FAULT DIAGNOSIS METHOD OF ROTATING SYSTEM BASED ON TRANSFER LEARNING

摘要

A failure diagnosis method of a transfer learning-based rotating body system according to a preferred embodiment of the present invention includes a first extraction model implementation step of implementing a first extraction model for extracting a first characteristic factor from source data of a source domain, the first A first learning step of implementing a first state diagnosis model by learning about a characteristic factor, a second extraction model implementation step of implementing a second extraction model of extracting a second characteristic factor from target data of a target domain, the second characteristic A second learning step of implementing a second state diagnosis model by learning about factors, a transfer learning step of transferring learning of the first extraction model or the first state diagnosis model, and learning information increased according to the transfer learning result and a state diagnosis step of diagnosing the state of the target domain using the second extraction model and the second state diagnosis model.