Harmonic/Interharmonic Detection and Estimation based-SimPowerSystems for Machine Learning

This paper presents new methodologies for the detection and estimation of harmonics and interharmonics in data sets, using specialized technology libraries from Simscape Power Systems (SPS). A dedicated MATLAB script has been developed to analyze each phase by simulating short-circuit faults on a standard IEEE reference base. Various direct, inverse and homopolar harmonic components are generated and randomly injected at different points of the test network. During simulation, current and voltage signals from the buses are recorded at a frequency of 20 kHz, creating a rich database for automatic learning. This method, which can be extended to more complex networks and higher harmonic orders, extracts precise characteristics in terms of frequency, amplitude and onset/disappearance time. It is also adaptable to non-standard networks modeled in Simscape. The proposed approach thus modernizes and enhances advanced harmonic phase detection (HPD) and estimation (HPE) techniques, while implementing an efficient time-delay (TD) method for fault detection. Finally, the resulting database provides a reliable basis for training robust models and implementing critical protection functions.