Implementasi Algoritma Long Short-Term Memory dalam Prediksi Konsentrasi Gas Metana (CH4) di Kota Salatiga

Abstract

Implementation of the Long Short Term Memory (LSTM) algorithm is done to build a prediction model that can handle complex time series data. Model development uses training and testing data and combines multiple time series to improve prediction accuracy. Model testing is done by measuring the root mean square error (RMSE) value as a performance indicator. The test results show that the application of the LSTM algorithm to the (CH4) sensor provides an optimal RMSE value, namely with a value for training data of 20% (0.09) and test data of 80% (0.14), indicating the prediction accuracy of methane gas (CH4) concentration is potentially unexploded, the results obtained have important implications for safety monitoring. This test contributes to the development of predictive methods to monitor and manage potential risks associated with (CH4) concentrations. The application of LSTM to (CH4) sensors not only improves prediction accuracy but also opens up opportunities for the development of safety systems that can more effectively predict and prevent potentially harmful phenomena due to methane gas.