ANALISIS PERKIRAAN BEBAN LISTRIK JANGKA MENENGAH RUMAH TANGGA ULP MATTIROTASI PAREPARE AREA SOREANG DENGAN METODE EXTREME LEARNING MACHINE (ELM)

Abstract

Electricity load estimation is one way to reduce the risk of unstable electricity supply by predicting the electricity load for the next day by utilizing a number of data that depend on the model prepared. In writing this thesis, electricity load forecasting aims to obtain accurate electricity load predictions using artificial neural networks using the Extreme Learning Machine (ELM) method. Extreme Learning Machine (ELM) is a new learning method in artificial neural networks with a single layer feedforward neural networks model. In predicting electrical loads, the data will be trained and the most optimal weights will be found. Next, by carrying out the training process on trained data, it will be known how well the pattern is recognized by the network so that the error value obtained reaches the minimum value. With the validation test, a value will be obtained from the estimated electricity load for the following month using the optimal weights from the training process. Based on the implementation carried out on electricity load data for the city of Parepare in the Soreang area, which uses data from July 2017 to July 2023, it is known that from the two inputs tested, the Extreme Learning Machine (ELM) provides electricity load estimation results with Mean Absolute Percentage Error (MAPE) values in 1 input parameter is 0.389% training and 0.807% testing and 2 input parameters are 0.403% training and 0.833% testing. The value of the Mean Absolute Percentage Error (MAPE) is based on the standard range of Mean Absolute Percentage Error (MAPE) values, indicating that the value range below 10% has very good performance.