ALSTOM气化炉的多模型预测控制  

Multiple model predictive control for ALSTOM gasifier

作　　者：于冲[1] 吕剑虹[1] 吴科[1] 

机构地区：[1]东南大学能源与环境学院,南京210096

出　　处：《东南大学学报（自然科学版）》2010年第4期783-788,共6页Journal of Southeast University：Natural Science Edition

基　　金：国家高技术研究发展计划(863计划)资助项目(2006AA05A107);江苏省科技成果转化专项资金资助项目(BA2007008)

摘　　要：为了克服ALSTOM气化炉多变量、非线性的特性给系统控制带来的困难,提出了一种多模型预测控制(multiple model predictive control,MMPC)算法.根据气化炉在多个工况下的局部线性模型,设计了一组简化状态估计的子预测控制器.通过子模型输出误差项的无量纲化处理,改进了递推贝叶斯加权策略,并对各子预测控制器的输出进行了加权计算,得到了系统的实际控制量.将该多模型预测控制算法用于ALSTOM气化炉的控制,仿真结果表明:在压力扰动测试中,即使各控制量受到严格的约束,各输出量的偏差仍能维持在规定的范围内;在变负荷测试中,系统能够迅速跟踪负荷设定值的变化,从而满足了基准测试的各项要求.In order to deal with the effect of ALSTOM gasifier＇s properties of multi-variable and non-linearity on the control quality of the system,a type of multiple model predictive control（MMPC） algorithm is proposed in this paper.Based on the local linear models of the gasifier under different working conditions,a set of simplified state estimation sub-predictive controllers are designed.Through the dimensionless treatment of the output errors of the sub-model,the recursive Bayes weighting strategy is improved.The weightings of the sub-predictive controller outputs are then calculated to obtain the actual control variables of the system.The multiple model predictive control algorithm is used to control the ALSTOM gasifier.The simulation results show that all the output biases can be kept within specified ranges in the pressure disturbance tests even if the inputs are subjected to strict constrains,and the system can rapidly track the changes of the load settings in the load change test,which meets the requirements of the benchmark tests.

关 键 词：ALSTOM气化炉 非线性 多模型预测控制 基准测试 

分 类 号：TP237.4[自动化与计算机技术—检测技术与自动化装置]