An Improved Sequential Quadratic Programming Method Based on Positive Constraint Sets.

Li Xia; Jianyang Ling; Rongshan Bi; Wenying Zhao; Xiaorong Cao; Shuguang Xiang

SQP (sequential quadratic programming) is an effective method to solve nonlinear constraint problems, widely used in chemical process simulation optimization. At present, most optimization methods in general chemical process simulation software have the problems of slow calculation and poor convergence. To solve these problems, an improved SQP optimization method based on positive constraint sets was developed. The new optimization method was used in the general chemical process simulation software named optimization engineers, adopting the traditional Armijo type step rule, L-BFGS (Limited-Memory BFGS) algorithm and rules of positive definite matrix. Compared to the traditional SQP method, the SQP optimization method based on positive constraint sets simplifies the constraint number of corresponding subproblems. L-BFGS algorithm and rules of positive definite matrix simplify the solution of the second derivative matrix, and reduce the amount of storage in the calculation process. The new optimization method saves the computer memory, and makes the optimal calculation easier to converge, using precise penalty function and step rule. The example shows that the optimization function of optimization engineers based on the improved SQP optimization method can meet the needs of chemical process optimization calculation. The iteration times can be reduced by about 10 %, and the calculation time can be reduced by 5.3 %. It is suitable for the simulation optimization calculation of general chemical process.

CHEMICAL processes; COMPUTER software; ALGORITHMS; COMPUTER storage devices; QUADRATIC programming

CET Journal - Chemical Engineering Transactions, 2020, Vol 81, p157

1974-9791

Academic Journal

10.3303/CET2081027