摘要
基于BP神经网络建立了热管式真空管集热器热性能的预测校正模型。该模型采用了LM和BR两种网络训练方法,经实验数据验证,预测校正模型输出结果的最大相对误差为2.8%,平均相对误差为1.2%,而数学模型输出结果的最大相对误差为6.2%,平均相对误差为4.3%,证明此预测校正模型的预测校正效果较好。应用该集热器热性能预测校正模型,可较精确地预测出不同运行状态、不同环境下集热器的出口温度,提高了系统的仿真精度。
A predictive correction model for thermal characteristics of heat-pipe evacuated tubular solar collector was established based on BP neural network. The model was trained with the experimental data by means of LM and BR. The maximum relative error output of the predictive correction model is 2.8%, the average relative error is 1.2%. The maximum relative output error of the mathematical modle is 6.2%, the average relative error is 4.3%. The result proves that the predictive correction model is more accurate. The application of the predictive correction model can improve the simulation accuracy and accurately predict the outlet temperature of solar collector in the different running state and environment.
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2008年第6期690-693,共4页
Acta Energiae Solaris Sinica
基金
青岛市科技发展计划项目(03-2-sh-08)
关键词
热管式真空管太阳集热器
集热效率
BP神经网络
数学模型
预测校正模型
heat-pipe evacuated tubular solar collector
heat-collecting efficiency
BP neural network
mathematical model
predictive correction model