Chaotic Time Serise Prediction Based On Improved Particle Swarm Algorithm

Chaos is a widely existing phenomenon in nature and a kind of motion state producedby nonlinear dynamics system. With the improvement of chaos theories, chaotic timeseries prediction has become the focus of research for scholars in various fields, and iswidely used in social science, natural science and other fields, such as economicprediction, prediction of hydrological forecasting, prediction of stock market, prediction ofsunspots, prediction of power, etc, it has important application value in this paper, westudy the prediction of chaotic time series, and puts forward three kinds of research plan.The article on the theory of chaos is briefly introduced, prediction of chaotic timeseries, and several prediction models were discussed, the main research contents are asfollows.Firstly, the article explores the theory of chaos, and chaos time series forecasting. Itgeneral put forward several kinds of chaotic time series prediction scheme, points out theresearch purpose of this subject.Secondly, the article analyzes chaotic time series prediction model of BP neuralnetwork, particle swarm optimization is introduced to optimize BP neuralnetwork, according to analysis of particle swarm disadvantages, particle swarm algorithmbased on fitness function is put forward to optimization BP neural network model, and thescheme is applied to Logistic, Henon, Lorenz system, comparing the experimental results,show the rationality of scheme design, feasibility.Thirdly, the article analyzes chaotic time series prediction model of RBF neuralnetwork, particle swarm optimization is introduced to optimize RBF neuralnetwork, according to analysis of particle swarm algorithm disadvantages, particle swarmalgorithm based on chaotic initialization and chaotic perturbation location is put forwardto optimize the RBF neural network model, and the scheme is applied to the prediction ofchaotic time, comparing the experimental results, the optimization model hashigher convergence speed and prediction accuracy.Finally, the article analyzes chaotic time series prediction model of SVM, particle swarm optimization is introduced to SVM, according to analysis of particleswarm algorithm disadvantages, particle swarm algorithm based on chaotic inertia weightand acceleration factor is put forward, using this algorithm to optimize SVM. Prediction ofchaotic time series using the optimized model, achieved good results.