A fault diagnosis strategy for track circuits based on data-knowledge collaboratively driven
With the rapid expansion of China's railway operation scale,track circuits and their related railway equipment have shown explosive growth,and the difficulty of maintaining track circuits increases daily.To satisfy the market demand for track circuits operation and maintenance demands,as well as overcome the challenge of inadequately exploiting the multi-dimensional failure characteristic information reflected by multi-source status monitoring data under a single model training mechanism,an interdisciplinary data-knowledge collaboratively driven fault diagnosis strategy was proposed based on the working principle of track circuits and existing monitoring data.First,a fault diagnosis algorithm based on multi-source data fusion was proposed for data level diagnostic decision-making.The output voltage and current,the voltage and current at the sending(receiving)end outdoor side and indoor side,the main track output voltage,the short track input voltage,the current at E1E2 of the tuning and matching unit at the sending and receiving ends,the current at V1V2 of tuning and matching unit at the sending and receiving ends of the track circuits of this section were selected as multi-source characteristic variables.The SMOTE data augmentation algorithm was used to solve the problem of imbalanced distribution of data samples under various fault modes.The CatBoost based multi feature ensemble learning algorithm was used to construct preliminary a track circuits fault diagnosis model,and the grid search algorithm and the 10-fold cross validation method were adopted to optimize the hyperparameters of the fault diagnosis model to implement fault area localization and fault type identification in track circuit systems.On this basis,combined with multi-disciplinary monitoring data,knowledge level expert decision-making was made based on expert knowledge to achieve Interdisciplinary fault diagnosis of track circuit systems.Finally,the algorithm was applied to the ZPW-2000A track circuits for example analysis,interdisciplinary fault diagnosis was achieved,which verifies the effectiveness of the algorithm.The research results provide effective solutions for interdisciplinary fault diagnosis of track circuits and promote the transformation of maintenance mode towards preventive"condition based maintenance".
fault diagnosisdata drivenInterdisciplinarytrack circuitsrailway signal basic equipment
万方数据
维普
