一种时变加权系数广义预测控制算法鲁棒性分析

A Time Varying Weighting Generalized Predictive Control Strategy and Robustness Analysis

针对标准广义预测控制算法,本文提出一种时变加权系数的广义预测控制性能指标,和标准算法相比,减小计算量并获得更好的性能指标。同时,文中讨论了两种典型的系统不确定性,并针对非结构型系统不确定性讨论该文算法的鲁棒性,给出了鲁棒控制域。通过对受控自回归积分滑动平均模型的仿真实例表明选择不同时变系数β和预测时域N,控制量加权系数λ,对系统性能有很大的影响。因此,该算法为现场操作人员的参数选择提供一个衡量的依据。最后,仿真结果验证了算法的有效性。

A time varying weighting generalized predictive control algorithm, which can reduce the amount of computation and provide better performance compared with a standard generalized predictive control scheme, is presented in this paper. Two typical system uncertainties are also discussed, and the robustness of proposed algorithm is analyzed. Based on CARIMA model, simulation results indicate that different performance lies on different parameters β, N and λ. Therefore, the strategy is to help the engineer in selecting suitable parameters without being entangled in trial and error. Simulation verifies the efficiency of the discussed strategy.