Unit Three Kartini, Bambang Suprianto, Nurhayati, Sayyidul Aulia Alamsyah, Parama Diptya Widayaka, Tri Wahyu Yulianto
A novelty using the modified Long Short Term Memory (LSTM) modeling for optimized prediction of very short-term power electrical using input variable weather data with a development LSTM method. This modeling is a development-modified LSTM method. The modified LSTM model is designed to forecast power electrical 1 hours ahead of basic meteorology data for the Photovoltaic (PV) generation station on-grid. The proposed research of this modified LSTM model is taken into account using meteorology data. The method development modified is a preprocessing data to the LSTM method. In the error value of the development-LSTM simulation method, the value mean absolute percentage error value is 0.5 0/0. The modified Long STM model prediction is then compared to actual data, development decomposition, and results indicate that the Modified-LSTM model presented in the novelty from the proposed study can predict 60 minutes power electrical with optimal accuracy value. © 2023 IEEE.
Universitas Negeri Surabaya, Dept. Electrical Engineering, Surabaya, Indonesia
