FUZZY RULE BASE MODELING FOR RESERVOIR OPERATION

Abstract

Reservoir operation to find an optimal release has been researched for many years. Various techniques have been developed and adopted for reservoir operation rule incorporating the uncertainty due to stochastic nature of objectives in the reservoir operation and due to imprecise goal values in the reservoir purposes. Relatively little of the research on reservoir operating procedures has found its way into actual practice. One reason is that operators are uncomfortable with complex optimization models and reluctant to use procedures that they do not fully understand. Fuzzy logic approach seems to offer a way to improve on existing operating practices, which is relatively easy to explain and understand. In the present study, fuzzy stochastic dynamic programming model is used to calculate an optimal release from the operation of multipurpose single reservoir. Fuzzy stochastic dynamic programming uses a fuzzy approach to reflect an imprecise information in goals. In fuzzy approach a membership function is used which has the capability to incorporate the uncertainty involved in the objective. Markov chain process is adopted to generate transition probability matrices that are used by fuzzy stochastic dynamic programming. Transition probability matrices are derived from generated inflow data to the reservoir on monthly basis. The methodology is applied for suggesting the reservoir operation rules for Bendo reservoir system. To derive operation rule for a multipurpose single reservoir a fuzzy rule base model is constructed. Operation rule are generated based on optimal release for reservoir operation resulted from fuzzy stochastic dynamic programming method. The fuzzy rule base model operates on an "IF-THEN" principle, where the "IF" is a vector of fuzzy explanatory variable or premises and "THEN" of fuzzy consequences. The reservoir storage volume, estimated inflows, and demand are used as the premise and release from the reservoir is taken as consequences. Fuzzy Rule base !Modeling for 1(gservoir Operation Fuzzy associative memory (FAM) method is used in this study to derive a IF-THEN fuzzy rule base. Fuzzy associative memory used a fuzzy stochastic dynamic programming result as input-output data and a set of triangular membership function of storage, inflow, demand and release to derive a fuzzy rule base. The fuzzy rule base is constructed for the three set of membership function. Fuzzy rule base operation procedures for reservoir that developed by fuzzy associative memory are simulated with the principle of water balance equation. After that, the results from the simulation are compared and evaluated. Performance indices namely reliability, repairabilitiy, incident period and vulnerability are used to evaluate performance of the reservoir operation rules. As comparison, the result of fuzzy stochastic dynamic programming is also simulated and evaluated with same procedure.