Moamar Sayed-Mouchaweh

Digitalized virtual power plants (DVPPs) are advanced software platforms that aggregate and optimally orchestrate a collection of distributed energy resources (DERs) to support grid balance and stability to alleviate stressed transmission and distribution lines and provide critical grid services. DVPPs could reach a total capacity equal to one or several nuclear power plants.

Nowadays, machines are everywhere, and while their design and construction are a major challenge, maintenance has to be one of the main problems to solve during their life cycle. Reactive maintenance is simply waiting for a component to fail before taking action, which is not only costly but also leads to downtime and can result in lost revenue. Predictive maintenance, on the other hand, is proactive and can both minimise downtime and extend the life of the equipment. However, predictive maintenance relies on reliable forecasts to avoid over-scheduling. This problem is particularly interesting for wind farms, as they are usually located in remote areas, making access to repairs difficult and time-consuming. In this paper, we discuss predictive maintenance in the context of a wind farm and propose and develop a diagnostic and recommendation model based on the random-cut forest approach that can both assist an operator in his decision-making process and provide him with information to understand why he should make a particular decision. The methodology's effectiveness is verified and applied to a realm dataset of wind farms.

Explainable Artificial Intelligence (XAI) fills the role of a critical interface fostering interactions between sophisticated intelligent systems and diverse individuals, including data scientists, domain experts, end-users, and more. It aids in deciphering the intricate internal mechanisms of ``black box'' Machine Learning (ML), rendering the reasons behind their decisions more understandable. However, current research in XAI primarily focuses on two aspects; ways to facilitate user trust, or to debug and refine the ML model. The majority of it falls short of recognising the diverse types of explanations needed in broader contexts, as different users and varied application areas necessitate solutions tailored to their specific needs. One such domain is Predictive Maintenance (PdM), an exploding area of research under the Industry 4.0 \& 5.0 umbrella. This position paper highlights the gap between existing XAI methodologies and the specific requirements for explanations within industrial applications, particularly the Predictive Maintenance field. Despite explainability's crucial role, this subject remains a relatively under-explored area, making this paper a pioneering attempt to bring relevant challenges to the research community's attention. We provide an overview of predictive maintenance tasks and accentuate the need and varying purposes for corresponding explanations. We then list and describe XAI techniques commonly employed in the literature, discussing their suitability for PdM tasks. Finally, to make the ideas and claims more concrete, we demonstrate XAI applied in four specific industrial use cases: commercial vehicles, metro trains, steel plants, and wind farms, spotlighting areas requiring further research.

This paper presents a new fault prognosis approach applied to wind turbine system based on self-excited induction generator (SEIG) for offshore and isolated areas. This generator is very sensitive to wind speed variation and excitation source. The SEIG is excited by a capacitor bank with an appropriate value to ensure the good operating of the production system. Capacitor bank faults are usually related to chemical aging, electrical and thermal stress conditions. These abnormalities can affect one or more properties of the system, which can lead to failures or even complete breakdown of the production system. Specifically, in this paper, we propose a saturated flux model for the SEIG and develop a hybrid monitoring method that detects faults occurrence gradually and estimates the remaining useful life (RUL). Such monitoring method applies data mining techniques in order to identify and track the faults using only useful data that captures the dynamics of the degradation. Moreover, to deploy efficient maintenance schedules, RUL is estimated by exploiting wind speed (variable and max speed) information. The proposed hybrid fault prognosis method is tested under variable excitation capacitors degradation scenarios. The obtained results confirm the robustness and accuracy of the proposed method.