automatically comparing real-time switch action power curves with historical data using grey relational analysis and fuzzy clustering. The system independently identifies abnormalities and determines fault types without relying on worker experience, thus maintaining operational simplicity while dramatically improving fault identification accuracy and eliminating false or missed determinations
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present application relates to the technical field of fault monitoring for rail transit equipment, and in particular, to a method and device for monitoring a state of a switch power curve. The method includes: acquiring first feature information of a real-time switch action power curve; acquiring second feature information of switch action power curves in different operating states; based on a grey relational analysis method and/or a fuzzy clustering method, calculating a similarity between the first feature information and the second feature information corresponding to each fault type; and determining an operating state corresponding to the real-time switch action power curve based on the similarity. The present application provides an intelligent method and system for diagnosing an abnormality in an electrical characteristic curve of railway equipment, which have higher accuracy, degree of automation, and robustness. By using a fault diagnosis method based on digital signal processing, fault features of switch action power curve data are effectively extracted, thereby meeting various fault detection requirements, simplifying a structure of a classifier, and improving the precision and efficiency of fault diagnosis.