With the rapid development of the internet of things and communication technology， a large amount of real-time operation data in modern industrial equipment are monitored， promoting the equipment prognostics and health management into the era of big data. Facing the challenge of the uncertainty， low value density， multi-source heterogeneous characteristics of the monitoring data， it is difficult to adaptively capture the weak fault characteristics contained in the data. In recent years， as a research hotspot in the field of machine learning， deep learning theory has received widespread attention from academia and industry， and the related researches and applications on industrial equipment prognostics and health management （PHM） have emerged， injecting fresh blood to this field. To this end， this paper reviews the development of industrial equipment prognostics and health management technology； and reviews the research results achieved by deep learning from three aspects： anomaly detection， fault diagnosis and fault prediction； discusses the hot topics of deep learning in the current industrial equipment prognostics and health management； analyses the challenges faced in engineering practice， and discusses the effective measures to cope with these challenges.Finally， the future development trends to address these challenges are discussed.