A Multioffspring Genetic Algorithm Based on Sorting Grouping Selection and Combination Pairing Crossover
Xin Jin,Fulin Wang
DOI: https://doi.org/10.1155/2022/4203082
IF: 1.43
2022-01-01
Mathematical Problems in Engineering
Abstract:A multi-offspring genetic algorithm based on sorting grouping selection and combination pairing crossover was proposed in this paper. First, individuals in the population were sorted according to their objective function values, and the remaining individuals were divided into several groups after eliminating the worst. Then, the selection operation, which has the advantage of a simplified calculation that can be easily performed, was implemented. Second, a combination pairing crossover operator was designed. Individuals from different groups were selected for new combinations. In a combination, if a crossover condition was met, individuals pairing and transposons crossover operation were made. Otherwise, offspring copies of the individuals in the combination were used, which increase the opportunity of generating better offspring by producing multi-offspring. Finally, a new population was formed by adopting basic bit mutation operator, elitism policy and the strategy of multi-offspring population competition. Moreover, the catastrophe mechanism has been introduced into improved algorithm to avoid premature convergence. The test results on the functions of CEC 2017 test suites shown that the algorithm proposed in this paper has better search performance, stability and faster convergence to the global optimal solution. These results thus verified the effectiveness and feasibility of the algorithm proposed in this paper. Applying the improved algorithm to the location optimization problem of agricultural product logistics facilities, it shown that the improved algorithm is an effective method to solve the location optimization.