Within the power industry engineers are increasingly faced with demands for improvements in performance and reliability of the electrical equipment under their supervision. Power grids are made up of a multitude of electrical components, combined with protection equipment in place to ensure expensive assets are protected from damage. Attempting to effectively manage the sheer volume of electrical plant becomes an impossible task without remote monitoring and automated analysis of the huge volumes of data produced. During major storm conditions the information produced from network monitoring can also become unmanageable. Volumes of operational alarms, for example SCADA (Supervisory, Control and Data Acquisition) data, can reach up to 65,000 alarms within 24 hours during storm conditions. This thesis presents the use of model-based reasoning for the analysis of SCADA data for the validation of protection performance and the identification of network incidents. The base requ irements were determined in conjunction with a transmission system operator within the United Kingdom with a major task identified in reducing software maintenance whilst maintaining an accurate diagnostic result. The novel approaches taken in this system include the provision of dynamically configurable protection device models for reuse on a multitude of different network connections and the modification of the GDE (General Diagnostic Engine) to identify the occurrence of missing alarms, previously not having been applied to data of this form. The approach taken ensures that the system is easy to update as the network evolves, new equipment is added and data formats change. Models of protection devices combined with the GDE provide the ability to diagnose the maloperation of protection equipment, including multiple faults, missing alarms and new or previously unknown faults. The implemented system, PSDiagnosis, is examined including its performance as an on-line system de veloped for a utility partner.