Study On SVM Prediction Of Chaotic Time Series Based On Information Granulation And Application

Time series forecasting has always been an important research direction was predicted inthe field, due to the continuous monitoring of the application of new technologies, leading tomore and more accumulated data, the number of extremely large. How to effectively use thesedata to identify which rules, and then the data to predict the value of a very importantpractical and theoretical significance. Because life has a lot of data chaos, traditionalstatistical analysis methods ineffective in the application of data processing, to find a newmethod of data processing has become a new hot spot. The SVM because of its excellentnon-linear characteristics, ideal for chaotic time series analysis and processing of data. Giventhe massive nature of data, information granulation theory into data to predict areas graduallyattracted people’s attention.This article will introduce information granulation theory predicting chaotic time seriessupport vector machine, and its research, the main contents include; first chapter introducesthe background and purpose of the system is the significance of the topic, expounded onchaos theory, support vector machine chaotic time series studies and application statusinformation granulation; second chapter introduces the theory used herein; Chapter IIIdescribes the experimental procedure, the reconstruction of chaotic time series and granulated,and then build a prediction model; Chapter IV, the data simulation, the experimental results,and the results were analyzed; Chapter V, apply real-life simulation of urban traffic flow data,the model was verified, given analysis of results.Through a number of experiments show that the information granulation is applied topredict the chaotic time series support vector machine is feasible, capable of short-rangeforecast, but the prediction accuracy to be improved.