The building sector, primarily through heating, ventilation, and air conditioning (HVAC) systems, accounts for over 30% of global final energy consumption and 26% of energy-related emissions, highlighting the urgency for efficient energy management and effective fault detection. Optimizing HVAC system performance is crucial for energy conservation and sustainability. This study introduces a hybrid modeling methodology to enhance HVAC systems’ fault detection and isolation (FDI). Using feature extraction through principal component analysis (PCA) and autoencoder (AE), the proposed approach integrates first-principles knowledge with data to improve the performance of different clustering algorithms (K-means, density-based spatial clustering of applications with noise (DBSCAN), and ordering points to identify the clustering structure (OPTICS)) to distinguish datasets of different operating conditions (normal and faulty conditions). The proposed approach is applied to detect common faults in HVAC systems, demonstrating superior performance compared to purely data-driven methods.