大型过程优化问题降维分解的研究

Study of Dimension Decrement Decomposition for Solution of Large-Scale Process Optimization Problems

在过去10年里,SQP算法由于研究了一些降维分解策略,如BLW分解、LEW分解和RND分解,在求解大型过程优化问题中获得了成功的应用.本文提出一种新的降维分解(LMD)方法,它包括LEW分解和RND分解为特例,并在LMD基础上应用数值相关理论处理可能出现的矩阵病态问题,给出一个求解大型问题的快速稳健的算法.

For the development of dimension decrement decomposition stragies such as BLW, LEW and RND decomposition, SQP algorithm have got successful application on the solution of large-scale process optimization problems in the past decade. A new dimension decrement decomposition (LMD) method is presented in this paper, which includes LEW and RND decomposition as its special cases. And on the base of LMD method, Jacobian matrix stiff problem encountered possibly can be approached using the theory of numerical linear dependence, so a fast and robust algorithm for solution of large-scale problem is given.