基于多模型加权调度的锅炉过热汽温广义预测串级控制  

Generalized Predictive Cascaded Control of Boiler Superheated Steam Temperature Based on Multi-model Weighted Scheduling

作　　者：胡宇航 徐振华 王林 程功 刘锦廉 朱策 HU Yuhang;XU Zhenhua;WANG Lin;CHENG Gong;LIU Jinlian;ZHU Ce(State Grid Fujian Electric Power Research Institute,Fuzhou,Fujian 350007,China)

机构地区：[1]国网福建省电力有限公司电力科学研究院,福建福州350007

出　　处：《广东电力》2023年第10期121-129,共9页Guangdong Electric Power

基　　金：国家电网有限公司总部科技项目(52130422001C)。

摘　　要：针对采用常规串级比例积分微分(proportion-integral-derivative,PID)控制器调节锅炉过热汽温系统存在不易整定参数、自适应性不佳的问题,设计一种基于多模型加权调度的锅炉过热汽温广义预测串级控制器。首先,根据典型工况下喷水量扰动时的过热汽温动态特性建立模型子集,采用递推最小二乘法辨识可控自回归积分滑动平均模型参数;其次,根据辨识得到的模型参数求解各子控制器最优输出,采用高斯隶属度函数得出各子控制器输出的加权系数;然后,通过单一工况下考虑减温水调节阀特性的预测控制仿真试验,比较广义预测串级控制器和常规串级PID控制器的鲁棒性;最后,通过变负荷仿真试验及抗干扰测试,验证多模型加权调度广义预测串级控制器对变负荷时模型失配的适应性。试验结果表明,相比常规串级PID控制器,广义预测串级控制器拥有更好的鲁棒性,设定值扰动试验中最大超调量控制在2℃左右。多模型加权调度广义预测串级控制器对模型失配具有良好的适应性,并具有良好的抗干扰性能。A generalized predictive cascaded controller for boiler superheated steam temperature based on multi-model weighted scheduling is designed to address the issues of difficult parameter tuning and poor adaptability when using conventional cascade PID controllers to regulate the boiler superheated steam temperature system.Firstly,a subset of the model is established according to the dynamic characteristics of superheated steam temperature under typical operating conditions with water spray disturbance,and the parameters of the controllable autoregressive integral moving average model are identified using the recursive least squares method.Secondly,based on the identified model parameters,the optimal output of each sub controller is solved and the Gaussian membership function is used to obtain the weighted coefficients of each sub controller output.Then,through a predictive control simulation experiment considering the characteristics of the desuperheating water regulating valve under a single operating condition,the robustness of the generalized predictive cascaded controller and the conventional cascaded PID controller is compared.Finally,through variable load simulation experiments and anti-interference tests,the adaptability of the multi-model weighted scheduling generalized predictive cascaded controller to model mismatch during variable load is verified.The experimental results show that compared to conventional cascaded PID controllers,the generalized predictive cascaded controller has better robustness,with the maximum overshoot controlled at around 2℃in the set point disturbance test.The multi-model weighted scheduling generalized predictive cascaded controller has good adaptability to model mismatch and has good anti-interference performance.

关 键 词：过热汽温 广义预测控制 多模型 高斯隶属度 

分 类 号：TM621.27[电气工程—电力系统及自动化] TK32[动力工程及工程热物理—热能工程]