计算机工程与应用 ›› 2016, Vol. 52 ›› Issue (23): 55-59.

• 理论与研发 • 上一篇    下一篇

多学科协同优化算法的分析和改进

王  强,郑  松,徐  傲,葛  铭   

  1. 杭州电子科技大学 自动化学院,杭州 310018
  • 出版日期:2016-12-01 发布日期:2016-12-20

Analysis and improvement of multidisciplinary collaborative optimization method

WANG Qiang, ZHENG Song, XU Ao, GE Ming   

  1. College of Automation, Hangzhou Dianzi University, Hangzhou 310018, China
  • Online:2016-12-01 Published:2016-12-20

摘要: 针对协同优化算法计算量大、优化结果多为局部最优解的问题,提出了一种改进的协同优化算法。首先,在系统级一致性等式约束中采用改进的松弛因子,使系统级优化的可行域是存在的,且可行域的范围逐步减小,以保证子学科间的一致性;其次,在子学科中,将目标函数分为一致性目标函数和子学科最优目标函数两个部分,以不同的权重相加作为子学科的目标函数,既考虑了一致性,又兼顾了子学科独立性。最后,以各子学科级独立优化结果作为初始点进行优化。采用两个经典案例对改进算法进行验证,优化结果表明,改进的算法具有更好收敛速度和可行性。

关键词: 协同优化, 松弛因子, 一致性, 独立性

Abstract: To solve the problem that CO is large in computing capacity and the results are always local optimal solution, an improved CO method is presented. Firstly, a new slack factor is introduced in system optimization to ensure the existence of system feasibility that gradually reduces to keep consistent. Secondly, the objective function of subsystem is divided into two parts:consistent objective function and subsystem optimal objective function. Those two functions are added with different weights as subsystem objective function. This not only considers the consistence, but also takes into account the independence of subsystem. Thirdly, taking results that subsystem independently optimized is as initial point for optimization. Finally two classic examples are optimized by the improved CO method, and the results show that it has better convergence rate and feasibility.

Key words: Collaborative Optimization(CO), slack factor, consistency, independence