Research on rolling bearing fault diagnosis based on improved beluga whale optimization algorithm

Abstract

Aiming at the problem of low diagnostic accuracy caused by manually setting hyperparameters in LSTM network in bearing fault diagnosis, a rolling bearing fault diagnosis method based on the improved beluga optimization algorithm to optimize the attention bi-directional long and short-term memory network is proposed. Firstly, the beluga population is initialized by introducing a semi-uniform initialization distribution system to improve the uniformity and randomness of the spatial distribution of the population and avoid the problem of local optimization to a certain extent; after that, a selection strategy based on the balance of fitness and distance is adopted to improve the optimization stage of the algorithm to maintain the diversity of the population and prevent the algorithm from falling into the local optimization, and then the optimization algorithm is used for the hyper-parameter setting of the attention bidirectional long- and short-term memory network. Then the optimization algorithm is used to optimize the hyperparameters of the attention bidirectional long and short-term memory network, construct a fault diagnosis model to realize the classification and diagnosis of the bearing fault signals, and validate the fault diagnosis effect of the proposed method by using the CWRU (Western Reserve University, USA) dataset. The experimental results show that the diagnostic model optimized by this algorithm is significantly better than the diagnostic model optimized by other swarm intelligence algorithms in terms of fault classification effect and search global performance.