摘要
分别采用灰色模型、神经网络以及串联灰色神经网络对某机组凝汽器水侧清洁系数进行预测。结果表明,串联灰色神经网络模型优于单一预测模型,其预测值更接近真实值,在网络训练过程中实现了误差可控,将该模型用于凝汽器水侧清洁系数的预测可行。在实际应用中,可根据前3个清洗周期某时刻的数据来预测下1个周期该时刻的清洁系数,依据所需的输入参数,即可实现在线预测凝汽器的清洁系数。
The gray model,neural network model and series gray neural network model were employed to predict the water side clean coefficient of the condenser tubes in a power plant.The results showed that,the series gray neural network model was superior to the single prediction model.Its predicted value was closer to the true value,and the error was controllable during the network training process.So it is feasible and effective to predict the water side clean coefficient of the condenser tubes by applying this model.In practical applications,we can predict the clean coefficient at some time according to the one at the time in the first three cleaning cycles.By the required input parameters,the online prediction of water side clean coefficient of the condenser tubes can be realized.
出处
《热力发电》
CAS
北大核心
2013年第9期95-99,共5页
Thermal Power Generation
关键词
凝汽器
清洁系数
灰色理论
人工神经网络
灰色神经网络
预测
condenser
clean coefficient
gray theory
artificial neural network
gray neural network
prediction