论文部分内容阅读
为解决预测控制综合方法中的模型不确定问题,不同于以往利用多胞模型描述的方法,提出了一种新的基于递推子空间的自适应预测控制综合方法.通过在每一步中加入当前输入输出数据重新构建Hankel矩阵,对广义能观矩阵进行更新,从而获得对应的状态空间模型;然后将新获得的模型应用于预测综合的优化求解过程,得到当前时刻的控制律.为提高算法的收敛速度,在辨识的过程中引入了基于模型匹配误差的时变遗忘因子.最后,在慢时变与线性时不变两种情况下进行仿真,验证了所提出算法的有效性.
In order to solve the problem of model uncertainty in the method of predictive control synthesis, a new method of adaptive predictive control based on recursive subspace is proposed, which is different from the method of multicomponent model described in the past. Input and output data to reconstruct the Hankel matrix, and update the generalized observational matrix to obtain the corresponding state space model; and then apply the newly obtained model to the optimization and prediction process of the comprehensive prediction to obtain the current control law. Convergence rate and the time-varying forgetting factor based on the model matching error are introduced in the process of identification.Finally, the simulation is carried out under both the slow time-varying condition and the linear time-invariant case, which verifies the effectiveness of the proposed algorithm.