Power systems are naturally exposed to several fault conditions, namely short-circuits. Automatic protection and control systems are responsible for selective tripping of faulted devices and their subsequent reconnection. An application to automatically diagnosing abnormal conditions in power distribution network substations is described in this work. It is based on a model of the power network and the associated protection and control system. Its main purpose is to assist in the analysis of protection device records, explain the observed operations and detect possible misbehavior. This application is specified and implemented using a combination of Logic Programming tools and techniques, in particular non-monotonic reasoning. The model is structured in several abstraction levels and a set of strategies and preferences is used to guide the diagnosis process. The developed application was tested with several substation configurations, including sequences of events recorded at real installations.