Design and Implementation of a Neural Network for Thermo-Mechanical Simulation in Additive Manufacturing
The WAAM (Wire Arc Additive Manufacturing) process is a metal additive manufacturing method that allows for the production of large parts with a high deposition rate. However, this process results in highly stressed and deformed parts, making it complex to predict their geometric and mechanical characteristics. Thermomechanical modeling is crucial for predicting these deformations, but it requires significant computational resources and long calculation times. The NEUROWAAM project aims to develop a precise and fast thermomechanical numerical model using neural networks to predict the physical phenomena of the WAAM process. An internship in 2025 will provide a database through thermomechanical simulations using the CAST3M software. The post-doc's objective is to develop a neural network architecture capable of learning the relationship between the manufacturing configuration and the thermomechanical characteristics of the parts. Manufacturing tests on the CEA's PRISMA platform will be conducted to validate the model and prepare a feedback loop. The CEA List's Interactive Simulation Laboratory will contribute its expertise in accelerating simulations through neural networks and active learning to reduce training time.
Digital Correction of the Health Status of an H/F Electrical Network
Cable defects are usually detected when communication is interrupted, resulting in significant repair costs and time. In addition, data integrity is becoming a major issue due to the increased threats of attacks and intrusions on electrical networks that can disrupt communication. Being able to distinguish a disturbance due to the degradation of the physical layer of an electrical network or to an ongoing attack on the energy network will guide decision-making regarding correction operations, including network reconfiguration and predictive maintenance to ensure network resilience. In this post-doc, you will study the relationship between emerging cable defects and their impact on data integrity in power line communication (PLC). Your work will be based on the deployment of instrumentation using electrical reflectometry, combining distributed sensors and AI algorithms for online diagnosis of emerging faults in electrical networks. In the presence of certain defects, advanced AI methods will be applied to digitally correct the health of the physical layer of the power grid and thus ensure its reliability.
Study of potential attacks on data and model and their countermeasures in the context of distributed AI for intelligent energy systems
This post-doctoral fellowship is part of the AI-NRGY research project, which aims to propose an AI-based distributed architecture for intelligent energy systems made up of a large number of dynamic components (e.g. smart grids, electric vehicles, renewable energy sources). More specifically, the aim of this post-doc is to protect AI-based services against malicious disruptions that could affect the essential functionality of energy systems. Given the ubiquity of AI systems in modern digitised systems, their potential corruption poses a major threat to critical infrastructures. Two types of threats can be investigated: privacy threats (such as pattern reversal or data mining) and security threats (such as evasion attacks or data poisoning).
Privacy threats have been widely addressed by the scientific community and the CEA has conducted extensive work on integrating and optimising robust cybersecurity primitives. However, emerging security such as model poisoning (which arises from data poisoning) and adversarial attacks now require additional processing. Data poisoning is a cyber attack that can be used to simply compromise the convergence of the learning phase and result in underperforming models, but it can also be used to embed a ‘backdoor’ into the learned model that allows the expected result to be manipulated.
This post-doctoral position will enable the candidate to carry out theoretical and applied research in the field of privacy and security in distributed machine learning, particularly in the context of intelligent energy systems. More specifically, the candidate will study the potential threats of distributed/federated learning, and propose solutions to defend against the attacks identified as the most relevant.
Digital correction of the health status of an electrical network
Cable faults are generally detected when communication is interrupted, resulting in significant repair costs and downtime. Additionally, data integrity becomes a major concern due to the increased threats of attacks and intrusions on electrical networks, which can disrupt communication. Being able to distinguish between disruptions caused by the degradation of the physical layer of an electrical network and an ongoing attack on the energy network will help guide decision-making regarding corrective operations, particularly network reconfiguration and predictive maintenance, to ensure network resilience. This study proposes to investigate the relationship between incipient faults in cables and their impact on data integrity in the context of Power Line Communication (PLC). The work will be based on deploying instrumentation using electrical reflectometry, combining distributed sensors and AI algorithms for online diagnosis of incipient faults in electrical networks. In the presence of certain faults, advanced AI methods will be applied to correct the state of the health of the electrical network's physical layer, thereby ensuring its reliability.
Advanced reconstruction methods for cryo-electron tomography of biological samples
Cryo-electron tomography (CET) is a powerful technique for the 3D structural analysis of biological samples in their near-native state. CET has seen remarkable advances in instrumentation in the last decade but the classical weighted back-projection (WBP) remains by far the standard CET reconstruction method. Due to radiation damage and the limited tilt range within the microscope, WBP reconstructions suffer from low contrast and elongation artifacts, known as ‘missing wedge’ (MW) artifacts. Recently, there has been a revival of interest in iterative approaches to improve the quality and hence the interpretability of the CET data.
In this project, we propose to go beyond the state-of-the-art in CET by (1) applying curvelet- and shearlet-based compressed sensing (CS) algorithms, and (2) exploring deep learning (DL) strategies with the aim to denoise et correct for the MW artifacts. These approaches have the potential to improve the resolution of the CET reconstructions and facilitate the segmentation and sub-tomogram averaging tasks.
The candidate will conduct a comparative study of iterative algorithms used in life science, and CS and DL approaches optimized in this project for thin curved structures.
Simulation of thermal transport at sub-Kelvin
Thermal management in quantum computers is an urgent and crucial task. As the number of qubits rapidly scales, more electric circuits are placed close to qubits to operate them. Joule-heating of these circuits could significantly warm the qubit device, degrading its fidelity. With intensive activity in quantum computing at Grenoble, we (CEA-LETI, Grenoble, France) are looking for an enthusiastic post-doc researcher to study thermal transport at cryogenic temperature (sub-Kelvin).
The post-doc will apply the finite-element non-equilibrium Green’s function [1], developed in the group of Natalio Mingo at CEA-Grenoble, to simulate phonon transport in various designed structures. The simulation result promotes comparison with on-going experiments and constructive discussions in order to optimize the thermal management.
[1] C. A. Polanco, A. van Roekeghem, B. Brisuda, L. Saminadayar, O. Bourgeois, and N. Mingo, Science Advances 9, 7439 (2023).
Development of optoelectronic systems for quantum sensor technologies
The main mission of CEA LETI's Autonomy and Sensor Integration Laboratory (LAIC) is to develop sensor systems, and in particular quantum sensors for high-precision magnetic field measurement applications. The team's activities are at the interface of hardware (electronics, optronics, semiconductors), software (artificial intelligence, signal processing) and systems (electronic architecture, mechatronics, multiphysics modeling). The Swarm project (https://swarm.cnes.fr/en/), which put our quantum sensors for measuring the Earth's magnetic field into orbit in 2013, is one of our track records, and a new program with similar objectives gets underway this year.
Quantum technologies are strategic for the development of sensors with unrivalled performances, as we have demonstrated in magnetometry. Our challenge today is to adapt these developments and this know-how to new physics.
To support our developments in quantum sensors, we are looking for an opto-electronics post-doc researcher to design new quantum sensors and develop the associated optical benches. This post-doc position will have a significant experimental component.
Your main mission will be to participate to the development of these new sensors and their associated characterization benches, interfacing with CEA experts in the field.
More specifically, your mission will revolve around the following actions:
• Design and assembly of quantum sensors (optical fibers, RF sources, photodetectors)
• Participation in modeling the physical phenomena involved
• Design and build the optical characterization benches
• Development of the control electronics
• Publication of results in scientific journals
• Presentation of work in international conferences
• Patents proposal
Exploring microfluidic solutions for manufacturing targets for fusion power generation
As part of a call for projects on "innovative nuclear reactors", the TARANIS project involves studying the possibility of energy production by a power laser-initiated inertial confinement fusion power plant. The current context, which encourages the development of low-carbon energies, and the fusion experiments carried out by the NIF's American teams, make it very attractive to conduct high-level research aimed at eventually producing an economically attractive energy source based on inertial fusion.
Among the many technical hurdles to be overcome, the production of fusion targets with a suitable reaction scheme compatible with energy production is a major challenge. The CEA has the know-how to produce batches of capsules containing the fusible elements of the reaction. However, the current process is not suitable for mass production of hundreds of thousands of capsules per day at an acceptable cost.
One high-potential avenue lies in the use of microfluidic devices, for which the Microfluidic Systems and Bioengineering Laboratory (LSMB) of the Health Technologies and Innovation Department (DTIS) of CEA's DRT has recognized expertise.
Development of noise-based artifical intellgence approaches
Current approaches to AI are largely based on extensive vector-matrix multiplication. In this postdoctoral project we would like to pose the question, what comes next? Specifically we would like to study whether (stochastic) noise could be the computational primitive that the a new generation of AI is built upon. This question will be answered in two steps. First, we will explore theories regarding the computational role of microscopic and system-level noise in neuroscience as well as how noise is increasingly leveraged in machine leaning and artificial intelligence. We aim to establish concrete links between these two fields and, in particular, we will explore the relationship between noise and uncertainty quantification.
Building on this, the postdoctoral researcher will then develop new models that leverage noise to carry out cognitive tasks, of which uncertainty is an intrinsic component. This will not only serve as an AI approach, but should also serve as a computational tool to study cognition in humans and also as a model for specific brain areas known to participate in different aspects of cognition, from perception to learning to decision making and uncertainty quantification.
Perspectives of the postdoctoral project should inform how future fMRI imaging and invasive and non-invasive electrophysiological recordings may be used to test theories of this model. Additionally, the candidate will be expected to interact with other activates in the CEA related to the development of noise-based analogue AI accelerators.
Study of the specific features of highly distributed architectures for decision and control requirements
Our electricity infrastructure has undergone and will continue to undergo profound changes in the coming decades. The rapid growth in the share of renewables in electricity generation requires solutions to secure energy systems, especially with regard to the variability, stability and balancing aspects of the electricity system and the protection of the grid infrastructure itself. The purpose of this study is to help design new decision-making methods, specially adapted to highly distributed control architectures for energy networks. These new methods will have to be evaluated in terms of performance, resilience, robustness and tested in the presence of various hazards and even byzantines.