Data-driven monitoring and optimisation of decentralised machining processes (ROADMAP)

You are here


Machining, data analytics, decentralised manufacturing, monitoring and optimisation

Project description

The machining industry is highly competitive and extremely challenging. Upcoming regions are challenging established companies while new machining centers and technologies are becoming available everywhere. Furthermore, machining operations are also very complex processes. Next to the high variability in manufacturing strategies of a particular part, there is also a myriad of possible tools, machines and settings that needs to be appropriately selected. This makes it difficult to thoroughly understand and standardise the process. Together with this high complexity, companies are also faced with increasing challenges regarding customer demand, in which they need to live up to the expectation of high responsivity, flexible production possibilities, the highest quality and everything at a low cost.

In line with the move towards Industry 4.0, in which the aim is to increasingly digitalise the manufacturing process, there is an increasing attention for the data-driven monitoring of machining processes. In a decentralised manufacturing setting, companies manufacture only those products that are in demand in a specific region or country or close to regional resource suppliers. Whereas the products that are produced across these different manufacturing sites often differ, the machining processes that are undertaken to produce these products (or subcomponents thereof) are often very similar. By extracting insights from the data that is gathered in these decentralised manufacturing environments, knowledge can be shared between different manufacturing sites, resulting in a more optimal manufacturing process that contributes to coping with these challenges.

However, the introduction of data analytics for optimising these processes currently mainly remains limited to the monitoring of particular aspects of the machining operation. Next to the lack of data science experience and expertise in standard manufacturing companies, the main reasons for this are the lack of:

  1. an appropriate comprehensive measurement methodology (both with respect to hardware and software to capture all relevant information from machining processes in the production flow, from raw material to end product)

  2. corresponding data-driven techniques to extract useful insights from the gathered data

  3. a principled approach to integrate this information with the available domain knowledge and physics-based approaches to arrive at a multi-modal methodology. 

Next to that, existing state-of-the-art approaches in this direction often fail to be sufficiently reliable and robust in order to enable their industrial applicability in multi-machine settings in decentralised manufacturing environments.

The aerospace sector is in front of these advancements and currently facing the aforementioned challenges, mainly driven by the increasing competition and more stringent customer demands. In this context, SABCA, one of the leading aerospace companies active in the design, manufacturing and assembly of structural components, teamed up with the VUB and Sirris in order to address these challenges in the ROADMAP project.

In particular, the project aims to create a multi-modal methodology for monitoring and optimising machining processes in decentralised manufacturing environments. This overall strategic objective can be split into the following detailed project objectives:

  1. The construction of a comprehensive measurement approach (HW/SW) to enable the monitoring of machining process characteristics

  2. The definition of a multi-modal methodology for analysing individual machining processes (multi-physics + expert knowledge + data analytics)

  3. The exploitation of the multi-machine environment in decentralised manufacturing environments for machining process optimisation

  4. The validation of the resulting methodology both in in-the-lab (with the Precision Manufacturing lab of Sirris) and in-the-field (at SABCA) by means of two industrial validation cases, namely:

    • guaranteeing product quality at a constant time 

    • autonomous parameter optimisation of the machining process


Project partners

SABCA – Société Anonyme Belge de Constructions AéronautiquesScreenshot 2021-06-15 at 13.18.42_0_1.png

SABCA is one of the main aerospace companies in Belgium. Founded in 1920 with its headquarters in Brussels, the company has built an extensive and varied know-how in designing, building and upgrading large and complex elements for aircraft (civil and defense) and space applications. Its customers and partners belong to the elite of the aerospace world and are spread worldwide.

In Belgium, SABCA also has offices in Charleroi and a subsidiary company “SABCA Limburg” situated in Lummen. SABCA has around 840 employees, amongst which 650 are located in Brussels. 

VUB – Vrije Universiteit BrusselVUB.png

The VUB is a university in Brussels. Within this project, the Electronics and Informatics Department (ETRO) and the department of Mechanical Engineering (MECH) are involved. Their expertise encompasses: 

The department of Electronics and Informatics (ETRO) at VUB conducts research structured around three major directions: 1) conception and design of sensors in multiple spectral bands of the electromagnetic spectrum and the improvement of the sensor’s behaviour given the knowledge of the subsequently devised data processing and modelling methodologies; 2) data modelling methodologies which includes concepts such a sparse representations, (interpretable) deep learning, and distributed data models to address the distributed and vast nature of multi-sensor systems, including communications and resource-efficient processing aspects; 3) data analysis and fusion of multi-sensorial data to allow for knowledgeable rendering solutions.

The Acoustics and Vibrations Research Group (AVRG) within the department of Mechanical Engineering of the VUB has the central goal to conduct fundamental and applied research in the broad field of Acoustics & Vibration. AVRG has a track record in state-of-the-art development of advanced testing and experimental modelling methodologies with application in the broad mechanical engineering domain. The current core activities of AVRG are in the field of Experimental and Operational Modal Analysis, Damage Assessment and Structural Health Monitoring, Offshore Wind Turbine Monitoring, Additive Manufacturing Modelling and Control. The development of novel system identification and data processing techniques as well as optical measurement techniques are a key contributors in the reached milestones.

Sirris - the collective centre of the Belgian technology industrySirris.png

Sirris is a private collective research center with around 2.500 member companies (95 % SMEs). Its mission is to support large and smaller companies in the Belgian technology industry make the right technological choices and implement them, in order to achieve sustainable economic growth. Sirris helps companies to bring new technological innovations to the market. 

Within this project two departments of Sirris will be involved: the Precision Manufacturing Program from the Advanced Manufacturing business unit and the Data and AI Competence Lab (the EluciDATA Lab) from the Mechatronics & ICT business unit. 

The Precision Manufacturing program of Sirris exists since 2014 and bundles Sirris’ expertise in manufacturing processes (including the practical use of simulation tools in manufacturing). The program also owns two environmental chambers in Diepenbeek, where detailed and robust measurements will be realised within this project. Different machines are equipped with sensors and data capturing systems.

The Data and AI Competence Lab of Sirris (EluciDATA Lab) bundles Sirris' expertise in advanced AI methods for real-world challenges. The Lab's mission is to stimulate data innovation and the uptake of AI within the Belgian technological industry. It has a proven track record in realising data-driven proof-of-concepts for industrial partners in diverse application scenarios and domains.

Project details 

May 2019 - October 2022

Project presentations & publications

  • Sirris presented a paper on  ‘A Machine Learning-based Approach for Predicting Tool Wear in Industrial Milling Processes’ at the ECML-PKDD workshop on IoT Stream for Data Driven Predictive Maintenance in Wurzburg (Germany), September 16th, 2019

  • Sirris presented an extended abstract of the paper  ‘A Machine Learning-based Approach for Predicting Tool Wear in Industrial Milling Processes’ at the BENELEARN/BNAIC conference in Brussels (Belgium), November 6-8, 2019

  • Sirris presented the project idea as well as an overview of the current results on tool wear prediction during the ‘AI for Industry 4.0 Research Projects in Belgium’ seminar of the Belgian AI Week (virtual conference), March 16, 2021

Projet subsidié par la Region de Bruxelles-Capital - Innoviris/project gesubsidieerd door het Brussels Hoofdstedelijk (Project subsidised by the Brussels Capital Region - Innoviris):

Innoviris & region_0.png