Classical time series model can efficiently handle many forecasting problems, but these models can not solves the forecasting problem in which values of the time series are represented by language values or fuzzy sets. Song and Chissom and many other scholars put forward many models, and these models can only forecast research about historical data. This paper presents a new fuzzy time series forecasting model which can predict the data of unknown years.

Since Song and Chissom proposed fuzzy time series forecasting theory, already exceed in the 20 years. Scholars have proposed many fuzzy time series forecasting models, the prediction accuracy of historical simulation data continues to improve. Unfortunately has not hitherto given for fuzzy time series forecasting model about the data of unknown years. This paper presents an improved forecasting model of fuzzy time series. It may predict the historical simulation data, but also may predict the unknown year data.

Method based on vague optimization evaluation is vague pattern recognition. There are six detailed steps of application. The first, Set up Techno-economic indicator system. Secondly set up preparative optimization scheme sets. Thirdly set up optimal scheme in theory. It is made up of each Techno-economic indicator optimal data. Fourthly transform techno-economic input data into vague data. The fifth, Calculating similarly measures. Similarity measures will be evaluated between preparative optimization scheme vague sets and optimal scheme in theory. The last is vague optimization evaluation. The weight of each preparative optimization scheme is given. The data of weighted similarity measures by the weight factors are obtained. And applying them we obtain the good and bad sort of vague optimization scheme. The new similarity measures formula between vague sets is given. The formula is indispensable in the method of vague optimization evaluation. Application examples show that the Vague optimization evaluation method to the conclusion is reliable.

In order to amend the defects of existing similarity measure formula between vague sets, a new definition of similarity measure between vague sets is proposed and a new formula with higher resolution and highlighted uncertainty is presented on the basis of data mining vague value method. A general fault diagnosis method of Vague sets (GFDMVS) is proposed. The same practical case is studied with three methods and the results demonstrate the validity and reasonability of the method proposed in this paper.

Definition of conversion from a single value data to the Vague value data is given; two conversion formulas from a single value data to the Vague value data are given; a similarity measure formula between Vague sets are given; Vague pattern recognition algorithm is given. The algorithm is applied to irrigation system design, application examples show that theVague pattern recognition algorithms and formulas are all useful.