LISI GROUP - Integrated report 2023

43 2 0 2 3 I N T E G R A T E D R E P O R T STRATEGY Investing in the future QUALITYCONTROL Foundry defects tracked by AI The Mélisey site (LISI AUTOMOTIVE) has commissioned two quality control applications assisted by artificial intelligence. The first of these applications, launched in May 2022, was designed to detect cracks – foundry defects – and impacts on one of our parts. The second application deals with detecting impacts on dark-colored parts, for which the conventional control system, based on the reflection of light, was ineffective. “We turned to artificial intelligence when we couldn’t get satisfactory results with the conventional optical system,” explained Yves Jassey, in charge of sorting and packaging at the Mélisey site (north-eastern France). “In the first case, the two anomalies sought are very close visually and led to very high leak rates – “false” non-compliant ones.” AI-assisted learning-based control has significantly reduced these “false negatives”. “To achieve that, we had to set up a huge image database by checking tens of thousands of parts, compiling the defects, classifying them, and then teaching the algorithm how to recognize the parts to be discarded,” continued Yves Jassey. “It is sometimes necessary to plan several iterations before finding the right level of adjustment; with each new defect encountered, the model must be corrected and enriched.” While here artificial intelligence plays a very important supporting role, the deployment of a self-learning solution involves developing a robust database to achieve the desired results. However, the results are commensurate with the invested energy: the leak rate matches the degree of vigilance of a trained human controller. #LISI AUTOMOTIVE #HSE #PEOPLE “The Smart Factory project not only benefits us, it also has a direct positive impact on the local economy, with the development of new skills in the local job market.” –MarkCapell Rugby plant General manager. “We turned to artificial intelligence when we couldn’t get satisfactory results with the conventional optical system.” –Yves Jassey Automatic Control APU Manager at the Mélisey site (north-eastern France). SMART FACTORY Data-driven production at Rugby in the United Kingdom LISI AEROSPACE Rugby’s Smart Factory project includes the development of a new generation of machines and a skills reengineering plan. Launched in 2017, Rugby’s Smart Factory project has entered its final phase. An integral part of the Group’s continuous improvement plan from the outset, it is designed to project the site into the industry of the future by meeting market requirements and societal and environmental challenges. The project includes the installation of autonomous production cells controlled by sensors whose data feed machine learning algorithms. The first generations of these “self-learning” machines are already in production on site. The first complete new-generation line, integrating tool handling and adjustment, will be operational by the end of 2024. Other lines will be deployed progressively between now and 2030. The project, developed in partnership with Advanced Manufacturing Research Centre (AMRC) at the University of Sheffield, has received support from the UK government as part of InnovateUK and from the Aerospace Technology Institute (ATI). It is accompanied by an extensive training program to develop in-house skills in data analysis (data literacy, software development) and production line management. These learning approaches have been developed in partnership with with the universities of Cambridge iFM, Birmingham and De Montfort. #LISI AEROSPACE #PLANT 4.0 Rugby, UK plant (LISI AEROSPACE).