In case of a permanent fault at the electrical distribution system (EDS), a centralized self-healing scheme (SHS) is deployed to automatically identify the location of the fault and to restore the electrical service to as many users as possible. The service restoration is carried out considering topological and operational constraints, in a short amount of time and with minimal human intervention. A true SHS is only possible within a smart grid context, wherein electrical and topological variables are supervised and remotely controlled via SCADA and smart metering. Thus, this paper designs and simulates a centralized SHS that, after a permanent fault, executes the following set of functions: a three-phase state estimator via nonlinear programming (NLP), a short-term load forecasting method via autoregressive integrated moving average (ARIMA), a bus-matrix-impedance-based fault location method, and a service restoration method via mixed-integer nonlinear programming (MINLP). The proposed SHS has been developed as a geographic information system (GIS) with a graphical user interface (GUI) in python. Results were simulated using real unbalanced three-phase EDS, and the proposed SHS has shown to be adaptive, robust, scalable and fast for real-world applications.