ﭘﻴﺶ ﺑﻴﻨﻲ ﺁﻧﻮﻣﺎﻟﻲ ﺛﻘﻞ ﺩﺭ ﺷﺒﻜﻪ ﻫﺎﻱ ﺛﻘﻞ ﺳﻨﺠﻲ ﻧﻔﺖ ﻭﮔﺎﺯ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺷﺒﻜﻪﻫﺎﻱ ﻋﺼﺒﻲ ﻣﺼﻨﻮﻋﻲ

Prediction of gravity anomalies in oil an‎d gas gravimetry networks using artificial neural networks

مهندسي نقشه برداري

Accurate prediction of gravity anomalies in inaccessible areas is one of the fundamental challenges in geophysical studies an‎d exploration of subsurface resources such as oil an‎d gas reservoirs. Geographical constraints such as mountains, lakes, an‎d valleys prevent direct acquisition of gravity data in some locations. In such circumstances, the use of numerical interpolation methods an‎d predictive modeling to estimate gravity an‎d elevation values is necessary an‎d inevitable. In this study, using gravity an‎d elevation data within the geographical area of Iran, a set of machine learning an‎d deep learning models were developed to interpolate these values. The models used included deep neural network (DNN), XGBoost, Ran‎dom Forest, an‎d classical methods such as IDW, Kriging, MLS, Gaussian, an‎d Collocation. The performance of these models was eva‎luated based on RMSE, MAE, an‎d R² coefficient of determination for both gravity an‎d elevation components. Numerical results showed that the DNN model provided the most accurate performance in estimating gravitational anomalies an‎d height, with a significant difference compared to other models. This model was able to reduce the root mean square error (RMSE) for gravitational anomaly to about 29 units an‎d for height to about 30 units, while its coefficient of determination in both components was more than 0.998. The XGBoost an‎d Ran‎dom Forest models also had acceptable performance with errors of about 110 to 122 units an‎d coefficient of determination of about 0.974 to 0.978. In contrast, classical methods such as IDW, Kriging, an‎d Collocation provided lower accuracy in estimating values, with errors higher than 126 to 167 units an‎d coefficient of determination less than 0.971. These results indicate that deep learning models, especially neural networks, have a higher ability to extract nonlinear patterns an‎d provide accurate estimates when faced with complex an‎d irregular data. The use of these models not only reduces interpolation errors, but also allows generalization to data-poor areas. As a result, the framework proposed in this study can be effectively used to reduce field sampling costs, increase the accuracy of subsurface modeling, an‎d improve decision-making in the exploration stages of underground resources.

6

143717