An Improved BP Neural Network Fault Detection Based on Time Series Modeling

Combined with data-driven and artificial intelligence, in this paper, we propose a two-layer fault detection mode based on time series modeling and BP neural network. The method is used to realize the autonomy and intelligence of system fault detection. Firstly, we introduce the non-stable characteristics of the system measurement data to construct a non-stationary time series model. Secondly, we use the parameter characteristics of the time series model to determine the node number of input layer, and the neural network parameters will be adaptively determined. Thus, the training data can be used to test the network structure, and then the fault detection can be realized. The simulation data shows that the method is effective and can improve the fault detection performance.