The Prediction Of Chaotic Time Series Based On Support Vector Machines

Chaos widely exists in natural phenomena and human life.With the rise of nonlinear science,chaos research has become a popular topic in recent years.Among them,chaotic time series has also become a popular topic of research,and it has been widely used in many different scientific fields.Such as traffic flow forecasting,stock market forecasting,economic forecasting,wind forecasting and power forecasting,etc.Therefore,exploring more effective chaotic time series forecasting methods has theoretical and practical significance for the development of chaotic research.Therefore,this paper is based on the support vector machine model,under different models and parameters,to predict the one-dimensional time series of hyperchaotic systems,mainly completed the following tasks:1.The support vector regression model is used to predict the chaotic time series.First,The same dimension and different time delays are used to reconstruct the phase space,and the reconstructed one-dimensional time series are predicted.Second,in order to analyze the different support vector regression models for the influence of parameters on the prediction results,three different kernel functions are selected,different regular term parameters and gamma parameters are compared the prediction results of the model.From the prediction results,it can be seen that the support vector regression model has a good prediction effect,and it is better to select the parameters with small delay for phase space reconstruction prediction.2.The unbiased least squares support vector machine model is used to predict the chaotic time series.The least squares support vector machine is added with a fixed value parameter to eliminate the bias term,thereby simplifying the algorithm.The unbiased least squares support vector machine is used to predict the chaotic time series.The one-dimensional time series of the chaotic system is predicted,and the particle swarm optimization algorithm is used to optimize the kernel parameters and regular terms in the model.Different time delays are selected for comparative analysis,and the coefficient of determination is used to evaluate model.At the same time,the unbiased least squares support vector machine model is compared with other machine learning models.It proves the effectiveness of unbiased least squares support vector machine.