Two Stages Multi-operator Hybrid Constraint Handling Strategy for CMTOPs

Abstract

The difficulty of constrained multitask optimization algorithms (CMTOPs) is to balance the constraint violation and the objective function, which means to make the objective function optimal in the feasible region. In this paper, we propose a two stages multi-operator hybrid constraint handling mechanism (TSMOH) to solve the constrained multitask optimization problems. The primary contributions of the proposed algorithm are as follows: (1) Based on the considerations of population evolution and the number of feasible solutions, we design a two stages multi-evaluation criterion strategy to achieve a balance between constraint violations and the objective function. (2) We utilize the advantages of different crossover operators to achieve knowledge transfer between the tasks. (3) To address the problem of low probability assignment of the crossover operator, we divide the population evolution into multiple stages and execute the preprocessing algorithm at the beginning of each stage. In the experiments, we validate the performance of TSMOH on CMTOPs benchmark suite, and the experimental results demonstrate the effectiveness of the proposed strategy and the superior performance of the algorithm in overall performance.