摘要
分析了BP神经网络、多元线性回归两种机器学习算法针对发电机定子线圈出水蒲莹温度模型、指纹模型实现预测分析的理论依据及实现方法;针对300 MW、660 MW汽轮发电机进行了验证计算及效果分析;提出了实时监测与综合分析相结合的应用模式以及基于发电机水电回路温度测点分布的综合分析方法,介绍了两个典型应用案例。
The theoretical basis and implementation method of two machine learning algorithms,BP neural network and multiple linear regressions,for the Puying temperature model and the fingerprint model of the generator stator coil outlet water to achieve predictive analysis were analyzed.Verification calculation and effect analysis are carried out for 300 MW and 660 MW steam turbine generators.An application mode combining real-time monitoring and comprehensive analysis and a comprehensive analysis method based on the distribution of temperature measurement points in the hydroelectric circuit of generators are proposed,and two typical application cases are introduced.
作者
王晓华
WANG Xiao-hua(ZhejiangZheneng Wenzhou Power Generation Co.Ltd.,Wenzhou 325600,China)
出处
《能源工程》
2022年第5期40-47,共8页
Energy Engineering
关键词
定子线圈
蒲莹温度模型
指纹模型
BP神经网络
多元线性回归
stator coil
Puying temperature model
fingerprint model
BP neural network
multiple linear regression