摘要
针对铝电解槽中氧化铝浓度无法实时测量、传统控制方法过分依赖专家经验且难以精准控制的问题,提出了基于非线性模型预测控制方法(Nonlinear model predictive control,NMPC)的氧化铝浓度精确控制策略。首先,通过铝电解生产机理分析,确定了NMPC内部模型(简称内模)的输入输出变量;其次,采用基于最小二乘支持向量机的非线性Hammerstein系统子空间辨识方法,建立了数据驱动的氧化铝浓度控制的状态空间模型,同时,为了降低模型计算的复杂度,用多项式拟合模型中表示非线性特性的核函数;再次,根据拟合多项式的反函数,设计了非线性控制系统结构和性能指标函数。最后,基于某铝厂实际生产数据验证了所提控制方法的有效性和正确性。
An accurate control strategy for alumina concentration based on nonlinear model predictive control(NMPC) was proposed aiming at the problems that the alumina concentration in aluminum reduction cells cannot be measured in real time,while the traditional control methods rely too much on expert experience and are difficult to control accurately.Firstly,the input and output variables of the internal model of NMPC are determined through analysis on the production mechanism of aluminum electrolysis.Secondly,the data-driven state space model of alumina concentration control is established based on the non-linear Hammerstein system subspace identification method with the least squares support vector machine,meanwhile,in order to reduce the complexity of model calculation,the kernel function representing the nonlinear characteristics in the polynomial fitting model is adopted.Thirdly,the nonlinear control system structure and performance index function are designed according to the inverse function of the fitting polynomial index.Finally,the validity and correctness of the control method are verified based on the actual production data of an aluminum plant.
作者
阎群
刘波
路辉
崔家瑞
黄若愚
王义轩
李擎
曹斌
Yan Qun;Liu Bo;Lu Hui;Cui Jiarui;Huang Ruoyu;Wang Yixuan;Li Qing;Cao Bin(School of Automation and Electrical Engineering,University of Science and Technology Beijing,Beijing 100083,China;Guiyang Aluminum Magnesium Design&Research Institute Co.,Ltd.,Guiyang 550003,China)
出处
《轻金属》
北大核心
2022年第8期21-25,32,共6页
Light Metals
基金
贵州省科技成果应用及产业化项目(黔科合成果[2021]一般085)
中国博士后基金项目(2021M690798)。
关键词
铝电解
氧化铝浓度
下料控制
非线性子空间辨识
非线性模型预测控制
aluminum electrolysis
alumina concentration
feeding control
nonlinear subspace identification
nonlinear model predictive control
