An Effective Modified Differential Evolution Algorithm for Reliability Problems

We propose an effective modified differential evolution algorithm (EMDE) to solve reliability problems in this paper. The proposed algorithm modifies the mutation operator and crossover operator of the original differential evolution algorithm (DE). The modified mutation operator enables all solution vectors to mimic the global best solution vector with linearly increased probability, and the modified crossover operator enables the EMDE algorithm to carry out large-scale searching and small-scale searching throughout the whole iteration. We can obtain satisfactory feasible solutions for reliability problems by combining the EMDE algorithm and the penalty function method. Experimental results demonstrate that the EMDE is an efficient method on solving reliability problems.