基于GPAD的多变量预测控制及其在PLC上的应用  

Multivariable Predictive Control Based on GPAD and Its Application in PLC

作　　者：顾卫东 尹军 杨晓亮 程发 郭骁 陈小强 GU Weidong;YIN Jun;YANG Xiaoliang;CHENG Fa;GUO Xiao;CHEN Xiaoqiang(Huanggang Dabieshan Power Generation Co.,Ltd.,Macheng 438300;CLP Huachuang Power Technology Research Co.,Ltd.,Suzhou 215123,China)

机构地区：[1]黄冈大别山发电有限责任公司,湖北麻城438300 [2]中电华创电力技术研究有限公司,江苏苏州215123

出　　处：《常熟理工学院学报》2020年第5期63-67,共5页Journal of Changshu Institute of Technology

摘　　要：为了应对工业应用中对模型预测控制算法的时效性要求,提出了基于对偶加速梯度投影法(GPAD)的多变量预测控制算法,通过GPAD算法加速预测控制中优化问题的求解.首先采用CARMA模型递推预测输出,形成QP问题,并通过对偶加速梯度投影法,对带有约束的目标函数进行最优化求解,再在求得对应控制时域的控制率后,取第一个时刻的控制量作为本次采样时刻的控制输入量,然后在下一个时刻重新形成优化问题并进行求解.通过仿真并与内点法形成对比验证算法的有效性及优越性.最后针对双输入双输出的水箱模型在可编程逻辑控制器(PLC)平台上应用,结果显示出算法在PLC平台上应用的有效性,及对约束以及干扰处理能力的高效性.In order to meet the timeliness requirements of model predictive control algorithms in industrial applications,a multivariable predictive control algorithm based on accelerated dual gradient projection method(GPAD)is proposed,and the optimization problem of predictive control is accelerated by the GPAD algorithm.Firstly,the CARMA model is used to recursively predict the output to form a QP problem.Secondly,the accelerated dual gradient projection method is used to optimize the constrained objective function.Thirdly,after the control law corresponding to the control time domain is obtained,it is taken as the control input at this sampling time at the first moment,and the optimization problem is formed again and solved at the next moment.The effectiveness and superiority of the algorithm are verified through simulation and comparison with the interior point method.Finally,the tank model for double input and double output is used.It is applied on the programmable logic controller(PLC)platform,and the results show the effectiveness of the algorithm applied on the PLC platform,as well as the high efficiency of constraints and interference handling capabilities.

关 键 词：对偶加速梯度投影法(GPAD) 多变量预测控制 模型预测控制(MPC) 可编程逻辑控制器(PLC) 最优化问题 QP问题 

分 类 号：TP19[自动化与计算机技术—控制理论与控制工程]