State Identification and Prediction of Nonlinear Dynamic Systems Based on Mixed Spiking Neural Networks

Nonlinear dynamical systems exist widely in nature and engineering fields, and the study of their dynamic behavior characteristics such as oscillation, chaos and bifurcation is still a hot topic in academic circles. The chaotic state of nonlinear dynamic system can be judged by images. Dynamics problems usually involve the interaction of multiple variables and complex nonlinear relationships, which may not be well dealt with by traditional statistical methods. The impulsive neural network and brain-like intelligence have the advantages of self-adaptability, nonlinear processing ability and parallel computing ability, and can deal with these complex dynamic problems better. In this paper, a mixed spiking neural network is adopted to classify dynamic system images, and an image-based mixed spiking neural network classification and prediction model is established. The performance of the model is analyzed by using nonlinear dynamic system cases, and the model generation parameters are modified to realize the classification and prediction of nonlinear dynamic system. This scheme has the characteristics of low complexity and strong anti-noise ability, and can play an important role in production and life.