IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (IEEE SmartGridComm 2021) — recording passcode: SGC-VC21

Session K3: Enabling the Data-driven Future
Session COS4: Smart grid communications Session CYS4: Distribution grids and privacy Session COS5: Optimisation and AI in smart grids Session CYS5: Power Systems Application Security Session GAS3: Data-drive computational methods for grid operation
Session W2: Workshop on Charging Solutions for Electro-mobility of the Future

Session COS4

Smart grid communications

Conference
9:00 AM — 10:30 AM CEST
Local
Oct 28 Thu, 3:00 AM — 4:30 AM EDT

Data Communication Interfaces in Smart Grid Real-time Simulations: Challenges and Solutions

Mehrdad Sheikholeslami; Zuyi Li

0
This paper presents the challenges and also suggests solutions associated with developing data communication interfaces between real-time digital simulator (RTDS) and hardware or software devices under study. While RTDS supports a wide range of standard and well-established communication protocols, employing such communication protocols generally increases the cost of the educational project as these standard communication protocols require licenses as well as third-party hardware and software devices to act as gateways. The need for these licenses and third-party hardware and software devices adds to the total cost of the project and also requires additional training. This paper provides two sets of cost-effective data interface solutions for local and remote networks based on the lessons learned from different projects that the authors were involved with. These practical solutions are especially useful for projects that involve multiple partners located remotely that are facing logistic challenges due to the Covid-19 pandemic.

A novel load distribution strategy for aggregators using IoT-enabled mobile devices

Nitin Shivaraman; Jakob Fittler; Saravanan Ramanathan; Arvind Easwaran; Sebastian Steinhorst

0
The rapid proliferation of Internet-of-things (IoT) as well as mobile devices such as Electric Vehicles (EVs), has led to unpredictable load at the grid. The demand to supply ratio is particularly exacerbated at a few grid aggregators (charging stations) with excessive demand due to the geographic location, peak time, etc. Existing solutions on demand response cannot achieve significant improvements based only on time-shifting the loads without considering the device properties such as charging modes and movement capabilities to enable geographic migration. Additionally, the information on the spare capacity at a few aggregators can aid in re-channeling the load from other aggregators facing excess demand to allow migration of devices. In this paper, we model these flexible properties of the devices as a mixed-integer non-linear problem (MINLP) to minimize excess load and the improve the utility (benefit) across all devices. We propose an online distributed low-complexity heuristic that prioritizes devices based on demand and deadlines to minimize the cumulative loss in utility. The proposed heuristic is tested on an exhaustive set of synthetic data and compared with solutions from a solver/optimization tool for the same runtime to show the impracticality of using a solver. A real-world EV testbed data is also tested with our proposed solution and other scheduling solutions to show the practicality of generating a feasible schedule and a loss improvement of at least 57.23%.

Effect of 5G communication service failure on placement of Intelligent Electronic Devices in Smart Distribution Grids

Romina Muka; Michele Garau; Besmir Tola; Poul E. Heegaard

0
Information and Communication Technology (ICT) is fundamental to guarantee efficient monitoring, control and protection of smart distribution grids by interconnected Intelligent Electronic Devices (IEDs). The impact of failures in the IEDs communication service, and the dependency between the communication network and the power grid, needs to be understood and taken into account when determining the optimal placement of IEDs. In this paper, the main objective is to investigate how loss of the communication service that connects the IEDs to Distribution Management System (DMS), will affect the placement of IEDs for smart grid monitoring and control. It is assessed the impact of 5G communication service failure on the IEDs placement with the objective to minimize the interruption costs (Cost of Energy Not Supplied), and the yearly expenses of the IEDs installed. The method is tested on the IEEE 33-bus radial distribution system, with a 5G communication network, covering both rural and urban areas. The results suggest a need for more IEDs per bus in the rural area because the power lines are longer, and the failure rates are higher than in the urban area. Furthermore, when introducing sub-regions that have higher power line failure rates and less reliable communication service, we observe that more IEDs are suggested to be placed in these regions. This demonstrates that methods for IEDs placement should take into consideration the ICT communication service failures, especially in sub-regions with higher power line failure rates and/or unstable ICT communication service that comes as result of failures in the power grid.

Graph Convolution Networks-Based Island Partition for Energy and Information Coupled Active Distribution Systems

Qiyue Li; Shengquan Dai; Ximing Li; Weitao Li; Wei Sun

0
When a fault occurs in the active distribution network, it's important to divide islands according to the real-time operating status of the grid to form multiple independent microgrid systems. However, existing methods of island partition ignore the actual communication requirements of the active distribution network, so it is difficult to adapt to the impact of fluctuations in the communication quality of grid nodes, which may cause the performance of the system to deteriorate. This paper proposes an active distribution network island partition method based on graph convolutional network combined with autoencoder, which considers grid communication delay constraints and multi-objective optimization. Detailed simulation and experimental results show that the method can divide the partitions reasonably and effectively which can meet the power grid's energy and information requirements and achieve the established multiple optimization goals.

Modelling framework for study of distributed and centralized smart grid system services

Fredrik B Haugli; Poul E. Heegaard

0
The increased complexity of modern smart grids require new methods for dependability analysis, as system services depend on other services as well as components both in the power grid and ICT domain. This paper describes a method for modeling such a system with its direct and implicit dependencies. A tool has been developed for defining system models in an object oriented manner in the Python programming language and extract dependability metrics for the different system services through Discrete Event Simulation. Finally, an example case is shown illustrating the trade-off in performance and complexity between a centralized and decentralized control scheme.

Play Recording
Session CYS4

Distribution grids and privacy

Conference
9:00 AM — 10:30 AM CEST
Local
Oct 28 Thu, 3:00 AM — 4:30 AM EDT

Poisoning Attack against Event Classification in Distribution Synchrophasor Measurements

Mohasinina Kamal; Alireza Shahsavari; Hamed Mohsenian-Rad

0
Distribution-level phasor measurement units (DPMUs), a.k.a., micro-PMUs, have received a growing attention in recent years to support various applications in power distribution systems. Many of the applications of micro-PMUs work based on the analysis of events in the stream of synchrophasor measurements to achieve situational awareness. A key step in almost every event-based method in this emerging field is to classify the type of the event, where classification can be done with respect to various factors. However, if the task of event classification is compromised, then an adversary can highly affect the perception of the utility operator and undermine any event-based application that makes use of the event classification results. In this paper, we explore a new cyber-threat against data-driven event classification in micro-PMU measurements. In particular, we model the poisoning attack against support vector machine (SVM) as the method of event classification; which has been used in practice to study distribution synchrophasors. We apply the new attack model to an event classifier that uses real-world micro-PMU data. In addition to conducting vulnerability analysis, we also propose a novel attack detection method which can detect and evaluate the changes in the decision boundary of the SVM due to the poisoning attack. The proposed attack detection method is also able to identify the number of poisoned data points in the training dataset.

Learning Sparse Privacy-Preserving Representations for Smart Meters Data

Mohammadhadi Shateri; Francisco Messina; Pablo Piantanida; Fabrice Labeau

2
Fine-grained Smart Meters (SMs) data recording and communication has enabled several features of Smart Grids (SGs) such as power quality monitoring, load forecasting, fault detection, and so on. In addition, it has benefited the users by giving them more control over their electricity consumption. However, it is well-known that it also discloses sensitive information about the users, i.e., an attacker can infer users' private information by analyzing the SMs data. In this study, we propose a privacy-preserving approach based on non-uniform down-sampling of SMs data. We formulate this as the problem of learning a sparse representation of SMs data with minimum information leakage and maximum utility. The architecture is composed of a releaser, which is a recurrent neural network (RNN), that is trained to generate the sparse representation by masking the SMs data, and an utility and adversary networks (also RNNs), which help the releaser to minimize the leakage of information about the private attribute, while keeping the reconstruction error of the SMs data minimum (i.e., maximum utility). The performance of the proposed technique is assessed based on actual SMs data and compared with uniform down-sampling, random (non-uniform) down-sampling, as well as the state-of-the-art in privacy-preserving methods using a data manipulation approach. It is shown that our method performs better in terms of the privacy-utility trade-off while releasing much less data, thus also being more efficient.

Electricity Theft Detection in the Presence of Prosumers Using a Cluster-based Multi-feature Detection Model

Arwa Alromih; John Clark; Prosanta Gope

1
Data driven approaches have been widely employed in recent years to detect electricity thefts. Although many techniques have been proposed in the literature, they mainly focus on electricity thefts by consumers of power from the grid. Existing studies do not consider electricity thefts by prosumers, who act as both supplier and consumer in the energy system. This is of great importance as inaccurate reports of prosumers' behaviours can disturb power system operation. Here we examine the role prosumers may play in subverting the energy system and propose a novel means of detecting such malfeasance. Specifically, we introduce a new electricity theft attack scenarios called balance attacks, where an attacker concurrently modifies his readings along with neighbouring meters in an attempt to balance the total aggregated reading. Such attacks can be difficult to detect by existing solutions that reach detection decisions based on aggregated readings. We propose a novel electricity theft detector that can detect thefts in the presence of prosumers. Current approaches use either a single model for all users across the system or else a model for each user. Here we adopt a half-way house approach and propose a cluster-based detection model. For users in a cluster, we decompose a power time series for a user into trend, cyclical and residual components. Residual data, along with different features from multiple data sources, are fed in an ML classification algorithm to detect anomalous readings. Simulations have been conducted using a newly generated dataset and results have shown that the proposed model can detect electricity theft with high detection and low error rates. The results also shows that the model can detect thefts with great accuracy from new users.

Exploiting DLMS/COSEM Data Compression To Learn Power Consumption Patterns

Marcell Fehér; Daniel E. Lucani; Morten Tranberg Hansen; Flemming Enevold Vester

2
Smart electricity meters are widely deployed report power consumption automatically and frequently. However, the current compression methods have been suspected to leak information about the times when consumers are active, by mirroring spikes of power consumption in the compressed message size. In this paper we show that, compressed message sizes are indeed highly correlated with the underlying power use. We present a clustering-based method that allows a passive adversary monitoring encrypted network traffic to build and exploit power consumption profiles of homes. We evaluate the vulnerability of legacy compressors of the DLMS/COSEM standard as well as a recently proposed algorithm. Our results show that the novel algorithm not only provides higher compression potential, but results in the least information leakage. We evaluate our results on an real, anonymized dataset spanning 9 months and 95 households.

Adversarial Machine Learning Against False Data Injection Attack Detection for Smart Grid Demand Response

Guihai Zhang; Biplab Sikdar

1
Distributed demand response (DR) is used in smart grids to allow utilities to balance the power supply with the demand by modulating the consumer's behavior by varying the price according to consumption patterns and forecasts. False data injection (FDI) attacks of DR can cause large economical losses for utilities, equipment damage, and issues with power flows. Recently, FDI attack detection methods based on deep learning models have been proposed and these methods have better detection performance as compared to traditional approaches. However, deep learning based models may be vulnerable to adversarial machine learning (AML) attacks. In this paper, we demonstrate the vulnerability of state-of-the-art deep learning based FDI attack detectors in DR scenarios to AML attacks. We propose a new black-box FDI attack framework to fabricate power demands in distributed DR scenarios that is capable of deceiving deep learning based FDI attack detection. The evaluation results show that the proposed AML framework can significantly decrease the FDI detection model's accuracy and outperforms other AML techniques proposed in literature.

Play Recording
Session W2

Workshop on Charging Solutions for Electro-mobility of the Future

Conference
9:00 AM — 1:00 PM CEST
Local
Oct 28 Thu, 3:00 AM — 7:00 AM EDT

Play Recording
Session COS5

Optimisation and AI in smart grids

Conference
11:00 AM — 12:30 PM CEST
Local
Oct 28 Thu, 5:00 AM — 6:30 AM EDT

Low-complexity Risk-averse MPC for EMS

Johannes Philippus Maree; Sebastien Gros; Venkatachalam Lakshmanan

0
A data-driven stochastic MPC strategy is presented as an EMS for the Skagerak Energilab microgrid. Uncertainties, introduced due to the intermittent nature of RES and load demands, are systematically incorporated into the MPC problem via adaptive chance-constraints. These chance-constraints promote admissible probabilistic operation of the microgrid within the stipulated SOC bounds of a ESS. For computational tractability, these chance-constraints are approximated by solving the inverse cumulative distribution function of a disturbance innovation sequence. This disturbance innovation sequence defines the difference between forecast and realized disturbances, and is sampled for a sliding window as disturbances are revealed over closed-loop operation. No a-prior assumptions are made on the distribution function of the disturbance innovation sequence, instead, solving the Maximum Spacings Estimation problem off-line, we adapt some parametrized distribution function to fit this disturbance innovation sequence. The proposed strategy has computational complexity comparable to nominal deterministic MPC, promote the satisfaction of constraints in a probabilistic sense, and, decrease closed-loop operational costs by 26%.

A World Model Based Reinforcement Learning Architecture for Autonomous Power System Control

Magnus Tarle; Mårten Björkman; Mats Larsson; Lars Nordström; Gunnar Ingeström

0
Renewable generation is leading to rapidly shifting power flows and it is anticipated that traditional power system control may soon be inadequate to cope with these fluctuations. Traditional control include human-in-the-loop-control schemes while more autonomous control methods can be categorized into Wide-Area Monitoring, Protection and Control systems (WAMPAC). Within this latter group of more advanced systems, reinforcement learning (RL) is a potential candidate to facilitate power system control facing these new challenges.

In this paper we demonstrate how a model based reinforcement learning (MBRL) algorithm, which learns and uses an internal model of the world, can be used for autonomous power system control. The proposed RL agent, called the World Model for Autonomous Power System Control (WMAP), includes a safety shield to minimize risk of poor decisions at high uncertainty. The shield can be configured to permit WMAP to take actions with the condition that WMAP asks for guidance, e.g. from a human operator, when in doubt. As an alternative, WMAP could be run in full decision support mode which would require the operator to take all the active decisions.

A case study is performed on a IEEE 14-bus system where WMAP is setup to control setpoints of two FACTS devices to emulate grid stability improvements. Results show that improved grid stability is achieved using WMAP while staying within voltage limits. Furthermore, a disastrous situation is avoided when WMAP asks for help in a test scenario event that it had not been trained for.

Towards Strategic Local Power Network Decarbonization: A Stackelberg Game Analysis

Qisheng Huang; Jianwei Huang

0
Many governments have implemented the Renewable standard portfolio (RPS) policy to enforce power utilities to procure a minimum amount of energy supply from renewable resources. We construct a two-stage Stackelberg game to explore the strategic behaviors of the power utility, the solar farm, and the prosumers under a given RPS policy. The power utility acts as the leader to decide the capacity subsidy to incentivize his prosumers and the solar farm to invest in renewables, with the objective of profit-maximization. When facing the power utility's decisions, the prosumers and the solar farm compete with each other to make the renewable investment decisions. Each prosumer seeks to minimize the total cost of energy consumption and renewable investment. The objective of the solar farm is to maximize his profit. We completely characterize the equilibrium of the dynamic game considering different capital costs. Surprisingly, we find that the prosumers are more willing to invest in renewable than the solar farm. In particular, when the prosumers and the solar farm have the same capital costs, the prosumers' total renewable investment is no less than that of the solar farm. Numerical experiments based on real-world data show that a higher market competition leads to a higher total renewable investment and a lower overall system cost.

Performance Evaluation of an Advanced Distributed Energy Resource Management Algorithm

Jing Wang; Jeff Simpson; Rui Yang; Bryan Palmintier; Soumya Tiwari; Yingchen Zhang

0
This paper presents performance evaluation of a new distributed energy resource management system (DERMS) algorithm via an advanced hardware-in-the-loop (HIL) platform. The HIL platform provides realistic testing in a laboratory environment, including the accurate modeling of sub-transmission and distribution networks, the DERMS software controller, and 84 power hardware solar photovoltaic (PV) inverters, standard communication protocols, and a capacitor bank controller. The DERMS algorithm is also called, Grid-Optimization of Solar (GO-Solar) platform which includes predictive state estimation (PSE) and online multiple objective optimization (OMOO) to dispatch the legacy devices and distributed energy resources (e.g., PV). The voltage regulation performance is evaluated under three scenarios, volt-var smart inverter (baseline), and DERMS control for 100% and 30% of PV. The results show that controlling 30% of PV systems with the GO-Solar platform may provide the best balance of control performance and implementation cost.

Benchmarking a Decentralized Reinforcement Learning Control Strategy for an Energy Community

Niklas Ebell; Marco Pruckner

0
The energy transition towards a more sustainable, secure and affordable electrical power system consisting of high shares of renewable energy sources increases the energy system's complexity. It creates an energy system in a more decentralized pattern with many more stakeholders involved. In this context, new data-driven operation control strategies play an important role in order to provide fast decision support and a better coordination of electrical assets in the distribution grid. In this paper, we evaluate a novel Multi-Agent Reinforcement Learning approach which focuses on cooperative agents with only local state information and aim to balance the electricity generation and consumption of an energy community consisting of ten households. This approach is compared to a rule-based and an optimal control policy. Results show that independent Q-learner achieve performance 35% better than rule-based control and compensate high computational effort with adaptability, simplicity in communication requirements and respect of data-privacy.

Play Recording
Session CYS5

Power Systems Application Security

Conference
11:00 AM — 12:30 PM CEST
Local
Oct 28 Thu, 5:00 AM — 6:30 AM EDT

Modeling of Cyber Attacks Against Converter-Driven Stability of PMSG-Based Wind Farms with Intentional Subsynchronous Resonance

Hang Du; Jun Yan; Mohsen Ghafouri; Rawad Zgheib; Marthe Kassouf; Mourad Debbabi

0
Subsynchronous resonance (SSR) is among the most severe instability conditions that may happen when grid-tied inverter-based renewable energy sources (RESs), like wind power, connect to a weak transmission grid. The potential impact of SSR includes loss of wind power generation, physical equipment damage, or instability that could spread to a larger area. Such risks make the subsynchronous stability of PMSG-based wind farms a potential target for adversaries. To this end, this paper investigates and models two new cyber attack schemes targeting SSR in PMSG-based wind farms, which have high energy output and less maintenance. Considering the major causes and different damping controls for SSR in PMSG-based wind farms, this paper demonstrates the feasibility of the threat from the two proposed cyber attacks and compares them using the IEEE 9-bus benchmark. The results show that smartly craft cyber attacks can successfully degrade SSR damping, trigger an SSR, and even destabilize the power grid.

Vulnerabilities of Power System Operations to Load Forecasting Data Injection Attacks

Yize Chen; Yushi Tan; Ling Zhang; Baosen Zhang

1
We study the security threats of power system operations from a class of data injection attacks on load forecasting algorithms. In particular, we design an attack strategy on input features for load forecasting algorithms which can be implemented by an attacker with minimal system knowledge. System operators can be oblivious of such wrong load forecasts, which lead to uneconomical or even insecure decisions in commitment and dispatch. This paper brings up the security issues of load forecasting algorithms and shows that accurate load forecasting algorithm is not necessarily robust to malicious attacks. If power grid topology information is exploited, more severe attacks can be designed. We demonstrate the impact of load forecasting attacks on two IEEE test cases. We show our attack strategy is able to cause load shedding with high probability under various settings in the 14-bus test case, and also demonstrate system-wide threats in the 118-bus test case with limited local attacks.

Securing SCADA networks for smart grids via a distributed evaluation of local sensor data

Verena Menzel; Johann Hurink; Anne Remke

0
Within smart grids the safe and dependable distribution of electric power highly depends on the security of Supervisory Control and Data Acquisition (SCADA) systems and their underlying communication protocols. Existing network-based intrusion detection systems for Industrial Control Systems (ICS) are usually centrally applied at the SCADA server and do not take the underlying physical process into account. A recent line of work proposes an additional layer of security via a process-aware approach applied locally at the field stations. This paper broadens the scope of process-aware monitoring by considering the interaction between neighboring field stations, which facilitates upcoming trends of decentralized energy management (DEM). Local security monitoring is lifted to monitoring neighborhoods of field stations, therefore achieving a broader grid coverage w.r.t. security. We provide a distributed monitoring algorithm of the generated sensory readings for this extended setting. The feasibility of the approach is shown via a prototype simulation testbed and a scenario with two subgrids.

Distort to Detect, not Affect: Detecting Stealthy Sensor Attacks with Micro-distortion

Suman Sourav; Binbin Chen

0
In this paper, we propose an effective and easily deployable approach to detect the presence of stealthy sensor attacks in industrial control systems, where (legacy) control devices critically rely on accurate (and usually non-encrypted) sensor readings. Specifically, we focus on stealthy attacks that crash a sensor and then immediately impersonate that sensor by sending out fake readings. We consider attackers who aim to stay hidden in the system for prolonged period. To detect such attacks, our approach relies on continuous injection of "micro distortion" to the original sensor's readings. In particular, the injected distortion should be kept strictly within a small magnitude (e.g., 0.5% of the possible operating value range), to ensure it does not affect the normal functioning of the ICS. Our approach uses a pre-shared secret sequence between a sensor and the defender to generate the micro-distortions. One key challenge is that the micro-distortions injected are often much lower than the sensor's actual readings, hence can be easily overwhelmed by the latter. To overcome this, we leverage the observation that sensor readings in many ICS (and power grid in particular) often change gradually in a significant fraction of time (i.e., with small difference between consecutive time slots). We devise a simple yet effective algorithm that can detect stealthy attackers in a highly accurate and fast (i.e., using less than 100 samples) manner. We demonstrate the effectiveness of our defense using real-world sensor reading traces from two different smart grid systems.

Play Recording
Session K3

Enabling the Data-driven Future

Conference
1:30 PM — 2:30 PM CEST
Local
Oct 28 Thu, 7:30 AM — 8:30 AM EDT

Enabling the Data-driven Future

Steven Blair

0
This talk does not have an abstract.

Play Recording
Session GAS3

Data-drive computational methods for grid operation

Conference
3:00 PM — 4:30 PM CEST
Local
Oct 28 Thu, 9:00 AM — 10:30 AM EDT

Physics-Informed Neural Networks for Minimising Worst-Case Violations in DC Optimal Power Flow

Rahul Nellikkath; Spyros Chatzivasileiadis

0
Physics-informed neural networks exploit the existing models of the underlying physical systems to generate higher accuracy results with fewer data. Such approaches can help drastically reduce the computation time and generate a good estimate of computationally intensive processes in power systems, such as dynamic security assessment or optimal power flow. Combined with the extraction of worst-case guarantees for the neural network performance, such neural networks can be applied in safety-critical applications in power systems and build a high level of trust among power system operators. This paper takes the first step and applies, for the first time to our knowledge, Physics-Informed Neural Networks with Worst-Case Guarantees for the DC Optimal Power Flow problem. We look for guarantees related to (i) maximum constraint violations, (ii) maximum distance between predicted and optimal decision variables, and (iii) maximum sub-optimality in the entire input domain. In a range of PGLib-OPF networks, we demonstrate how physics-informed neural networks can be supplied with worst-case guarantees and how they can lead to reduced worst-case violations compared with conventional neural networks.

Energy Blockchain for Demand Response and Distributed Energy Resource Management

Mikhak Samadi; Henry Schriemer; Sushmita Ruj; Melike Erol-Kantarci

1
The high impact of demand reduction on the energy grid management and the importance of reducing loss of distributed energy resources (DERs), in addition to the necessity of a secure distributed data storing system motivate us to propose an energy blockchain solution. This paper presents a demand response (DR) solution utilizing energy blockchain to reduce demand, save the extra DERs, and efficiently incorporate customers block mining ability. In this work, a real dataset of customer demand profiles and PV generation in the Ottawa region is used to deploy a DR Stackelberg game between a control agent (CA) and local customers to negotiate demand reduction by integrating the block mining method as DERs saving. This article presents a novel and well-suited consensus algorithm, Proof of Energy Saving (PoES), that is used to incentivize the customers to reduce their demand, discharge their electric vehicle (EV) and maximize their chance for block mining to earn monetary rewards and DER savings. This results in lower peak demand, customer bill reduction, and transforms energy savings into monetary resources. Furthermore, the results show that our proposed consensus algorithm is robust and secure against malicious actions of users.

Distributed Weighted Least-Squares and Gaussian Belief Propagation: An Integrated Approach

Dino Zivojevic; Muhamed Delalic; Darijo Raca; Dejan Vukobratović; Mirsad Cosovic

0
The purpose of a state estimation (SE) algorithm is to estimate the values of the state variables considering the available set of measurements. The centralised SE becomes impractical for large-scale systems, particularly if the measurements are spatially distributed across wide geographical areas. Dividing the large-scale systems into clusters (\ie subsystems) and distributing the computation across clusters, solves the constraints of centralised based approach. In such scenarios, using distributed SE methods brings numerous advantages over the centralised ones. In this paper, we propose a novel distributed approach to solve the linear SE model by combining local solutions obtained by applying weighted least-squares (WLS) of the given subsystems with the Gaussian belief propagation (GBP) algorithm. The proposed algorithm is based on the factor graph operating without a central coordinator, where subsystems exchange only ``beliefs", thus preserving privacy of the measurement data and state variables. Further, we propose an approach to speed-up evaluation of the local solution upon arrival of a new information to the subsystem. Finally, the proposed algorithm provides results that reach accuracy of the centralised WLS solution in a few iterations, and outperforms vanilla GBP algorithm with respect to its convergence properties.

Learning without Data: Physics-Informed Neural Networks for Fast Time-Domain Simulation

Jochen Stiasny; Samuel Chevalier; Spyros Chatzivasileiadis

0
In order to drastically reduce the heavy computational burden associated with time-domain simulations, this paper introduces a Physics-Informed Neural Network (PINN) to directly learn the solutions of power system dynamics. In contrast to the limitations of classical model order reduction approaches, commonly used to accelerate time-domain simulations, PINNs can universally approximate any continuous function with an arbitrary degree of accuracy. One of the novelties of this paper is that we avoid the need for any training data. We achieve this by incorporating the governing differential equations and an implicit Runge-Kutta (RK) integration scheme directly into the training process of the PINN; through this approach, PINNs can predict the trajectory of a dynamical power system at any discrete time step. The resulting Runge-Kutta-based physics-informed neural networks (RK-PINNs) can yield up to 100 times faster evaluations of the dynamics compared to standard time-domain simulations. We demonstrate the methodology on a single-machine infinite bus system governed by the swing equation. We show that RK-PINNs can accurately and quickly predict the solution trajectories.

Matrix Completion for Improved Observability in Low-Voltage Distribution Grids

Marija Markovic; Anthony Florita; Bri-Mathias Hodge

0
This paper considers the problem of recovering a partially observed matrix from relatively few measurements (i.e., the so-called matrix completion problem) with the aim of increasing the presently limited observability of low-voltage distribution grids. To this end, the partially observed matrix is formed using voltage magnitude measurements while accounting for their spatial information. Voltages readings are assumed to be collected from distribution utility sensors and/or geographically-distributed cable television network sensors located in immediate proximity to distribution grid nodes. A matrix completion approach built on the parameter-less singular value shrinkage technique is used to estimate voltage magnitudes at otherwise non-observable low-voltage nodes using a small number of single- or multiple-snapshot data. The effectiveness of the proposed approach is demonstrated using a U.S.-style distribution test system from the synthetic SMART-DS data set under very low- to moderate-observability conditions.

Play Recording

Made with in Toronto · Privacy Policy · SmartGridComm 2020 · © 2021 Duetone Corp.