Semi-Supervised Learning Optimization Based on Generative Models to Identify Type Of Electric Load at Low Voltage

Electrical energy is a fundamental requirement for a modern society to operate the various electrical equipment used. So that the service providers must always maintain the quality of service, one measure is to maintain the harmonic content in the system to comply with the standards set. Various kinds of electrical appliances are use energy-saving features that cause high harmonic values that can cause damage to the transformer. This study identifies the harmonic value of various types of load combinations. To obtain the load harmonics data, surveys and measurements have been carried out on household consumers served by a distribution transformer. To detect the type of electrical load based on harmonics, semi-supervised learning method is used with generative model algorithm. Method optimization is performed to produce better results from previous studies. This method yields an average of 83.5% accuracy with various experimental scenarios.