Utilizing the Bounding Method Through FFT and SVM for Classification of Electric Power System

Abstract

This paper presents a hybrid technique that integrates the Bounding method, Fast Fourier Transform (FFT), and Support Vector Machine (SVM) for detecting and classifying disturbances in a power system under N-1 contingency scenario. By using the IEEE 30 bus system as a test case, each transmission line was taken out one at a time. The Bounding Method examined the status of the lines based on power flow limits and categorized them into three groups: safe, unsafe, and critical. We applied FFT to the voltage angle data to identify the main frequency components that indicate system instability. An SVM classifier was trained using these spectral and flow data, achieving 97% accuracy with good precision and recall. Significant angular deviations and thermal overloads occurred when lines $6-8$ and $28-27$ were disconnected, which showed that there were serious weaknesses. The results show that the FFT % SVM method works well for catching both dynamic oscillations and static overloads. This makes it a strong approach for real-time monitoring and control in modern power systems.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint