Research on Reducer Fault Diagnosis Method Based on Artificial Immune Neural Network

This paper establishes one method based on artificial immune and neural network to resolve the problem of Reducer fault diagnosis. Firstly, a neural network model that has one hidden layer and no connection among the nodes in the same layer is constructed, and the HI-IT method is adopted to extract the feature vectors of original signal which are taken as the network input. Then, the structure of neural network is optimized through regulating the number of hidden layers, connection weights, etc., under the artificial immune algorithm. The case results shows that artificial immune neural network performs pretty better than BP neural network in convergence steps, time and the relevant effor.