【摘要】 提出一种采用BP人工神经网络预测单U型地埋管换热器埋管与钻孔壁面间的无量纲热阻(2πK_g·R_b)的新方法。将具有2个隐藏层的BP人工神经网络计算结果与2D数值模型计算结果、函数拟合公式计算结果进行了全面对比分析,结果发现:对于2D数值模型所获得的单U型地埋管换热器钻孔内无量纲热阻2πK_g·R_b而言,基于本文所构建的包含2个隐藏层的BP人工神经网络所计算的结果较文献中函数拟合公式具有更好的计算精度。更多还原

【Abstract】 An artificial neural network( ANN) model has been proposed to evaluate the effective pipe-to-borehole thermal resistance for vertical single U-tube ground heat exchanger based on the primary data of 2D numerical model of the ground heat exchanger. The comprehensive comparisons of the dimensionless effective pipe-to-borehole thermal resistance( i. e., 2πK_g·R_b) have been conducted between the available best-fit correlations in the literature,2D numerical model and the present ANN model. It is found that the proposed two-hidden layers of ANN model is more accurate than the best-fit correlations available to evaluate the effective pipe-to-borehole thermal resistance for vertical single U-tube ground heat exchanger.更多还原