Data science and industrial performance

DiagRAMS: a fast-growing start-up

Since 2019, the start-up DiagRAMS, the result of work on data science carried out by the Modal project team at the Inria Lille – Nord Europe centre, has been developing predictive maintenance software. Its aim is to anticipate breakdowns and malfunctions in machines by analysing industrial data that has, until now, been under exploited. In spite of the health crisis in 2020, the deep tech company is doing well. It has attracted new clients and is working with major groups in the energy, agri-food and automotive sectors. It has just completed a round of fundraising with €1.7 million.

“So far, all the feedback has been positive and our partnerships are constantly growing. The results really improve our clients' processes and it is very satisfying to see that our work is appreciated,” says Quentin Grimonprez, a Doctor in statistics at Inria and now Head of Data Science at DiagRAMS. This success can be explained in particular by the company’s objective: offer disruptive innovation in industrial data science using artificial intelligence methods. To achieve this aim, it focuses on developing both the algorithmic code at the heart of its software and the visualisation of the results it produces. The result is a genuine turnkey tool for maintenance teams in the field.

A strong scientific and strategic structure

DiagRAMS technology aims to optimise industrial performance for its clients. To do so, the start-up has to take into account complex and heterogeneous data from varied sources and contend with the high level of variation in industrial processes (different operational modes of industrial equipment over time, different materials etc.) “This is why our anomaly detection and remaining useful life (RUL) prediction methods have to take this complex industrial context into account. The speed of calculation and explainability of our methods are also important to ensure that the utilisation of our tool is optimised for our clients,” explains Quentin Grimonprez.

These are all challenges that DiagRAMS and its scientific committee are addressing. Created in 2021, the committee offers a new framework for working with Inria researchers specialised in the company’s key themes. They include Christophe Biernacki, an expert in clustering (algorithmic methods for structuring data) and Cristian Preda, a researcher specialised in processing and analysing temporal data series like those produced by industrial machines. The company is keen to innovate by adding new software packages to its range and thus extend its expertise. The researchers, as consultants, help guide solutions and resolve problems in the field.

In parallel, the start-up has also created a strategic committee to ensure the products stay in line with the market. “These two committees share a common purpose, which is to save us time. The structuring of our discussions, focussed on matters concerning the market and technical issues, will help ensure better development of our product,” says Margot Corréard, Co-founder DiagRAMS.

Strengthening know-how

“As a deep tech company, the technical development of our product and its ergonomics requires us to invest in expertise. Fundraising will therefore enable us to strengthen our scientific team,” explains Margot Corréard. In addition to consolidating its scientific expertise internally, DiagRAMS is counting on its close ties with Inria to complement it. This has led to the regular and joint production of scientific publications and the presentation of their results at conferences.

Thanks to this unique connection with Inria, new forms of exchange are emerging such as the co-supervision of trainees and partnership contracts. “We have already carried out a joint R&D project and would like to repeat the experience,” says Quentin Grimonprez.

More generally, the use of artificial intelligence in industry is still recent and the full extent of its potential has yet to be discovered. Maintaining a close link with research therefore offers a key advantage. The technological future of the start-up should also depend on the needs identified by its clients. The DiagRAMS team remains proactive and intends, in time, to increase the size of its scientific committee by including other experts in the operation of industrial machines. These new contributions will be complemented by the company’s continual monitoring of scientific advances in data science.