对具有未知方差的线性二次高斯(LQG)问题,本文提出了具有学习特点的对偶自适应控制方法.控制器一方面能够控制系统朝着期望的目标运行,另一方面又能对未知参数进行学习,仿真结果表明了控制方法的有效性.
For the control problem of linear-quadratic-Gaussian(LQG) with unknown variances of noises, a dual adaptive control approach with learning features is proposed. The controller, on the one hand, drives the system toward a desired state; on the other hand, it performs the learning process to reduce the uncertainties of unknown parameters. Simulation results validate the proposed approach.