kılıç, heybet (TU Delft), khaki, Behnam (New York Power Authority), Gümüş, Bilal (Dicle University), Yılmaz, Musa (Batman University), Palensky, Peter (TU Delft)

A Robust Data-Driven Approach for Fault Detection in Photovoltaic Arrays

Scheduled for presentation during the Regular session "Renewable energy integration 7" (WeA7), Wednesday, October 28, 2020, 14:10−14:30, Online discussion room 7

2020 IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe), October 26-28, 2020, Virtual

This information is tentative and subject to change. Compiled on April 24, 2025

Abstract

In this paper, a robust data-driven method for fault detectionin photovoltaic (PV) arrays is proposed. Our method is based onthe random vector functional-link networks (RVFLN) which has theadvantage of randomly assigning hidden layer parameters with no tuning. To eliminate the effect of measurement noise and overfitting in thetraining process which reduce the fault detection accuracy, the sparseregularization method is utilized which uses l2−norm with loss weighting factor to compute the output weights. To attain a strong robustnessagainst the outlier samples, the non-parametric kernel density estimationis employed to assign a loss weighting factor. Through rigorous simulation studies, we validate the performance of our proposed method in detectingthe short and open circuit faults based on only the output current andvoltage measurements of PV arrays. In addition to a stronger robustnesscomparing with the least square-support vector machine, we also showthat our proposed method provides 80% and 100% average detection accuracy for short circuit and open circuit, respectively.