TY - JOUR AU - Khang Nguyen AU - Nhan Nguyen PY - 2022/12/31 Y2 - 2024/03/29 TI - Fault classification for photovoltaic module based on maximum power point and machine learning techniques JF - Science & Technology Development Journal: Natural Sciences JA - STDJNS VL - 6 IS - 4 SE - Original Research DO - https://doi.org/10.32508/stdjns.v6i4.1221 UR - http://stdjns.scienceandtechnology.com.vn/index.php/stdjns/article/view/1221 AB - Photovoltaic (PV) module is the key component in the solar energy system. Fault classification for photovoltaic module is necessary for safety, efficiency and reliability of photovoltaic systems. When faults of PV module occurred, the current-voltage characteristics (I-V curves) of PV module are changed leads to the shift in voltage and current of the maximum power point (MPP). In this paper, fault classification for photovoltaic module based on MPP and machine learning techniques is proposed. Through analysing the location of the MPP in I-V parameters plane, faults of PV module during performance of could be identified and classified. The parameters of voltage and current of PV module in the faults (line-to-line fault, shading, open circuit in one array, open circuit, short circuit during different irradiation levels) are collected by simulations in Simulink/MATLAB software. Then, these parameters would be processed and built into a dataset. This dataset would be fed into fault classification model for PV module using machine learning algorithms included Support Vector Machine (SVM) and k-means. The desired results of machine learning models would be accurately classified faults using parameters of voltage and current of the MPP and survey the fault diagnosis through the borderline of each classification group of the models. The result shows that the fault classification capability and accuracy of the machine learning model using SVM algorithm (98,9%) are better than the k-means algorithm (61,1%). ER -