Determination of an optimal unit of hydrograph of river basin using genetic algorithm

摘要

Unit hydrograph (UH) is used as a conceptual rainfall-runoff model for estimating peak discharge, timing of peak, and/or flood volume from design storms of a river basin, which are key inputs in the planning, design, operation, and management of various water resources projects in the basin and is the focus of hydrological modeling. In the past, majority of the optimal programs for determining the ordinates of a UH have employed objective functions and constraints that are linear in nature. Recently, evolutionary computing methods such as genetic algorithm (GA) have provided the researchers and modelers with tools that can model a highly non-linear, dynamic, and complex physical system with great ease. Most of the GA applications in engineering use binary-coded genetic algorithm, which has certain shortcomings such as precision, hill climbing etc., to overcome such problems of binary coded GA, real-coded GA (RGA) can be employed. The objectives of the present study are to (a) develop an optimal program consisting of non-linear objective function and constraints for the determination of the ordinates of a UH, and (b) solve the developed problem using real-coded GA. A new optimal program is proposed, which consists of fewer decision variables and constraints, and an objective function that is non-linear in nature. This paper presents the findings of probably the first attempt of employing the new solution technique of real-coded GA to the problem of determination of an optimal UH using the historical rainfall and runoff data from a basin. The rainfall-runoff data derived from Nenagh River basin, for twenty two storms have been employed to determine optimal UHs and test the performance of the new solution technique. Certain standard statistical performance evaluation measures was employed to evaluate the performance of the methodology developed. The preliminary results obtained are highly encouraging and indicate that the new technique of real-coded GA can be successfully applied to the problem of determining an optimal UH of a river basin.