Nanpeng Yu

Yuanqi Gao, Nanpeng Yu

To facilitate the development of reinforcement learning (RL) based power distribution system Volt-VAR control (VVC), this paper introduces a suite of open-source datasets for RL-based VVC algorithm research that is sample efficient, safe, and robust. The dataset consists of two components: 1. a Gym-like VVC testing environment for the IEEE-13, 123, and 8500-bus test feeders and 2. a historical operational dataset for each of the feeders. Potential users of the dataset and testing environment could first train an sample-efficient off-line (batch) RL algorithm on the historical dataset and then evaluate the performance of the trained RL agent on the testing environments. This dataset serves as a useful testbed to conduct RL-based VVC research mimicking the real-world operational challenges faced by electric utilities. Meanwhile, it allows researchers to conduct fair performance comparisons between different algorithms.

Brandon Foggo, Koji Yamashita, Nanpeng Yu

This paper introduces pmuGE (phasor measurement unit Generator of Events), one of the first data-driven generative model for power system event data. We have trained this model on thousands of actual events and created a dataset denoted pmuBAGE (the Benchmarking Assortment of Generated PMU Events). The dataset consists of almost 1000 instances of labeled event data to encourage benchmark evaluations on phasor measurement unit (PMU) data analytics. PMU data are challenging to obtain, especially those covering event periods. Nevertheless, power system problems have recently seen phenomenal advancements via data-driven machine learning solutions. A highly accessible standard benchmarking dataset would enable a drastic acceleration of the development of successful machine learning techniques in this field. We propose a novel learning method based on the Event Participation Decomposition of Power System Events, which makes it possible to learn a generative model of PMU data during system anomalies. The model can create highly realistic event data without compromising the differential privacy of the PMUs used to train it. The dataset is available online for any researcher or practitioner to use at the pmuBAGE Github Repository: https://github.com/NanpengYu/pmuBAGE.

Brandon Foggo, Koji Yamashita, Nanpeng Yu

We present pmuGE (phasor measurement unit Generator of Events), one of the first data-driven generative model for power system event data. We have trained this model on thousands of actual events and created a dataset denoted pmuBAGE (the Benchmarking Assortment of Generated PMU Events). The dataset consists of almost 1000 instances of labeled event data to encourage benchmark evaluations on phasor measurement unit (PMU) data analytics. The dataset is available online for use by any researcher or practitioner in the field. PMU data are challenging to obtain, especially those covering event periods. Nevertheless, power system problems have recently seen phenomenal advancements via data-driven machine learning solutions - solutions created by researchers who were fortunate enough to obtain such PMU data. A highly accessible standard benchmarking dataset would enable a drastic acceleration of the development of successful machine learning techniques in this field. We propose a novel learning method based on the Event Participation Decomposition of Power System Events, which makes it possible to learn a generative model of PMU data during system anomalies. The model can create highly realistic event data without compromising the differential privacy of the PMUs used to train it. The dataset is available online for any researcher to use at the pmuBAGE Github Repository - https://github.com/NanpengYu/pmuBAGE. Part I - This is part I of a two part paper. In part I, we describe a high level overview of pmuBAGE, its creation, and the experiments used to test it. Part II will discuss the exact models used in its generation in far more detail.

Mohammad Ostadijafari, Anamika Dubey, Yang Liu et al.

Owing to the call for energy efficiency, the need to optimize the energy consumption of commercial buildings-- responsible for over 40% of US energy consumption--has recently gained significant attention. Moreover, the ability to participate in the retail electricity markets through proactive demand-side participation has recently led to development of economic model predictive control (EMPC) for building's Heating, Ventilation, and Air Conditioning (HVAC) system. The objective of this paper is to develop a price-sensitive operational model for building's HVAC systems while considering inflexible loads and other distributed energy resources (DERs) such as photovoltaic (PV) generation and battery storage for the buildings. A Nonlinear Economic Model Predictive Controller (NL-EMPC) is presented to minimize the net cost of energy usage by building's HVAC system while satisfying the comfort-level of building's occupants. The efficiency of the proposed NL-EMPC controller is evaluated using several simulation case studies.

Wenyu Wang, Nanpeng Yu

Accurate phase connectivity information is essential for advanced monitoring and control applications in power distribution systems. The existing data-driven approaches for phase identification lack precise physical interpretation and theoretical performance guarantee. Their performance generally deteriorates as the complexity of the network, the number of phase connections, and the level of load balance increase. In this paper, by linearizing the three-phase power flow manifold, we develop a physical model, which links the phase connections to the smart meter measurements. The phase identification problem is first formulated as a maximum likelihood estimation problem and then reformulated as a maximum marginal likelihood estimation problem. We prove that the correct phase connection achieves the highest log likelihood values for both problems. An efficient solution method is proposed by decomposing the original problem into subproblems with a binary least-squares formulation. The numerical tests on a comprehensive set of distribution circuits show that our proposed method yields very high accuracy on both radial and meshed distribution circuits with a combination of single-phase, two-phase, and three-phase loads. The proposed algorithm is robust with respect to inaccurate feeder models and incomplete measurements. It also outperforms the existing methods on complex circuits.

Jingtao Qin, Yuanqi Gao, Mikhail Bragin et al.

Unit commitment (UC) is a fundamental problem in the day-ahead electricity market, and it is critical to solve UC problems efficiently. Mathematical optimization techniques like dynamic programming, Lagrangian relaxation, and mixed-integer quadratic programming (MIQP) are commonly adopted for UC problems. However, the calculation time of these methods increases at an exponential rate with the amount of generators and energy resources, which is still the main bottleneck in industry. Recent advances in artificial intelligence have demonstrated the capability of reinforcement learning (RL) to solve UC problems. Unfortunately, the existing research on solving UC problems with RL suffers from the curse of dimensionality when the size of UC problems grows. To deal with these problems, we propose an optimization method-assisted ensemble deep reinforcement learning algorithm, where UC problems are formulated as a Markov Decision Process (MDP) and solved by multi-step deep Q-learning in an ensemble framework. The proposed algorithm establishes a candidate action set by solving tailored optimization problems to ensure a relatively high performance and the satisfaction of operational constraints. Numerical studies on IEEE 118 and 300-bus systems show that our algorithm outperforms the baseline RL algorithm and MIQP. Furthermore, the proposed algorithm shows strong generalization capacity under unforeseen operational conditions.

Yuanbin Cheng, Koji Yamashita, Jim Follum et al.

The global deployment of the phasor measurement units (PMUs) enables real-time monitoring of the power system, which has stimulated considerable research into machine learning-based models for event detection and classification. However, recent studies reveal that machine learning-based methods are vulnerable to adversarial attacks, which can fool the event classifiers by adding small perturbations to the raw PMU data. To mitigate the threats posed by adversarial attacks, research on defense strategies is urgently needed. This paper proposes an effective adversarial purification method based on the diffusion model to counter adversarial attacks on the machine learning-based power system event classifier. The proposed method includes two steps: injecting noise into the PMU data; and utilizing a pre-trained neural network to eliminate the added noise while simultaneously removing perturbations introduced by the adversarial attacks. The proposed adversarial purification method significantly increases the accuracy of the event classifier under adversarial attacks while satisfying the requirements of real-time operations. In addition, the theoretical analysis reveals that the proposed diffusion model-based adversarial purification method decreases the distance between the original and compromised PMU data, which reduces the impacts of adversarial attacks. The empirical results on a large-scale real-world PMU dataset validate the effectiveness and computational efficiency of the proposed adversarial purification method.

Ali Jahanshahi, Sara Rashidi Golrouye, Osten Anderson et al.

The rapid growth of AI/ML data centers has led to higher energy consumption and carbon emissions. The shift to renewable energy and growing data center energy demands can destabilize the power grid. Power grids rely on frequency regulation reserves, typically fossil-fueled power plants, to stabilize and balance the supply and demand of electricity. This paper sheds light on the hidden carbon emissions of frequency regulation service. Our work explores how modern GPU data centers can coordinate with power grids to reduce the need for fossil-fueled frequency regulation reserves. We first introduce a novel metric, Exogenous Carbon, to quantify grid-side carbon emission reductions resulting from data center participation in regulation service. We additionally introduce EcoCenter, a framework to maximize the amount of frequency regulation provision that GPU data centers can provide, and thus, reduce the amount of frequency regulation reserves necessary. We demonstrate that data center participation in frequency regulation can result in Exogenous carbon savings that can outweigh operational carbon emissions

Jingtao Qin, Nanpeng Yu

Unit commitment (UC) problems are typically formulated as mixed-integer programs (MIP) and solved by the branch-and-bound (B&B) scheme. The recent advances in graph neural networks (GNN) enable it to enhance the B&B algorithm in modern MIP solvers by learning to dive and branch. Existing GNN models that tackle MIP problems are mostly constructed from mathematical formulation, which is computationally expensive when dealing with large-scale UC problems. In this paper, we propose a physics-informed hierarchical graph convolutional network (PI-GCN) for neural diving that leverages the underlying features of various components of power systems to find high-quality variable assignments. Furthermore, we adopt the MIP model-based graph convolutional network (MB-GCN) for neural branching to select the optimal variables for branching at each node of the B&B tree. Finally, we integrate neural diving and neural branching into a modern MIP solver to establish a novel neural MIP solver designed for large-scale UC problems. Numeral studies show that PI-GCN has better performance and scalability than the baseline MB-GCN on neural diving. Moreover, the neural MIP solver yields the lowest operational cost and outperforms a modern MIP solver for all testing days after combining it with our proposed neural diving model and the baseline neural branching model.

Zhentong Shao, Jingtao Qin, Xianbang Chen et al.

The integration of distributed energy resources (DERs) into wholesale electricity markets, as mandated by FERC Order 2222, imposes new challenges on system operations. To remain consistent with existing market structures, regional transmission organizations (RTOs) have advanced the aggregation of transmission-node-level DERs (T-DERs), where a nodal virtual power plant (VPP) represents the mapping of all distribution-level DERs to their respective transmission nodes. This paper develops a real-time economic dispatch (RTED) framework that enables multi-transmission-node DER aggregation while addressing computational efficiency. To this end, we introduce a spatio-temporal graph convolutional network (ST-GCN) for adaptive prediction of distribution factors (DFs), thereby capturing the dynamic influence of individual T-DERs across the transmission system. Furthermore, an iterative constraint identification strategy is incorporated to alleviate transmission security constraints without compromising system reliability. Together, these innovations accelerate the market clearing process and support the effective participation of T-DER aggregators under current market paradigms. The proposed approach is validated on large-scale test systems, including modified 118-, 2383-, and 3012-bus networks under a rolling RTED setting with real demand data. Numerical results demonstrate significant improvements in reducing operational costs and maintaining transmission network feasibility, underscoring the scalability and practicality of the proposed framework.

Zhentong Shao, Jingtao Qin, Nanpeng Yu

The increasing integration of intermittent distributed energy resources (DERs) has introduced significant variability in distribution networks, posing challenges to voltage regulation and reactive power management. This paper presents a novel neural two-stage stochastic Volt-VAR optimization (2S-VVO) method for three-phase unbalanced distribution systems considering network reconfiguration under uncertainty. To address the computational intractability associated with solving large-scale scenario-based 2S-VVO problems, a learning-based acceleration strategy is introduced, wherein the second-stage recourse model is approximated by a neural network. This neural approximation is embedded into the optimization model as a mixed-integer linear program (MILP), enabling effective enforcement of operational constraints related to the first-stage decisions. Numerical simulations on a 123-bus unbalanced distribution system demonstrate that the proposed approach achieves over 50 times speedup compared to conventional solvers and decomposition methods, while maintaining a typical optimality gap below 0.30%. These results underscore the method's efficacy and scalability in addressing large-scale stochastic VVO problems under practical operating conditions.

Zhentong Shao, Jingtao Qin, Nanpeng Yu

Two-stage stochastic unit commitment (2S-SUC) problems have been widely adopted to manage the uncertainties introduced by high penetrations of intermittent renewable energy resources. While decomposition-based algorithms such as column-and-constraint generation has been proposed to solve these problems, they remain computationally prohibitive for large-scale, real-time applications. In this paper, we introduce a Neural Column-and-Constraint Generation (Neural CCG) method to significantly accelerate the solution of 2S-SUC problems. The proposed approach integrates a neural network that approximates the second-stage recourse problem by learning from high-level features of operational scenarios and the first-stage commitment decisions. This neural estimator is embedded within the CCG framework, replacing repeated subproblem solving with rapid neural evaluations. We validate the effectiveness of the proposed method on the IEEE 118-bus system. Compared to the original CCG and a state-of-the-art commercial solver, Neural CCG achieves up to 130.1$\times$ speedup while maintaining a mean optimality gap below 0.096\%, demonstrating its strong potential for scalable stochastic optimization in power system.

Zhentong Shao, Jingtao Qin, Nanpeng Yu

This paper proposes a neural stochastic optimization method for efficiently solving the two-stage stochastic unit commitment (2S-SUC) problem under high-dimensional uncertainty scenarios. The proposed method approximates the second-stage recourse problem using a deep neural network trained to map commitment decisions and uncertainty features to recourse costs. The trained network is subsequently embedded into the first-stage UC problem as a mixed-integer linear program (MILP), allowing for explicit enforcement of operational constraints while preserving the key uncertainty characteristics. A scenario-embedding network is employed to enable dimensionality reduction and feature aggregation across arbitrary scenario sets, serving as a data-driven scenario reduction mechanism. Numerical experiments on IEEE 5-bus, 30-bus, and 118-bus systems demonstrate that the proposed neural two-stage stochastic optimization method achieves solutions with an optimality gap of less than 1%, while enabling orders-of-magnitude speedup compared to conventional MILP solvers and decomposition-based methods. Moreover, the model's size remains constant regardless of the number of scenarios, offering significant scalability for large-scale stochastic unit commitment problems.

Shaorong Zhang, Yuanbin Cheng, Nanpeng Yu

This paper presents a novel physics-informed diffusion model for generating synthetic net load data, addressing the challenges of data scarcity and privacy concerns. The proposed framework embeds physical models within denoising networks, offering a versatile approach that can be readily generalized to unforeseen scenarios. A conditional denoising neural network is designed to jointly train the parameters of the transition kernel of the diffusion model and the parameters of the physics-informed function. Utilizing the real-world smart meter data from Pecan Street, we validate the proposed method and conduct a thorough numerical study comparing its performance with state-of-the-art generative models, including generative adversarial networks, variational autoencoders, normalizing flows, and a well calibrated baseline diffusion model. A comprehensive set of evaluation metrics is used to assess the accuracy and diversity of the generated synthetic net load data. The numerical study results demonstrate that the proposed physics-informed diffusion model outperforms state-of-the-art models across all quantitative metrics, yielding at least 20% improvement.

Zuzhao Ye, Yuanqi Gao, Nanpeng Yu

The rapid adoption of electric vehicles (EVs) calls for the widespread installation of EV charging stations. To maximize the profitability of charging stations, intelligent controllers that provide both charging and electric grid services are in great need. However, it is challenging to determine the optimal charging schedule due to the uncertain arrival time and charging demands of EVs. In this paper, we propose a novel centralized allocation and decentralized execution (CADE) reinforcement learning (RL) framework to maximize the charging station's profit. In the centralized allocation process, EVs are allocated to either the waiting or charging spots. In the decentralized execution process, each charger makes its own charging/discharging decision while learning the action-value functions from a shared replay memory. This CADE framework significantly improves the scalability and sample efficiency of the RL algorithm. Numerical results show that the proposed CADE framework is both computationally efficient and scalable, and significantly outperforms the baseline model predictive control (MPC). We also provide an in-depth analysis of the learned action-value function to explain the inner working of the reinforcement learning agent.

Yinglun Li, Nanpeng Yu, Wei Wang

This paper develops a machine learning-driven portfolio optimization framework for virtual bidding in electricity markets considering both risk constraint and price sensitivity. The algorithmic trading strategy is developed from the perspective of a proprietary trading firm to maximize profit. A recurrent neural network-based Locational Marginal Price (LMP) spread forecast model is developed by leveraging the inter-hour dependencies of the market clearing algorithm. The LMP spread sensitivity with respect to net virtual bids is modeled as a monotonic function with the proposed constrained gradient boosting tree. We leverage the proposed algorithmic virtual bid trading strategy to evaluate both the profitability of the virtual bid portfolio and the efficiency of U.S. wholesale electricity markets. The comprehensive empirical analysis on PJM, ISO-NE, and CAISO indicates that the proposed virtual bid portfolio optimization strategy considering the price sensitivity explicitly outperforms the one that neglects the price sensitivity. The Sharpe ratio of virtual bid portfolios for all three electricity markets are much higher than that of the S&P 500 index. It was also shown that the efficiency of CAISO's two-settlement system is lower than that of PJM and ISO-NE.

Wenyu Wang, Nanpeng Yu

Accurate estimates of network parameters are essential for modeling, monitoring, and control in power distribution systems. In this paper, we develop a physics-informed graphical learning algorithm to estimate network parameters of three-phase power distribution systems. Our proposed algorithm uses only readily available smart meter data to estimate the three-phase series resistance and reactance of the primary distribution line segments. We first develop a parametric physics-based model to replace the black-box deep neural networks in the conventional graphical neural network (GNN). Then we derive the gradient of the loss function with respect to the network parameters and use stochastic gradient descent (SGD) to estimate the physical parameters. Prior knowledge of network parameters is also considered to further improve the accuracy of estimation. Comprehensive numerical study results show that our proposed algorithm yields high accuracy and outperforms existing methods.

Jie Shi, Brandon Foggo, Nanpeng Yu

Online power system event identification and classification is crucial to enhancing the reliability of transmission systems. In this paper, we develop a deep neural network (DNN) based approach to identify and classify power system events by leveraging real-world measurements from hundreds of phasor measurement units (PMUs) and labels from thousands of events. Two innovative designs are embedded into the baseline model built on convolutional neural networks (CNNs) to improve the event classification accuracy. First, we propose a graph signal processing based PMU sorting algorithm to improve the learning efficiency of CNNs. Second, we deploy information loading based regularization to strike the right balance between memorization and generalization for the DNN. Numerical studies results based on real-world dataset from the Eastern Interconnection of the U.S power transmission grid show that the combination of PMU based sorting and the information loading based regularization techniques help the proposed DNN approach achieve highly accurate event identification and classification results.

Yuanqi Gao, Wei Wang, Nanpeng Yu

Volt-VAR control (VVC) is a critical application in active distribution network management system to reduce network losses and improve voltage profile. To remove dependency on inaccurate and incomplete network models and enhance resiliency against communication or controller failure, we propose consensus multi-agent deep reinforcement learning algorithm to solve the VVC problem. The VVC problem is formulated as a networked multi-agent Markov decision process, which is solved using the maximum entropy reinforcement learning framework and a novel communication-efficient consensus strategy. The proposed algorithm allows individual agents to learn a group control policy using local rewards. Numerical studies on IEEE distribution test feeders show that our proposed algorithm matches the performance of single-agent reinforcement learning benchmark. In addition, the proposed algorithm is shown to be communication efficient and resilient.

Brandon Foggo, Nanpeng Yu

We derive the closed-form expression of the maximum mutual information - the maximum value of $I(X;Z)$ obtainable via training - for a broad family of neural network architectures. The quantity is essential to several branches of machine learning theory and practice. Quantitatively, we show that the maximum mutual information for these families all stem from generalizations of a single catch-all formula. Qualitatively, we show that the maximum mutual information of an architecture is most strongly influenced by the width of the smallest layer of the network - the "information bottleneck" in a different sense of the phrase, and by any statistical invariances captured by the architecture.

Brandon Foggo, Nanpeng Yu

This paper considers the problem of Phase Identification in power distribution systems. In particular, it focuses on improving supervised learning accuracies by focusing on exploiting some of the problem's information theoretic properties. This focus, along with recent advances in Information Theoretic Machine Learning (ITML), helps us to create two new techniques. The first transforms a bound on information losses into a data selection technique. This is important because phase identification data labels are difficult to obtain in practice. The second interprets the properties of distribution systems in the terms of ITML. This allows us to obtain an improvement in the representation learned by any classifier applied to the problem. We tested these two techniques experimentally on real datasets and have found that they yield phenomenal performance in every case. In the most extreme case, they improve phase identification accuracy from $51.7\%$ to $97.3\%$. Furthermore, since many problems share the physical properties of phase identification exploited in this paper, the techniques can be applied to a wide range of similar problems.

Brandon Foggo, Nanpeng Yu

In this paper, we present and illustrate some new tools for rigorously analyzing training data selection methods. These tools focus on the information theoretic losses that occur when sampling data. We use this framework to prove that two methods, Facility Location Selection and Transductive Experimental Design, reduce these losses. These are meant to act as generalizable theoretical examples of applying the field of Information Theoretic Deep Learning Theory to the fields of data selection and active learning. Both analyses yield insight into their respective methods and increase their interpretability. In the case of Transductive Experimental Design, the provided analysis greatly increases the method's scope as well.

Brandon Foggo, Nanpeng Yu, Jie Shi et al.

This paper considers the subject of information losses arising from the finite datasets used in the training of neural classifiers. It proves a relationship between such losses as the product of the expected total variation of the estimated neural model with the information about the feature space contained in the hidden representation of that model. It then bounds this expected total variation as a function of the size of randomly sampled datasets in a fairly general setting, and without bringing in any additional dependence on model complexity. It ultimately obtains bounds on information losses that are less sensitive to input compression and in general much smaller than existing bounds. The paper then uses these bounds to explain some recent experimental findings of information compression in neural networks which cannot be explained by previous work. Finally, the paper shows that not only are these bounds much smaller than existing ones, but that they also correspond well with experiments.