Using MSNN to Identify State of Duffing Dynamical System

Duffing oscillator systems are widely found in nature. Due to the easy occurrence of period bifurcation phenomenon, the identification and prediction of its period state is of great significance. The traditional theoretical solution has problems such as excessive computation, and the numerical solution is also more complicated. The development of deep learning provides a new solution to this problem. By using a deep neural network with strong feature extraction capability to extract the relevant features in the phase diagrams of different states of the duffing oscillator and fitting and learning them, the states of the power system can be categorized with less computational complexity. In this paper, we propose a hybrid neural network combining spiking neural network and convolutional neural network to classify the phase diagram of the duffing oscillator, and the hybrid neural network outperforms the convolutional neural network on this task by measuring the energy efficiency ratio of the model.