基于PSO-SVR模型的凝汽器真空目标值预测  

Target Value Prediction of Condenser Vacuum Degree Based on PSO-SVR Model

作　　者：李建强[1] 陈星旭[1] 赵凯[1] 

机构地区：[1]华北电力大学能源动力与机械工程学院,河北保定071003

出　　处：《电力科学与工程》2017年第2期66-72,共7页Electric Power Science and Engineering

基　　金：中央高校基本科研业务费专项资金(916021007)

摘　　要：针对凝汽器真空目标值确定问题的研究现状和存在的不足,应用粒子群与支持向量机相结合的算法建立了凝汽器真空目标值预测模型,在历史运行数据库的基础上,利用关联规则算法对循环水系统优化参数进行挖掘,为真空预测模型提供训练样本,引入粒子群算法的寻优功能对支持向量机模型的参数进行优化,使模型具有一定的有效性和泛化性,并以某600 MW机组凝汽器运行数据为例,对不同负荷下凝汽器真空目标值进行了预测并通过凝汽器真空实测值与模型预测目标值对比,实现对凝汽器真空运行状态的评估,为凝汽器运行优化及故障诊断提供了参考依据。Contraposing the researches and deficiencies of target value of the condenser vacuum, the prediction model of condenser vacuum is established by using the combined algorithm of particle swarm optimization and support vector machines. On the basis of the historical operation of the database, this paper used the association rule algorithm to optimize the parameters of the circulating water system to provide training samples for the prediction model of condenser vacuum and the optimization function of PSO was used to optimize the parameters of SVM model to ensure the validity and generalization of this model. Taking a 600 MW unit condenser operation data as an example, this paper realized the evaluation of the operating state of the condenser via the comparison between the measured vacuum value of the condenser and the predicted vacuum value of the model and provided a reference basis for the operation optimization and fault diagnosis of condenser.

关 键 词：真空目标值 支持向量机回归 粒子群 关联规则 软测量 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]