Research On The Fuzzy Time Series Forecasting Model Based On Wolf Pack Algorithm

With the rapid development of science and technology in modern society and the continuous improvement of human living standard,prediction plays an increasingly important role in human life.The traditional time series prediction model has very strict requirements on the data.It can not deal with the data expressed by the language value,and the prediction effect on the small-scale data set is not ideal.Therefore,the traditional time series model will not meet the needs of people's life.In order to solve this problem,fuzzy time series model came into being.It is of great significance to study fuzzy time series models.This thesis aims to improve the prediction accuracy of the fuzzy time series model,and will improve the model from two aspects: First,the wolf pack algorithm(WPA)is introduced and improved,which is used to optimize the division of the universe of discourse.Second,the ordered weighted average(OWA)operator is introduced to propose a more reasonable fuzzy prediction method.In the application,the improved model is applied to two typical data sets and some good results are obtained.The details are as follows:Firstly,in order to improve the optimization accuracy and accelerate the convergence speed of WPA algorithm,the chemotactic behavior of bacterial foraging optimization(BFO)is employed to optimize the scouting behavior of WPA,so that the wolves can find the optimal target value faster.In order to enhance the ability of WPA algorithm to jump out of local extremum,the elimination-dispersal behavior of BFO is introduced into the scouting behavior of WPA.In summary,an improved wolf pack algorithm(HWPA)is proposed.The simulation confirmed that the HWPA algorithm has faster convergence speed and higher convergence accuracy.Secondly,for the aspect of the partition of the universe of discourse,the improved wolf pack algorithm is used to search for the global optimal position,that is,the optimal interval division point.By comparing with multiple existing models,the introduction of this algorithm can effectively improve the prediction accuracy of the model.Thirdly,for the aspect of the fuzzy prediction method,this thesis proposes a fuzzy prediction method based on OWA operator,which not only considers the importance of each fuzzy set,but also considers the importance of different positions.Compared withother prediction methods,this method is more reasonable in data aggregation and more accurate in prediction.In summary,this thesis proposes a new fuzzy time series prediction model based on the HWPA algorithm and OWA operator.The model is used to predict the two typical data sets of Alabama University enrollment and TAIFEX sequence.The prediction accuracy of the new model is higher,which verifies the effectiveness of the new model.