论文部分内容阅读
为解决预测控制中遇到的被控对象的不确定性及模型失配等问题,提出了一种基于遗忘因子的递推子空间自适应预测控制策略。同时结合子空间辨识与广义预测方法,形成了完全数据驱动预测控制器,为了提高递推子空间算法的实时计算效率还采用了Givens旋转方法,只更新受影响的列而不是整个矩阵,避免了在每一步都要进行QR分解。通过这种优化策略,在很大程度上节约了计算时间。最后利用电炉模型进行了仿真实验,验证了设计方法的有效性。
In order to solve the problem of uncertainty and model mismatch of the controlled object encountered in predictive control, a novel predictive subspace adaptive predictive control strategy based on forgetting factor is proposed. At the same time, a complete data-driven predictive controller is formed by combining subspace identification and generalized prediction. In order to improve the real-time computation efficiency of the recursive subspace algorithm, a Givens rotation method is also adopted to update only the affected columns instead of the entire matrix, thus avoiding QR decomposition at every step. Through this optimization strategy, to a large extent, saves the calculation time. Finally, the electric furnace model was used to simulate the experiment to verify the effectiveness of the design method.