A transformer health assessment ranking method: Use of model based scoring expert system

Electric utilities need to answer three questions in order to effectively manage their transformer fleets: How many are going to fail each year? When is each unit going to fail? How to allocate available resources to minimize risk? To answer these questions, transformer health diagnosis and prediction is critical. This paper proposes an attempt of combining assessment results of different health indicators using a scoring expert system to obtain a single measure that reflects overall failure time expectation of each unit. To overcome the difficulty presented by limited information, expert knowledge is incorporate to produce reasonable estimates and assumptions during the scoring process. Results of such an analysis can aid in risk or reliability based decision-making processes for effective allocation of resources.