Harmonics Detector in Distribution Systems based on Compressive Sensing

Harmonics identification in distribution systems has always been a difficult task, due to the variety of the loads, more and more often nonlinear and unbalanced, and the spread of the harmonics along the grid. Typically, few real measurements are available in distribution systems, commonly in the HV-MV substations, whereas monitoring of power quality phenomena requires specific methodologies and measurement devices. Due to the costs of dedicated devices, the problem can be faced by estimating the harmonic sources existing in the network, rather than the harmonic status of the entire grid. In the complex scenario of modern distribution grids, it can be reasonable to consider a distributed monitoring system in charge of observing specific parts of the grid where the number of harmonic sources is lower than the number of loads connected to the grid. In this regard, this paper proposes an identification approach based on Compressive Sensing (CS). The proposed CS method is applied to a small distribution grid, where few loads behave as harmonic sources, thus leading to the sparsity that is exploited in the mathematical formulation of the problem. The obtained results are presented and discussed, in order to show the features of this promising methodology.

I. Introduction

Modern distribution grids are characterized by a large variety of loads. The majority of them are power-electronics based, which causes power quality (PQ) phenomena along the grid. One major PQ-related issue is the harmonic pollution [1]. The harmonics reduce the life time of the components, incrementing costs [2], [3], and their accurate identification is still an open challenge.