CAJ | 학술논문

SQP法是求解非线性规划问题最有效的方法之一,在求解过程中,一般需要对惩罚函数进行线性搜索.惩罚因子的选择会带来一些问题,filter-SQP是Roger Fletcher和Sven Leyffer提出的一种不用惩罚函数的算法.本文红模块环境下应用改进的filter-SQP对化工过程优化进行了研究,提出了相应的算法.采用的优化策略是不可行路径法,filter中的约束目标是由断裂流方程、设计规定及不满足的不等式约束线性组合得到.使用filter检验是否接受QP步长作为下次迭代的出发点,避免了对惩罚函数进行线性搜索带来的弊端.当filter搜索失败时,提出了相应的处理策略,提高了算法的稳定性.用于判断优化是否收敛的判据不冉是K-T误差,而是目标函数和约束条件地同时收敛.提出了一个逐步规格化策略,提高了计算效率.计算实例表明,filter-SQP法优于传统的SQP法,本文提出的策略提高了算法的效率和稳定性.
SQP법시구해비선성규화문제최유효적방법지일,재구해과정중,일반수요대징벌함수진행선성수색.징벌인자적선택회대래일사문제,filter-SQP시Roger Fletcher화Sven Leyffer제출적일충불용징벌함수적산법.본문홍모괴배경하응용개진적filter-SQP대화공과정우화진행료연구,제출료상응적산법.채용적우화책략시불가행로경법,filter중적약속목표시유단렬류방정、설계규정급불만족적불등식약속선성조합득도.사용filter검험시부접수QP보장작위하차질대적출발점,피면료대징벌함수진행선성수색대래적폐단.당filter수색실패시,제출료상응적처리책략,제고료산법적은정성.용우판단우화시부수렴적판거불염시K-T오차,이시목표함수화약속조건지동시수렴.제출료일개축보규격화책략,제고료계산효솔.계산실례표명,filter-SQP법우우전통적SQP법,본문제출적책략제고료산법적효솔화은정성.
Sequential Quadratic Programming (SQP) is one of the most efficient algorithms for nonlinear optimization. A penalty fimetion is usually used when linear search, that produces some problems. Filter-SQP has been developed by Roger Fletcher and Sven Leyffer which avoids using penalty function. An improved filter-SQP in modular simulator environment for process optimization is studied. Infeasible path strategy is applied. The constraint function in filter-SQP is composed of tear equations, specific designs and unsatisfied inequations. K-T error is no longer used as the convergence criterion. The objective function and constraints converging at the same time is used to check if the optimization finishes. A successive sealing strategy improving the efficiency of optimization is proposed for filter-SQE Case study is encouraging.