Fuzzy Time Series Forecasting Model Research And Its Application In Wastewater Treatment

With the development of science and technology, prediction are playing a more and more important role in people’s lives. The traditional forecasting method is commonly according to the actual observation statistics and various transcendental model, with material to fitting model.However, sample collection of data error and the interaction between variables often causes material uncertainty and fuzziness. Second, because of the complexity of the objective world and changeful, many dynamic material can not use accurate numerical to say, for example, the hot weather, tall. The material is difficult to use traditional model structure and analysis method.Based on the above reasons, we introduces fuzzy theory to study the relevant statistical model. Fuzzy forecasting method and the traditional forecasting method of the most essential difference in:the traditional prediction method research objects of the elements in the collection for classic, and fuzzy forecasting method of fuzzy concentration of research is elements. Time series prediction method is a kind of traditional forecast method, it is based on the understanding predicts the development trend of the target and ignore other effects based on the requirements of the original material relatively clear forecast. Fuzzy time series prediction method and the classic time series prediction of the difference in:fuzzy time series prediction into the membership function and the concept of fuzzy relations, this played a important role in time series prediction.This paper on the basis of previous studies, first for standard fuzzy support vector machine (SVM) algorithm was improved, analyzes the advantages and disadvantages of the three kinds of membership functions of three kinds of membership functions of the weighted to construct a new membership functions, construct a new fuzzy support vector machine (SVM) model, and with the new model to sample for training. Second, considering the fuzzy modeling of fuzzy set interval to the influence of interval prediction accuracy, differentiated afresh the fuzzy set, improved the Song and Chissiom the proposed fuzzy time series prediction model. Based on the sample domain son interval, this paper first fuzzy-C mean clustering methods, will get in the center of the cluster as sample domain in the center of the son interval. Secondly, based on sample domain, the sort of two adjacent after the midpoint of the center as dividing point, making the son, the division of the interval is more reasonable. Again, in the original sample blur, established directly sample data and the corresponding relation of fuzzy set and simplify the operation. Finally, in the division of the sample set up the new field of fuzzy support vector machine and fuzzy time sequence combination forecast model, and applied it to the sewage treatment process, to get a good effect. This paper method provide a new idea for prediction of dense data.