In-service decision modelling for equipment maintenance: a data-driven approach

This paper confronts this gap in research by introducing and evaluating a data-driven modelling approach of in-service maintenance decisions for industrial equipment. In order to accomplish this, we have used data from real-world operational systems to develop data-driven models (DDMs) and demonstrate how they are used for in-service decision making. A revelatory case will be provided on in-service decisions for wind turbines – complex equipment and devices that offer proven sources of clean and renewable energy. Wind turbines are designed to capture kinetic energy from the wind (by means of rotors and blades) and this energy is subsequently converted to mechanical and then electrical energy (by means of the gearbox, generator and converters respectively). We specifically focus on data from wind turbine gearboxes in operation and use insights from our data-driven approach to inform the maintenance policy of a wind turbine manufacturer. Our view is that the use of real-world data ensures that practitioners and in-service decision makers can adopt the approaches developed or research for modelling in-service decisions for their own industrial equipment.
Overall, in this research we seek to make two main contributions. First, we offer a different take on equipment maintenance by using DDMs, which is an approach that leverages available data to develop models that aid in making in-service decisions. This is unlike the classical and more common maintenance philosophies which rely on a problem which is then modelled based on the problem parameters before data is employed in the model to make decisions. Second, we evaluate how DDMs can be used to aid in making in-service decisions for equipment maintenance. For this, we present a case study where we advance equipment service decision modelling research and practice through the use of real data from an operational wind farm to evaluate the DDMs.