Combining Penalty Function with Modified Chicken Swarm Optimization for Constrained Optimization

摘要

In many mechanical designs, such as airborne electro-optical platform, optical lenses, mechanical containers, speed reducer, and so on, lightweight design has always been our goal. Under various constraints, obtaining the minimum of some parameter is the optimization problem we often encounter in the engineering works. Chicken Swarm Optimization (CSO), a new bio-inspired algorithm, is namely applied to deal with these kinds of problems. This paper firstly describes the origin and the basic model of the CSO and shows the result of applying the CSO to the algorithm test functions and a fair statistical comparison of the CSO with Bat Algorithm (BA) and modified Bat Algorithm based on Differential Evolution (DEBA) on the same test functions. Then, the CSO algorithm is modified. After that, the modified CSO is used to do the test on the previous test functions in order to be compared with the basic CSO, BA and DEBA. Finally, the modified CSO is combined with a dynamic penalty function to solve nonlinear constrained optimization problems and compared with other algorithms. From the results of all the tests, we can see that the CSO outperforms many other algorithms or their modified ones in terms of both optimization accuracy and stability. However, the modified CSO gets better performances than the CSO. As well, the modified CSO combined with penalty function is better than the CSO and many other optimization algorithms for constrained optimization problems.