Improvement and application of hybrid real-coded genetic algorithm
Haohao Song,Jiquan Wang,Li Song,Hongyu Zhang,Jinling Bei,Jie Ni,Bei Ye
DOI: https://doi.org/10.1007/s10489-021-03048-0
IF: 5.3
2022-04-02
Applied Intelligence
Abstract:When solving constrained optimization problems (COPs) with high-dimension and multi-extreme problems, genetic algorithm (GA) has the issue of trapping into local optimum. Therefore, this paper proposes a hybrid real-coded genetic algorithm (HIRCGA). First, a sorting group selection (SGS) is given, which is a simple operation and easy to implement. Second, a combinational crossover (CX) operator is developed. It consists of a heuristic normal distribution and direction crossover based on the optimal individual (HNDDX-BOI) and a sine cosine crossover (SCX), which enhances the exploration ability of the algorithm. Third, an operation of eliminating the similarity of different variables in the same dimension (ES) is added, which significantly avoids premature convergence and maintains the population diversity. Fourth, a combinational mutation (CM) operator is proposed, where the global and local search abilities of HIRCGA are fully considered. Fifth, the chaotic search (CS) based on Tent map is introduced to enhance the search power of HIRCGA. Moreover, 28 benchmark test functions in CEC 2017 and two complex real-world optimization problems are selected to demonstrate the effectiveness and superiority of HIRCGA. The computational results and statistical analysis indicate that HIRCGA can improve the solution accuracy compared with other algorithms. The effectiveness of HIRCGA is verified in theory and practice.
computer science, artificial intelligence