Author(s)

Mengsha Qu

Corresponding Author

Mengsha Qu

Abstract

In modern society, the use of petroleum resources and natural gas resources is on a continuous upward trend. The existence of the oil and gas industry is of crucial importance. Monitoring the normality of daily oil production in an efficient and convenient way can effectively improve production efficiency. Therefore, in this paper, by combining the mathematical model with the actual resources, an improved particle swarm optimization algorithm with an improved inertia weight (IPSO, Improved Particle Swarm Optimization), a random forest model (RF, Random Forest), and an interpretable analysis (SHAP, SHapley Additive exPlanations) are combined to construct a parallel integrated algorithm, and it is applied to the construction of the daily oil production prediction model in the oil and gas industry, that is, the IPSO-RF-SHAP model predicts the daily oil production. The daily transportation efficiency and whether there are problems in the process can be determined by comparing the accurate prediction results with the actual oil production. The IPSO-RF-SHAP coupling model can be applied in the oil and gas field, not only improving the exploration and development efficiency of oil and gas resources, but also having the potential for outward expansion; Through precise data analysis and interpretation, it enhances the scientific nature, effectively reduces labor costs and saves resources, and ensures the in-depth development of the combination of artificial intelligence and the oil and gas field.

Keywords

Oil and gas, Machine learning, Optimization algorithm, Interpretable machine learning