Solar energy photovoltaic is an essential renewable source. However, its intermittent nature, combined with potential failures of photovoltaic systems, represents a major challenge to guarantee reliable and continuous production. In this context, early fault detection is crucial to maintaining system performance and avoiding energy losses. Artificial intelligence (AI) and deep learning (DL) approaches are emerging as a promising solution for automating this detection, improving diagnostic accuracy, and thus boosting the reliability of photovoltaic installations. The objective of this study is to compare the performance of artificial intelligence (AI) models for the detection and classification of photovoltaic (PV) faults based on electrical measurements. Our strategy uses the analysis of the physical parameters. The models used for this comparative analysis include K-Nearest Neighbors (KNN), Logistic Regression (LR), Multi-Layer Perceptron (MLP), AdaBoost, Random Forest (RF), Extreme Gradient Boosting (XGBoost), Long Short-Term Memory (LSTM), and Bidirectional LSTM (BiLSTM). The scope of this analysis is to determine the best and most precise models for the automated intelligent fault classification of solar panels. Like every single model, BiLSTM and XGBoost showed good performance and stood out for their respective applications as well. With respect to Balanced Accuracy it delivered its best result at 0.830 while BiLSTM achieved substantially greater metrics at 0.871. Furthermore, in addition to these metrics for some fault classes, the BiLSTM model achieved value which can be considered near perfect in recall F1-score and precision, which demonstrates that BiLSTM is capable of detecting the most critical faults in photovoltaic systems with high confidence.