PREDICTION OF THE ELECTRIC POWER BY OSCILLATING WATER COLUMN WAVE POWER PLANTS ON BAWEAN ISLAND USING LSTM

Abstrak

The demand for electricity in Indonesia continues to increase in line with population growth and the expansion of economic development. This increase is not matched by the diminishing electricity resources, as fossil fuels, which are non-renewable, are being used. Therefore, there is a need for renewable energy sources that can be utilized as long-term electricity resources. The abundant marine areas in Indonesia make it a potential source of alternative energy, one form of its utilization is the Ocean Wave Power Plant using the Oscillating Water Column (OWC) method. Bawean Island in Gresik is one of the regions that has this potential, while also facing long-standing electricity supply limitations that have resulted in uneven electricity distribution among the community. The problem does not stop at power generation but also extends to the transmission system between supply and demand. This research is conducted to predict the electricity generated by the ocean wave power plant to help avoid mismatches when supplying electricity. This study uses time series data from January 1st, 2021, to May 5th, 2024, which includes wave height, length, period, and amplitude. Electricity prediction based on these parameters can be performed using deep learning-based methods that can effectively process sequential time series data, such as the Long Short Term Memory (LSTM) method, by experimenting with the number of neurons, epochs, and batch sizes. The best prediction results for the variables of height, length, period, and amplitude of the waves obtained MAPE values of 0.3657%, 0.1637%, 0.0888%, and 0.3480%, respectively. The electricity prediction results from the best parameters obtained a MAPE of 0.3549%.