arXiv subjects

Eric MSP Veith

Publications and source records attributed to Eric MSP Veith.

12 recordsLinked to original sources

arXiv2024

On zero-shot learning in neural state estimation of power distribution systems

This paper addresses the challenge of neural state estimation in power distribution systems. We identified a research gap in the current state of the art, which lies in the inability of models to adapt to changes in the power grid, such as loss of sensors and branch switching, in a zero-shot fashion. Based on the literature, we identified graph neural networks as the most promising class of models for this use case. Our experiments confirm their robustness to some grid changes and also show that a deeper network does not always perform better. We propose data augmentations to improve performance and conduct a comprehensive grid search of different model configurations for common zero-shot learning scenarios.

cs.LG↗

arXiv2024

Imitation Game: A Model-based and Imitation Learning Deep Reinforcement Learning Hybrid

Autonomous and learning systems based on Deep Reinforcement Learning have firmly established themselves as a foundation for approaches to creating resilient and efficient Cyber-Physical Energy Systems. However, most current approaches suffer from two distinct problems: Modern model-free algorithms such as Soft Actor Critic need a high number of samples to learn a meaningful policy, as well as a fallback to ward against concept drifts (e. g., catastrophic forgetting). In this paper, we present the work in progress towards a hybrid agent architecture that combines model-based Deep Reinforcement Learning with imitation learning to overcome both problems.

cs.AI↗

arXiv2022

Learning to Attack Powergrids with DERs

In the past years, power grids have become a valuable target for cyber-attacks. Especially the attacks on the Ukrainian power grid has sparked numerous research into possible attack vectors, their extent, and possible mitigations. However, many fail to consider realistic scenarios in which time series are incorporated into simulations to reflect the transient behaviour of independent generators and consumers. Moreover, very few consider the limited sensory input of a potential attacker. In this paper, we describe a reactive power attack based on a well-understood scenario. We show that independent agents can learn to use the dynamics of the power grid against it and that the attack works even in the face of other generator and consumer nodes acting independently.

cs.CR↗

arXiv2022

Sampling Strategies for Static Powergrid Models

Machine learning and computational intelligence technologies gain more and more popularity as possible solution for issues related to the power grid. One of these issues, the power flow calculation, is an iterative method to compute the voltage magnitudes of the power grid's buses from power values. Machine learning and, especially, artificial neural networks were successfully used as surrogates for the power flow calculation. Artificial neural networks highly rely on the quality and size of the training data, but this aspect of the process is apparently often neglected in the works we found. However, since the availability of high quality historical data for power grids is limited, we propose the Correlation Sampling algorithm. We show that this approach is able to cover a larger area of the sampling space compared to different random sampling algorithms from the literature and a copula-based approach, while at the same time inter-dependencies of the inputs are taken into account, which, from the other algorithms, only the copula-based approach does.

cs.LG↗

arXiv2021

Comparison of Random Sampling and Heuristic Optimization-Based Methods for Determining the Flexibility Potential at Vertical System Interconnections

In order to prevent conflicting or counteracting use of flexibility options, the coordination between distribution system operator and transmission system operator has to be strengthened. For this purpose, methods for the standardized description and identification of the aggregated flexibility potential of distribution grids are developed. Approaches for identifying the feasible operation region (FOR) of distribution grids can be categorized into two main classes: Random sampling/stochastic approaches and optimization-based approaches. While the latter have the advantage of working in real-world scenarios where no full grid models exist, when relying on naive sampling strategies, they suffer from poor coverage of the edges of the FOR due to convoluted distributions. In this paper, we tackle the problem from two different angles. First, we present a random sampling approach which mitigates the convolution problem by drawing sample values from a multivariate Dirichlet distribution. Second, we come up with a hybrid approach which solves the underlying optimal power flow problems of the optimization-based approach by means of a stochastic evolutionary optimization algorithm codenamed REvol. By means of synthetic feeders, we compare the two proposed FOR identification methods with regard to how well the FOR is covered and number of power flow calculations required.

eess.SY↗

arXiv2021

Pointing out the Convolution Problem of Stochastic Aggregation Methods for the Determination of Flexibility Potentials at Vertical System Interconnections

The increase of generation capacity in the area of responsibility of the distribution system operator (DSO) requires strengthening of coordination between transmission system operator (TSO) and DSO in order to prevent conflicting or counteracting use of flexibility options. For this purpose, methods for the standardized description and identification of the aggregated flexibility potential of distribution grids (DGs) are developed. Approaches for identifying the feasible operation region (FOR) of DGs can be categorized into two main classes: Data-driven/stochastic approaches and optimization based approaches. While the latter have the advantage of working in real-world scenarios where no full grid models exist, when relying on naive sampling strategies, they suffer from poor coverage of the edges of the FOR. To underpin the need for improved sampling strategies for data-driven approaches, in this paper we point out and analyse the shortcomings of naive sampling strategies with focus on the problem of leptocurtic distribution of resulting interconnection power flows (IPFs). We refer to this problem as convolution problem, as it can be traced back to the fact that the probability density function (PDF) of the sum of two or more independent random variables is the convolution of their respective PDFs. To demonstrate the convolution problem, we construct a series of synthetic 0.4 kV feeders, which are characterized by an increasing number of nodes and apply a sampling strategy to them that draws set-values for the controllable distributed energy resources (DERs) from independent uniform distributions. By calculating the power flow for each sample in each feeder, we end up with a collapsing IPF point cloud clearly indicating the convolution problem.

eess.SY↗

arXiv2020

Analyzing Power Grid, ICT, and Market Without Domain Knowledge Using Distributed Artificial Intelligence

Modern cyber-physical systems (CPS), such as our energy infrastructure, are becoming increasingly complex: An ever-higher share of Artificial Intelligence (AI)-based technologies use the Information and Communication Technology (ICT) facet of energy systems for operation optimization, cost efficiency, and to reach CO2 goals worldwide. At the same time, markets with increased flexibility and ever shorter trade horizons enable the multi-stakeholder situation that is emerging in this setting. These systems still form critical infrastructures that need to perform with highest reliability. However, today's CPS are becoming too complex to be analyzed in the traditional monolithic approach, where each domain, e.g., power grid and ICT as well as the energy market, are considered as separate entities while ignoring dependencies and side-effects. To achieve an overall analysis, we introduce the concept for an application of distributed artificial intelligence as a self-adaptive analysis tool that is able to analyze the dependencies between domains in CPS by attacking them. It eschews pre-configured domain knowledge, instead exploring the CPS domains for emergent risk situations and exploitable loopholes in codices, with a focus on rational market actors that exploit the system while still following the market rules.

cs.CY↗

arXiv2020

The Adversarial Resilience Learning Architecture for AI-based Modelling, Exploration, and Operation of Complex Cyber-Physical Systems

Modern algorithms in the domain of Deep Reinforcement Learning (DRL) demonstrated remarkable successes; most widely known are those in game-based scenarios, from ATARI video games to Go and the StarCraft~\textsc{II} real-time strategy game. However, applications in the domain of modern Cyber-Physical Systems (CPS) that take advantage a vast variety of DRL algorithms are few. We assume that the benefits would be considerable: Modern CPS have become increasingly complex and evolved beyond traditional methods of modelling and analysis. At the same time, these CPS are confronted with an increasing amount of stochastic inputs, from volatile energy sources in power grids to broad user participation stemming from markets. Approaches of system modelling that use techniques from the domain of Artificial Intelligence (AI) do not focus on analysis and operation. In this paper, we describe the concept of Adversarial Resilience Learning (ARL) that formulates a new approach to complex environment checking and resilient operation: It defines two agent classes, attacker and defender agents. The quintessence of ARL lies in both agents exploring the system and training each other without any domain knowledge. Here, we introduce the ARL software architecture that allows to use a wide range of model-free as well as model-based DRL-based algorithms, and document results of concrete experiment runs on a complex power grid.

cs.AI↗

arXiv2020

Large-Scale Co-Simulation of Power Grid and Communication Network Models with Software in the Loop

Power grids are transitioning from an infrastructure model based on reactive electronics towards a smart grid that features complex software stacks with intelligent, pro-active and decentralized control. As the power grid infrastructure becomes a platform for software, so does the need for a reliable roll-out of software updates on a large scale. In order to validate resilient large-scale software roll-out protocols, corresponding test beds are needed, which mirror not only ICT networks, but also include the actual software being deployed, and show the interaction between the power grid and the ICT network during the roll-out, and especially during roll-out failures. In this paper, we describe the design implementation of a large-scale co-simulation test bed that combines ICT and power grid simulators. We pay specific attention to the details of integrating containerized software in the simulation loop.

cs.SE↗

arXiv2020

Explainable Reinforcement Learning: A Survey

Explainable Artificial Intelligence (XAI), i.e., the development of more transparent and interpretable AI models, has gained increased traction over the last few years. This is due to the fact that, in conjunction with their growth into powerful and ubiquitous tools, AI models exhibit one detrimential characteristic: a performance-transparency trade-off. This describes the fact that the more complex a model's inner workings, the less clear it is how its predictions or decisions were achieved. But, especially considering Machine Learning (ML) methods like Reinforcement Learning (RL) where the system learns autonomously, the necessity to understand the underlying reasoning for their decisions becomes apparent. Since, to the best of our knowledge, there exists no single work offering an overview of Explainable Reinforcement Learning (XRL) methods, this survey attempts to address this gap. We give a short summary of the problem, a definition of important terms, and offer a classification and assessment of current XRL methods. We found that a) the majority of XRL methods function by mimicking and simplifying a complex model instead of designing an inherently simple one, and b) XRL (and XAI) methods often neglect to consider the human side of the equation, not taking into account research from related fields like psychology or philosophy. Thus, an interdisciplinary effort is needed to adapt the generated explanations to a (non-expert) human user in order to effectively progress in the field of XRL and XAI in general.

cs.LG↗

arXiv2020

Robust and Deterministic Scheduling of Power Grid Actors

Modern power grids need to cope with increasingly decentralized, volatile energy sources as well as new business models such as virtual power plants constituted from battery swarms. This warrants both, day-ahead planning of larger schedules for power plants, as well as short-term contracting to counter forecast deviations or to accommodate dynamics of the intra-day markets. In addition, the geographic distribution of renewable energy sources forces scheduling algorithms with a hugely different communication link qualities. In this paper, we present an extension to the Lightweight Power Exchange Protocol (LPEP), dubbed LPEP++. It draws on the strength of the LPEP to find the optimal solution of the combinatorial power demand-supply problem with string guarantees in acceptable time and extends it with facilities for long-term planning, parallel negotiations and reduces its memory footprint. We furthermore show its robustness towards volatile communication link quality.

cs.MA↗

arXiv2018

Adversarial Resilience Learning - Towards Systemic Vulnerability Analysis for Large and Complex Systems

This paper introduces Adversarial Resilience Learning (ARL), a concept to model, train, and analyze artificial neural networks as representations of competitive agents in highly complex systems. In our examples, the agents normally take the roles of attackers or defenders that aim at worsening or improving-or keeping, respectively-defined performance indicators of the system. Our concept provides adaptive, repeatable, actor-based testing with a chance of detecting previously unknown attack vectors. We provide the constitutive nomenclature of ARL and, based on it, the description of experimental setups and results of a preliminary implementation of ARL in simulated power systems.

cs.AI↗