| Summary: | Voltage sags or dips are one of the power qualities (PQ) disturbances in distribution system (DN). This problem can disrupt sensitive equipment and, if severe enough, result in power outages. Extraction of voltage sag features, such as amplitude, duration, and frequency of sags, can assist power system operators and engineers in better understanding the causes and impacts of these events, as well as developing mitigation strategies. However, the device for monitoring the PQ disturbances is very expensive and cannot be affordable. This research paper focused on classification of voltage sag on different types of bulbs using low-cost microcontroller of Arduino Uno and one versus one support vector machine (OVOSVM) learning. So, AC Voltage module, Arduino Uno, and MATLAB software are the apparatus used to record the real-time signal of voltage sag. Then, advanced signal processing of S-transform (ST) is applied to extract significant features of voltage sag that used as an input for classifier tools of OVOSVM. After extracting the features, OVO-SVM will be performed using Linear Kernel and Radial Basis Function (RBF). The accuracy of these two Kernel SVMs will be compared and evaluated to determine the best method for classifying voltage sag characteristics. Result shows the classification OVO-SVM using RBF Kernel is the best compared to the Linear Kernel, where its accuracy is 90.0%. © 2024 IEEE.
|
|---|