Analisa Model Machine Learning dalam Memprediksi Laju Produksi Sumur Migas 15/9-F-14H


  • Devy Ayu Rhamadhani University of Jember
  • Eriska Eklezia Dwi Saputra


Artificial intelligence, Machine learning, Prediction, SVR, Elastic-net, Linear regression


AI algorithm learns various data streams from various sources sensors and engines to extract the analytics resulting in sound advice smart based on business needs. This deep insight makes it possible for oil and gas companies to have better visibility of the whole process and operations, thereby enabling them to make strategic decisions better. This of course leads to increased operating efficiency, cost reduction, and even reduce the risk of failure. Application of artificial intelligence using machine learning to production of oil and gas wells needs to be done to get predictive results perfect. With the support of existing field data so obtained simulation results that provide an overview of the prediction of production wells can optimizing the implementation of production performance for wells that have same production history. The simulation is carried out using the development of machine learning models, Support Vector Regression (SVR), Elastic Net, dan Linear Regression. The data which contains informations about the well production will be divided into two parts, 70% for training and 30% for testing. Of the three models will be seen which one is the best in predicting the production rate of the well 15/9-F-14H based on the RMSE and R2 score. SVR is the best model for predicting oil by producing RMSE 5.48 and R2 0.88 when testing. Elastic-Net is the best model for predicting gas by producing RMSE 966.82 and R2 0.85 when testing. There is no model that fits to predict the water production.