Research on Fault Diagnosis Based on Artificial Neural Network

According to classification framework of classical neural network, contemporary neural network, and soft computing, the basic concepts of the feedforward neural network (MLP, BP, and RBF), the feedback neural network (Hopfield, Boltzmann, Elman), the self-organizing neural network (SOM, ART, CPN), deep and extreme neural network (Deep learning, extreme learning), the novel neural network (SVM, PNN), and soft computing neural networks combined with various methods are introduced, and the research progress and typical applications of the neural network in fault diagnosis are given. The existing problems and future development directions of fault diagnosis are also discussed.