The challenge of diagnostic accuracy in bearing fault diagnosis is affected by domain differences in cross-domain data. This study proposes an adaptive diagnostic method that integrates One-Dimensional Convolutional Neural Networks (WDCNN) with Domain-Adversarial Training (DANN). This method employs WDCNN for hierarchical feature extraction, integrating the gradient reversal layer and domain classifier from DANN to minimize the feature distribution differences between the source and target domains. During the domain-adaptive training, the method dynamically adjusts the gradient reversal strength to balance the classification task and the domain-adversarial task. To validate the effectiveness of the proposed method, comparative experiments with multiple load transfers were conducted on the CWRU bearing dataset. This research provides an efficient domain-adaptive solution for the field of bearing fault diagnosis, which is significant for improving the diagnostic accuracy of rotating machinery.