Abstract

During the shale gas extraction process, affected by the internal pressure of the geological layer and other factors, the internal pressure will gradually decrease with time, and the production will also decrease, and it is necessary to rely on artificial pressurization and other ways to keep the production stable. Accordingly, we analyzed the high-frequency data of shale gas production obtained from a block in the southwest shale gas field, and proposed a data-driven prediction model combining Principal Component Analysis (PCA), Particle Swarm Optimization (PSO) and Long Short-Term Memory (LSTM), which can determine whether artificial pressurization is needed from the predicted results. The model adopts multi-variate input and uni-variate output, firstly, the PCA algorithm for processing the characteristic parameters (casing pressure, oil pressure, pre-valve temperature) and labeling parameters (instantaneous production), secondly, the PSO algorithm is used to iteratively search for the optimal hyper-parameters of the LSTM to find the most suitable hyper-parameter configuration, and finally, established a data-driven shale gas production prediction model. The combined model is analyzed through an example study, and the accuracy is higher in shale gas production prediction.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call