A Comparison of Partition of Initial population for Multi-objective Genetic Algorithm using by Optimal Drainage

Abstract

This paper proposes a new Multi-Objective Optimization Problems (MOOP), which is called a Partition of Initial Population for Multi-Objective Genetic Algorithm (PIMOGA). It applys a clustering algorithm, which is K-Means clustering algorithm, for partition an initial population into small groups aiming to achieve global solutions efficiently. This proposed work is of the drainage suitability model. This is derived from the results of the predictive value with Holt-Winters’ Exponential Smoothing method, which was time series forecasted. The equation of appropriateness of all objectives 7 and 5 independent variables. The results of PIMOGA's experiment showed that 1) k = 3 had the lowest number of terminated generation. Therefore, it was suitable for PIMOGA experiment. 2) PIMOGA had time measurement Test Activity Measured Time, which can be used to increase the efficiency of searching for the last generation of solutions. MOGA and MOPSO have the ability to improve search performance. The Global Optimal Solution is better than MOGA and MOPSO, except MOEA. And this research provides an application to support the estimation of water volume in the network drainage using PIMOGA method by specifying the defaults as date and season influences. It was found that the efficiency in the estimation of water volume in the drainage.