Oil well productivity prediction in tight reservoirs based on machine learning

Abstract

The complex seepage laws and high production costs in tight oil reservoirs have prompted scholars to apply machine learning methods to optimize construction plans and predict development results. The machine learning method uses artificial intelligence algorithm to make data "speak" to reveal the internal relationship and change rule among parameters in the process of system operation. In this paper, principal component analysis, k-means clustering, time series and other methods are used to predict the production capacity of tight oil reservoirs, reveal the development law of tight oil reservoirs, and guide the efficient and rapid development of unconventional resources in China.