摘要
为了解决选择性催化还原(Selective Catalytic Reduction,SCR)烟气脱硝系统的控制和运行优化问题,基于某1 000 MW超超临界火电机组SCR系统的历史运行数据,采用系统辨识方法建立了该SCR系统的传递函数模型。在建模过程中,采用偏互信息(Partial Mutual Information,PMI)变量选择方法筛选确定传递函数模型的输入变量;将自然选择机制引入到差分进化算法中,提出基于自然选择的差分进化(Natural Selective Differential Evolution,NSDE)算法,并用于SCR系统模型的参数估计。研究结果表明:利用PMI变量选择方法筛选确定SCR系统模型的输入变量是可行且有效的,该方法的使用可以有效地降低模型的复杂度,提高模型的泛化能力;相比基本DE算法,采用NSDE算法辨识得到的SCR系统模型具有更高的精度。
Based on the historical operating data of the Selective Catalytic Reduction( SCR) system of a1000 MW ultra-supercritical coal-fired power generation unit,a transfer function model of the SCR system was established by a system identification method in order to optimize the control and operation of the SCR system. In the modeling process,the Partial Mutual Information( PMI) variable selection method was used to select and determine the input variables of the transfer function model. The natural selection mechanism was introduced to the differential evolution algorithm,and the Natural Selective Differential Evolution( NSDE) algorithm was proposed and used for parameter estimation of the SCR system model.The results showed that the proposed PMI variable selection method is feasible and effective for filtrating and determining input variables of the SCR system model,which can effectively reduce the complexity and improve the generalization ability of the model. Compared with the basic DE algorithm,the SCR system model identified by the NSDE algorithm has higher accuracy.
作者
康英伟
刘向伟
郑鹏远
杨平
KANG Ying-wei;LIU Xiang-wei;ZHENG Peng-yuan;YANG Ping(College of Automation Engineering,Shanghai University of Electric Power,Shanghai,China,Post Code :200090)
出处
《热能动力工程》
CAS
CSCD
北大核心
2019年第2期75-81,共7页
Journal of Engineering for Thermal Energy and Power
基金
国家自然科学基金(61573239)
上海市自然科学基金(15ZR1418600)
上海发电过程智能管控工程技术研究中心项目(14DZ2251100)~~
