Smart machine-tool and HSM process with Emma – SmartEmma

To address these issues, the idea of ??the ANR SmartEMMA project is to combine artificial intelligence tools (automatic data mining and unsupervised learning) with knowledge engineering methods, to provide operators and various stakeholders in operational management (methods office, programming, quality, maintenance, etc.) with performance indicators (KPIs) that are useful for their daily activities and understandable in their technical languages. These indicators are obtained by scientific methods of multi-scale aggregation based on knowledge. Knowledge management consists of the use of methodological and computer means to represent the problem, capture the history of past events and problems. It also involves extracting the business rules of experts to understand how they solved the problems, with a view to reusing the solutions in the future. Thanks to a knowledge base and inference engines, the AI ??system. hybrid integrating “artificial intelligence & knowledge engineering” makes it possible to automatically analyze large volumes of heterogeneous manufacturing data, detect manufacturing anomalies and help understand the causes of events.

In order to propose a decision support system based on mining machining monitoring data, significant formalization work was carried out, which made it possible to define the software architecture of the demonstrator following an Edge Computing strategy, based on less voluminous SmartData with more physical meaning. An AI was developed for incident detection, in an unsupervised manner, with integration of manufacturing knowledge. The foundations of a digital twin and a diagnosis by knowledge management were initiated. The production of aluminum and titanium aeronautical parts by MECACHROME made it possible to propose adapted SmartData and to validate the data management strategy, on 6 machine tools with different productions (flexible/large series, machined aluminum/titanium parts). The LS2N laboratory developed the SmartEmma decision support demonstrator and deployed it on the AIRBUS Nantes site. It automatically analyzes the ½ GB of machining monitoring data collected daily from production machine tools, and reports dedicated KPIs to each department's offices. Their use has validated the entire approach and its great value. The project has thus jointly resulted in high levels of scientific publications and technological maturity (TRL 7). Following maturation, commercialization is under discussion.

This project has given great satisfaction and many perspectives are underway as evidenced by the interest of Airbus which joined the project during the course and allowed new advances towards Edge Computing. These perspectives concern on the one hand the continuation of investigations on other indicators and monitoring criteria, and on the other hand the generalization of the approach towards other technologies such as additive manufacturing (2 European projects have been submitted for this). Finally, the other axes of decision support will also be the subject of advanced research in the future.

Ladj, A.; Wang, Z.; Meski, O.; Belkadi, F.; Ritou, M.; et al. A knowledge-based Digital Shadow for machining industry in a Digital Twin perspective. Journal of Manufacturing Systems. 2021, 58, 168-179.

Meski, O.; Belkadi, F.; Laroche, F.; Ritou, M.; Furet, B. A generic knowledge management approach towards the development of a decision support system. International Journal of Production Research. 2021, 59 (22), 6659-6676.

Wang, Z.; Ritou, M.; da Cunha, C. M.; Furet, B. Contextual classification for smart machining based on unsupervised machine learning by Gaussian mixture model. International Journal of Computer Integrated Manufacturing. 2020, 33 (10-11), 1042-1054.

Ritou, M.; Belkadi, F.; Yahouni, Z.; da Cunha, C.; Laroche, F.; Furet, B. Knowledge-based multi-level aggregation for decision aid in the machining industry. CIRP Annals. 2019, 68 (1), 475-478.

Demonstrator installed at the JVMA (Jules Verne Manufacturing Academy, factory-school 4.0) for students from Nantes University and Pays de la Loire.

The objective of the SmartEmma project is to develop innovative smart and connected machine-tool for High Speed Machining (HSM). The aim is to contribute to the digital Factory of the Future. Signals from close to the process instrumentation and from the connection to the CNC (Computer Numerical Control) of the machine-tool are collected in large process database. Research will be carried out to define Key Performance Indicators (KPI) and methods for data analysis, in order to improve HSM process efficiency, through decision support tool. The approach requires an adapted modelling and efficient management of HSM process data and knowledge. KPI will be identified by Knowledge Discovery in Database and Data Mining. Continuous process improvement will be eased by an expert system and by long term learning. It will conduct to new leverages for decision making for the operational management of a machining company. Smart machines will also result from advanced perception and new real-time adaptive control capabilities.