An Adaptive Detection and Clustering Method of Harmonic From Noisy Signal Based on Prony

The increasing penetration of renewable energy resources based on power electronic devices results in increasing levels of harmonic distortion. It is necessary to improve the harmonic monitoring and control ability. The performance of the Prony algorithm can be promised if the order of Prony is set to half of the window length, but this will introduce many pseudo harmonic modes which interfere the harmonic mechanism screening. To solve the problem above, this paper proposes a detection and clustering method of harmonic from noisy signal. The exactly existed harmonics are adaptively detected using the estimated threshold that can reflect the level of pseudo harmonic modes caused by the time-varying background noise. Then, the harmonics of different mechanisms are classified by the agglomerative nesting (AGNES) hierarchical clustering method in the unsupervised sense. The performance of the proposed method has been verified by the numerical simulation.