Kalman filter–based approach for detection of series arc fault in photovoltaic systems

Abstract

Summary Series arc fault is one of the main failures of photovoltaic (PV) systems that should be detected as soon as possible after occurrence. Electrical shock hazard and fire in PV systems are the main consequences of delayed series arc fault detection. In this paper, an approach based on analyzing high‐frequency components of voltage signal derived by Kalman filter is presented to detect series arc fault occurrence. A criterion is defined using energy variations of the estimated harmonics for this purpose. The performance of the proposed method is evaluated by plenty of simulations and experiments in different conditions. These cases include different arc lengths, electrodes movements, temporary shadings, and different types of noise. The results confirm that the proposed technique is efficient from speed response, reliability, and accuracy points of view.

Kalman filter variant intelligent control for power quality improvement in photovoltaic active power filter system

“…Other techniques such as discrete Fourier transform, wavelet transform etc. have been analyzed for extraction of the fundamental component of the load current in Singh and Singh and Masoum et al These methods are based on direct extraction whereas KF‐based approach first estimates and then extracts the fundamental frequency component. A KF‐based reference current generation scheme under various steady‐state and dynamic conditions was proposed in Ahmed et al Another control technique based on KF has been discussed in Kanieski et al combined with a linear quadratic regulator for power quality improvement.…”

Series arc fault detection in photovoltaic system by small‐signal impedance and noise monitoring

“…In Lu et al, 7 the authors provided a detailed review of SEA detection methods, and classified them as Fast Fourier Transformation (FFT), short time fast Fourier transformation (STFFT), wavelets transformation (WT), artificial intelligence (AI), statistical, and other methods. The Kalman filter-based approach is presented in Ahmadi et al, 14 which also relies on the monitoring of the high frequency components. 8,9 Pink noise, or 1/f noise, is characterized by a spectral density that is inversely proportional to the frequency.…”

“…Moreover, reported detection time is less than 10 ms, which is faster than in methods summarized in Ahmadi et al 14 The paper is organized as follows. Moreover, reported detection time is less than 10 ms, which is faster than in methods summarized in Ahmadi et al 14 The paper is organized as follows.…”

“…The analysis conducted in Johnson et al 13 demonstrated that the frequency band between 100 Hz and 100 kHz provides most of the information about SEA induced noise. The Kalman filter-based approach is presented in Ahmadi et al, 14 which also relies on the monitoring of the high frequency components. The accuracy, speed, and reliability of the proposed technique were reported as satisfactory.…”

“…When compared with FFT and WT, the method presented in the paper does not need previous information on the spectrum of the switching noise, since it automatically synchronizes with the switching harmonic and attenuates the higher order harmonics. Moreover, reported detection time is less than 10 ms, which is faster than in methods summarized in Ahmadi et al 14 The paper is organized as follows. Section 2 describes a typical grid-connected PV system structure and operation.…”