I. Introduction
Timely and effective maintenance actions are vital in the operation of industrial equipment since they can significantly improve the reliability, availability, and safety of equipment and minimize the breakdown loss [1–2]. Intelligent predictive maintenance (PdM) is an advanced maintenance strategy which has the capacity to predict potential failures and take maintenance actions in a timely and appropriate manner [3]. It has gradually replaced the traditional maintenance strategies, including run-to-failure maintenance (also known as breakdown maintenance), reliability-centered maintenance, preventive maintenance (PM, also known as scheduled maintenance), condition-based maintenance (CBM) and so on, in modern industry. Over the past few years, with the rapid progress of sensor and network technology, there has been a dramatic increase in the availability of vibration, temperature, pressure and other types of condition monitoring data of electrical and mechanical equipment [3–5]. Faced with the industrial big data, artificial intelligence (AI) techniques, especially machine learning, deep learning and transfer learning, have been widely applied in current PdM systems. The PdM system in the era of big data is shown in Fig. 1 [2, 6, 7]. As shown in Fig. 1, it mainly consists of four parts: data acquisition and preprocessing, fault diagnostics, fault prognostics and maintenance decision-making.