A novel FPGA-based system for real-time calculation of the Spectral Kurtosis: A prospective application to harmonic detection

Abstract

Autonomous measurement units are indispensable for the development of the smart grid in order to get faster processing responses, stability, and to provide the user with information regarding power quality (losses, harmonics, outages, steady-state and transient events). In this paper, the development of a novel FPGA-based real-time analyzer is presented. The signal processing nucleus is based on higher-order statistics, introducing the Spectral Kurtosis as an online assessment. Indeed, this fourth-order spectrum enhances non-Gaussian behavior in very unfavorable noise conditions. In the Power Quality assessment frame, the outline of this spectrum can be used to reveal the presence of underlying electrical perturbations. In this paper it is presented an autonomous analyzer that uses the Spectral Kurtosis to detect the presence of low level harmonics. These harmonics are difficult to detect using conventional techniques and can generate problems in power distribution when they are associated to a resonant response in the powered load. The proposed system has been described in synthesizable VHDL and performs an indirect estimator of a poly-spectrum slide, based on the FFT. This synthesizable description has been implemented using a low cost FPGA, enabling a continuous monitoring of the power signal (without dead times) with a high oversampling rate. The proposed solution has been conceived to be introduced in smart grids with high penetration of Distributed Energy Resources (DER).