摘要
综合运用模糊数学和神经网络知识构建了一个模糊神经网络模型,用以预测电站燃煤锅炉的结渣特性．通过引入反映煤灰特性的4个常用指标以及反映锅炉运行情况的两个指标,使所建模型综合考虑了煤灰特性和锅炉运行因素对结渣的影响．以实际电厂燃煤锅炉为样本,基于改进的BP(back-propagation)算法对网络模型进行了训练．为验证模型的准确性,对7台电站燃煤锅炉的结渣特性进行预测,并将该模型与只考虑煤灰特性指标的常规 BP网络模型进行比较．验证结果表明,模糊神经网络模型的预测结果与实际相符,效果优于常规BP网络模型．
In order to forecast the slagging properties of coal-fired boiler, a fuzzy neural network model using fuzzy mathematics and neural network knowledge was set up. The effect on the slagging of boiler resulting from coal ash＇ s properties and operational factors was considered in the model by introducing four usual indices which reflect the coal ash＇ s properties and two indices related with the boiler＇ s operational status. With actual coal-fired boilers in power plant as training samples, this fuzzy neural network model was trained based on improved BP （back-propagation） algorithm. Slagging properties of seven boilers were forecasted by this model to validate its accuracy. At the same time, this model was compared with the usual BP network model which only considers the effect on the slagging of boiler resulting from coal ash＇ s properties. All these show that the forecasting results of this model were consistent with the actual status and the forecasting effect was better than that of the usual BP network model.
出处
《燃烧科学与技术》
EI
CAS
CSCD
北大核心
2006年第2期175-179,共5页
Journal of Combustion Science and Technology
关键词
模糊神经网络
锅炉
结渣
预测
改进误差反向传播算法
fuzzy neural network
boiler
slagging
forecasting
improved back-propagation algorithm
