Anticipatory Troubleshooting

Troubleshooting is the process of diagnosing and repairing a system that behaving abnormally. It involves performing various diagnostic repair actions. Performing these actions may incur costs, traditional troubleshooting algorithms aim to minimize costs incurred until fixed. Prognosis deals with predicting future failures. We propose incorporate prognosis diagnosis techniques solve problems. This integration enables (1) better fault isolation (2) more intelligent decision making respect employ over time. In particular, we consider an anticipatory challenge in which fix time, while reasoning about both current Anticipatory raises two main dilemmas: fix–replace dilemma replace-healthy dilemma. The question how faulty component: fixing it or replacing new one. whether healthy component should be replaced one order prevent from failing future. dilemmas by modeling them as Markov problem failures using survival analysis literature. resulting algorithm was evaluated experimentally, showing proposed yield lower overall compared do not reason faults.