Abstract: In the context of data privacy and security,domain adaptive methods is unavailable due to inaccessibility of source domain data. A multi-source domain adaptive fault diagnosis method under no-accessing source data constraints is proposed. Firstly,the source and target domain data are aligned in the feature space by information maximization loss. Then the feature representation information of the target domain data is further mined using the self-supervised pseudo-label strategy,and the influence of noise pseudo-labels is suppressed using the entropy filtering strategy. Finally,the knowledge of multiple source domains is effectively utilized and the influence of negative transfer is suppressed to realize the fault diagnosis of wind turbine gearbox under no-accessing source data constraints through adaptive weighting. This method is applied and verified using the drivetrain dynamic simulator test bench data and the wind turbine CMS data of a wind farm. The results show that the proposed method can effectively realize the fault diagnosis of wind turbine gearbox in the target domain using only the pre-trained source domain model and the unlabeled data of the target domain.