Novel Maritime Condition monitoring technology by acoustic emission and machine learning processing for Lifetime Optimisation.

Ship propulsion systems, like engines and thrusters, are crucial for the safe operation of sea vessels. Other rotating machinery, such as anchor winches and pumps, also play vital roles. Ensuring the safety and reliability of these components is essential. Typically, ships employ maintenance routines and sensors to monitor operational conditions like vibrations and temperature. However, this approach often detects significant damage only when it's well-developed, like a fractured gear tooth or a large bearing surface damage. By this point, a total breakdown may be imminent, allowing little time for preventive measures. The MarCoPolo research project introduces an innovative approach. It combines advanced machine learning (ML) and acoustic emission (AE) technologies in ship thrusters and other machinery. The goal is not just to detect subsurface damage early but also to characterize damage in real-time and integrate it into a digital twin. This integration improves the assessment of Remaining Useful Life (RUL) and enables better predictive maintenance planning and measures for extending the lifetime, such as power restrictions or a "limp mode." The RUL estimation process involves using signals from remote AE sensors to determine the size and locations of fatigue cracks. These signals are then analysed to assess fracture crack growth numerically. The technology also considers the effects of component materials, heat treatments, and surface modifications on lifetime, including fatigue crack initiation and growth. By accounting for resulting stress states and material properties, this approach aims to optimize the lifetime of gears and bearings through tailored manufacturing processes.

Ship propulsion systems, e.g., engines, thrusters, and shaft lines, as well as other rotating machinery, e.g., anchor winches and pumps (illustrated in Figure 1), are fundamental for the operation and safety of all sea going vessels. Common for these applications is that safety and reliability of the machinery cannot be compromised. In modern ships, there are often both comprehensive maintenance regimes as well as sensors providing input on operational conditions, such as vibrations and temperature. However, this approach to damage detection requires that a significant damage (usually a fractured gear tooth or a large damage in a bearing surface) is already developed and causing elevated temperatures and increased vibrations. At this point, total breakdown of the component is often imminent, leaving little time for lifetime extending actions and maintenance planning. The innovation in the current research project, MarCoPolo, consists of the employment of advanced machine learning (ML) and acoustic emission (AE) technologies in ship thrusters and other rotating machinery, not only to enable very early detection of subsurface material damage and degradation, but also through real-time damage characterisation and integration in a digital twin establish improved RUL assessment, as well as predictive maintenance planning and lifetime extension measures, such as power restrictions or a “limp mode”. The RUL estimation involves determination of the size and locations of fatigue cracks by processing of signals from remote AE sensors and linking this to numerical assessment of fracture crack growth. In addition, the effects of component materials, heat treatments and surface modification on lifetime (fatigue crack initiation and growth) will be included by also considering the resulting stress states and material properties. This technology will be used to optimise lifetime of the gears and bearings by tailoring the manufacturing processes.