论文部分内容阅读
针对丙烯腈聚合过程的强时滞和较大参数不确定等特性,该文提出一种基于二阶段自适应多模型的广义预测控制方法。该方法首先根据系统的参数范围,建立多个自适应模型,应用最小二乘算法分别进行参数估计。再利用各自适应模型的参数估计值和预报误差计算模型的权值,将各参数估计值加权求和得到最终参数估计值。将该参数估计值作为参数的真值,利用广义预测控制算法确定各时刻的控制作用。仿真结果显示:该方法能使系统未知参数快速收敛到真值,同时系统的动态性能和对理想温度的跟踪精度较常规多模型自适应控制有明显的提高。
Aiming at the characteristics of strong delay and uncertain parameters of acrylonitrile polymerization process, a generalized predictive control method based on two-stage adaptive multi-model is proposed. The method first establishes a plurality of adaptive models according to the parameter range of the system, and estimates the parameters respectively by using a least squares algorithm. Then, the weights of the model parameters and the prediction errors are calculated, and the final parameters are estimated by weighted summation. The parameters estimated value as the true value of the parameter, the use of generalized predictive control algorithm to determine the role of control at all times. The simulation results show that this method can make the unknown parameters of the system converge to the true value quickly, meanwhile the system dynamic performance and tracking accuracy of the ideal temperature are obviously improved compared with the conventional multi-model adaptive control.