Study On Runoff Forecasting Methods Based On Variable Fuzzy Set Theory

An efficient flood-control decision-making is an important non-structural measure to relieve water resources shortage and fully utilize wate r resources. It is necessary to fully understand the objective hydrology regularities so as to make scientific and reasonable flood-control decision. Therefore, hydrology forecasting is the basis of reservoir operation. It is also a very important premise to improve the watershed runoff ability and make full use of the comprehensive benefit of reservoir. Based on the variable fuzzy set theory, this dissertation mainly studies the fuzzy reasoning methods for mid-long term runoff forecasting, which can provide the technical support for improving the reservoir operation lever and relieving the water resource shortage. On the other hand, the classified and forecasting methods for basin flood are also studied here.The main objectives and results for this research are as follows.(1) The relative membership degree and relative membership function are the most important definitions in the fuzzy set theory. In the variable fuzzy set theory, the relative membership function based on interval pattern value can be used to determine the relative membership degree of any element in a fuzzy set. The difference and connection between the relative membership function based on interval pattern value and the existing fuzzy distribution function are found, i.e. the existing fuzzy distribution is a special case of the relative membership function based on interval pattern value. The relative membership function based on interval pattern value has universality and extensive adaptablity. Finally, the noticeable problem for this relative membership function is considered.(2) The fuzzy reasoning method with single factor for mid-long runoff forecasting based on rank feather value is proposed. This method applies the interval pattern value to determine the relative membership degree by relative membership function and enhances the theory basis of existing fuzzy reasoning method. The reasoning model in the proposed method is established by the rank feather value of forecasting runoff, which can avoid a lot of repeated calculation. The fuzzy reasoning method principles are presented in detail and verified by case studies in this thesis. It shows that the fuzzy reasoning method can improve the existing hydrology and meteorology fuzzy reasoning method.In order to enrich and improve the fuzzy reasoning method for runoff forecasting, a fuzzy reasoning method with factor weight for runoff forecasting is developed on the basis of approximate fuzzy reasoning similarity relationship. This method is applied to study the Dahuofang reservoir yearly runoff forecasting. The results with factor weight are better than those with equal weigh, which shows the method with factor weight is necessary.(4) The fuzzy variable reasoning method for mid-long runoff forecasting is proposed on the basis of approximate fuzzy reasoning similarity relationship. The variable parameters in the variable method are found out. A case study of Dahuofang reservoir yearly runoff forecasting is given by six combination conversion of variable parameters. The final forecast result can be obtained by analyzing the relationship among different forecasting information and used to guide the schedule of reservoir operation.(5) A classified forecasting method for basin floods is proposed based on the variable fuzzy sets theory. This method can be used to classify the basin floods according to the early stage characteristics of floods. Then the parameters in flood forecast model are optimized respectively and applied to forecast the corresponding type of basin floods. The application study shows that the proposed method is reasonable and valuable.(6) The fuzzy variable evaluation method is applied to assess the water resources renewability for nine administrative divisions in the Yellow River basin. Based on variable fuzzy set theory and artificial neural network, an innovative evaluation model is proposed and used to study water resources renewability.