Following the call for projects “Physical devices and software for intelligent supervision of biomethane production facilities” launched by GRDF in May and June 2020, some 50 innovative companies, Research and start-up laboratories made proposals for improving monitoring of methanisation sites. 9 projects were selected following the analysis of the files, and continue the selection process.
Three categories were created, three “sensor” projects, three “man-machine software or interfaces” projects and three “predictive maintenance” proposals. We invite you to discover each of these innovative companies in a series of articles.
THE SOLUTION as you would explain it to your grandmother
It’s simple: machines can fall ill, so instead of their usual “click click” noise, they start saying “click clack”. That’s what our sensors hear: a strange noise will be a sign that the machines are not feeling well.
That’s Grandma’s version!
In summary, we listen to sounds and vibrations, and we are experts in the analysis of this data to see if everything is working well or if a maintenance operation needs to be scheduled.
To do this, we have developed Predict, an off-the-shelf sensor that is ready to use. These small sensors are affixed to a machine near a bearing or on the motor, for example. They are very quick and easy to install: you just need enough time for the glue to dry, and that’s it!
These sensors can be controlled with a smartphone app. They use the LoRa network, which sends the data collected to the cloud. A summary of this data is received in the app several times a day, and if irregularities appear in the summary, you can bring the smartphone near the machine and receive the raw data via a Bluetooth connection in order to make a more detailed analysis.
Wavely is also developing a customised offering to identify a specific sound automatically. This offering is structured in several stages: feasibility study, data acquisition, annotation, adaptation and driving of a machine learning model based on the annotated data set, deployment on a prototype, etc., right up to the industrialisation of a tailor-made solution.
- The key word (concept) to understand
It’s hard to pick just one key word!
The key concept is ‘predictive’, which invokes machine learning and therefore artificial intelligence.
Predict is a communicating object. It can send data up to the cloud, with Artificial Intelligence embedded in the sensors to process the data acquired by the microphone, the accelerometers, etc.
This sensor is easy to use whilst including several embedded technologies, for efficient predictive maintenance.
- The key selling point, THE real difference in the market
Our main added value is based on the combination of two possible modes of use in a single sensor: the simple mode with irregularity detection to monitor machines without any knowledge of vibration analysis, and expert mode to access the raw data and analyse it yourself. All of this is contained in a single sensor, without the need to take additional special measurements.
It’s a real difference, and it aligns with our philosophy: “Artificial intelligence is good; human intelligence is better.”
This is a solution designed to serve people, to allow them to understand the factors that lead the sensor to identify a malfunction. Artificial intelligence can identify the irregularity, but the possibility of retrieving the data allows the technicians to get to grips with the analysis.
- Journeys, associates, partners, etc.
In late 2015, Alexis Vlandas, CNRS researcher, Nicolas Côté, lecturer at an engineering school, started a research project on a system of acoustic sensors to detect noise in the context of the smart city.
A few months later, Marion Aubert, a graduate of Sciences Po and HEC, took an interest in their research, believing that it could give rise to the creation of a startup. She doesn’t have a scientific background, which gives her a different perspective on Wavely’s strategic development. What sparked her interest? An encounter, a good feeling, and the desire to take a ‘hard technology’ and industrial path, outside the areas where women often tend to work.
Alexis casts a scientific eye over the development of the solution. Nicolas specialises in the technical aspects of data processing and acoustics. Marion handles the sales and financial aspects.
In 2017, Wavely became a startup following a re-think about the market it wanted to target: the smart city is not yet a mature market, at least not with regard to noise issues. On the other hand, detection methods using noise are a real issue for the manufacturing sector. Industrial predictive maintenance therefore became Wavely’s core business.
The world of research is fascinated by technology, in-depth understanding, etc., but it stops at feasibility. We wanted to be able to transform our expertise into technologically and technically complex products in response to a customer need.
Wavely fine-tuned its positioning during its incubation at Euratechnologies, Lille, and then raised funds via Finovam and Fira, with additional support from Bpifrance.
The real acceleration came in 2017, when we signed a contract with acoustics design consultancy Sim Engineering for Total, for the acoustic detection of gas leaks.
The aim is to convert laboratory innovation into a market-facing product. And also to create a company that reflects our values.
All three of us had the same vision of how a company should be, from a social and ecological standpoint, in terms of its approach to the work–life balance, etc. We wanted to show that we can create a company that respects those values without falling into the old cliché of a startup where people work hard, play hard, and then suffer from burnout. We place a lot of trust in our colleagues, giving them a great deal of autonomy and responsibility, and flexible working hours compatible with their personal lives. A game of table football in the afternoon and then finishing work at 10 PM might be very nice, but it’s not our way of doing things: going home to enjoy your personal life is important for everybody. We also want to look at some fundamental topics, and to become an example to others in a way that we aspire to but still need to work towards: the product life cycle, for example, and particularly the fate of the products, the origin of the components, the conditions for processing data obtained from AI, etc.
We want to operate our company in a way that limits our footprint…
- What excites them
Carrying a research project right through to its conclusion and providing a response to a real customer problem
- The popular misconception that irritates them:
There is an aura surrounding Artificial Intelligence leading to a belief that it is somehow magical, like a black box that we don’t really understand, and we wonder what it’s for… whereas in fact, it’s just a tool—nothing more and nothing less.
We want to succeed in showing that it can be applied to concrete problems that technicians would face, and above all that it will help them, not replace them.
- What being selected by GRDF would mean for the development of the business
The opportunity to experiment on a use case with GRDF would be a great example; an interesting reference. We have done a lot of work with companies in the energy sector, so working with GRDF would be perfectly consistent. In addition, biomethane is a forward-looking topic, and Wavely would find it very interesting to contribute to that.
- The main challenges in the sector as a whole
For maintenance, the real subject is instrumentation for equipment monitoring: less leakage (saving energy), machine lifetime, etc. The issues are both economic and ecological.
For energy, the ecological impact of production facilities is important, and the issues of leak reduction and safety are particularly so. For biomethane, in view of the decentralisation of units, technicians cannot be everywhere at all times, so delocalised solutions are required.
- And, more particularly, the main challenges for the future?
We have done a lot of custom development for key accounts, whilst refining the Wavely Predict solution. Our challenge now is to market Wavely Predict whilst continuing to develop it to provide new features in accordance with specific needs. Our ambition is to go from detecting irregularities to diagnosis (identifying the source of the problem). That’s what we are working on.
Our international growth is currently happening via our French customers and their export sales.
RAPID-FIRE QUESTIONS // THE ENTREPRENEURIAL EXPERIENCE
- They are soon going to pitch in front of the GRDF panel; no doubt it’s not the first time. Any anecdotes to share with us?
OK, so, speaking from experience, I’d appreciate it if we could avoid any specifically female-oriented questions! I want to talk about the project like other people (and as men would), not about the difficulties faced by women in industry or about combining multiple roles! (Marion Aubert)
- Your first thought each morning?
I’m a lucky person!
- Your first real professional satisfaction
The first contract we signed with Sim Engineering for Total! (Marion)
The first transfer of data from the sensors, which brought the project to life (Nicolas)
- The best pep talk you have ever received?
What our mentors said to us during our incubation period. When experienced entrepreneurs remind us that it’s tough for everyone, that it’s normal to go through phases where things come in waves, and that you have to stick with it…that really lifts your spirits!
- Mantra or phrase that you often repeat
“ Forget the mistake, remember the lesson”
- A personal quality that sums you up
- A short message to convince the panel?
We’re here to listen to you!