Focus on Artificial Intelligence - "Ask SPIE Nucléaire"

Deep dive into artificial intelligence! Let's explore the ‘Ask SPIE Nucléaire’ platform with Frédérique WEBER, Technical Director of SPIE Nucléaire (France), and take inspiration from a selection of projects developed by our subsidiaries. Get ready for the sprint to 2025!

Artificial Intelligence (AI) encompasses technologies that enable machines to mimic human cognitive functions. It includes machine learning for pattern recognition and prediction, natural language processing for understanding and generating human language, robotics for performing physical tasks, and computer vision for interpreting visual information. AI systems learn from data, improve over time, and make decisions, simulating human intelligence while operating with speed and precision beyond human capabilities.

“Ask SPIE Nuclear” is an AI-enhanced, peer-to-peer Q&A platform jointly developed by SPIE Nucléaire and the French start-up Ask For The Moon. This Q&A mode is accessible to users working on-site and enables experts to only respond to questions once, allowing them to focus their expertise where it is most needed. The quality of the data is ensured through an ontological model designed by the start-up and by having answers validated by humans. Meanwhile, integrated measurement tools evaluate performance, including the time saved by each user. Currently, the platform is undergoing a Proof of Value with 200 employees from two different SPIE departments, specifically on sensitive operations like welding and earthquake-reinforced electricity.

How does the “Ask SPIE Nucléaire” platform benefit our colleagues and customers?

The Q&A platform makes it easier for colleagues to access technical knowledge and highlights our specialists, showcasing the expertise present in our various professions. Colleagues can now effortlessly tap into a readily available and reusable bank of questions and answers. Responses are tagged by theme and site, which enables a really targeted Q&A structure. The platform is for internal use only as of the testing phase, but this kind of knowledge transmission adds value to our customer offerings, especially considering that the nuclear sector is undergoing renewal while facing a tight job market.

Which tasks or processes have been improved by the platform’s implementation?

“ASK SPIE Nucléaire” injects new vitality into the expert network established in 2021, which was originally based on a static list of Q&As. It adds a more human touch. Also, the format for asking questions is much simpler now, which helps us capitalise on our know-how. It’s quicker and easier to find structured, comprehensive answers than e.g. by searching the entire document library.

Have you experienced time and/or financial savings with this solution?

To begin with, it significantly enhances our ability to capitalise on technical knowledge. We’re also seeing time savings for both the experts, who only need to answer questions once, and the users, who can indicate how much time they saved when they find the answers they need. This time is tracked by the application so we can fully evaluate the return on investment as we prepare further developments.

Discover a selection of initiatives using Artificial Intelligence.

Wind Tunnel Maintenance - Germany and Switzerland

SPIE Efficient Facilities is responsible for the Technical Facility Management of a wind tunnel, a complex technical system with numerous potential points of failure. The system performs continuous vibration analysis through battery-powered sensors to detect wear and tear that could, in time, lead to failures. Four times a day, sensor data is transmitted through a 4G gateway and mesh network to the cloud, and made available for analysis via an online dashboard, on mobile and desktop devices.

Railroad crossing security - France

SPIE France is developing a solution aiming to bolster safety at railroad crossings. The solution involves implementing computer vision technology with robust AI-based video and image recognition at railroad crossings, in order to scrutinise the environment and generate immediate alerts when necessary. The system uses deep learning, a subset of AI ideally suited to simulate how human brains process data and create patterns, to identify various elements featured at railroad crossings: bicycles, pedestrians, cars, etc. This level of granularity in detection serves a critical purpose: categorising potential risks. The goal is that, through computer vision, the system will be able to identify e.g. a pedestrian or a car on a railroad, recognise the different associated risks, and recommend mitigation actions accordingly.

AI video analytics - Netherlands

By adding the advanced capabilities of the Vaidio AI Analytics platform to its security portfolio, SPIE offers customers a complete solution to make video analytics more accurate, automated, and affordable.

Land Subsidence Management - Germany and Austria

Because land subsidence can seriously jeopardise the stability of overhead power lines, monitoring movements in the earth’s surface is crucial to detect potential damage early on. To this end, SPIE has deployed a remote sensing method carried out via satellite. By comparing radar images taken at different times, operators can calculate displacement rates with an accuracy in the millimetre-range. This evaluation specifically focuses on images providing constant signal to the satellite over a long period of time (e.g. buildings, roads, rails, or overhead line masts), which enables analyses over a period of several years. Such long-term forecasting allows for timely detection and corrective actions if needed.

Modudal virtual power plant - Netherlands

SPIE’s modular Virtual Power Plant (iVPP) offer gives customers control over their energy transition, from generation, storage, off-take, and release of energy flows around their properties.