Harmonic Source Identification Based on VGG-MIX-KAN Model
Wendong Jiang, Qiutong Dong · 2024
Harmonic source identification is an essential part of harmonic research in power systems, aiming to accurately determine the sources of harmonic currents or voltages within the power system. With the widespread application of power electronic technology and the increasing complexity of power grid structures, harmonic issues have become increasingly prominent, significantly impacting the safety, stable operation, and power quality of power systems. Therefore, harmonic source identification technology has garnered extensive attention and research. This study proposes a model based on Kolmogorov-Arnold Networks (KAN) for identifying harmonic sources. The model includes three core components: a feature extraction module, a cross-domain feature fusion module, and a classification module. The feature extraction module uses a convolutional neural network to extract key features from the time series waveforms and frequency domain spectra of harmonic sources, forming their respective feature vectors. Subsequently, we have constructed a cross-domain feature fusion mechanism that merges feature vectors from both the time and frequency domains to enhance the expressive power of the features. Ultimately, these enhanced feature vectors are fed into the KAN for final classification prediction. Experimental results demonstrate that the model proposed in this paper outperforms current mainstream algorithms in the task of identifying harmonic sources in terms of performance.