基于权重因子自校正的主蒸汽温度外挂广义预测串级控制  

External Generalized Predictive Cascade Control for Main Steam Temperature Based on Weight Factor Self-regulating

作　　者：王懋譞 王永富[1] 柴天佑 张晓宇 WANG Mao-Xuan;WANG Yong-Fu;CHAI Tian-You;ZHANG Xiao-Yu(School of Mechanical Engineering and Automation,North-eastern University,Shenyang 110819;State Key Laboratory of Synthetical Automation for Process Industries,Northeastern University,Shenyang 110819;China Energy Investment Cor-poration Limited,Beijing 100011)

机构地区：[1]东北大学机械工程与自动化学院,沈阳110819 [2]东北大学流程工业综合自动化国家重点实验室,沈阳110819 [3]国家能源投资集团,北京100011

出　　处：《自动化学报》2022年第2期418-433,共16页Acta Automatica Sinica

基　　金：国家自然科学基金(51775103)资助。

摘　　要：针对电厂目前普遍采用PI-PI串级控制器调节锅炉主蒸汽温度系统,不能有效克服惯性、时滞和参数时变等问题的影响,本文提出了一种理想GPC(Generalized predictive control)-PI串级控制器.首先,该理想串级控制器不仅能抑制一次和二次扰动,而且外环GPC通过对主蒸汽温度的多步预测,并结合滚动优化技术能有效克服主蒸汽温度系统的惯性和时滞问题.另外,针对主蒸汽温度系统参数时变的特性,该理想控制器采用了T-S(Takagi-Sugeno)型模糊神经网络(Fuzzy neural network,FNN)作为主蒸汽温度模型,该模型能够通过反馈校正技术实时更新模型参数.同时,为了改善主蒸汽温度系统动态响应品质和稳定性,对外环GPC中的权重因子进行了模糊自校正设计,通过理论分析和对比仿真验证了该理想GPC-PI串级控制器优于权重因子固定的GPC-PI和PI-PI串级控制器.最后,考虑到直接将电厂集散控制系统(Distributed control system,DCS)中的PI-PI串级控制器升级为理想GPC-PI串级控制器存在安全以及风险责任等问题,故将电厂的传统PI-PI串级控制器升级成外挂的GPC-PI-PI串级控制器,既改善了锅炉主蒸汽温度的控制效果又规避了风险责任,实际应用验证了该方法的有效性.The PI-PI cascade controller,which is widely used in power plants for adjusting main steam temperature,cannot effectively overcome the negative effects caused by inertia,time delay and time-varying parameters,therefore an ideal generalized predictive control(GPC)-PI cascade controller is proposed in this paper.Firstly,the GPC-PI cascade controller could attenuate the primary and secondary disturbances.To solve the problems of inertia and time delay in the main steam temperature system,the GPC-PI cascade controller predicts the multi-step main steam temperature integrating with rolling optimization technique.In addition,facing the time-varying parameters of the main steam temperature system,the GPC-PI cascade controller adopts the T-S(Takagi-Sugeno)fuzzy neural networks(FNN)as the main steam temperature model,whose parameters can be identified and updated in real time.Meanwhile,in order to further improve the dynamic response speed and stability of the main steam temperature system,fuzzy self-regulating of weight factor in outer-loop GPC is designed.The theoretical analysis and comparative simulations verify that the ideal GPC-PI cascade controller is superior to the GPC-PI cascade controller with fixed weight factor and the PI-PI cascade controller.Finally,considering the safety and risks due to the substitution of PI-PI cascade controller by the ideal GPC-PI cascade controller in distributed control systems(DCS),the PI-PI cascade controller is upgraded to the GPC-PI-PI controller for power plant,which not only improves the control effects but also avoids liabilities for risks.The practical application demonstrates the effectiveness of the GPC-PI-PI controller.

关 键 词：主蒸汽温度 广义预测控制 模糊神经网络 权重因子 自校正 串级控制 

分 类 号：TM621[电气工程—电力系统及自动化]