Nowadays, the energy sector is experiencing a profound transition. Among all renewable energy sources, wind energy is the most developed technology across the world. To ensure the profitability of wind turbines, it is essential to develop predictive maintenance strategies that will optimize energy production while preventing unexpected downtimes. With the huge amount of data collected every day, machine learning is seen as a key enabling approach for predictive maintenance of wind turbines. However, most of the effort is put into the optimization of the model architectures and its parameters, whereas data-related aspects are often neglected. The goal of this paper is to contribute to a better understanding of wind turbines through a data-centric machine learning methodology. In particular, we focus on the optimization of data preprocessing and feature selection steps of the machine learning pipeline. The proposed methodology is used to detect failures affecting five components on a wind farm composed of five turbines. Despite the simplicity of the used machine learning model (a decision tree), the methodology outperformed model-centric approach by improving the prediction of the remaining useful life of the wind farm, making it more reliable and contributing to the global efforts towards tackling climate change.

Fault detection and diagnosis for industrial systems has been an important field of research during the past years. Among these systems, the Tennessee Eastman process is extensively used as a realistic benchmark to test and compare different fault detection and diagnosis strategies. In this context, data-driven approach has been widely applied for fault detection and diagnosis of the Tennessee Eastman process, by exploiting the massive amount of available measurement data. However, only few published works had attempted to deal with the dynamic behavior of the whole system including the mixing zone, circulating pumps, the reactor, the separator, the stripper, and so on, because of the difficulty of modeling physical phenomena that may occur in such complex system. In this article, an accurate model of the Tennessee Eastman process, properly tailored for fault detection and diagnosis purposes, is provided. This model shows better fault detection and diagnosis performances than all the others proposed in the literature and gives better or comparable results with the data-driven approaches. This work uses the bond graph methodology to systematically develop computational and graphical model. This methodology provides a physical understanding of the system and a description of its dynamic behavior. The bond graph model is then used for monitoring purposes by generating formal fault indicators, called residuals, and algorithms for fault detection and diagnosis. Hence, abnormal situations are detected by supervising the residuals’ evolution and faults are isolated using the nature of the violated residuals. Therefore, the dynamic model of the Tennessee Eastman process can now be used as a basis to achieve accurately different analysis through the causal and structural features of the bond graph tool.

The digitalization of the economy in the past decades has made data availability grow and become more important. From the maintenance point of view, clients are more demanding, wanting systems that will not have breakdowns while reducing exploitation costs. This challenging scenario has pushed companies in the direction of more intelligent maintenance solutions that involve choosing the best course of action in terms of system availability. Nowadays, these solutions are usually called prescriptive maintenance. This term is vaguely defined and its use is often unjustified. In this article we will discuss what really characterizes prescriptive maintenance, review some of the work published with this term and propose a generic framework to guide the development of such solutions. In the end, we will illustrate the use of the generic framework in a practical case.