Fault Diagnosis of Draft Tube Based on Stochastic Resonance, Multi-Dimensional Entropy and Optimized SVM Model

Aiming at the difficulty of extracting turbine vibration fault features under strong noise background, a fault diagnosis method based on stochastic resonance, multi-dimensional permutation entropy and optimized Support Vector Machine diagnostic model was proposed. Firstly, the signal is denoised by stochastic resonance, then multidimensional permutation entropy is used to extract the signal feature vector, and finally the fault is identified by optimized SVM model. Simulation examples show that this method has higher diagnostic accuracy.