Classification of Health Index of Distribution Substations using Supervised Learning Analysis with SVM Method

Abstrakt

As the only electricity provider in Indonesia, PLN is required to be reliable in distributing electrical energy to customers, this is greatly influenced by several PLN assets in the form of distribution substations. The function of this distribution substation is quite crucial in carrying out PLN's business processes to distribute electrical energy. In this study, efforts were made to improve the reliability of distribution substations by knowing the health index in accordance with EDIR PLN No. 017 concerning Distribution Transformer Maintenance Methods Based on Asset Management Principles as the Basis of the Health Index. By knowing the health level of the transformer at the distribution substation, the substation that has substandard criteria can be prioritized for maintenance. The research carried out was to take a sample in 1 month, namely March 2024, from a total of 239 substations, which were then classified using the Support Vector Machine (SVM) method which was compiled in the Python programming language which had been labeled with criteria on each substation. The criteria used in accordance with PLN EDIR No. 017 PLN are Good, Sufficient, Less and Poor. By using Machine Learning according to the Support Vector Machine (SVM) method with Supervised Learning, after the data samples were labeled, then from 239 sample data, it was divided into 2 data, namely training data and test data. In this study, the experiment was carried out with changes in training data by 60%, 70%, 80% and 90% which were then evaluated for accuracy using libary from Python.