This research aims to investigate the application of the Artificial Neural Network (ANN) with Backpropagation algorithm in forecasting natural gas production. The methodology involves modeling the prediction of natural gas production with the objective of achieving the lowest Mean Absolute Percentage Error (MAPE). The data collected for this study encompasses production data from natural gas wells during the period from January to December 2022. This dataset will be utilized for both training and testing the artificial neural network model. With sigmoid method the research results show a neural architecture of 16-4-1 neurons with the best accuracy value of 99.97808219% and the lowest MAPE percentage value of 2.191780822%. This predictive model for gas well production provides insight into the complexity of historical daily production data from production wells. Additionally, the model serves as a useful forecasting tool to enhance the productivity of natural gas from existing production wells.
Keywords: Artificial Neural Network, Backpropagation Algorithm, Forecasting, Mean Absolute Percentage Error (MAPE), Sigmoid Method, Natural Gas Production.