Intelligent predictive maintenance for industry 4.0

Predicting equipment failure is a major preoccupation for maintenance managers, enabling them to define the most technically and economically appropriate strategies. The spread of new digital technologies using connected objects, the Internet of Things, the cloud, big data, artificial intelligence and data science have led to the development of a new maintenance concept known worldwide as intelligent predictive maintenance for Industry 4.0. This article presents its challenges, origins, objectives, methods and tools, highlighting its advantages and limitations. The first section describes the challenges of intelligent predictive maintenance for Industry 4.0, which can be seen as extensions of those of conventional predictive maintenance insofar as failure prediction implements the components of Industry 4.0. The definition of industry 4.0, also known as the industry of the future or the smart factory, is proposed in the second section. It also provides a description of the industrial revolutions that have led to what corresponds to Industry 4.0. It also presents a state of the art of Industry 4.0 for large and medium-sized companies, describing German, French, American and Chinese initiatives to support their industrial sectors. Given that Industry 4.0 concepts vary according to the field of application, a generic example of architecture with its essential components is proposed.

The third section presents the terminology required to develop a predictive maintenance program. Important definitions include RUL (remaining useful life) or DVUR (durée de vie utile restante) and DEFAD (durée estimée de fonctionnement avant défaillance). Particular emphasis is placed on the definition of prognosis and its metrics, which are essential for assessing confidence in failure prediction. Links with CBM (condition-based monitoring) and PHM (prognostics and health management), which use the same tools as intelligent predictive maintenance, are also briefly described.

As numerous initiatives to develop this innovative maintenance strategy have emerged, this section concludes with an overview of the evolution of this concept. Given the development of several hundred tools over the last few decades, thanks to the contributions of artificial intelligence, learning techniques and new data storage and processing techniques (data mining, big data, cloud computing, deep learning, machine learning, conversational agents similar to ChatGPT...), the fourth section is dedicated to a succinct and obviously non-exhaustive presentation of these main tools. As the intelligent predictive maintenance of Industry 4.0 leads to technico-economic optimization, the main optimization algorithms based on distributed intelligence (swarm intelligence) will be briefly described (genetic algorithms, ant and...