摘要
为了提高简约空间序列二次规划算法的求解效率和稳定性,使其适于求解更大规模过程系统优化问题,提出了一种改进的大规模过程系统简约空间序列二次规划算法.算法采用新的基变换规则,根据基变换规则实时调整基变量,利用一种综合过滤线性搜索方法求取搜索步长,该搜索方法结合了传统搜索方法和过滤搜索方法的优点.改进算法被应用于部分小规模Benchmark算例和三个经典大规模算例求解,计算结果表明,本算法可以较大地减少迭代和函数估计次数,其求解效率远高于标准序列二次规划算法,且算法的稳定性也有较大提高.
To improve the efficiency and stability of reduced sequential quadratic programming （rSQP） algorithm, and extend it to solve much larger scale problems of process systems, an improved rSQP algorithm for optimization of large-scale process systems was presented. In the improved algorithm, a new rule for selection of basis was adopted. Basis was adjusted at every iteration according to the rule, so the stability can be improved obviously. Also, an integrated line search of filter method, with the advantages of normal line search method and filter method, was incorporated into the algorithm to obtain steplength. Numerical results of some benchmark examples and three large examples with variable dimension show that the proposed algorithm can reduce the number of iterations and function evaluations, and is much more effective than standard sequential quadratic programming （SQP） algorithm. Also the stability of rSQP algorithm is improved.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2005年第10期1470-1474,1506,共6页
Journal of Zhejiang University：Engineering Science
基金
国家"863"高技术研究发展计划资助项目(2002AA412110)
国家重点基础研究发展规划项目(2002CB312200)
