基于证据网络的多变量MPC经济性能评估

摘要

MPC控制系统作为先进控制策略,已经被广泛地应用于工业生产中.但在实际工业中,MPC控制系统的变量的软约束往往设定得比较保守,使系统无法达到最优经济性能.针对有约束的MPC控制系统,采用二次型经济性能指标函数来评价系统的经济性能,将最优工作点的求解问题转化为一个典型的有约束的线性规划问题.进而根据历史数据和二次型经济性能指标函数所得最优运行结果建立多变量MPC的证据网络模型,通过证据网络的反向推理和决策,得到造成MPC控制系统性能下降的可能原因,并提出改善控制系统性能的策略.最后通过仿真实验,验证了基于证据网络的经济性能评估的有效性.%Model predictive control (MPC), as an advanced control strategy, has been widely applied to process industry. But the soft constrain limits of MPC is often set in a conservative way in industrial application, which leads to poor economic performance. For MPC system, the quadratic economic performance index is used to assess the economic performance in the paper. After that the problem of solving the optimal solution is transformed into a typical linear programming problem with constraints. Then according to the historical data and optimal operation results obtained from quadratic economic performance function, evidential network modeling of multivariable MPC can be built. Through reasoning and decision-making, the reason which causes the system performance decline can be got. The strategy to improve the control system performance is proposed. Finally, the effectiveness of assessment economic performance based on evidential network is verified through the simulation experiments.