低磷酸盐处理的LM-Smith控制器  

LM-Smith controller for low phosphate treatment

作　　者：吴龙庭[1,2] 曹顺安[3] 胡家元[3] 

机构地区：[1]长沙理工大学化学与生物工程学院电力与交通材料保护湖南省重点实验室,湖南长沙410004 [2]中南财经政法大学会计学院,湖北武汉430073 [3]武汉大学动力与机械学院,湖北武汉430072

出　　处：《计算机与应用化学》2014年第4期438-444,共7页Computers and Applied Chemistry

基　　金：长沙理工大学电力与交通材料保护湖南省重点实验室开放基金项目(2011CL07);中南财经政法大学基本科研业务费项目(31541111121);国家社会科学青年基金课题(编号:10CGL010)

摘　　要：电厂给水水质的提高使得低磷酸盐处理成为炉水调节的常用手段。为便于管理,炉水水样一般通过取样管传送至化学车间进行统一检测和分析。由于水样的传送和分析需要耗时,炉水磷酸根控制成为典型的时滞系统控制。现有的自动加药系统多采用PI控制,该方案能基本保证加药控制的稳定性,但在取样管较长,时滞明显的场合,不能对磷酸根浓度进行精确调节,这一缺陷在系统调峰和变负荷运行时更加明显。为改善磷酸根浓度的控制效果,本文提出了炉水低磷酸盐处理的LM-Smith控制方案。该方案根据经典的Smith控制,在取样点实施管道改造,增设现场磷酸根表,通过电信号反馈补偿管道传送的时延。为克服传统Smith控制抗干扰性差的弱点,引入Levnberg-Marquardt神经网络算法,在线辨识Smith预估模型,实现对被控对象的准确跟踪,仿真结果表明该方法响应迅速,辨识精度高,输出信号准确,鲁棒性较强,改善了Smith控制的效果,提高了磷酸盐自动加药系统的控制性能。The quality improvement of power plant feed water makes low phosphate treatment become a common means for boiler water quality control. For management convenience, water samples are often transmitted, by sampling tubes, to chemical plant for unified detection and analysis. Since the transmission is often time-consuming, the phosphate control of boiler water is a typical time-delay control problem. Currently most automatic dosing system use PI control. This scheme generally could guarantee control stability, but in case of long sampling tube, which means large time-delay, it cannot make accurate and immediate regulation of phosphate concentration. Such defect is more obvious in occasion of peak modulation and load adjustment. To improve the control effect, this paper presents an LM-Smith controller for boiler water low phosphate treatment. Based on the classic Smith control, the scheme requires pipeline renovation and phosphate table adding at sampling points. Electrical signal feedback is used to compensate transmission time-delay completely. In order to overcome the weakness of resisting interference of traditional Smith control, Levenberg-Marquardt neural network is introduced to identify Smith prediction model and track system object online. Simulation results show that this controller is of rapid response, high identification precision and excellent output accuracy, which improves the performance of both traditional Smith control and phosphate automatic dosing system.

关 键 词：磷酸盐处理 神经网络 Smith控制 水质调节 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]