On 3 July 2020, DCbrain organised a webinar on the theme “How does hybrid AI tackle new gas industry challenges?”
Report written by Maxime Chevreton
The objective of the webinar was to show the benefit of AI in the gas sector, allowing to preview anomalies, optimising your flows, and make you gain lot of time.
The event was structured around several interventions:
Doudja Kartobi (Business developer), who explained the benefit of AI in the optimisation of the gas sector.
François Lacombe (Lead Data Engineer) who presented the Data “aspect”, and the engineer vision during this webinar.
The objectives and interests of hybrid AI
Hybrid AI brings a lot of new possibilities:
- It’s based on data and machine learning tools to build historical behavior models
- It Integrates simplified theoretical models when data is scarce
- It can learn from historical data and follow the progress of the network but can’t extrapolate
“Classic AI” doesn’t offer you the same possibilities as we can see:
- Based on a statistical model (built around physical rules of consumptions and propagations to predict results)
- Can operate without data but hard and costly to maintain / adapt
- Traditional approach for network design
- Not used for day to day optimization
That’s why Hybrid AI adapts better to the new challenges brought by today’s gas industry.
Hybrid AI: DCbrain’s new approach
As we learnt during the webinar, hybrid AI is a mix of static and dynamic models necessary to hold up with the fast evolution of every sector, especially gas. Here are the reasons why DCbrain uses it in its new approach:
- Use of Artificial Intelligence to enhance the physical models
- Automatic learning and/or configuration by experts
- Few parameter models
- Easy to understand
- Efficient models for optimisation
Our tool INeS is based on AI and hybrid AI for those reasons, and is particularly effective to optimise and previsualise flows on your network as we saw during the rest of the webinar.
Demonstration of our tool
We also had a demonstration of INeS during this webinar. We saw the quickness and the easy feasibility of the creation of an entire network and its digital twin. In fact, you just have to create the topology and measures of your network and DCbrain creates your entire network. We can do that on every existing network to visualise, preview and optimise every type of physical flow.
Conclusion with use cases
Our business developer Doudja Kartobi showed some of our use cases to prove the benefits of Hybrid AI and INeS. These examples demonstrate how Hybrid AI can be used to tackle new challenges in the gas sector in order to optimise your costs, ressources and flows.
So, what are you waiting for to inject AI into your gas networks?