This thesis develops probabilistic models to formally verify a fault detection and service restoration system in a smart grid using a probabilistic model checker. It develops discrete time Markov chain models for fault detection, isolation and supply restoration for a Tianjin electric power network with 3 and 6 load switches. It also develops a communication protocol model. Temporal logic properties are defined and the models are verified. Results show the effect of wireless communication on failure probabilities compared to an ideal network.