Performance Analysis of Weighted Cat Swarm Optimization with Sobol Mutation

Abstract This paper introduces an analysis by suggesting three types of modification to an existing well known algorithm for optimization i.e. Cat swarm optimization. The motivation behind the proposed modification is to provide a superior convergence along with the production of finer results in diverse real world applications. The first modification introduces a modified weight factor by the replacement of traditional weight factor in the original algorithm. In the second proposed modification, a versatile mutation process i.e. sobol mutation operator is embedded in the execution process of cat swarm optimization for further enhancement in the performance of the optimization technique. The third proposed modification uses the sobol mutation operator with the modified version weighted cat swarm optimization algorithm. This paper provides a detailed analysis of the three proposed techniques for establishing the superiority over the traditional algorithm. Standard benchmark functions are used in this paper for analysing the effectiveness of the modification proposals. The third proposed modification combines the best feature of both the first and second modifications and compensates the disadvantages to some extent. This third modified version provides the best results in almost all categories of functions along with higher precision. Moreover, all the modifications are providing superior results in comparison to the standard cat swarm algorithm, but the weighted cat swarm optimization algorithm with sobol mutation is dominating in all the cases.