摘要
燃气轮机电厂实际运行中,涡轮排气温度是一个重要的热参数,其传感器的状态直接影响到排气温度观测值。应用广义回归神经网络(general regression neural network,GRNN)构建了涡轮排气温度传感器状态自动检测用的网络,并从网络的最优化设计、误差控制及网络实际效果检测等方面进行了分析和研究。同时,提出传感器状态判断阈值的建立方法。以某厂实际使用的排气温度传感器为实例进行了验证,证明所建立的GRNN网络对机组传感器的状态检测有较好的工程应用价值。
Turbine exhaust temperature is an important thermal parameter in gas turbine power plants and its measured value is affected by sensors state. General regression neural network （GRNN） was used to construct an auto-detection network for turbine exhaust temperature sensors. Optimizing design of network, error controlling and effect testing were studied, and also a method of threshold for sensor detection was advanced. The network is verified by practical data from a power plant and is proved to be with good value of engineering application for sensor state detection of unit.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2009年第32期92-97,共6页
Proceedings of the CSEE
关键词
燃气轮机电厂
涡轮排气温度
传感器
广义回归
神经网络
阈值
gas turbine power plants
turbine exhaust temperature
sensor
general regression neural network
threshold
