Publications
Modeling Tools and Datasets
Projects
Meta-Learning-Based Adaptive Stability Certificates for Dynamical Systems
Transient stability assessment (TSA) is an indispensable part in power system operation as it provides an understanding of system’s transient behavior. This helps prevent a power system from experiencing catastrophic incidents, such as a blackout. Under parametric stochasticity, the TSA process becomes very challenging in the sense that it must be performed multiple times as the system doesn’t remain stationary. In case the TSA isn’t repeated each time the system alters, it results in a very conservative stability region which isn’t very informative to a power system operator. In this context, we propose a meta-learning-based adaptive Lyapunov function that adapts to any parametric shifts and performs a TSA on the new system. To achieve this, we integrate neural Lyapunov functions (NLFs) with model agnostic meta-learning (MAML) framework to propose meta-neural Lyapunov functions (meta-NLFs). These meta-NLFs train across a set of deterministic instances of a system under parametric stochasticity. During deployment, a fully trained meta-NLF swiftly adapts (in one gradient update) to a test-time system and correspondingly, generates a meaningful stability region. Through rigorous simulation results on five different dynamical systems (two power systems and three benchmark control systems), we showcase the superiority in performance of meta-NLF over other baseline TSA methods.
Blockchain and Energy
Blockchain technologies are considered one of the most disruptive innovations of the last decade, enabling secure decentralized trust-building. However, they have raised concerns about their sustainable operation in electric grids due to the rapid increase in energy consumption for cryptocurrency mining. In response, we investigated the tri-factor impact of mining loads in the U.S. on carbon footprint, grid reliability, and electricity market price. We created open-source high-resolution data and modeling tools for high-resolution modeling of influencing factors such as location and flexibility. Additionally, we designed electricity market structures that utilize the demand flexibility of mining loads in electric energy systems and offer ancillary services such as frequency regulation, alleviating concerns about the sustainable operation of cryptocurrency mining data center. Explore our findings here and here.
Resiliency Against Emerging Challenges (Covid-19, blackouts)
As climate change intensifies the frequency and severity of extreme weather events, the risk of power grid blackouts increases, emphasizing the need for grid resilience and stability. The Covid-19 pandemic was another challenge impacting grid consumption patters and posing substantial risks on electric grids. To address these concerns, it is crucial to model and analyze the impact of these challenges on the safe and economic operation of the power grid. By utilizing real-world cross-domain datasets and providing synthetic grid models, our research enables the quantitative assessment of these emerging challenges. Furthermore, we study the potential impact of various corrective measures to enhance grid resiliency, ensuring a robust response to these evolving threats. Explore our findings here, here and here
Power Electronics Intelligence at Network Edge (PINE)
A self-organizing power electronic converter with control intelligence at the edge of the distribution network is proposed. The proposed converter is called power electronics intelligence at the network edge (PINE), and it has the potential to add intelligence at the network edge to the electricity delivery system of the present and in the future. The proposed approach consists of a power electronic converter (rectifier/dc-link/inverter) termed PINE to supply residential loads. The rooftop solar and battery energy storage system is connected to the dc link. With the bidirectional characteristic of the PINE, the load voltage is regulated via feedback, whereas input distribution voltage can be allowed to vary in a range. This type of configuration allows for control of input power factor to be unity and reactive power to be injected at the grid edge to regulate the voltage and also enable energy budgeting, i.e., limit the amount of power to the residential load under disaster situations. Explore our findings here.
Data Sets and Simulation Tools for Machine Learning in Power System
The electric grid is a key enabling infrastructure for the ambitious transition towards carbon neutrality as we grapple with climate change. With deepening penetration of renewable resources, the reliable operation of the electric grid becomes increasingly challenging. We present PSML, a first-of-its-kind open-access multi-scale time-series dataset, and OpenGridGym, an AI-based distribution market simulator to aid in the development of data-driven machine learning (ML)-based approaches towards reliable operation of future electric grids. The dataset is synthesized from a joint transmission and distribution electric grid to capture the increasingly important interactions and uncertainties of the grid dynamics, containing power, voltage and current measurements over multiple spatio-temporal scales. Explore our findings here and here.
Reinforcement Learning for Grid Control
We use reinforcement learning to improve the control of distribution grid, as well as the protection of distribution systems with many distributed energy resources (DERs). Explore our findings here, here and here.
Chance Constrained and Scenario Optimization
We introduce an empirical approach to dispatch resources in real-time power system operation with growing levels of uncertainties emerging from intermittent and distributed energy resources in the supply and the demand side. It is shown that by taking empirical data of specific sizes, the dispatch results can lead to a quantifiable and rigorous bound on the risk of violating constraints at the implementation stage. In particular, we formulate the look-ahead real-time economic dispatch problem using the scenario approach. This approach takes empirical data as input and guarantees a tunable probability of violating the constraints according to the input data size. By exploiting the structure of the economic dispatch, we show that in the absence of transmission constraints, the number of samples that is required by the theory does not grow with the size of the problem. In the more general case with transmission constraints, it is shown that the posterior bound on the risk of dispatch can be quantified and can be much smaller than the risk bound before solving the dispatch. Numerical examples based on a standard test system suggest that the scenario approach can provide a practically attractive solution with theoretically rigorous properties for risk-limiting power system operations. More information can be found here, here and here.
Demand Response: Design and Analysis
In competitive electricity market systems such as Texas, Load Serving Entities (LSEs) purchase energy in a wholesale market and then sell it in a retail market. In wholesale market, LSEs see dynamic real-time prices, whereas in retail market, LSEs typically provide flat rate contracts to end-consumers. An intuitive idea to induce savings for LSEs is to shift loads from high-price hours to low-price hours of the wholesale market by incentive-based demand response. We design and implement such a system, called EnergyCoupon, to provide coupon incentives to users and collect their responses. We run the experiment over household participants through the summer of 2016. The experimental results suggest behavioral changes in energy consumption can be achieved, which could be beneficial to both end users and LSEs. We present the system set up, key algorithms, as well as experimental results analysis. We also worked on aggregating thermal inertial loads to the demand response program in wholesale electricity markets. More information can be found here, here and here.
Power System Dynamics and PMU Analysis
We propose data-driven algorithms for locating the source of forced oscillations and suggest physical interpretations for the method. By leveraging the sparsity of forced oscillations along with the low-rank nature of synchrophasor data, the problem of source localization under resonance conditions is cast as computing the sparse and low-rank components using Robust Principal Component Analysis (RPCA), which can be efficiently solved by the exact Augmented Lagrange Multiplier method. Based on this problem formulation, an efficient and practically implementable algorithm is proposed to pinpoint the forced oscillation source during real-time operation. Furthermore, theoretical insights are provided for the efficacy of the proposed approach, by use of physical model-based analysis, specifically by highlighting the low-rank nature of the resonance component matrix. Without the availability of system topology information, the proposed method can achieve high localization accuracy in synthetic cases based on benchmark systems and real-world forced oscillations in the power grid of Texas. More information can be found here and here.
Economics and Market Construct of the Future Power System
We study the market construct and pricing mechanism for the future power systems with lots of uncertainty and inter-temporal ramping constraints. More information can be found here, here and here.
Cyber Physical Security of the Energy Systems
We pioneered and demonstrated an active defense for defending renewable-rich microgrids against cyber attacks. Cyber vulnerabilities in such microgrid settings are identified. A defense mechanism based on the dynamic watermarking is proposed for detecting cyber anomalies in microgrids. The proposed mechanism is shown to be readily implementable and it has theoretically provable performance in term of detecting cyber attacks. The effectiveness of the proposed mechanism is tested and validated in a number of realistic system settings. Explore our findings here and here.