摘要
针对存在噪声干扰的线性系统,提出了一种基于闭环子空间辨识的自适应预测控制器设计方法。结合PI控制律,通过采用比例积分形式的目标函数,推导具有比例积分结构的闭环子空间预测控制算法,结合递推方法和滑动窗口方法将算法在线实施,并分别采用双曲变换法与快速滑窗算法进行快速Cholesky分解更新。该方法整合模型辨识与控制器设计,是一种数据驱动的方法,能够有效抑制超调,控制作用更加平滑,具有较高的在线计算效率。仿真结果表明,该方法能够对存在噪声的系统起到良好的控制效果。
A approach to adaptive predictive control design is proposed based on closed-loop subspace identification for the lineal system existing noise. The PI type predictive control method based on closed-loop subspace identification by using the PI type objective function, which and is implemented online by coalescenting the recursive algorithm and moving window algorithm, and the hyperbolic transformation method and fast moving window method are used to update Cholesky decompositions. The proposed method is a data-driven method, which units model identification and controller design and improves the computational efficiency and reduces the overshoot .A number example shows the efficiency of the proposed method.
出处
《科技通报》
北大核心
2015年第11期159-162,189,共5页
Bulletin of Science and Technology
基金
中国科学院知识创新工程国防科技创新重要项目(YYYJ-1122)
关键词
闭环子空间辨识
预测控制
自适应控制
closed-loop subspace identification
predictive control
adaptive control
