Two Kinds Of Prediction Approaches For The Fuzzy Random Times Series

In this thesis the researches of two kinds of forecasting approaches for the fuzzy random time series in economical system are made. The characteristics of the foreca- sting approaches are analyzed and then several kinds of linear fuzzy random model are established. The main contents of the thesis are as follows: I .The forecasting approach for the fuzzy random time series is presented based on the fl.tzzy stochasic processes. (1).The definitions of the fuzzy auto-correlation function and the fuzzy partial correlation function are given and their basic properties are discussed. (2).The forecasting approach for the stationary fuzzy random time series is res- earched. Taking advantage of scattered point diagram, the way of judging the station- nary property of the fuzzy random time series is presented; Three kinds model of lin- ear fuzzy random梩he fuzzy autoregressive model, the fuzzy moving average m- odel and the fuzzy autoregressive moving average model are introduced; The appro- aches of recognition, determining order, parameter estimation and examination of th- ese models are presented. An application example is given in light of the fuzzy auto- regressive model. 2.The forecasting approach is given on the basis of the fuzzy system of linear equations. (1).One type of the fuzzy random time series model is advanced through using a fuzzy system of linear equations with real coefficients and fuzzy varibles. The soluti- on of the model is transformed into a quadratic programming problem. (2).The other type of the fuzzy random time series is presented by using a fuzzy system of linear equations with fuzzy coefficients and real varibles. The solution of the model is changed into a linear programming problem.