Research On The Basin Runoff Prediction System Of Yunnan Province Based On The Composed Intelligent Optimized Algorithm

Flood and Drought damages frequently occur in Yunnan province. In order to furthest reduce the loss of flood and drought damage; it is necessary to enhance the non-engineering measure construction. The runoff prediction system is an integration system for many non-engineering flood control measures that based on computer technology, Intranet & communication technology and 3S technology. By automaticly acquisiting, real-time transmissing, comprehensively analyzing and intelligent processing a great variety of flood control informations, the flood control decision support system constructs a complete and effective consultation system to obtain the real time simulation and emulation of the rainfall-runoff-flood process. At the same time, the scientific decision-making scheme is formed to conduct the flood control and emergency protection. This paper establishes a runoff prediction system by dealing with the key technology problems of the runoff prediction system of Yunnan province. The main innovations of this paper are as follows:1) This paper built a runoff forecast model by assembly using different types of intelligent modeling algorithm. It is not only has the excellence of artificial neural network but also has advantages of genetic algorithm and wavelet analysis. These characters overcome the drawbacks of the traditional ANN algorithm such as the samples are not precise, it takes a long time to train, we get the initial weights random, it falls into the local minimum easily and"excessive learning". These advantages provide a new way of thought in the area of complex, non-linear, high-dimensional runoff forecast. Compared with the BP algorithm, simulated results show that no matter the runoff forecast precision or the forecasting rate improved a lot.2) This paper presents an effective and practical design of generalized data base management system. Database is the basis of the flood control decision support system. It can store and manage a great variety of Multi-Source and massive data what supply each application subsystem with reliable data support service. When we design a database, no only the comparatively flexibility, expandability and maintainability of the structure are essensial but also the reliability, consistency, intergrity and security of the data. At the same time we must make sure that database meets the demands such as query, display, scientific computation, forecast and decision and so on. The client/serve structure is adopted considering of the reasons above. We take object-relation data model to manage the spatial data and attribute data by intergrating ArcSDE and the large-scaled relational database system Oracle. Results have demonstrated that this method can promote the flexibility of updataing data, improve the information processing date, reduce the data useless and enhance the data safety.3) In this paper we built a rainfall-runoff forecasting model based on geographical and spatial structure by closely combinating the runoff forecast model and GIS. No only the model's function of nynamic simulating and deciding to the runoff process is fully used, but also the ability of GIS in managing spatial data, spatial analysis and visual expression is made the most too.