Neural Network Approaches for Computation of Soil Thermal Conductivity

The effective thermal conductivity (ETC) of soil is an essential parameter for the design and unhindered operation underground energy transportation storage systems. Various experimental, empirical, semi-empirical, mathematical, numerical methods have been tried in past, but lack either accuracy or are computationally cumbersome. recent developments computer science provided a new computational approach, neural networks, which easy to implement, faster, versatile, reasonably accurate. In this study, we present three classes networks based on different network constructions, learning strategies predict ETC soil. A total 384 data points collected from literature, Artificial (ANN), group method handling (GMDH) gene expression programming (GEP), constructed trained. best each measured with coefficient determination (R2) found be 91.6, 83.2 80.5 ANN, GMDH GEP, respectively. Furthermore, two sands 80% 99% quartz content measured, performing class GEP independently validated. model estimate sand 80%.