期刊名称：International Journal of Advances in Soft Computing and Its Applications
印刷版ISSN：2074-8523
出版年度：2015
卷号：7
期号：1
出版社：International Center for Scientific Research and Studies
摘要：Many hydrologists proclaimed Tank model is able to achieve comparable or better forecasting results than more sophisticated models even with its simple concept and computation. With the development of Artificial Intelligence (AI) in recent years, various Global Optimization Methods (GOMs) had been adopted to calibrate Tank model parameters automatically. However, these GOMs are only able to search optimal result for a single objective function. The calibration and validation processes need to be repeated for each objective function in searching the optimal solution and this consumes a lot of time and effort. Hence, multiobjective particle swarm optimization (MOPSO) is adapted in this study to allow PSO be able to deal with a few objective optimization functions simultaneously. The selected study area is Bedup basin, Samarahan, Sarawak, Malaysia. Input data used for model calibration are hourly and daily rainfall and runoff. Two sets of objective functions are investigated. The first set of optimization function consists of ordinary least square (OLS) and root mean square error (RMSE). Where else the second set objective function consists of OLS, RMSE and coefficient and correlation (R). The accuracy of the simulation results are measured using R and Nash-Sutcliffe Coefficient (E 2 ). Results revealed that the performance of MOPSO with 3 objective functions is slightly better than MOPSO with 2 objective functions for both hourly and daily Tank model. Results also proved that MOPSO is able to discover a set of optimal nondominated solution through the true Pareto optimal solutions for the test problems considered
