Improving the performance of the optimization technique using chaotic algorithm

Abstract

Optimizing the operations of a multi-reservoir systems are complex because of their larger dimension and convexity of the problem. The advancement of soft computing techniques not only overcomes the drawbacks of conventional techniques but also solves the complex problems in a simple manner. However, if the problem is too complex with hardbound variables, the simple evolutionary algorithm results in slower convergence and sub-optimal solutions. In evolutionary algorithms, the search for global optimum starts from the randomly generated initial population. Thus, initializing the algorithm with a better initial population not only results in faster convergence but also results in global optimal solution. Hence in the present study, chaotic algorithm is used to generate the initial population and coupled with genetic algorithm (GA) to optimize the hydropower production from a multi-reservoir system in India. On comparing the results with simple GA, it is found that the chaotic genetic algorithm (CGA) has produced slightly more hydropower than simple GA in fewer generations and also converged quickly.