This research focuses on a critical challenge in modern power systems: the effective coordination between transmission system operators (TSOs) and distribution system operators (DSOs) to harness distributed flexibility for grid stability and efficiency. As energy systems undergo a fundamental transformation toward decentralized, decarbonized, and digitalized operations, the ability to accurately estimate and activate aggregated flexibility in distribution systems is becoming a cornerstone of secure and sustainable grid management..... ... Read more

Rapid decarbonisation and large-scale deployment of inverter-based renewables are turning power systems into low-inertia, weakly meshed grids where stability can no longer be guaranteed by a few synchronous machines. At the same time, Distributed Energy Resources (DERs) such as storage, electric vehicles, rooftop PV, and flexible loads are proliferating in distribution networks, creating substantial but fragmented flexibility. Virtual Power Plants (VPPs) aggregate these DERs into controllable portfolios, yet ancillary services are still often treated as static reserve capacities or power set-points, neglecting the dynamic behaviour required for Fast Frequency Response (FFR), inertia emulation, and other short-timescale services. This thesis, therefore, adopts the concept of Dynamic Virtual Power Plants (DVPPs), which explicitly model internal device dynamics. The thesis develops a coordination framework and associated operation models that enable aggregated DERs to provide reliable dynamic ancillary services under scalability constraints, limited disclosure of local information, and behavioural considerations. It is organised around three themes: modelling DVPP operation for dynamic services, designing distributed and scalable solution methods, and integrating prosumer motivation into ancillary-service provision. ... Read more

As power systems increasingly rely on renewable energy, grid services traditionally supplied by central plants must increasingly be sourced from distributed energy resources (DERs). Virtual power plants (VPPs) aggregate DERs to act as a single entity, but coordination is complicated by information asymmetry, possibly resulting in strategic behaviour. This thesis studies how we can design a mechanism for a commercial VPP, having to satisfy a fixed commitment, while optimising the revenue from the VPP operator.
We first develop a tractable, multi-period VPP model with linear costs, local and temporal constraints for DERs and soft system-wide commitments enforced via deviation penalties. On top of this model we design and compare four mechanisms: first-price sealed bid (FPSB), uniform pricing, Vickrey–Clarke–Groves (VCG) and Arrow–d’Aspremont– Gerard-Varet (AGV). We evaluate them on revenue optimality, weak budget balance, incentive compatibility, individual rationality and scalability. Furthermore, we investigate how the composition of a VPP’s portfolio could inform mechanism design choices. FPSB realises payments equalling costs under truthful reports and remains competitive for small strategic fractions, but overpayment grows with the share of strategic agents and with cost dispersion. Uniform pricing is comparatively insensitive to the strategic fraction but highly sensitive to cost dispersion, often leading to large overpayments. VCG is strategy-proof and insensitive to strategic behaviour, yet externality payments increase with cost dispersion and raise total payouts. AGV keeps the payment-to-cost ratio near or below one by relying on expected externalities and scaling, improving operator viability but potentially violating individual rationality in instances. These results yielded the following guidelines regarding the suitability of mechanisms. FPSB for low strategic participation, uniform pricing for homogeneous portfolios, VCG when truthfulness is vital and external funding is possible, and AGV when operator viability is the hard constraint with safeguards for individual rationality.F ... Read more

Intraday electricity markets (IDM), which is designed to correct forecast error of renewable energy generations and enable energy trading, are characterized by high volatility and rapid price fluctuations, which not only provide market participants with strong motivation to make accurate price predictions, but also present significant challenges. The use of machine learning methods for price prediction has become a major trend in recent research. However, in previous studies, only a few specific features, such as Volume-Weighted Average Price (VWAP) and last transaction price \cite{abstract167, lasso39, abstract168}, have been applied, while the rich features embedded in orderbooks have not received sufficient attention. Furthermore, while quantile regression tasks, which provide richer information for trading strategies, have been employed in IDM price prediction, they have generally been confined to statistical models \cite{sta131, lasso38}. Deep learning-based quantile regression, capable of capturing nonlinear relationships and incorporating uncertainty, has yet to be applied. Additionally, in current research, IDM price prediction is often based on a specific orderbook, and thus, the generalization of prediction methods across different orderbooks, as well as their structural similarities across different markets and product types, has not been convincingly addressed.

To address the challenges mentioned above, this report focuses on the German and Austrian markets over continuous trading periods from January 2022 to January 2025, considering both hourly and quarter-hourly products. A total of 384 feature candidates were extracted, including percentiles, momentum, and volatility of prices and trading volumes on both buy and sell sides across multiple time windows. For the extracted feature candidates, we propose an innovative feature selection approach based on their correlation with normal and extreme price labels. Comparative experiments demonstrate that this algorithm outperforms L1-based selection and Principal Component Analysis (PCA) compression in quantile forecast evaluations. Based on the selected features, Quantile LightGBM (QLGBM), Quantile Extreme Gradient Boosting (QXGB), Quantile Multilayer Perceptron (QMLP), and Quantile Kolmogorov–Arnold Network (QKAN) were used to predict the labels, providing a comprehensive set of benchmarks. Additionally, generalization studies across markets and products were conducted using transfer learning, with multiple strategies such as zero-shot, fine-tuning, and joint learning applied. The results reveal valuable insights into the relationships between different markets and product types. ... Read more

This work proposes a new Modelling-to-Generate Alternatives (MGA) method for Energy System Optimisation Models (ESOMs) using a Genetic Algorithm (GA).
Instead of generating each alternative one by one, the GA aims to optimise for a diverse set of alternatives, meaning they cover the space of possible alternatives as evenly as possible.
Such a diverse set of alternatives has the potential to improve the decision-making process by accelerating the extraction of stakeholder requirements and finding more agreeable compromises.
Before designing the algorithm, we investigate what diversity metric is most suitable to optimise.
The components of the GA are designed to exploit useful properties of ESOMs to increase efficiency.
The performance of the GA is tested in terms of output quality and scalability for increasingly large ESOMs, showing promising performance in terms of output quality for a similar computational burden as state-of-the-art MGA methods.
A potential issue caused by the curse of dimensionality is formulated, requiring further investigation on its impact on the quality of the method's output.
We show the generated output of applying the proposed method to the European power system, which encourages further testing of the method on increasingly large ESOMs. ... Read more

Distributed energy resources challenge the situational awareness of power flows. Many distribution grid (DG) operators have not yet implemented state estimation (SE) due to the expense or privacy constraints of measurements that lead to an unobservable system, as well as inaccurate grid parameters. A key concern with the latter is the presence of medium- and low-voltage transformers with off-load tap changers, whose tap positions critically influence voltage levels across the network. Identifying transformers where the registered position is likely incorrect — and, when on-site verification is impractical, estimating a plausible tap setting — constitutes a valuable contribution to improving network observability and operational accuracy. Although operators could manually inspect each transformer, this is impractical — there are, for instance, up to 20,000 transformers in the Southern Netherlands.

To address these challenges, this thesis proposes a novel topology-aware framework for estimating state and transformer tap positions in unobservable DGs. The framework comprises two key components. First, a generative adversarial network (GAN) is used to train a generative model conditioned on the network topology and synthetic power flow data, generating realistic measurements. Second, an integrated model, referred to as the TapSEGNN model, is proposed for estimating state and transformer tap positions.

Both of these components employ a core model architecture which combines graph and simplicial complex neural networks to capture spatial dependencies between nodes, edges, and higher-order structures. To train these components, an industrial-grade data-processing pipeline was developed using a real DG topology and simulating the exact available measurement locations. The results show that balanced adversarial training of GAN accurately imputes the missing active power injection measurements, but produces high variance in imputations for voltage magnitude and active power flow measurements. The performance of the TapSEGNN model demonstrates at least tenfold higher accuracy for SE compared to conventional methods, and it predicts transformer tap positions with 100% accuracy in a computationally efficient manner. TapSEGNN also shows promising scalability for larger networks; however, it struggles with generalisability across similar real networks. Finally, the suboptimal performance of both components in certain aspects warrants further investigation, which is recommended as future work. ... Read more

State estimation (SE) plays a critical role as a prerequisite for grid control and operation. However, the increasing penetration of distributed energy resources (DERs) and integrated energy systems (IES) introduces new challenges—such as unreliable pseudo-measurements and time-varying slack bus conditions—which make traditional methods like weighted least squares (WLS) increasingly difficult to apply. Moreover, DER integration leads to more frequent topological changes due to safety requirements and economic considerations. Yet, most existing machine learning methods for SE do not explicitly consider the topological changes. Therefore, this study evaluates three GNN-based models—GCN, GAT, and EvolveGCN—for state estimation in power grids under topological changes.

First, in the scenario without topological changes, where only the phase angle of the slack bus is fixed and noisy measurements are used, WLS performs worse than the three GNNs, indicating its susceptibility to interference under non-ideal conditions. Second, among the static GNNs, GAT performs best when topological changes are visible during training, but it cannot fully predict the voltage drops of unseen topological changes. GCN, on the other hand, demonstrates better generalization to unseen topologies and effectively suppress overfitting caused by noise. Third, although EvolveGCN is less accurate overall, it shows greater stability on nodes near PV buses, highlighting its potential to use historical information to enhance the time dimension when dealing with problems such as weak spatial correlation or local information loss.

These results suggest that GCN and GAT are potentially more suitable than WLS for SE in distribution grids with high DER penetration or IES. However, PV buses pose unique modeling challenges: they are physically but not numerically correlated with neighboring nodes, which make static GNNs hard to model their value change. Accurate estimation at PV buses is crucial, as they inject power into the system; in this context, EvolveGCN shows promise due to its stable predictions at these nodes. ... Read more

Cascading failures in power networks pose a significant threat, capable of escalating from isolated line outages to extensive blackouts with severe economic and societal impacts. The topic presents a probabilistic framework designed to assess and compute the risk of cascading failures within the power network and rank various cascade contingencies, utilizing the IEEE39 bus 10-machine New England Power System. Using DigSILENT PowerFactory, a detailed contingency analysis was conducted, focusing on line loading conditions following faults as a starting point. This approach operates under the foundational assumption that cascading failures can be effectively modelled through the sequential analysis of line contingencies. Central to this framework is assessing topological vulnerabilities in the grid and determining frequently occurring outage patterns called probabilistic contingency motifs (PCMs). By analyzing the characteristics of the grid, an impact metric is proposed using short-circuit analysis, electrical distance and LODFs for the identified cascade contingency. The outage probabilities and the proposed impacts are used to compute the risks of cascade cases.

After ranking based on their associated risks, high-risk contingencies are dynamically simulated through time-domain simulations to assess dynamic security. This simulation approach validates the model's predictions and ensures that ranked contingencies reflect realistic cascades. Use cases can show that the model will enable Transmission System Operators (TSOs) to implement preventive measures and simulate corrective actions effectively. By systematically identifying and incorporating PCMs and leveraging grid topology, the model estimates the likelihood and impact of cascading events and delivers actionable insights for improving power system robustness. Future work will expand the framework’s scalability to larger and more complex networks and integrate real-time data streams to facilitate dynamic risk assessment and proactive mitigation strategies. ... Read more

This thesis investigates the occurrence and mitigation of Sequential Social Dilemmas (SSDs) in Local Energy Communities (LECs) managed through Multi-agent Reinforcement Learning (MARL). LECs have great potential as pivotal elements in the green energy transition, yet the inherent conflict between individual incentives and community-wide objectives creates SSD scenarios that challenge learning processes. To address these issues, we propose an agent-centric approach and develop a custom MARL environment where agents interact via a communal battery system and a local trading mechanism.

We systematically investigate the impact of resource constraints and social interactions on the agents' learning. In non-cooperative settings, limited resources impede policy optimization, while the introduction of a shared battery reveals SSD dynamics driven by both greed and fear factors. Our experiments show that rescaling the training data leads agents to adopt more cooperative behaviors, and that reward function modifications incentivizing community-friendly battery use cause a significant increase in social welfare. These mitigation techniques are further validated in a realistic LEC environment with multiple, heterogeneous households engaging in trading and storage actions.

The contributions of this thesis are threefold: (1) the proposal of a new agent-centric MARL environment for LECs, (2) the demonstration of SSDs impacting MARL performance in these decentralized energy systems, and (3) the introduction of concrete strategies for aligning individual and community incentives. ... Read more

As the power system grows more complex and active, equivalent models have become a solution for modelling parts of the network that have limited observability or are confidential or too complex to simulate otherwise. In the past decade, this topic has also made its way to distribution networks because of its transition towards an active network, which was not the case before. The current grey- and black-box techniques for equivalent modelling create models that are fitted to the average dynamic response of the system or have limited applicability. Also, the existing methods lack extensive verification under different system conditions, and these works rarely focus on active distribution networks (ADNs) with multiple points of common coupling (PCCs). This thesis aims to develop an equivalent model that can estimate the dynamic response of an active distribution network with multiple PCCs and under diverse operating conditions and topological changes.

The proposed approach is based on a graph-time convolutional neural network (GTCNN) that relates available PMU measurements inside the distribution network on a graph structure 𝒢_GTCNN. The graph 𝒢_GTCNN is obtained by taking a modified line graph of the graph representation of the power system and is expanded using the Cartesian product graph rule to include the temporal dependencies of nodes on their past values. The inputs of the equivalent model are the voltage magnitude |V| and angle θ at the PCC and the initial power injections P₀ and Q at non-PCC nodes, while the model outputs the active P and reactive Q power at the PCC nodes. The GTCNN explicitly considers the initial power injections P₀ and Q₀ at non-PCC nodes to help the model learn how different operating conditions and topological changes impact the dynamic response. The DSO trains the equivalent model using simulation data or collected PMU measurements. The model is exchanged with the TSO every month, who can use the equivalent in co-simulation with their transmission network model to perform transient stability studies.

The GTCNN-based equivalent model showed promising performance as an equivalent model for transient stability. The GTCNN was benchmarked against two state-of-the-art Long Short-Term Memory (LSTM)-based equivalent models and a hybrid GTCNN-LSTM model. The evaluation was performed on a real Dutch distribution network using three datasets, each focussing on a different system condition: different fault events, different operating conditions and different hidden topological changes. The GTCNN-based equivalent model had a mean-squared error (MSE) below 0.02 for each dataset, which means it can accurately reproduce the dynamics. This accuracy is comparable to the LSTM-based equivalent models, but the GTCNN could train 4x faster. The GTCNN also showed good generalisation performance, as its accuracy did not decrease on the validation and test sets. A study on scaling performance suggested that the MSE of the GTCNN-based equivalent model increases slower than that of the LSTM-based models while its training time increases faster. Therefore, the GTCNN-based equivalent model trains faster for smaller ADNs but will be more accurate with more measurement nodes. However, the proposed GTCNN has difficulty learning the response at different close-by PCC terminals if the dynamics are different.

The developed GTCNN-based equivalent model can predict the dynamic response accurately under changing topologies and operating conditions at a similar performance level to existing LSTM-based approaches. However, its training time is much faster, which can result in a more accurate equivalent model by a more frequent model exchange between the DSO and TSO or a more extensive dataset being used to train the model. In future research, the GTCNN performance will be evaluated on a more comprehensive dataset containing all three system conditions to establish how much data is needed to train the equivalent model accurately. Also, the system frequency will be considered as an additional input. Moreover, its scaling performance will be evaluated more extensively and with a more efficient coding implementation. Furthermore, a heterogenous graph convolutional operator will be implemented to learn the connection per relational type (source node - edge type - target node). Finally, the co-simulation interface between the equivalent model and popular simulation tools will be explored.
... Read more

The rapid shift toward renewable energy has positioned solar power as a key player in reducing carbon emissions. Yet, the inherent variability of solar irradiance, in particular abrupt fluctuations caused by local cloud movements, poses significant challenges for grid stability and hinders the large-scale adoption of this technology. Accurate short-term forecasting of solar irradiance becomes crucial to mitigate these issues.
Traditional forecasting methods, such as Numerical Weather Prediction, lack the spatial and temporal resolution required to predict these sudden changes in real time. To address this gap, we propose a novel generative AI pipeline that employs diffusion models to predict future sky conditions using ground-based sky images. These predicted images are processed by a convolutional neural network to forecast the solar irradiance reaching the ground.
We benchmark our approach against existing machine learning and traditional forecasting techniques, showing promising improvements in predicting short-term irradiance, particularly during dynamic situations. Additionally, we explore the role of stochasticity in diffusion models and develop a probabilistic framework that generates full probability distributions rather than single-point predictions. This allows our method not only to deliver robust predictive performance but also to quantify the uncertainty associated with each prediction.
... Read more

To keep pace with increasing renewable energy penetration and consequent increase in inverter-based resources in the power grid, it is pertinent for present-day research to address the resulting drop in system inertia levels and its impact on frequency stability. With decreasing levels of inherent rotational inertia present in the system, any sudden disturbance causing an energy imbalance in the grid could lead to more drastic excursions of system frequency than those experienced hitherto. To ensure the resilience of the grid in such scenarios, advanced and competent frequency stability assessment and control methods are required. This thesis presents Neural Ordinary Differential Equations (NODE), a recently introduced family of neural networks, as an effective tool to achieve fast, real time estimates of the expected frequency response trajectory during an energy imbalance event.

Since high-impact frequency instability events are sparse in reality, both real-world grid data and synthetically generated data corresponding to different inertial conditions are used to train predictive NODE models. Firstly, NODE is adapted to frequency prediction applications through relevant data processing steps, and modification of network parameters and algorithmic aspects pertaining to the predictive model definition. Secondly, patterns corresponding to specific sections of the frequency response curve are used to selectively train NODE models. Pattern-specific training methods exhibit better prediction performance when the NODE model encounters frequency behaviour similar to the one it initially trained on. Thirdly, a pre-training approach to cut short on the real-time training time required by NODE models to achieve desired levels of prediction performance is presented. Fast estimates of critical frequency stability parameters like nadir could act as potential triggers for early stability control actions to achieve a more controlled frequency response.

Application of predictive NODE models for different frequency scenarios are presented using three test-cases: normal operating scenario, restoration post-system split scenario and synthetically generated high-impact frequency disturbance scenarios. Model tuning and training methods specific to each test-case are described, and prediction results are evaluated with relevant performance metrics. Finally, a comparison is made between the implementation of NODE among different test-cases and real-world implications of the frequency prediction outcomes from the test-cases are further discussed. ... Read more

This work seeks to resolve an outstanding problem in the use of reinforcement-learning methods for the simulation of economically-rational agents. We discuss the problem of non-stationarity, and how this subsequently limits market simulation capabilities. After explicating and isolating the source of the problem for a day-ahead electricity market, we demonstrate the application of methods which resolve this problem in simple test-cases, and prove conditions under which similar methods will work in general. Subsequently, we illustrate how these techniques can be used to solve a restricted market-design problem, in the process introducing a framework for discussing adversarial market-design for electricity markets in general. It is hoped that, insofar as they provide a new feedback-loop for market-design, these results will facilitate the design of more complex electricity-markets suitable for the energy transition. ... Read more

The growing demand for electricity, driven by widespread adoption of heat pumps, electric vehicles, and industrial electrification, strains power grids and introduces challenges for a reliable and secure supply amidst intermittent renewable energy integration. Network topology control offers flexibility, altering connections to redirect power flows and mitigate transmission line overloads. This thesis aims to investigate an ML and AI approach to overcome the computational complexity. The proposed approach merges a curriculum-trained machine learning agent with a Monte Carlo Tree Search (MCTS) to enhance power network action security. The MCTS guides the simulation of potential actions, considering future outcomes for improved long-term performance identification. A curriculum-based ML approach is used to pre-train an agent to propose grid actions. MCTS is then used to secure these actions, leveraging outcomes in the training algorithm for enhanced sample efficiency and reduced training times. The approach uses MCTS-verified, simulation-tested actions for immediate training feedback, eliminating the need to wait for scenario completion, enhancing sample efficiency. An electrically distance-guided search in the MCTS improves convergence by prioritising actions closer to overflows, often found to be most influential in reducing violations. ... Read more

In the electricity system, one barrier to the energy transition is the degradation of frequency stability due to the decrease of system inertia and frequency control ancillary services (FCAS), which is caused by the replacement of inertia-abundant and governor-based conventional power plants with zero-inertia and inverter-based renewable energy sources (RES).

There are already inverter technologies for RES and battery energy storage systems (BESS) to provide virtual inertia (VI) and fast frequency response (FFR) services, which are equivalent to physical inertia and conventional FCAS. Potential providers include wind, solar, battery energy storage systems (BESS), and other types of devices with the feature of energy storage. However, there are non-technical barriers to the actual implementation. For example, in most parts of the world, these services cannot participate in the electricity market and thus there is a lack of incentive for both the provision of and investment in the VI and FFR.

In the literature, there are already proposals of possible market designs for VI and FFR that procure the services, guarantee the frequency-stability requirements, and provide payments to the service provider. The main focal point is on the formulation of security-constrained unit commitment (SCUC) and security-constrained economic dispatch (SCED) problems. The formulation needs to be accurate in modeling, be solvable, and be with reasonable computational burden. Most of them only considered the allocation of ancillary services but did not price them, or price them but only by directly assigning a shadow price. Only a few works considered explicit prices in the bid.

In this project, we consider a market design with explicit bid prices. In the SCUC and SCED problems, we adopt the state-of-the-art formulation of frequency nadir constraint and method of modeling the frequency dynamics based on the linear ramp assumption of the dynamics of frequency response (FR). With these methodologies, we will investigate the features of such a market by identifying inter-dependencies of parameters in the market setups, including the interaction between the bid price and bid amount of FR and VI, and analyzing the underlying mechanisms. The results show that the amount of FR sold depends on the bid price of FR monotonously and the amount of VI sold depends on the bid price of VI monotonously, though with different patterns. The underlying reason for such dependencies is that FR, VI, and the size of the largest unit both help mitigate frequency drop and recover it and they are influencing each other. The price of FR or VI decides the relative worthiness of each option. The size of the largest unit is also related to the amount of FR sold, which is confirmed by the formulation of the analytical nadir and the QSS constraint. However, the way that the amount of available FR influences the amount of VI sold does not show a clear pattern.

These findings provide insights into the interactions between the FR and VI products and thus provide a reference for the design of the FCAS market. ... Read more

The transition to green energy is reshaping the energy landscape, marked by increased integration of renewable energy sources, distributed resources, and the electrification of other energy sectors. These changes challenge grid security, particularly regarding the N-1 security criterion, a crucial factor in preventing blackouts. Furthermore, climate change is contributing to the growing frequency of extreme weather events, which constitute the second major cause of blackouts. As grid complexity keeps on increasing, the need for N-k security, where k lines fail simultaneously, and increased resilience against extreme weather events is becoming increasingly evident. This necessitates studying the security constrained optimal power flow (SCOPF) problem considering multiple line outages (N-k). Current methods exhibit poor scalability as k increases. In response to the challenge of limited scalability, this thesis proposes a constraint-driven machine learning approach to approximate N-k SCOPFs.

The proposed approach relies on the linearized direct current optimal power flow. The approach utilizes a neural network to map power system loads to generator setpoints. A feasibility restoration layer is employed to restore base case infeasible predictions. By incorporating line outage distribution factors (LODFs), all post-contingency flows are computed. The loss function utilized to train the neural network draws inspiration from the penalty function method. Lastly, a copula analysis computes joint outage probabilities for k \textgreater 1 enabling a probabilistic security assessment. The first academic contribution of this thesis is the development of a constraint-driven approach to approximate N-k SCOPFs considering all contingencies using LODFs. The second academic contribution is the formulation of a N-k risk based security criterion, providing an alternative to the current deterministic N-1 security criterion.

The approach shows promise in its ability to scale effectively to N-k contingencies. Using LODFs, the approach effectively computes all post-contingency flows for up to k = 3. Moreover, case studies show the constraint-driven approach's effectiveness in identifying violating post-contingency cases, with up to 173$\times$ speedups and close to optimal dispatch costs. However, the consideration of N-k contingencies holds combinatorial complexity, and more efficient methods need to be developed for the computation and storage of all LODFs, and for the computation of all post-contingency flows. Additionally, the proposed constraint-driven approach can not enforce any post-contingency constraints, necessitating post-contingency feasibility checks when security against specific contingencies is required. Next, by incorporating probabilities, the approach shows promise in improving power systems security and resilience, but further research is necessary.

In this thesis, only line outages are considered. In the future, the approach could be modified to additionally account for other equipment outages (e.g. generator outages). Furthermore, future research could investigate the adoption of this approach in corrective control settings, where it is employed in the restorative phase of a contingency event. Another suggestion is centered around the incorporation of graph neural networks in the proposed approach, which could provide a more scalable alternative to fully connected linear neural networks. Furthermore, more scalable methodologies could be explored to construct the matrix containing all LODFs, and a more scalable methodology for computing all post-contingency flows could be developed. Finally, future work could investigate how to utilize the proposed approach under varying conditions like network topology changes or changing outage probabilities. ... Read more

Whereas in the past, Distribution Systems played a passive role in connecting customers to electricity, Distribution System Operators (DSOs) will have to take in the future a more active role in monitoring and regulating the network to deal with the new behaviors and dynamics of the system brought by the energy transition. State Estimation, a task traditionally reserved for Transmission System Operators (TSOs), is, therefore, a needed tool for DSOs to properly monitor the distribution grid in the future. However, the implementation of Distribution System State Estimation (DSSE) faces several challenges. The distribution system lacks observability to get satisfying estimation accuracy, the denser network increases the complexity of the estimation process, and the lack of labeled data makes training Machine Learning alternatives difficult. To tackle these issues, we propose the Deep Statistical Solver for Distribution System State Estimation (DSS2), a Deep Learning model based on the Graph Neural Network (GNN) architecture and the Physic-Informed Machine Learning (PIML) framework.

The DSS2 model is based on the Deep Statistical Solver (DSS) framework, which seamlessly models power systems into GNN using Hyper-Heterogeneous Multi Graphs (H2MG), and emphasizes semi-supervised learning by learning to optimize, using optimization problem as a loss function. This thesis extends the DSS framework to the DSSE problem, using the traditional State Estimation algorithm as an optimization problem to learn, and incorporating the power flow equations in the loss function. This model is trained through a semi-supervised approach to learn the physics of the problem and alleviate the need for labels and uses the Deep Learning tools to improve accuracy and robustness in the DSSE task.

Case studies on 14-bus, 70-bus, and 179-bus networks show promising results, with the model outperforming the traditional WLS algorithm while showing better robustness. The model also competed in performance against supervised models and showed to be more suitable for the semi-supervised learning approach than simpler GNN architectures. ... Read more

In this thesis, a new photovoltaic fault detection and classification method is proposed. It combines the generation of a synthetic photovoltaic training database and the use of a machine learning model to detect and classify faults in small-scale residential PV systems. The database was generated in Matlab, and the machine learning modeling was done with the scikit-learn library for Python. From the modeled PV systems, solely power yield is used as an indicator, combined with system age and meteorological conditions. Using these features, four types of machine learning models are used to detect malfunctioning PV systems and classify short-circuit faults and open-circuit faults. This thesis also shows the benefit of a synthetic PV training base as opposed to alternative methods, with increasing performance due to control of database balance.

The result of this thesis is a method that can be used to construct a model for detection and classification of photovoltaic faults, specific to a single residential PV system. Malfunctioning systems can be detected with an accuracy of 80.4% using a random forest algorithm. For fault type classification, an F1-score of 0.759 was achieved, also using a random forest.
... Read more

With the increasing environmental concerns, the world is moving towards rapid decarbonization, and to meet the growing energy demand, more renewable energy sources like solar, Wind are getting added to the existing Distribution grid. The addition of new loads like Electric vehicles (EV), Heat pumps, etc. is putting additional pressure on the grid leading to some technical problems. Due to this, it is important to have real-time monitoring of the system by the Distribution System Operators(DSO)s which can be obtained by performing the State Estimation. The Distribution System State Estimation along with the state estimator has been discussed in the literature.

However, the accuracy of the state estimation is highly affected by the absence of real measurements from the meter and the presence of high-variance pseudo measurements. But due to the cost factor, it is practically not possible to install meters at every node. To reduce the state estimation error, a new meter placement strategy has been proposed in this project with the help of open-source software called Power Grid Model, devolved by the Dutch DSO Alliander for performing steady state analysis of the network.

The work presented in this project includes a sensitivity analysis of the proposed meter placement algorithm which is applied to a test MV distribution network in the Netherlands for the different sample sizes of the load profile as well as the standard deviation of the pseudo measurements. The application of the meter placement algorithm results in a subsequent reduction of the state estimation error. To find a suitable number of meters that can be placed in the network, three-meter placement criteria have been proposed and applied to the test network. The results show the effectiveness of the proposed meter placement algorithm for the determination of a suitable path as well as the number of meters that can be placed in the distribution network. This is important for the grid operators as it will help them to make decisions for the proper maintenance and operation of the grid. ... Read more

The growth of renewable energy technologies is leading to energy systems that are more reliant than ever on renewables such as Wind and Photovoltaic (PV) power. Despite their benefits in terms of sustainability, their ubiquity poses challenges in maintaining grid stability given their intermittency, emphasising the prediction of power fluctuations. Physical models and statistical approaches, especially for nowcasting (forecasting for 0-6 hours in the future), have been superseded by Machine Learning (ML) methods in terms of forecast accuracy (below 3% Root Mean Squared Error (RMSE)). Within ML, Artificial Neural Network (ANN) methods seem to perform particularly well for nowcasting. This project focuses on predicting solar and wind meteorology with that level of accuracy, and on how to best use the prediction to minimize the cost of maintaining a balanced energy system, i.e. one where power consumption matches production at any moment. Producing accurate power predictions based on Multi-Modal (MM) data and the extent to which prediction accuracy reduces system cost are challenges to be addressed in this thesis. MM and End-to-End (E2E) training (with the system cost as the task of an ANN based algorithm) are investigated to this end. MM learning involves handling information from multiple types of input (audio and visual, for example) for performing a ML task such as regression or classification. It is of interest for this project because it has been shown to outperform other NN approaches in predicting sudden changes in solar irradiance. E2E learning entails an algorithm design which predicts the end goal of a ML process directly from the inputs. This is pursued because it addresses the true task (cost minimization) of system operators as the focus of the ML algorithm. The proposed method consists of a NN architecture that learns to fuse features from MM data (sky imagery and meteorological sensor data) at intermediate layers of the network in order to predict PV or Wind generation. This prediction is then used as an input to an Optimal Power Flow (OPF) problem (which seeks to minimize generation costs in a power system, considering power balance and transmission network constraints to ensure the twin goals of economic and secure system operation). The proposed model is trained E2E, therefore it is informed by the minimized cost solved by the optimization, rather than the intermediate power prediction (as conventional approaches would involve). In an IEEE 6-bus system with PV generation, a sequential training baseline results in costs 10% higher than a perfect forecast, while our proposed MM4-E2E approach achieves costs only 7% higher, a significant improvement. The intermediate prediction of PV power by MM4-E2E is also improved, with 18% lower RMSE by the proposed model compared to the baseline, explained by the enhancement of one modality by the other through MM learning. In a power system with two renewable sources, costs are reduced through the proposed model compared to a conventional approach (4% excess cost compared to 7%, measured against a perfect forecast), but power prediction accuracy is worse, sue to convergence to local minima. ... Read more