用BP网络求解土体的导热系数

以人工神经网络为基本工具,利用其强大的非线性映射能力,综合考虑土体的物理性质指标对其导热性能的影响,为求解土体的工程参数提供了一条新的途径。结果表明,ＢＰ网络能够充分体现土体物理指标之间的非线性关系和隐含关系,具有较高的求解能力,所求出的土体的导热系数与实测值接近,其精度远高于线性回归,且简单实用。

Solution of thermal conduction coefficient from BP network

In order to calculate the thermal conduction coefficient of earths, neural network technique is used to map the non-linearity among the physical indexes of earth and an expected result is obtained. It is shown that a BP-network is useful to map the non-linear relationship among the coefficient and other indexes and this technique is more accurate than linear regression and further applications are expected.