Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
SEMI, MIFTAUL JATZIA
Indrianto, Indrianto
Haris, Abdul
Subject
Teknik Informatika
Datestamp
2023-05-31 07:39:31
Abstract :
The increasing demand and number of consumers is directly proportional to the
increasing demand for electrical loads from time to time. If demand is greater than
supply or vice versa, there will be a waste of electricity and system failure. the
classification of the use of short-term electrical power loads is needed for control and
consideration in scheduling maintenance of the electric power system in an effort to
reduce problems that arise due to imbalances in demand and supply. distributed and
stored for short-, medium- and long-term needs. The Learning Vector Quantization
method is an artificial neural network that is used for grouping inputs to be given and
learning at the competitive layer that is automatically supervised and is one of the
supervised learning methods that can be implemented in the problem of classifying the
use of power loads at electrical substations. This system will manage data in the form of
electricity consumption in June 2020 as training data and test data using data in the
following month. The parameters used to get the best accuracy results are learning rate
0.01, hidden 10 and epoch 100 with a total of 30 days of data. Based on the test results,
the accuracy in June was 96,6667%. By using the June 2020 substation power load
data, it can be applied as a reference for classification in the Learning Vector
Quantization method for power load classification at electrical substations.