Modified Multi-Scale Symbolic Dynamic Entropy and Fuzzy Broad Learning-Based fast fault diagnosis of Railway Point Machines

Abstract Railway point machine (RPM) condition monitoring has attracted engineers’ attention for safe train operation and accident prevention. To realize the fast and accurate fault diagnosis of RPMs, this paper proposes a method based on entropy measurement and broad learning system (BLS). Firstly, the modified multi-scale symbolic dynamic entropy (MMSDE) module extracts dynamic characteristics from the collected acoustic signals as entropy features. Then, the fuzzy BLS takes the above entropy features as input to complete model training. Fuzzy BLS introduces the Takagi-Sugeno fuzzy system into BLS, which improves the model’s classification performance while considering computational speed. Experimental results indicate that the proposed method significantly reduces the running time while maintaining high accuracy.