Research On Fuzzy Time Series Forecasting Model Based On Particle Swarm Optimization Algorithm

The time series model is a classical forecasting method with perfect theory and wide application,but the model has no way to deal with fuzzy and missing data,and the fuzzy time series(FTS)forecasting model came into being.In order to give full play to the advantages of the model in processing fuzzy data and accurate prediction,this dissertation aims to improve the prediction accuracy,and improves the discourse interval division and fuzzy rule extraction in the modeling process.The first prediction model in this paper first introduces the fuzzy C-means algorithm(FCM)objective function in particle swarm optimization algorithm(PSO)to form an FPSO hybrid algorithm;At the same time,a nonlinear inertial weight ω iterative formula is proposed,which enhances the global search and local optimization capabilities of the hybrid algorithm and improves the accuracy of domain division.Secondly,the establishment of fuzzy rules is to assign different weights to the frequency and order of first-order fuzzy relationships,establish a fuzzy joint weight matrix,and predict the value of the next moment according to the prediction rules,which improves the prediction accuracy.Finally,the model validity is validated on the Alabama University enrollment and agricultural futures price index datasets.The second predictive model in this paper is fuzzy time series for large samples.Firstly,the disturbance sequence generated by chaotic motion is embedded in the search process of F-PSO algorithm,which increases particle diversity,improves the convergence speed and accuracy of the algorithm,and realizes the accurate division of the domain of discourse.Secondly,the powerful complex nonlinear approximation ability of BP neural network is used to obtain the mapping relationship between input and output values,and the BP neural network is trained to extract fuzzy rules,so that the network has the ability to classify,defuzzify the output value and predict the value of the next moment.Finally,the effectiveness of the model is verified by taking Taiwan-weighted stock price(TAIEX)as an example.