摘要
海水循环冷却系统中钙离子的检测问题已成为阻碍系统精细化管理的重要限制因素。文章将随机森林算法与海水循环水中钙离子浓度预测结合,充分总结运行经验,采用易实现在线检测的电导率值、pH值和系统日期作为模型特征,结合随机森林算法,实现海水循环水中钙离子浓度预测。利用随机森林算法和宁海电厂5#海水循环冷却系统2014年与2015年海水循环水水质监测数据,建立海水循环水中钙离子浓度预测模型。通过五数概括法进行数据清洗,将样本划分为训练集和测试集,采用五折交叉验证和网格搜索法优化模型参数。训练集和测试集上模型的验证和评价效果良好,该模型可用于预测海水循环水中钙离子浓度。
The detection of calciums in the seawater closed-cycle cooling system has be-come an important limiting factor that hinders the refined management of the system.This article combined the stochastic forest algorithm with the seawater closed-cycle cooling system,fully sum-marized the operating experience,used the conductivity value and pH value that can be easily de-tected online,and combined the stochastic deep forest algorithm to achieve the calcium concentra-tion prediction in the seawater closed-cycle water.Based on the random forest algorithm and the monitoring data of seawater closed-cycle water quality of the 5#seawater closed-cycle cooling sys-tem of Ninghai Power Plant from 2014 to 2015,a prediction model of calcium concentration of sea-water closed-cycle cooling system was established.The data is cleaned up by the five-number gen-eralization method,the samples are divided into training sets and test sets,and the model param-eters are optimized by the five-fold cross validation and grid search method.The validation and evaluation results of the model on the training set and test set are good,and can be used to predict the calcium concentration of seawater closed-cycle cooling system.
作者
张益
汤益琛
ZHANG Yi;TANG Yichen(ChinaGuoneng Zhejiang Ninghai Power Generation Co.,Ltd.,Ningbo 315000,)
出处
《盐科学与化工》
CAS
2024年第7期19-22,26,共5页
Journal of Salt Science and Chemical Industry
关键词
海水循环冷却
机器学习
随机森林算法
五折交叉验证
回归
Seawater closed-cycle cooling
Machine learning
Random forest algorithm
Five fold cross validation
Regression
