Research Of Multivariate Fuzzy Time Series Model Based On Fuzzy Information Optimization Technology

Fuzzy time series models have unique advantages when dealing with data with fuzzy attributes.The traditional fuzzy time series model mostly deals with the problem of univariate or large sample multivariables.In fact,there are a large number of small samples and the multivariate time series needs analysis and prediction in real life.To avoid model causal deviation or prediction error that may occur due to traditional models,it is of great theoretical and practical significance to establish a multivariable fuzzy time series model that can handle small samples.This paper aims to improve the prediction accuracy of multivariable fuzzy time series model for small samples,and combines the information distribution to identify the characteristics of small sample information effectively.Based on information distribution technology,a multivariable fuzzy time series model is constructed,which is carried by assigning sample points.The information is divided into two parts to fill in the gap of information existing in the small sample,so as to achieve the purpose of improving the prediction accuracy.Normal information diffusion can spread the information carried by knowledge sample points to multiple monitoring points,improve the ability of understanding the system,and can improve the accuracy of small-sample time series prediction more effectively.Therefore,this paper builds another multivariable fuzzy time series model on the basis of normal information diffusion.Select the time series data of total energy consumption,per capita GDP,and sulfur dioxide emissions from 2001 to 2017 for case analysis;discuss the influence of ambiguity on information distribution model and information diffusion coefficient on the normal information diffusion model;select classic Markov The fuzzy time series model is used as a comparison model to verify the validity of the model constructed in this paper.The main conclusions are summarized as follows:First,The IDMFTSM model can effectively identify small sample information,thereby improving the model prediction accuracy;the ambiguity affects the prediction accuracy of the IDMFTSM model.As the ambiguity decreases,the prediction accuracy increases,and the prediction accuracy is the highest.Second,The NDMFTSM model can identify small sample information more effectively,thereby further improving the prediction accuracy of the model;the prediction accuracy of the NDMFTSM model is better than the IDMFTSM model;the information diffusion coefficient affects the prediction result of the NDMFTSM model,and the prediction result is better.Third,The prediction result of IDMFTSM?1 is better than Markov?1,and the prediction result of NDMFTSMh0 is better than Markovh0,which fully proves the validity of the model constructed in this paper and the calculation process is convenient and concise.