Feedback process neuron network model and its learning algorithm

A feedback process neural network model based on weight function base expansion is put forward. Structurally, this model has three layers, namely the input layer, process neuron hidden layer and process neuron output layer. The input layer completes input of the time-varying process signal and feedback of output from the hidden layer to the system, the process neuron hidden layer is used to complete the space weight aggregation and incitation operation, at the same time the output signal is transferred to the output layer and fed back to the input layer after being weighted. The output layer completes space weight aggregation and time aggregation of output signal from the hidden layer and the system output. The learning algorithm is given. The effectiveness of the model and algorithm is proved by automatic diagnoses of a rotating machine.EI06801-8063