Sensitivity Analysis of Heat Extraction Performance and A Data-Mining Based BP Neural Network Prediction Model of Geothermal Doublet System

To find the most influential factor on heat extraction ratio of doublet geothermal system, injection temperature, injection flow rate and well spacing are analysed in this paper, the Box-Behnken response surface methodology is used and 13 numerical simulation tests are designed, and a geothermal field in Shandong is taken as the geological background. The results show that the injection flow rate influences the heat extraction performance most which also influences the selection of injection temperature and well spacing. Too small injection flow rate will reduce the heat extraction ratio of the system and make the performance fall short of expectations, while too larger injection flow rate will reduce the service life of geothermal system. Increasing the well spacing within a certain range under the same injection flow rate can improve the heat extraction performance. The effects of injection temperature on the heat extraction performance is not obvious. Furthermore, a BP neural network prediction model is established, and the training results show that after 43409 iterations, the error reaches a satisfactory result of 10
⁻⁴
. The research results can provide important reference for related projects.