工业控制网络与系统研究室知识库

为解决双层结构预测控制在工程应用中存在的通用化需求及参数配置问题,针对稳态优化层研究两阶段求解策略的标准化过程,形成一系列线性规划(LP)和二次规划(QP)问题;提出目标函数含绝对值及数值截断误差问题的解决策略;针对所形成的LP及QP问题,从几何关系及代数关系两方面分析了解的特性,并分析了模型失配对LP及QP解的影响。仿真验证结果表明:随着决策变量权重系数变化,QP比LP的解具有更好的连续性,QP的权重系数能更精确调节决策变量的偏向性;可行域边界的变化导致LP的解一定发生变化,而一直在约束空间内部的QP解则不随时间变化。

In order to solve the problems of generalized demand and parameter configuration in the engineering application of two-layer predictive control, a series of linear programming (LP) or quadratic programming (QP) problems are formed by studying the standardization process of two-stage solution strategy in the steady-state optimization layer. It proposed the solution strategy of the objective function including absolute value and numerical truncation error, and analyzed the characteristics of solutions from the geometric and algebraic relations aiming at the problems of LP and QP. The simulation results prove that QP had better continuity than LP when the weight coefficient of decision variables changed, and QP's weight coefficient can more accurately adjust the bias of decision variables. The change of the boundary of the feasible region results in the change of the solution of LP while the solution of QP which has been in the constraint space does not change with time.