Technical Sessions

Session CO3

Operation and Control of Microgrids

10:30 AM — 12:00 PM +08
Oct 27 Thu, 10:30 PM — 12:00 AM EDT

A Hybrid Submodular Optimization Approach to Controlled Islanding with Heterogeneous Loads

Radha Poovendran and Dinuka Sahabandu (University of Washington, USA); Andrew Clark (Washington University in St. Louis, USA); Luyao Niu (University of Washington, USA)

Cascade failures, in which the failure of generators or transmission lines causes neighboring generators or lines to trip offline, threaten power system stability. Controlled islanding mitigates cascade failures by deliberately removing a subset of transmission lines in order to partition the system into disjoint, internally stable islands. In this paper, we investigate algorithms for controlled islanding to ensure stability while minimizing power flow disruption and load-generator imbalance. We consider a scenario where there are heterogeneous loads with varying costs of load shedding and formulate a hybrid optimization problem of jointly selecting a set of transmission lines to remove (discrete variables) and how much load to shed at each bus (continuous variables). In order to solve this optimization problem with provable optimality bounds, we propose a new notion of hybrid submodularity. We develop a polynomial-time islanding algorithm that achieves a provable 1/2-optimality bound. We use IEEE 118-bus and ACTIVsg 500-bus case studies to demonstrate that our approach provides better islanding solutions compared to a Mixed-Integer Linear Program (MILP)-based approach.

Microgrid Fault Detection Utilizing State Observer and Multi-Agent System

Saad Alzahrani and Joydeep Mitra (Michigan State University, USA)

Microgrid protection continues to be an emerging research problem for several reasons, such as integrating various levels of distributed generation and connecting power electronic converters. This paper develops a new protection approach using Multi-Agent System with a state observer and fault current limiters. This approach has two functions: achieving the fault detection for multiple zones of microgrid and restoring the power to the affected load in case of persistent fault. Mainly, this integration framework comprises distributed agents, which will communicate, interact, and exchange data for detecting the fault through the residual current value of the state observer within a particular protection zone. On the other hand, the fault current limiter will prevent the interruption of distributed generators during the faults. The proposed protection framework in this paper has been tested and applied to a microgrid configuration and is demonstrated to be an effective means to detect the faults as well as restore the power for multiple protection zones of the system.

Distributed Data Recovery Against False Data Injection Attacks in DC Microgrids

Zexuan Jin, Mengxiang Liu and Ruilong Deng (Zhejiang University, China); Peng Cheng (Zhejiang University & Singapore University of Technology and Design, China)

With the development of information and communications technology (ICT) in DC microgrids (DCmGs), the threat of false data injection attacks (FDIAs) is becoming more and more serious. However, the existing literature mainly focuses on the detection and identification of FDIAs in DCmGs, while the data recovery after the perception of FDIAs has never been thoroughly investigated yet. In this paper, we propose a distributed data recovery scheme to eliminate the adverse impact caused by FDIAs in DCmGs. Firstly, by observing the point of common coupling (PCC) voltage under FDIAs, the injected constant bias can be roughly estimated. In order to obtain the precise constant bias, the mean filter (MF) is adopted to handle the measurement noises and small oscillations. Then, the estimated precise constant bias is compensated for the communicated signal to eliminate the attack impact. Furthermore, our proposed data recovery scheme, which only needs local information, is fully distributed. Finally, the accuracy and effectiveness of the distributed data recovery scheme are evaluated through systematical hardware-in-the-loop (HIL) experiments.

Insurance Contract for High Renewable Energy Integration

Dongwei Zhao (MIT, USA); Hao Wang (Monash University, Australia); Jianwei Huang (The Chinese University of Hong Kong, Shenzhen, China); Xiaojun Lin (Purdue University, USA)

This talk does not have an abstract.

Session Chair and Room

Sebastian Köhler (University of Oxford, United Kingdom)— Room TT2-3

Session SS

Interplay Between Communication and Computation in Wireless-empowered Smart Grids

10:30 AM — 12:20 PM +08
Oct 27 Thu, 10:30 PM — 12:20 AM EDT

Edge Computing supported Fault Indication in Smart Grid

Petra Raussi, Jorma Kilpi and Heli Kokkoniemi-Tarkkanen (VTT Technical Research Centre of Finland, Finland); Anna Kulmala (ABB Distribution Solutions, Finland); Petri Hovila (ABB, Finland)

The distribution of smart grid applications to different physical devices not interconnected with physical sensors has opened the possibility for virtualization allowing flexible localization. Harnessing wireless 5G technology enables edge computing and locating smart grid applications at the edge. In this paper, we studied edge computing supporting medium voltage grid fault location, the challenges and benefits of bringing smart grid applications to the edge, and demonstrated fault location operation on an edge device. The challenges and benefits highlight the various aspect which must be considered for a profitable business case. While the demonstration showed the total data rate to be the key parameter in urban areas and latency due to large distances and general availability of edge resources are the most significant issues in rural areas.

Near Real-Time Distributed State Estimation via AI/ML-Empowered 5G Networks

Ognjen Kundacina (University of Novi Sad, Serbia); Miodrag Forcan (University of East Sarajevo, Bosnia and Herzegovina); Mirsad Cosovic and Darijo Raca (University of Sarajevo, Bosnia and Herzegovina); Merim Dzaferagic (Trinity College Dublin, Ireland); Dragi拧a Mi拧kovi膰 (University of Novi Sad, Serbia); Mirjana Maksimovic (University of East Sarajevo, Bosnia and Herzegovina); Dejan Vukobratovi膰 (University of Novi Sad, Serbia)

Fifth-Generation (5G) networks have a potential to accelerate power system transition to a flexible, softwarized, data-driven, and intelligent grid. With their evolving support for Machine Learning (ML)/Artificial Intelligence (AI) functions, 5G networks are expected to enable novel data-centric Smart Grid (SG) services. In this paper, we explore how data-driven SG services could be integrated with ML/AI-enabled 5G networks in a symbiotic relationship. We focus on the State Estimation (SE) function as a key element of the energy management system and focus on two main questions. Firstly, in a tutorial fashion, we present an overview on how distributed SE can be integrated with the elements of the 5G core network and radio access network architecture. Secondly, we present and compare two powerful distributed SE methods based on: i) graphical models and belief propagation, and ii) graph neural networks. We discuss their performance and capability to support a near real-time distributed SE via 5G network, taking into account communication delays.

Integration of LSTM based Model to guide short-term energy forecasting for green ICT networks in smart grids

Hamid Malik (University of Oulu, Finland); Ari T. Pouttu (Centre for Wireless Communications University of Oulu, Finland)

Existing ICT networks are characterized by high level of energy consumption. In order to power up 5G base station sites, rising energy cost and high carbon emissions are major concerns that need to be dealt with. To achieve carbon neutrality, ICT sector needs to transform base station sites in a self-sustainable manner using renewable energy sources, local batteries and energy conservation techniques, even in adverse weather conditions and unexpected power outages. In this paper, short term-forecasting models are studied for accurate energy consumption and production forecast. The proposed architecture provides adaptive energy conservation technique using time series data analysis and Long Short-Term Memory for 5GNR base station site which is independent of traditional power sources and is completely powered by green energy. The accuracy analysis of this study was performed by the Mean Square Error (MSE) and Root Mean Square Error (RMSE). The results show high accuracy levels of LSTM model in guiding short-term energy forecasting for green ICT networks.

A Robust and Explainable Data-Driven Anomaly Detection Approach For Power Electronics

Alexander K Beattie (Lappeenranta-Lahti University of Technology LUT, Finland); Pavol Mulinka (Centre Tecnologic de Telecomunicacions de Catalunya, Spain); Subham Sahoo (Aalborg University, Denmark); Ioannis T. Christou (The American College of Greece, Greece & NetCompany-Intrasoft, Luxembourg); Charalampos Kalalas (Centre Tecnol貌gic de Telecomunicacions de Catalunya (CTTC), Spain); Daniel Gutierrez Rojas (LUT University, Finland); Pedro Henrique Juliano Henrique Juliano Nardelli (Lappeenranta University of Technology & University of Oulu, Finland)

Timely and accurate detection of anomalies in power electronics is becoming increasingly critical for maintaining complex production systems. Robust and explainable strategies help decrease system downtime and preempt/mitigate infrastructure cyberattacks.

This work first explains the types of uncertainty present in datasets and machine learning algorithm outputs. Three techniques for combating these uncertainties are then introduced and analyzed. We further present two anomaly detection and classification approaches, namely the Matrix Profile algorithm and anomaly transformer, which are applied in the context of a power electronic converter dataset.

Specifically, the Matrix Profile algorithm is shown to be well suited as a generalizable approach for detecting real-time anomalies in streaming time-series data. The STUMPY python library implementation of the iterative Matrix Profile is used for the creation of the detector. A series of custom filters is created and added to the detector to tune its sensitivity, recall, and detection accuracy.

Our numerical results show that, with simple parameter tuning, the detector provides high accuracy and performance in a variety of fault scenarios.

Smart Home/Office Energy Management based on Individual Data Analysis through IoT Networks

Guang-Li Huang, Jinho Choi, Adnan Anwar, Seng W Loke and Arkady Zaslavsky (Deakin University, Australia)

Sustainable use of energy requires to achieve optimal energy utilization in smart grid systems. It is possible by empowering the Internet of Things (IoT) based wireless connectivity through real-time energy monitoring and analyses of power consumption patterns. Modeling optimal energy utilization considering multi-user behaviors is particularly challenging in such context. To address the challenge of one-to-one mapping of energy disaggregation in device-sharing environments by multiple co-existing users, a new method based on data-driven machine learning (e.g., individual energy usage pattern analysis) is proposed in this paper that aims to accurately match the energy consumption of electrical appliances with specific users. In particular, the machine learning model with the best performance is selected for real-time energy/power disaggregation on the local server (i.e., small-scale home/office) to ensure comparable or better performance with state-of-the-art disaggregation algorithms. In addition, energy usage patterns and individual power consumption data are analyzed comprehensively to match overall energy consumption and label datasets by events. Distributed learning is also discussed to exploit other local servers' datasets for better disaggregation through IoT networks. The effectiveness of the proposed method is verified by using simulated datasets in a motivation scenario.

Session Chair and Room

Pavol Mulinka (Centre Tecnologic de Telecomunicacions de Catalunya, Spain); Petra Raussi (VTT Technical Research Centre of Finland, Finland) — Room LT2

Session CO4

Scheduling and Optimization in Smart Grids

2:00 PM — 3:30 PM +08
Oct 28 Fri, 2:00 AM — 3:30 AM EDT

Scheduling Electric Vehicle Fleets as a Virtual Battery under Uncertainty using Quantile Forecasts

Nico Brinkel, Jing Hu and Lennard Visser (Copernicus Institute of Sustainable Development, Utrecht University, The Netherlands); Wilfried Van Sark (Utrecht University & Copernicus Institute, The Netherlands); Tarek AlSkaif (Wageningen University, The Netherlands)

Electric vehicles have significant potential to reduce their charging costs by participating in electricity markets through electric vehicle smart charging. However, one of the main barriers to electric vehicle participation in an electricity market is the high uncertainty in their availability at the market gate closure time. Not accounting for this uncertainty when making market bids could result in high imbalance costs. This study proposes a method to determine the optimal bidding strategy for a fleet of electric vehicles under uncertainty using a scenario-based stochastic optimization algorithm. This model considers both the uncertainty in electric vehicle availability and uncertainty in imbalance prices in the electricity market, as well as the risk-aversiveness of aggregators to high charging costs using the conditional value-at-risk. It proposes to model the electric vehicle fleet as a virtual battery, and to use a set of quantile forecasts of the virtual battery parameters to account for the uncertainty in electric vehicle availability. The effectiveness of the proposed model is evaluated by testing it on an actual case study fleet. The results indicate that it is crucial to consider both the expected charging costs and the conditional value-at-risk when determining market bids for an electric vehicle fleet under uncertainty

Achieving Self-Configurable Runtime State Verification in Critical Cyber-Physical Systems

Abel O Gomez Rivera (Sandia National Laboratories & University of Texas at El Paso, USA); Deepak K Tosh (University of Texas, El Paso, USA)

Cyber-Physical Systems (CPS) commonly monitor and manage critical cyber-enabled services such as distributed power generation architectures. Traditional CPS consists of heterogeneous devices that monitor physical systems processes generally stochastic and complex to model through state-of-art methods such as time-series analysis. Due to the stochastic nature, assurance of continuous runtime state integrity is challenging. Furthermore, adversaries exploit the lack of robust security mechanisms to deploy false sequential attacks that target the physical state of system processes. Therefore, this work designs runtime-system-state integrity assurance techniques necessary to enhance the security of critical CPS such as Small Modular Reactors (SMR). In this work, we propose a Reinforcement Learning(RL)-based Runtime-system-state Integrity (RRI) framework that aims to enable self-configurable runtime-system-states in SMR. The RRI framework generally addresses false sequential attacks by enabling fine-grained detail continuous runtime state integrity assurance through state-of-art RL and Machine Learning (ML) methods. A proof-of-concept of the RRI framework has been evaluated in an emulated SMR. This work demonstrates the RRI framework's performance regarding the RL methods' convergence time. Overall, the state-of-art RL methods converge in 1,000 episodes. We implemented the emulated experimental SMR through the open-source OpenAI, scikit-learn, and Stable Baselines3 platforms. The open-source platforms enable the development and comparison of RL and ML methods by enabling standard communication between baselines algorithms and ecosystems. The experimental results discussed in this work provide essential information that help understand complex and stochastic environments. Furthermore, we demonstrated that the RRI framework could provide high-fidelity CPS models that can provide helpful insights into understanding the system-state behavior of complex system processes.

Optimization-Based Exploration of the Feasible Power Flow Space for Rapid Data Collection

Ignasi Ventura Nadal and Samuel Chevalier (Technical University of Denmark, Denmark)

This paper provides a systematic investigation into the various nonlinear objective functions which can be used to explore the feasible space associated with the optimal power flow problem. A total of 40 nonlinear objective functions are tested, and their results are compared to the data generated by a novel exhaustive rejection sampling routine. The Hausdorff distance, which is a min-max set dissimilarity metric, is then used to assess how well each nonlinear objective function performed (i.e., how well the tested objective functions were able to explore the non-convex power flow space). Exhaustive test results were collected from five PGLib test-cases and systematically analyzed.

A Game Approach for EV Brands' Investment Planning of Battery Swapping Stations

Heyu Ren (The Chinese University of Hong Kong Shenzhen, China); Chenxi Sun (The Chinese University of Hong Kong (Shenzhen), China); Xiaoying Tang (The Chinese University of Hong Kong, Shenzhen, China)

In the current market of energy supply for electric vehicles (EVs), charging stations (CSs) have a definitely high market share, while the battery swapping stations (BSSs) are rapidly developed in recent years due to their potential to save users' waiting time by replacing a depleted battery with a fully-charged one. In order to assess the benefits of battery swapping, an economic model is needed to analyze the market, and different EV brands need to determine the optimal investment plans of BSSs in order to maximize their individual revenues. This paper proposes a Stackelberg game to characterize the interactions between different EV brands, CSs, BSSs, and EV consumers.

By analyzing the Nash Equilibrium of the Stackelberg game, a pricing scheme for BSSs is proposed and EV brands' specific profit model from BSSs is formulated to analyze the optimal investment plan of BSSs. Experimental results show that our model and strategy improve EV brands' profits, especially for brands with moderate battery capacity, moderate car price and large sales volume.

Session Chair and Room

Bo Tu (Singapore University of Technology and Design, Singapore) — Room TT2-3

Session SP3

Attack Detection and Localization in Smart Grids

2:00 PM — 3:50 PM +08
Oct 28 Fri, 2:00 AM — 3:50 AM EDT

Detecting Hidden Attackers in Photovoltaic Systems Using Machine Learning

Suman Sourav (Singapore University of Technology and Design, Singapore); Partha P. Biswas (Advanced Digital Sciences Center, Singapore); Binbin Chen (Singapore University of Technology and Design, Singapore); Daisuke Mashima (Advanced Digital Sciences Center & National University of Singapore, Singapore)

In modern smart grids, the proliferation of communication enabled distributed energy resource (DER) systems has increased the surface of possible cyber-physical attacks. Attacks originating from the distributed edge devices of DER system, such as photovoltaic (PV) system, is often difficult to detect. An attacker may change the control configurations or various setpoints of the PV inverters to destabilize the power grid, damage devices, or for the purpose of economic gain. A more powerful attacker may even manipulate the PV system metering data transmitted for remote monitoring, so that (s)he can remain hidden. In this paper, we consider a case where PV systems operating in different control modes can be simultaneously attacked and the attacker has the ability to manipulate individual PV bus measurements to avoid detection.
We show that even in such a scenario, with just the aggregated measurements (that the attacker cannot manipulate), machine learning (ML) techniques are able to detect the attack in a fast and accurate manner. We use a standard radial distribution network, together with real smart home electricity consumption data and solar power data in our experimental setup. We test the performance of several ML algorithms to detect attacks on the PV system. Our detailed evaluations show that the proposed intrusion detection system (IDS) is highly effective and efficient in detecting attacks on PV inverter control modes.

Early Detection of GOOSE Denial of Service (DoS) Attacks in IEC 61850 Substations

Ghada Elbez (Karlsruhe Institute of Technology (KIT), Germany); Klara Nahrstedt (University of Illinois Urbana-Champaign, USA); Veit Hagenmeyer (Karlsruhe Institute of Technology, Germany)

The availability of the communication in IEC 61850 substations can be hindered by Denial of Service (DoS) that result from advanced Generic Object Oriented Substation Event (GOOSE) poisoning attacks. To the best of our knowledge, most of the available approaches in the literature are unable to detect similar attacks and none of them can offer the detection in an early manner. Thus, we develop the Early Detection of Attacks for GOOSE Network Traffic (EDA4GNeT) method that takes into account the specific features of IEC 61850 substations and offers a good trade-off between the detection performance and the detection time. To validate the efficiency of the novel anomaly detection method against those specific GOOSE poisoning attacks, a comparison with the closest works to ours is conducted on a similar use case representing a T1-1 substation. Results demonstrate the possibility of an early detection approximately 37 time samples ahead and an average detection rate of EDA4GNeT of more than 99% with a low false positive rate of less than 1%.

On The Efficacy of Physics-Informed Context-Based Anomaly Detection for Power Systems

Nouman Nafees, Neetesh Saxena and Peter Burnap (Cardiff University, United Kingdom (Great Britain))

The Automatic Generation Control (AGC), a fundamental frequency control system, is vulnerable to cyber-physical attacks. Coordinated false data injection attack, aiming to generate fake transient measurements, typically precedes unwarranted actions, inducing frequency excursion, leading to electromechanical swings between generators, blackouts, and costly equipment damage. Unlike other works that focus on point anomaly detection, this work focuses on contextual detection of stealthy cyber-attacks against AGC by utilizing prior information,
which is essential for power system operation and situational awareness. More specifically, we depart from the traditional deep learning anomaly detection that is thoroughly driven by black-box detection; instead, we envision an approach based on physics-informed hybrid deep learning detection 鈥楥LDPhy,' which utilizes the combination of prior knowledge of physics and system metrics. Our method, to the extent of our knowledge, is the first context-based anomaly detection for stealthy cyber-physical attacks against the AGC system. We evaluate our approach on an industrial high-class PowerWorld simulated dataset - based on the IEEE 37-bus model. Our experiments observe a 36.4% improvement in accuracy for coordinated attack detection with contextual information, and our approach clearly demonstrates the superiority in comparison with other baselines.

On Holistic Multi-Step Cyberattack Detection via a Graph-based Correlation Approach

Oemer Sen (RWTH Aachen University & Fraunhofer FIT, Germany); Chijioke Eze and Andreas Ulbig (RWTH Aachen University, Germany); Antonello Monti (RWTH Aachen University & Institute for Automation of Complex Power Systems, Germany)

While digitization of distribution grids through information and communications technology brings numerous benefits, it also increases the grid's vulnerability to serious cyber attacks. Unlike conventional systems, attacks on many industrial control systems such as power grids often occur in multiple stages, with the attacker taking several steps at once to achieve its goal. Detection mechanisms with situational awareness are needed to detect orchestrated attack steps as part of a coherent attack campaign. To provide a foundation for detection and prevention of such attacks, this paper addresses the detection of multi-stage cyber attacks with the aid of a graph-based cyber intelligence database and alert correlation approach. Specifically, we propose an approach to detect multi-stage attacks by leveraging heterogeneous data to form a knowledge base and employ a model-based correlation approach on the generated alerts to identify multi-stage cyber attack sequences taking place in the network. We investigate the detection quality of the proposed approach by using a case study of a multi-stage cyber attack campaign in a future-orientated power grid pilot.

Localization of Coordinated Cyber-Physical Attacks in Power Grids Using Moving Target Defense and Deep Learning

Yexiang Chen and Subhash Lakshminarayana (University of Warwick, United Kingdom (Great Britain)); Fei Teng (Imperial College London, United Kingdom (Great Britain))

As one of the most sophisticated attacks against power grids, coordinated cyber-physical attacks (CCPAs) damage the power grid's physical infrastructure and use a simultaneous cyber attack to mask its effect. This work proposes a novel approach to detect such attacks and identify the location of the line outages (due to the physical attack). The proposed approach consists of three parts. Firstly, moving target defense (MTD) is applied to expose the physical attack by actively perturbing transmission line reactance via distributed flexible AC transmission system (D-FACTS) devices. MTD invalidates the attackers' knowledge required to mask their physical attack. Secondly, convolution neural networks (CNNs) are applied to localize line outage position from the compromised measurements. Finally, model agnostic meta-learning (MAML) is used to accelerate the training speed of CNN following the topology reconfigurations (due to MTD) and reduce the data/retraining time requirements. Simulations are carried out using IEEE test systems. The experimental results demonstrate that the proposed approach can effectively localize line outages in stealthy CCPAs.

Session Chair and Room

Biplab Sikdar (National University of Singapore, Singapore) — Room LT2

Session DAC3

Data Analysis and Computation in Smart Metering

4:30 PM — 6:00 PM +08
Oct 28 Fri, 4:30 AM — 6:00 AM EDT

Flexibility Management for Residential Users Under Participation Uncertainty

Thanasis G. Papaioannou, George Stamoulis and Christos Krasopoulos (Athens University of Economics and Business, Greece)

Demand flexibility management, often by means of Demand Response (DR), can significantly enhance the stability of the electric grid and reduce the investment cost for infrastructure upgrades in case of dynamic energy mix with renewable sources. However, uncertainty in the consumer response to the DR signals may disrupt this goal. In this paper, we deal with the optimal management of the flexibility offered by residential users under uncertainty. We develop a probabilistic user model to account for the uncertainty in the actual provision of the flexibility by a user in conjunction with incentives' offered thereto, which we subsequently introduce in the Demand Response (DR) targeting process. We consider a suitable optimization framework to enable flexibility maximization and budget minimization as separate single-objective expressions with the appropriate constraints. We define representative problems and solve them numerically for a wide range of user parameters, in order to illustrate the applicability and accuracy of our method, and to extract valuable insights. Finally, we develop techniques to resolve practical issues and to enable real-world implementation of the proposed scheme in pilot sites; namely, a mathematical expression to estimate the confidence intervals of the attained flexibility and a learning algorithm for extracting the individual user parameters according to their participation patterns.

Automatic Differentiation of Variable and Fixed Speed Heat Pumps With Smart Meter Data

Tobias Brudermueller (ETH Zurich, Switzerland); Florian Wirth (University of St. Gallen, Switzerland); Andreas Weigert (University of Bamberg, Germany); Thorsten Staake (ETH Zurich, Switzerland)

With the increasing prevalence of heat pumps in private households, the need for optimization is growing. At the same time, the growing number of active smart electricity meters generates data that can be used for remote monitoring. In this paper, we focus on the automatic differentiation between fixed speed and variable speed heat pumps using smart meter data. This distinction is relevant because it is necessary for evaluating the state or cyclic behavior of a heat pump. In addition, identifying fixed speed heat pumps is important because they are known to be the less efficient systems and therefore may be preferred targets in energy efficiency or replacement campaigns. Our methods are applied to electricity data from 171 Swiss households with a resolution of 15 minutes. In this setting, a K-Nearest Neighbor model achieves a mean AUC of 0.976 compared to 0.5 of a biased random guess model.

Attention-guided Temporal Convolutional Network for Non-intrusive Load Monitoring

Huamin Ren (Kristiania University College, Norway); Xiaomeng Su (Norwegian University of Science and Technology, Norway); Robert Jenssen (University of Tromso, Norway); Jingyue Li (Norwegian University of Science and Technology, Norway); Stian Normann Anfinsen (UiT The Arctic University of Norway, Norway)

With the prevalence of smart meter infrastructure, data analysis on consumer side becomes more and more important in smart grid systems. One of the fundamental tasks is to disaggregate users' total consumption into appliance-wise values. It has been well noted that encoding of temporal dependency is a key issue for successful modelling of the relations between the total consumption and its decomposed consumption on an appliance historically, and therefore has been implemented in many state-of-the-art models. However, how to encode the varied long-term and short-term dependency coming from different appliances is yet an open and under-addressed question. In this paper, we propose an attention-guided temporal convolutional network (ATCN), which generates different temporal residual blocks and provides an attention mechanism to indicate the importance of those blocks with respect to the appliance. Ultimately, we aim to address these two questions: i) How to employ both long-term and short-term temporal dependency to better disaggregate future loads while maintaining an affordable memory cost? ii) How to employ attention during the training of an appliance to obtain a better representation of the consumption pattern? We have demonstrated the effectiveness of our approach through comprehensive experiments and show that our proposed ATCN model achieves state-of-the-art performance, particularly on multi-status appliances that are normally hard to cope with regarding disaggregation accuracy and generalization capability.

Behind-the-Meter Disaggregation of Residential Electric Vehicle Charging Load

Kang Pu and Yue Zhao (Stony Brook University, USA)

With the rapidly evolving penetration of electric vehicles (EVs) in power distribution systems, a major issue utilities face is the lack of visibility into the charging behaviors of the behind-the-meter (BTM) EVs. Knowing the BTM EV charging behaviors can greatly enhance utilities' system planning and operation efficacy. In this paper, the problem of disaggregating BTM EV load traces from smart meter data traces is studied. Based on the characteristics of typical EV charging traces, three interdependent sub-problems are formulated: a) Detecting the presence of BTM EVs, b) Estimating the EV charging rate, and c) Detecting the EV charging periods. A unified iterative algorithmic framework is developed to solve all three sub-problems. Importantly, the proposed algorithms do not assume or utilize the knowledge of ground truth EV load traces but estimate BTM EV load traces in an "unsupervised" fashion. Numerical evaluation is conducted based on real-world 15-minute interval smart meter data from Austin, TX, and demonstrates great performance achieved by the proposed algorithms.

Session Chair and Room

Suman Sourav (Singapore University of Technology and Design, Singapore) — Room TT2-3

Session SP4

Cyber Security, Risk Management and Digital Twins

4:30 PM — 6:00 PM +08
Oct 28 Fri, 4:30 AM — 6:00 AM EDT

Assessment of Cyber-Physical Intrusion Detection and Classification for Industrial Control Systems

Nils M眉ller, Charalampos Ziras and Kai Heussen (Technical University of Denmark, Denmark)

The increasing interaction of industrial control systems (ICSs) with public networks and digital devices introduces new cyber threats to power systems and other critical infrastructure. Recent cyber-physical attacks such as Stuxnet and Irongate revealed unexpected ICS vulnerabilities and a need for improved security measures. Intrusion detection systems constitute a key security technology, which typically monitors cyber network data for detecting malicious activities. However, a central characteristic of modern ICSs is the increasing interdependency of physical and cyber network processes. Thus, the integration of network and physical process data is seen as a promising approach to improve predictability in intrusion detection for ICSs by accounting for physical constraints and underlying process patterns. This work systematically assesses real-time cyber-physical intrusion detection and multi-class classification, based on a comparison to its purely network data-based counterpart and evaluation of misclassifications and detection delay. Multiple supervised machine learning models are applied on a recent cyber-physical dataset, describing various cyber attacks and physical faults on a generic ICS. A key finding is that integration of physical process data improves detection and classification of all attack types. In addition, it enables simultaneous processing of attacks and faults, paving the way for holistic cross-domain root cause analysis.

Investigating the Cybersecurity of Smart Grids Based on Cyber-Physical Twin Approach

Oemer Sen (RWTH Aachen University & Fraunhofer FIT, Germany); Florian Schmidtke (RWTH Aachen, Germany); Federico Carere and Francesca Santori (ASM Terni, Italy); Andreas Ulbig (RWTH Aachen University, Germany); Antonello Monti (RWTH Aachen University & Institute for Automation of Complex Power Systems, Germany)

While the increasing penetration of information and communication technology into distribution grid brings numerous benefits, it also opens up a new threat landscape, particularly through cyberattacks. To provide a basis for countermeasures against such threats, this paper addresses the investigation of the impact and manifestations of cyberattacks on smart grids by replicating the power grid in a secure, isolated, and controlled laboratory environment as a cyber-physical twin. Currently, detecting intrusions by unauthorized third parties into the central monitoring and control system of grid operators, especially attacks within the grid perimeter, is a major challenge. The development and validation of methods to detect and prevent coordinated and timed attacks on electric power systems depends not only on the availability and quality of data from such attack scenarios, but also on suitable realistic investigation environments. Common approaches for the creation of such investigation environments are based on purely virtual reconstruction of smart grids with simplified models of the infrastructure. However, to create a comprehensive investigation environment, a realistic representation of the study object is required to thoroughly investigate critical cyberattacks on grid operations and evaluate their impact on the power grid using real data. In this paper, we demonstrate our cyber-physical twin approach using a microgrid in the context of a cyberattack case study.

HA-Grid: Security Aware Hazard Analysis for Smart Grids

Luca Maria Castiglione, Zhongyuan Hau, Pudong Ge, Luis Muñoz-González, Kenneth T. Co and Fei Teng (Imperial College London, United Kingdom (Great Britain)); Emil Lupu (Imperial College, United Kingdom (Great Britain))

Attacks targeting smart grid infrastructures can result in the disruptions of power supply as well as damages to costly equipment, with significant impact on safety as well as on end-consumers. It is therefore of essence to identify attack paths in the infrastructure that lead to safety violations and to determine critical components that must be protected. In this paper, we introduce a methodology (HA-Grid) that incorporates both safety and security modelling of smart grid infrastructure to analyse the impact of cyber threats on the safety of smart grid infrastructures. HA-Grid is applied on a smart grid test-bed to identify attack paths that lead to safety hazards, and to determine the common nodes in these attack paths as critical components that must be protected.

A Reconfigurable and Secure Firmware Updating Framework for Advanced Metering Infrastructure

Prosanta Gope (University of Sheffield, United Kingdom (Great Britain)); Biplab Sikdar (National University of Singapore, Singapore)

Smart meters play an important role in modern power grids by providing fine-grained power consumption data and enabling services such as dynamic pricing and demand-side management. The smart metering devices are firmware-driven, where it is important that the devices be able to securely update their firmware on a regular basis to fix bugs, and improve as well as add services. In this paper, we propose a \emph{new} privacy-aware secure firmware-updating framework called PRSUF (Privacy-aware Reconfigurable Secure-Firmware Updating Framework) to securely update the firmware in smart metering devices. The proposed the framework allows an intrinsic hardware secret to being updated and stored in a secure and efficient way. One of its key differentiating features is that, unlike existing mechanisms, the proposed scheme does not require storing any keys in the meter's non-volatile memory (NVM), thereby making it is secure against a number of physical and side-channel attacks. As compared to state-of-the-art solutions, the proposed security framework has notable features such as reconfigurability, protection against cloning and downgrading, detection of theft of services and tampering with the firmware and the hardware, etc.

Session Chair and Room

Ghada Elbez (Karlsruhe Institute of Technology (KIT), Germany) — Room LT2

Made with in Toronto · Privacy Policy · SmartGridComm 2020 · SmartGridComm 2021 · © 2022 Duetone Corp.