论文部分内容阅读
针对序列二次规划法(SQP)在应用中存在的两个关键问题:目前广泛应用的线性搜索策略——Watchdog监控技术不能有效地避免在非最优解处因搜索方向过小引起的低效率收敛,和二次规划子问题用传统的有效集方法求解,寻优初始可行点不易确定,计算工作量大,提出了基于Han-Powel法(VF02AD程序)的改进算法。通过对两个典型考题的计算分析,证实了本方法可以提高SQP算法求解石油化工过程优化问题的整体计算效率和稳健性。
Aiming at the two key problems in application of quadratic programming (SQP), the widely used linear search strategy, Watchdog monitoring technology, can not effectively avoid the low efficiency caused by too small search direction in the non-optimal solution. Convergence and quadratic programming sub-problems are solved by the traditional effective set method. The initial feasible points are not easy to find and the computational workload is high. An improved algorithm based on Han-Powel method (VF02AD program) is proposed. Through the calculation and analysis of two typical test questions, it is proved that this method can improve the overall computational efficiency and robustness of the SQP algorithm in solving petrochemical process optimization problems.