Health Condition Assessment of Railway Turnout Based on Stacked Sparse Auto Encoder

摘要

Turnout is a railway facility that transfers running vehicles from one track to another. It is also one of the most vulnerable assets that are likely to be affected by hostile weather and mechanical wear. Once a failure occurs, it may cause train delay or even derailment, endangering the safety of life and property. Although many studies have been carried out on the identification of possible failures in railway turnout systems, few people put forward the assessment of turnout health conditions. This paper proposes a health condition assessment method based on Stacked Sparse Auto Encoder (SSAE) . Firstly current signals collected during the work process are mapped to a nonlinear spatial domain using SSAE to extract deep feature representation from original data. Secondly the features representation is used by a softmax classifier for health condition classification. Finally experiments were performed using a turnout condition-monitoring data set and was compared with state-of-the-art method. After parameter optimization, the result shows that the identification rate of our proposed method is over 20% higher than that of others which demonstrates its effectiveness.