基于学习的自校正控制算法

Design of Learning-Based Self-Tuning Controller

提出了一种基于学习的自校正控制算法 ,算法中包含一个自适应模型和多个固定模型 ,每一个模型都有一个相对应的控制器 .在每一采样时刻 ,将当前时段内具有最小预测误差的模型对应的控制器的输出作为控制输入 .在该算法中 ,自适应模型和自适应控制器的作用是确保闭环系统的稳定性和输出跟踪误差的渐近收敛性 ,而固定模型和固定控制器的作用是当被控对象的参数发生变化时 ,在自适应模型的参数估计收敛之前 ,暂时担当控制器的角色 ,以改善闭环系统的暂态响应 .文中证明了闭环系统的稳定性和输出跟踪误差的渐近收敛性 .仿真结果表明算法的有效性 .

A learning- based self- tuning control algorithm was proposed to improve the transient response when the variation of the unknown parameters of the plant is large during the operation.One adaptive model and several fixed models were included in the proposed algorithm.At every sampling instant,the controller corresponding to the model with least prediction error was selected as the final controller to the plant.The adaptive model and corresponding controllerwere used to ensure the closed- loop stability and to achieve zero tracking error,while the fixed models and corresponding controllers were used to improve the transient response when the unknown parameters of the plantvary.The closed- loop stability and the prop- erty ofzero tracking errorwere demonstrated.The two simulations reveal the effectiveness ofthe proposed scheme.