基于改进LS-SVM的制粉出力软测量建模  

Study of Soft Sensor Modeling Based on Improved FIR Filter and Least Squares Support Vector Machines

作　　者：冯磊华[1,2] 桂卫华[1] 杨锋 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]长沙理工大学能源与动力工程学院,湖南长沙410077 [3]湖南江河机电自动化工程有限公司,湖南长沙410013

出　　处：《控制工程》2012年第6期944-946,951,共4页Control Engineering of China

基　　金：国家973计划项目(2009CB219803-03);长沙理工大学"动力机械及工程"项目(201006);长沙理工大学"可再生能源"湖南省重点实验室项目

摘　　要：针对火电厂双进双出钢球磨煤机制粉出力较难直接测量的问题,拟建立其最小二乘支持向量机(LS-SVM)模型实现软测量。由于LS-SVM算法缺乏"稀疏性",采用经Remez算法优化后的FIR数字滤波器与LS-SVM相结合,形成一种新的软测量建模方法—FIR-LSS-VM。仿真结果表明,经Remez优化后的FIR滤波效果更好。将该软测量方法应用于火电厂双进双出钢球磨煤机直吹式制粉系统,通过现场实验对比可知,改进的软测量模型学习速度较快且误差较小,更加适用于在线学习。According to the problem that mill output of direct fired system with duplex inlet and outlet ball pulverizer in power plant is difficult to by direct measured, this paper proposes a soft sensor method based on the least squares support vector machines （LS-SVM） methods. Because of the lack of sparseness, ES-SVM is combined with optimized FIR by Remez algorithm to form a new soft sensor modeling method FIR-LSSVM. Simulation results show that the optimization effect is better. Lastly-FIR-I_SSVM is used to establish the soft sensor model of mill output of direct fired system with duplex inlet and outlet ball pulverizer in power plant. The field tests show that the learning speed of FIR-LSSVM is faster and the error is smaller which indicates that FIR-LSSVM is better suitable for online learning.

关 键 词：钢球磨煤机 直吹式制粉系统 最小二乘支持向量机 FIR滤波器 软测量 

分 类 号：TK323[动力工程及工程热物理—热能工程]