<![CDATA[Gas production forecasting plays a crucial role in supporting planning and management within a country's or region's oil and gas (O&G) sector. Artificial Neural Networks (ANN) are among the most effective methods for forecasting; however, the main challenge lies in determining the optimal number of neurons and hidden layers to achieve high prediction accuracy. This study aims to design an ANN architecture using the backpropagation algorithm to predict gas well production. The research stages include gas production data collection, data pre-processing, prediction processing, accuracy and error testing, and implementation. The dataset consists of daily natural gas well production records from January to December 2022, used for both training and testing the model. The results indicate that the ANN architecture 16-4-1 with a sigmoid activation function achieved the highest accuracy of 99.978% and the lowest MAPE of 2.192%. These findings demonstrate that the developed model can accurately represent the complexity of historical daily production data, making it a valuable tool for O&G field managers to plan production, optimize operations, and support strategic decision-making in the energy sector.]]>
Copyrights © 2025
