基于动态罚函数的PSO-CO算法

A PSO-CO Algorithm Based on Dynamic Penalty Factor

本文分析了协同优化算法中所存在的问题,采用动态罚函数的解决思路,对系统级中的一致性等式约束问题进行改造,使其成为一无约束问题.另外,提出不同学科分配不同的惩罚权重的方法,大大提高了计算精度.同时,以粒子群算法替代了原有的求解算法,消除了初始解对优化结果的影响,也改善了算法的整体求解速度.在Matlab软件中实现该算法的运行,同时通过两个典型算例对该算法进行验证,表明其具有较好的优化性能.

This paper analyzes the problem existing in the collaborative optimization algorithm,and modifies the consistency equality constraint problem in system level using dynamic penalty function as a solution,making it into an unconstrained problem. In addition,the proposed method that assigns different penalty weights for different disciplines will improves accuracy greatly. Meanwhile,using particle swarm optimization algorithm to replace the original algorithm,not only eliminates the impact of the initial solution for the optimization results,but also improves the overall speed of the algorithm for solving. The running of this algorithm is realized in the Matlab software. At the same time,using two typical examples to validate the algorithm shows that it has better optimization performance.