Gearbox Fault Diagnosis Based on Optimized Stacked Denoising Auto Encoder and Kernel Extreme Learning Machine

The gearbox is one of the key components many large mechanical transmission devices. Due to complex working environment, vibration signal stability gear box poor, fault feature extraction difficult, and diagnosis accuracy makes it difficult meet expected requirements. To solve this problem, paper proposes a method based on an optimized stacked denoising auto encoder (SDAE) kernel extreme learning machine (KELM). Firstly, particle swarm optimization algorithm in adaptive weight (SAPSO) was adopted optimize SDAE network structure, number hidden layer nodes, rate, noise addition ratio iteration times were adaptively obtained make obtain best structure. Then, structure used extract deep information weak faults original signal. Finally, extracted features are fed into KELM for classification. Experimental results show that classification proposed can reach 97.2% under condition low signal-to-noise ratio, which shows effectiveness robustness compared with other diagnostic methods.