一类工业过程子空间预测控制器的设计

Design of Subspace Predictive Controller for a Class of Industrial Processes

针对一类非线性时变、动态特性突变显著,且无法进行准确机理建模的工业过程,提出一种基于机理分析与数据驱动方法相结合的子空间预测控制方法。该方法通过比较预测误差在线更新预测模型,并能根据能够根据反馈误差调整滚动窗口长度,增强了控制器对非线性时变特征的适应能力以及对不可测干扰的抑制能力。最后,通过对废杂铜冶炼过程的实际运行数据进行仿真研究,验证了方法的有效性。

For a class of industrial processes which have characteristics of nonlinear time-varying, significant dynamic mutation and cannot be accurately and mechanically modeled, this paper proposes a predictive control method based on mechanism analysis and the data-driven method. The method compares the prediction error to update the prediction model online, and adjusts the length of the scroll window automatically according to the feedback error, which enhances the adaptability for nonlinear time-varying characteristics, and strengthens its ability to inhibit unpredictable interferences. Finally, the simulation through actual operating data of the scrap copper smelting process verifies the effectiveness of the method.