Multi-Objective Instance Weighting-Based Deep Transfer Learning Network for Intelligent Fault Diagnosis

Fault diagnosis is a top-priority task for the health management of manufacturing processes. Deep learning-based methods are widely used to secure high fault accuracy. Actually, it difficult and expensive collect large-scale data in industrial fields. Several prerequisite problems can be solved using transfer learning diagnosis. Data from source domain that different but related target increase performance domain. However, negative occurs degrades due when discrepancy between within domains large. A multi-objective instance weighting-based network proposed solve this problem successfully applied The method uses newly devised weight deal with practical situations where It adjusts influence on model training through two theoretically indicators. Knowledge performed differentially by sorting instances similar terms distribution useful information task. This optimization process maximizes learning. case study an robot spot-welding testbed conducted verify effectiveness technique. applicability observed detail same as actual field comparison. diagnostic accuracy robustness high, even few used. Thus, technique promising tool successful