基于系统响应特征量近邻等效的PID“拉回”式整定方法  

Research on Neighborhood Equivalent PID“Pull-Back”Tuning Method Based on System Response Characteristic Quantities

作　　者：徐铭 王培红[1] 王华广 Xu Ming

机构地区：[1]能源热转换及其过程测控教育部重点实验室,东南大学,江苏南京210096 [2]国核电力规划设计研究院有限公司,北京100000

出　　处：《工业控制计算机》2021年第8期60-62,66,共4页Industrial Control Computer

基　　金：国家自然科学基金资助项目(51976032)。

摘　　要：针对火力发电热工过程控制中PID整定复杂困难的问题,提出一种基于系统响应特征量近邻等效的PID“拉回”式整定方法。该方法分为离线训练和在线整定两部分:离线训练部分,在原期望系统(含被控对象与控制器结构与参数)的基础上,改变PID参数以及对象纯延迟时间,进行闭环设定值单位阶跃响应仿真试验,并提取上升时间、超调等响应特征量,形成训练集;在线整定部分,基于当前系统的响应特征量,根据欧氏距离最短原则,在训练集中锁定与当前系统响应特征量最近的邻域样本,将当前系统因对象变化造成的系统响应特征量变化等效为对象不变控制器参数变化引起的等效响应特征量变化,并根据期望系统与最近邻样本系统控制器参数的差异,“反向”调整当前系统的控制器参数,将当前系统的控制性能“拉回”至期望系统工况。该方法不需要精确辨识被控对象模型,不依赖于复杂的专家知识与整定规则。仿真结果表明,该方法具有较好的精度且大大简化了PID整定流程。Aiming at the complex and difficult PID tuning in the thermal control process of thermal power generation,a novel performance“pull-back”tuning strategy based on the neighborhood equivalent principle is proposed in this paper.Without precise process model and complex expert knowledge and tuning rules,in this method,based on the known process model and controller of the original expected system,training set is generated in the off-line training period by means of traversing PID parameters and delay time and extracting closed-loop characteristic quantities,such as rise time,overshoot and etc.,from set-point unit step response curves.With the help of the nearest-neighbor which is found in on-line tuning period,compared with original expected system,the influence on response quantities caused by the change of the current system process with constant controller is equivalent to the influence caused by the change of the nearest-neighbor controller with constant process.Then an anti-change of the nearest-neighbor controller is acquired and applied to the current system controller to recover the performance back to the original expected system condition.Simulation results showed that this method has good accuracy and greatly simplifies the PID tuning process.

关 键 词：近邻信息 PID整定 性能“拉回” 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TM621[自动化与计算机技术—控制科学与工程]