论文部分内容阅读
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。本文在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再是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。
SQP method is one of the most effective methods for solving nonlinear programming problems. In the process of solving, it is generally necessary to search the penalty function linearly. The choice of penalty factor poses some problems. The filter-SQP algorithm is a non-penalty function proposed by Roger Fletcher and Sven Leyffer. In this paper, the application of improved filter-SQP in the module environment for chemical process optimization are studied, and the corresponding algorithm is proposed. The optimization strategy adopted is not feasible path method. The constraint objective in filter is obtained by linear combination of fracture flow equations, design rules and inequality constraints. Using filter to test whether to accept the QP step as the starting point for the next iteration avoids the drawbacks of linear search of penalty functions. When the filter search fails, the corresponding processing strategy is proposed to improve the stability of the algorithm. The criterion used to determine whether the optimization converges is no longer the K-T error, but the convergence of the objective function and the constraints simultaneously. A step-by-step normalization strategy is proposed to improve computational efficiency. The calculation example shows that the filter-SQP method is superior to the traditional SQP method. The proposed strategy improves the efficiency and stability of the algorithm.