A5030058408- Power System Optimization and Stability
- Smart Grid Energy Management
- Optimal Power Flow Distribution
- Smart Grid Security and Resilience
- Energy Load and Power Forecasting
- Microgrid Control and Optimization
- Power Systems and Technologies
- Computational Physics and Python Applications
- Electric Power System Optimization
- Complex Network Analysis Techniques
- Human-Automation Interaction and Safety
- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Neural Networks and Applications
- Explainable Artificial Intelligence (XAI)
- Time Series Analysis and Forecasting
- Distributed and Parallel Computing Systems
- Transportation and Mobility Innovations
- Reinforcement Learning in Robotics
- ICT Impact and Policies
- Power System Reliability and Maintenance
- Scheduling and Optimization Algorithms
- Rough Sets and Fuzzy Logic
- Flexible and Reconfigurable Manufacturing Systems
- VLSI and FPGA Design Techniques
Électricité de France (France)
2017-2024
Raidió Teilifís Éireann (Ireland)
2018-2022
Institut national de recherche en informatique et en automatique
2020
Laboratoire de Recherche en Informatique
2017-2019
Université Paris-Sud
2017-2019
Université Paris-Saclay
2018
Power networks, responsible for transporting electricity across large geographical regions, are complex infrastructures on which modern life critically depend. Variations in demand and production profiles, with increasing renewable energy integration, as well the high voltage network technology, constitute a real challenge human operators when optimizing transportation while avoiding blackouts. Motivated to investigate potential of AI methods enabling adaptability power operation, we have...
We propose a neural network architecture that emulates the behavior of physics solver solves electricity differential equations to compute flow in power grids (so-called "load flow"). Load computation is well studied and understood problem, but current methods (based on Newton-Raphson) are slow. With increasing usage expectations infrastructure, it important find accelerate computations. One avenue we pursuing this paper use proxies based "graph networks". In contrast with previous...
We address the problem of assisting human dispatchers in operating power grids today's changing context using machine learning, with theaim increasing security and reducing costs. Power networks are highly regulated systems, which at all times must meet varying demands electricity a complex production system, including conventional plants, less predictable renewable energies (such as wind or solar power), possibility buying/selling on international market more actors involved Europeanscale....
We propose a new method to efficiently compute load-flows (the steady-state of the power-grid for given productions, consumptions and grid topology), substituting conventional simulators based on differential equation solvers. use deep feed-forward neural network trained with precomputed by simulation. Our architecture permits train so-called "n-1" problems, in which load flows are evaluated every possible line disconnection, then generalize "n-2" problems without retraining (a clear...
For power grid congestion management, lots of research have focused on using generation redispatching, load shedding or demand side management flexibilities. However, a less costly option would be topology reconfiguration. Branch switching has been previously explored, since it could formulated as linear programming optimization problem we can solve, and showed some benefits. This further extended to the broader class non-linear nodal reconfigurations at substations. In this paper, present...
System operators are faced with increasingly volatile operating conditions. In order to manage system reliability in a cost-effective manner, control room turning computerised decision support tools based on AI and machine learning. Specifically, Reinforcement Learning (RL) is promising technique train agents that suggest grid actions operators. this paper, simple baseline approach presented using RL represent an artificial operator can operate IEEE 14-bus test case for duration of 1 week....
We address the problem of maintaining high voltage power transmission networks in security at all time, namely anticipating exceeding thermal limit for eventual single line disconnection (whatever its cause may be) by running slow, but accurate, physical grid simulators. New conceptual frameworks are calling a probabilistic risk-based criterion. However, these approaches suffer from requirements terms tractability. Here, we propose new method to assess risk. This uses both machine learning...
High penetration of distributed energy resources (DERs) changes the flows in power grids causing thermal congestions which are managed by real-time corrective topology switching. It is crucial to consider voltage stability margin (VSM) as a constraint when modifying grid topology. However, it nontrivial exhaustively search using AC flow (ACPF) for all control actions with desired VSM. Sensitivity methods used solve this issue “power flow-free VSM estimation” screen candidate actions. due...
We propose a new adversarial training approach for injecting robustness when designing controllers upcoming cyber-physical power systems. Previous approaches relying deeply on simulations are not able to cope with the rising complexity and too costly used online in terms of computation budget. In comparison, our method proves be computationally efficient while displaying useful properties. To do so we model an framework, implementation fixed opponent policy test it L2RPN (Learning Run Power...
The segmentation of large scale power grids into zones is crucial for control room operators when managing the grid complexity near real time. In this paper we propose a new method in two steps which able to automatically do segmentation, while taking account time context, order help them handle shifting dynamics. Our relies on "guided" machine learning approach. As first step, define and compute task specific "Influence Graph" guided manner. We indeed simulate state chosen interventions,...
We address the problem of maintaining high voltage power transmission networks in security at all time. This requires that flowing through lines remain below a certain nominal thermal limit above which might melt, break or cause other damages. Current practices include enforcing deterministic "N-1" reliability criterion, namely anticipating exceeding for any eventual single line disconnection (whatever its may be) by running slow, but accurate, physical grid simulator. New conceptual...
We propose a novel neural network embedding approach to model power transmission grids, in which high voltage lines are disconnected and reconnected with one-another from time time, either accidentally or willfully. call our architeture LEAP net, for Latent Encoding of Atypical Perturbation. Our method implements form transfer learning, permitting train on few source domains, then generalize new target without learning any example that domain. evaluate the viability this technique rapidly...