扰动模型的准确性对模型预测控制算法的扰动抑制能力有重要影响,当前模型预测控制广泛采用的阶跃扰动模型不能准确描述进入系统的不可测扰动,扰动抑制能力有限。自适应扰动模型可以较好的描述不可测扰动,提高对扰动的预估和抑制能力。本文对采用自适应时间序列扰动模型的预测控制进行分析,研究了扰动自适应预测控制(DMCA)的闭环结构以及带宽、灵敏度函数等频域指标与控制器抗扰性能的关系。带宽大的系统抑制扰动的速度快,灵敏度函数幅值越小则对扰动的抑制能力越强。理论分析和仿真结果表明与动态矩阵控制(DMC)相比,采用自适应扰动模型的DMCA算法能够更好的预测和抑制扰动,被控变量偏离设定值的最大幅度降低60%,带宽是DMC的1.5倍、调节速度更快,在低频段有较小的灵敏度函数值。自适应扰动模型提升了DMCA控制器的扰动抑制性能,对保障系统安全平稳运行和增加效益有重要意义。
Accuracy of the disturbance model has important impact on disturbance rejection performance of model predictive control algorithm. The step disturbance model widely used in MPC at present can not estimate the unmeasured disturbance in the system accurately and has limited disturbance rejection ability. Adaptive disturbance model can estimate the unmeasured disturbance better and improve the ability of disturbance rejection. Predictive control with adaptive time series disturbance model is studied in the paper. The closed-loop structure of disturbance adaptation predictive control (DMCA) is analyzed, as well as the relationship between frequency domain indices such as bandwidth, sensitivity function and the controller's disturbance rejection ability. System with larger bandwidth is faster in disturbance rejecting while system with smaller amplitude of sensitivity function has better rejection performance. Theoretical analysis and simulation results show that DMCA with adaptive disturbance model can predict and reject the disturbance better than Dynamic Matrix Control (DMC), the maximum deviation of DMCA from set point reduces 60% and the bandwidth is 1.5 times of DMC, which means higher regulation speed. What's more, DMCA has smaller amplitude of sensitivity function in the lower frequency part. Adaptive disturbance model enhances DMCA's disturbance rejection performance and it's important for the safe operation and increase in profits.