摘要
发电机内冷水系统是发电机组安全运行的一个重要组成部分,其渗漏监测是在强干扰下弱信号的识别问题。针对多工况的运行环境,提出了一种决策树结合神经网络的混合建模方法,有效地对各种工况做出识别并剔除掉大的扰动因素,极大改善了诊断模型的泛化性能和动态辨识能力。这一混合策略为发电生产过程的在线监测和故障诊断问题提供了一种新的解决思路。
The water-cooling system is very important for the safe operation of generator. The water leakage detection of the system is typically a problem of weak signal extraction under strong disturbances. Considering the multi-course running environments, this paper puts forward a hybrid approach combining the artificial neural networks（ANN） method and the decision tree （DT） method. It can effectively recognize current operation mode and remove some disturbances which improves the model＇ s ability of generalization and dynamic recognition greatly. The hybrid strategy provides a new solution to the problem of condition monitoring and fault diagnosis in the course of power generation.
出处
《水力发电》
北大核心
2006年第8期50-52,共3页
Water Power
关键词
水冷系统
渗漏监测
神经网络
决策树
混合建模
water cooling system
leakage detection
neural networks
decision tree
hybrid modeling
