How Deep Learning Can Flip the Switch for Energy Efficiency
Saving energy is good for the planet — and our pocketbooks. But large companies can struggle even to understand the energy needs of their workers and facilities, let alone use that information to find savings.
Take Colas, one of the world’s top producers of asphalt concrete. The Paris-based group has 55,000 employees and 500 factories globally. For a huge operation like theirs, reducing energy waste is more complex than switching off a few lights.
To make energy efficiency easier for enterprises, French startup Energisme is applying GPU deep learning to the key ingredient: information.
Big Savings from Big Data
By analyzing how, when and what energy is being used, businesses can optimize their consumption. Some organizations simply lack this information. Others have detailed data, but lack the resources to make sense of it.
Energisme solves this problem by combining smart sensors with deep learning algorithms accelerated by NVIDIA Tesla GPUs in the Microsoft Azure cloud.
With GPU acceleration, Energisme can analyze energy usage information in real time. That means its platform not only gives an up-to-date picture, but can also alert customers instantly to problems or anomalies, allowing them to react quickly.
Energisme’s platform also predicts energy requirement by building a detailed model of energy usage. It then simulates a variety of situations to compare how energy consumption, production and pricing may vary. It can even suggest a more cost-effective energy plan, based on its analysis.
Benefits Across the Board
Energisme’s solution has dramatically improved Colas’ performance. Although its facilities were producing a large amount of data about energy consumption, Colas had no tool to turn this information into insights. With Energisme’s platform, it can better understand its energy consumption and identify efficiencies.
Employees can now monitor the temperature and performance of production machinery in real time, and react immediately to solve problems.
By linking the amount of energy being used with production levels, Colas’ regional managers are able to assess how each factory in their area is performing. This in turn provides the company’s leadership with insight into global performance by comparing the energy efficiency of different sites.
“In many cases, companies already have information about their energy consumption, but they don’t know how to turn that into insights,” says Pierre Vidal, commercial director at Energisme. “By using GPU deep learning to accelerate our platform, we’re able to help them put their data to work and ultimately reduce costs.”
Inception Startups at GTC Europe
Energisme is just one of around 2,000 AI startups worldwide that we’re supporting through our Inception program. Through this program, we help accelerate startups by providing them with access to technology, expertise and marketing support.