Dynamic Phasor Estimation Using ℓ1-norm Constraint Maximum-Correntropy Criterion

Two algorithms namely variable centre maximum correntropy based recursive Gauss Newton filter (VCMC-RGN) and variable centre maximum correntropy based
ℓ
1-norm constraint (zero attracting) recursive Gauss Newton filter (VCMC-ZARGN) are proposed for dynamic phasor and frequency estimation suitable for Phasor Measurement Units. The correntropy term leads to better performance in estimating phasor and frequency under dynamic conditions of a power system both in Gaussian as well as impulsive noise environment. Additionally, some assumptions have been considered to compute the inverse Hessian matrix that optimizes memory resources and time consumption. In addition, the underlying sparse behaviour of the adaptive filter has been exploited through norm to significantly improve the accuracy and the convergence rate of the phasor estimation. Moreover, this work demonstrates the feasibility of implementing the new VCMC-ZARGN estimator in real time using the real-world measurement acquired from the Ecole Polytechnique Federale de Lausanne (EPFL) campus grid and also experimental setup based on the Field-Programmable Gate Arrays (FPGAs) with optimum device utilization that is authenticated using a PV-Wind hybrid system.