<p>This article displays a novel strategy for real-time fault recognition and location in power distribution networks. The challenge modern energy networks face has been addressed in the work. This approach offers complete network observability by making voltage and current measurements from strategically located PMUs within the phase domain in a cost-effective way. A deep learning framework is applied that employs CNNs to identify faults, forming the basis of precise fault localization. The methodology encapsulates an optimization algorithm for PMU placement that ensures deployment with maximum network coverage while remaining minimal. Extensive simulations on the IEEE 13-node standard network within the EMTP-RV platform validate the approach and show robust performance for many fault scenarios, including high- and low-resistance faults. Results have been highlighted, showing fault localization errors below 0.74%, considerably beyond the creative tactics. This work extends the state-of-the-art by fusing PMU data with deep learning techniques for a scalable, adaptable, and efficient solution for improving power system reliability and resilience.</p>