Detection and classification of power quality disturbances using long short time memory

Abstract

The detection and diagnosis of power quality (PQ) issues are critical themes in the electrical power system's generation, transmission, and distribution. PQ issues indicate that the electric power system is not operating at rated power frequency at optimal voltage and current. Nonlinear loads, power electronic converters, system malfunctions, and switching events are the most common causes of PQ issues. The major purpose of this thesis is to use a new artificial intelligence (AI) technique based on automatic feature extraction to discover and identify PQ difficulties. Simple and difficult PQ problems are the two types of PQ problems. Voltage interruption, sag, flicker, swell, and surge are examples of simple PQ issues. Complex PQ problems are like sag with harmonic distortions and swell with harmonic distortions. The proposed AI technique consists of a dedicated architecture of the Long Short-Term Memory (LSTM) network, which is a special type of Recurrent Neural Networks (RNNs). This LSTM approach concentrates on long and short-term PQ problems, with time intervals ranging from a few milliseconds to hours and days. Deep Neural Networks (DNNs), which have emerged as a powerful tool for machine-learning challenges and are known for real-time data processing, parallel computations, and the ability to work with a big dataset with improved accuracy.