Abstract:

Aimming at low accuracy solutions and the premature convergence problem in the reliability redundancy optimization problems, we proposed an adaptive differential evolution algorithm, which modified mutation operator and crossover operator on the basis of the original differential evolution algorithm. In the process of evolution, the scale factor F and crossover probability CR were adaptively adjusted by trigonometric function respectively to improve the diversity of the feasible solution and convergence rate of the algorithm. It solved the low accuracy solutions and premature convergence problems of the reliability redundancy optimization problems. Experimental results show that the algorithm not only improves the accuracy of solution, but also has better stability and faster convergence rate for solving the reliability redundancy optimization problem.

Key words: nonlinear programming, adaptive differential evolution, reliability optimization, redundancy allocation, constrained optimization