【摘要】 深度学习技术被广泛应用于旋转机械故障诊断领域，与传统的机器学习方法相比，深度域适应拥有更强的泛化能力，具有更好的可迁徙性，能解决训练集和测试集不满足独立同分布条件下的问题.本文分别对基于领域分布差异、对抗、重构、样本生成方法的深度域适应方法进行综述，首先对深度域适应方法的概念进行介绍，其次对四种不同深度域适应方法的应用性能进行综述，最后对基于深度域适应方法的旋转机械故障诊断技术的发展趋势和有待解决的问题进行了分析，总结并探索出具有良好泛化能力的多信息融合深度域适应技术，该技术是本领域的研究热点.更多还原

【Abstract】 Deep learning technology has been widely used in rotating machinery fault diagnosis. Compared with traditional machine learning methods,deep domain adaptation has more vital generalization ability and better mobility and can solve the problems when training sets and test sets do not meet the independent and homogeneous distribution conditions. This paper reviews deep domain adaptation based on domain distribution difference,adversarial,reconstruction,and sample generation methods. Firstly,the concept of the deep domain adaptation method is introduced,and then the application performance of four different deep domain adaptation methods is summarized. Finally,the development trend and problems to be solved of rotating machinery fault diagnosis technology based on the deep domain adaptation method are analyzed,and it is concluded that exploring deep domain adaptation technology of multi-information fusion with good generalization ability is the research hotspot in this field.更多还原