Photovoltaic (PV) panels can develop defects due to harsh operating conditions, environmental factors. These defects not only reduce performance but also increase the risk of fire, making it essential to promptly identify and address such issues to ensure system safety and reliability. This paper introduces a fault classification method utilizing the Tree-Based Pipeline Optimization Tool (TPOT) to assess the effectiveness of automated machine learning (AutoML) in detecting two types of line-to-line faults in series-parallel configurations. Accurate classification of these faults facilitates timely maintenance, enhancing both the efficiency and safety of PV installations. The proposed approach leverages irradiance, temperature, and current-voltage (I-V) curves as input parameters to optimize the algorithm for precise fault identification. To improve the quality of input data, preprocessing techniques are integrated into the TPOT pipeline. During the training phase, performance metrics such as accuracy, precision, recall, and standard deviation are evaluated. The validation phase demonstrates the advantages of the proposed method in classifying line-to-line faults. Comparative analysis with other algorithms, including Support Vector Machines (SVM), k-Nearest Neighbors (kNN), and Decision Trees (DT), highlights the superior accuracy and robustness of the TPOT-based approach. These findings illustrate the potential of AutoML tools in enhancing fault classification in PV systems, offering a reliable and efficient solution for identifying line-to-line defects and ensuring the safe operation of photovoltaic panels.