Interpretability and causal discovery of the machine learning models to predict the production of CBM wells after hydraulic fracturing

Machine learning approaches are widely studied in the production prediction of CBM wells after hydraulic fracturing, but rarely used in practice due to the low generalization ability and the lack of interpretability. A novel methodology is proposed to discover the latent causality existed in the observed data of CBM wells, which is aimed at finding an indirect way to interpret the machine learning models. Based on the theory of causal discovery, a causal graph is derived with explicit variables, including the input, output and treatment variables. The proposed method can capture the underlying nonlinear relationship between the factors and the output, which remedies the limitation of the traditional machine learning routines based on the correlation analysis of factors. The experiment on the data of a CBM reservoir shows that the detected causal relationship between the production and the geological/engineering factors, is coincident with the actual physical mechanism. Meanwhile, compared with the traditional methods, the interpretable machine learning models have better performance in predicting production capability, averaging 5%–31% improvement in accuracy. An application is presented to optimize the fracturing scheme and validated by numerical simulation, which shows the ineterpretable method can improve the stimulated production in a high extent.