<![CDATA[Several islands in East Nusa Tenggara Province (NTT) are underdeveloped areas with insufficient electrification. Therefore, renewable energy power plants are needed, namely Oscillating Water Column Technology Ocean Wave Power Plants (PLTGL-OWC). The objective of this study is to determine the performance of the bidirectional long short-term memory (Bi-LSTM) method in predicting the potential power generated from the height, length, and period of the Sawu Sea waves in NTT using PLTGL-OWC. This study utilises Sawu Sea wave data collected every 12 hours over 9 months. Bi-LSTM is used in this study because it can overcome the vanishing Gradient problem by utilising both the forward layer and the backward layer, making it more effective in solving complex issues, such as time series prediction. This study conducted tests on hyperparameter batch size and hidden layer node configurations. The smallest mean absolute percentage error (MAPE) prediction values obtained were 9.1943% for the wave height parameter, 11.3585% for the wave length parameter, and 7.1485% for the wave period parameter. It means that the Bi-LSTM method is suitable for predicting the electrical power generated by the PLTGL-OWC in the Sawu Sea, as the height and period parameters fall within the MAPE < 10% category, and the length parameter falls within the MAPE 10-20% category. The average electrical power generated is 2,639,865.948 watts per day over a 31-day period. The Sawu Sea has the potential to serve as a renewable energy source in the NTT region.]]>
Copyrights © 2025
