Prognosis of Electrical Faults in Permanent Magnet AC Machines using the Hidden Markov Model.

摘要

Failure prognosis and prediction of future state of operation is important to ensure continued operation and to exercise condition base maintenance. Most of the work in machines is focused on the diagnosis instead of prognosis. But time to failure, or remaining useful life is important. Generally prognosis needs large historic data sets to extract fault progression trends, which are not available in most of the cases. The problem under study: (1) A method for prognosis of electrical failures in a PM synchronous motor is presented; (2) Transient increased contact resistance faults are investigated. Similar results for turn-turn and turn to frame faults. Objective: Determine the probability of failure at the next step.