Nonlinear system modeling using convolutional neural networks

With the benefit of convolutional operation, the convolutional neural networks (CNN) has been successfully applied in classification, regression and time series modeling. For neural modeling of dynamic systems, CNN also should have many advantages over other neural models, such as avoiding local minima and the noises and outliers affections. In this paper, the dynamic system identification is addressed by CNN. The system identification with CNN is divided into two cases: feedforward CNN and backpropagation training. The proposed deep learning methods for dynamic system identification are validated with two benchmark data sets.