基于稳定平台的广义预测控制算法

广义预测控制算法(GPC)大都属于间接算法,即通过辨识被控对象的参数,进行多步预测和在线滚动优化来设计控制律,所以需要求解逆矩阵,计算量大。提出一种改进的广义预测控制算法,这是一种直接算法,直接估计控制器的参数,避免求逆矩阵。这种控制器能任意逼近多自由度稳定平台这种不确定对象,不用知道系统的精确结构,同时由于预测控制的加入,使系统在线计算方便,控制质量提高了。通过MATLAB软件的仿真结果可以看出,传统的PID对于不确定对象不能很好的解决,而采用提出的改进广义预测控制算法,系统的鲁棒性和快速性都得到了改善,并且具有较好的控制效果。

Generalized predictive control algorithm based on stable platforms

Generalized Predictive Control （GPC） algorithms are indirect algorithm mostly. In other words, they design control laws by means of identifying the parameters of the controlled objects to complete multi-step projections and rolling optimization on line. So large amount of calculation need to be done to solve inverse matrix. In order to deal with this problem, a direct algorithm -modified GPC is proposed in this paper, which estimates controller parameters directly so as to avoid solving inverse matrix. This kind of controller can approach the uncertain objects like the multi-freedom stable platforms without knowing system structures in detail. At the same time, the calculation on line is convenient and the control quality is improved due to the joining of prediction controllers. The simulation results of MATLAB show that the traditional PID can not solve a uncertain object well and the GPC presented in this paper improves not only the robustness and rapidity of the system but also the control effect.