Decision Tree Prediction Algorithm Based on Variable Structure Neural Network Control

In order to improve the ability to predict the time sequence statistics, excessive computational overhead for the linear current index prediction method, and easy to fall into local optimum, this paper proposes a prediction algorithm of decision tree structure based on neural network control, treat the statistical data of time series nonlinear characteristics of phase space reconstruction, decision making tree training in high dimensional feature reconstruction of phase space, principal component feature extraction of statistical data on the flow of information, the extracted characteristic parameters of the variable structure neural network controller with tracking training, data clustering search and forecast the pointing error of small disturbance suppression, realize the accurate prediction of statistical data of time series. The simulation results show that the algorithm has higher precision and better convergence performance, and the performance of the data flow is better than that of the traditional method.