AI4PM

Predictive maintenance integrates criteria such as the state of the equipment, the company's production imperatives and the possible critical nature of production in fields such as the automotive, rail, agri-food or pharmaceutical industries, for example. The stakes and challenges here are manifold.

On the one hand, it is necessary to reduce the impact on the environment by determining the optimum time to carry out maintenance operations, while at the same time limiting them to what is strictly necessary, without impacting production.

It is also necessary to take into account production imperatives when planning these operations.

The spin-offs for regional companies, therefore, are potentially very significant given the high density of industrial activity in our territory.

Using data collected via I-CARE sensors installed on the equipment to be monitored, the aim of the AI4PM project is to study the evolution over time of vibratory phenomena on this equipment and to identify any faults encountered by this equipment. In the context of this project, the equipment targeted is non-linear machines.

To this end, a major challenge is to be able to determine the machine's rotational speed at each instant using algorithms, and in particular deep learning models, in order to automate diagnosis without systematically calling on an expert.