Machine Learning and Big Data Anlytics in Smart Grid

Background & Motivation

Penetration of advanced sensor systems such as advanced metering infrastructure (AMI), phasor measurement units (PMUs) and high-frequency overhead and underground current and voltage sensors have been increasing significantly in smart grid over the past few years. To unleash full value of the complex data sets, innovative machine learning algorithms need to be developed to transform the way we operate and plan for the smart grid.

Presentations

2024 Seminar Series: Physics-informed Machine Learning for Power Systems

2021 Cornell/WSU Seminar: Machine Learning for Smart Grid: From Pure Data-Driven to Physics-Informed Methods

2019 IEEE SmartGridComm Tutorial: Machine Learning and Big Data Analytics in Power Distribution Systems

IEEE PES Big Data Analytics SubCommittee Tutorial: Machine Learning and Big Data Analytics in Power Distribution Systems

Intellectual Merit

Synergistic combination of machine learning algorithms and physical power system models.

Interpretable machine learning algorithms.

Applications

Learning to Model, Monitor, Control, and Plan Power Distribution Systems

Data-driven Modeling of Power Distribution Systems

Distribution network topology identification (Phase Connectivity Identification, Ref1, Ref2, Ref3, Ref4)

Data-driven Distribution network parameter estimation

Spatio-temporal load forecast (Ref1) and renewable generation forecast

Data-driven Monitoring of Power Distribution Systems

Energy theft detection (Ref1)

Three-phase state estimation in power distribution systems (Ref1)

Predictive equipment maintenance (Ref1)

Estimation of behind-the-meter Solar PV Generation (Ref1)

Real-time visualization

Reinforcement Learning-based Control of Power Distribution Systems

Reinforcement Learning-based Volt-VAR control (Ref1, Ref2)

Reinforcement Learning-based Distribution Network Reconfiguration (Ref1)

Data-driven Planing of Power Distribution Systems

Valuation and optimization of DERs (battery, Ref1) in power distribution network

Diversification factor and load factor estimation (Ref1)

Solar PV adoption forecast (Ref1) and EV adoption forecast

Spatio-temporal load forecasting (Ref1)

Applications of Machine Learning in Transmission Systems

Learning to Design, Evaluate and Trade in Electricity Market

Algorithmic Trading with Virtual Bids in Electricity Markets (Ref1)

PMU Data Analytics

Discover and Label Power System Events with PMU data (Ref1)

Information Theoretic Machine Learning

Background

In order to advance the field of machine learning, we need to develop a unified theory to explain and quantify the performance bounds of deep neural networks.

Research Summary

We study the subject of information losses arising from the finite datasets used in the training of deep nerual classifiers. We proved a relationship between such losses as a 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. We then bound 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. We ultimately obtain bounds on information losses that are less sensitive to input compression and in general much smaller than existing bounds.

Research Highlight

We developed new bounds on informationm losses from finite data. This began in the form of a relationship between these losses, the expected total variation of the neural model, and the information held in the hidden representation of the feature space. Then, by bounding the total variation term without invoking any more dependence on model complexity, we obtained bounds that are much tighter and less sensitive to I(X;Z) than previous theory.

Electricity Market Design and Optimization

Motivation

In the past 20 years, wholesale power markets operating in transmission systems have been effective at coordinating the operations of thousands of centralized power plants. This coordination needs to be extended to the operations of millions of DERs. To do this efficiently, a Distribution system operator (DSO) managed electricity market seems to be a viable solution. Although the concept of a DSO-managed electricity market has been introduced, a key algorithm for operating the market is still in its infancy. This algorithm is three-phase optimal power flow (OPF), and it needs significant development.

Research Summary

Design Integrated wholesale and retail market. The integrated market architecture is shown in the figure below.

Develop DSO market, three-phase DCOPF and ACOPF.

Research Highlight

The proposed three-phase ACOPF algorithm is not only computationally efficient but also guarantees global optimality on all IEEE distribution test circuits

Energy Efficient Smart Cities

Motivation

The widespread adoption of information and communication technologies facilitates the integration of sensors, networked communications, and computing hardware and software into physical infrastructure systems such as transportation systems and electric power grid. These enhancements are transforming traditional passive infrastructure - from vehicles to traffic lights to solar panels - into proactive components capable of self-monitoring, communication, and control. The large volume of complex and heterogeneous data generated from these interdependent infrastructure systems can be leveraged to significantly improve energy efficiency, reduce travel time, and improve air quality in smart cities.

Research Projects

Ride-sharing with Electric Vehicle (Improve Mobility and Reduce Energy Consumption)

Smart Buildings (Improve Occupants' Comfort and Reduce Electricity Cost)

Energy Efficient Data Center

Coordinate the Operations of DERs and Provide Services to Power Grid

Water Energy Climate Nexus

Motivation

Water and energy are intrinsically interconnected. Water is required for nearly all forms of energy production and electricity generation. On the other hand, energy is needed for the treatment, desalination, recycling, transportation, and distribution of water. Climate change and increased demand for water and energy are creating scarcity, variability, and uncertainty in water and energy systems. The strong interdependence between the systems means that disturbance in one of the systems will likely lead to vulnerabilities within the other system. To mitigate these vulnerabilities, it is imperative to closely study the interplay among the water, climate, and energy systems.

Research Summary

Hydropower generation is a crucial link in the climate-water-energy nexus. It has been discovered that natural and anthropogenic aerosols have a great influence on meteorological variables such as temperature, snowpack, and precipitation, which, in turn, impact the inflows into hydropower reservoirs. This paper takes the next logical step to explore the impact of aerosols on hydropower generation and revenue. A comprehensive framework is developed to quantify the impact of aerosols on hydropower generation and revenue by integrating the Weather Research and Forecasting Model with Chemistry, a statistical hydrologic forecasting model, and the hydropower operation optimization toolbox. A case study is performed in the Big Creek Hydroelectric Project in California. The simulation results show that aerosols reduce inflows into the reservoirs of Big Creek hydroelectric system by 1%-10%. This leads to a 6% reduction of annual hydropower generation, causing a $2.8 million loss in annual revenue.

Distributed Energy Resources Optimization and Valuation

Motivation

Driven by environmental regulations and rapidly falling renewable prices, the share of renewable generation in global electrical energy mix is expected to increase significantly over time. The intermittency of renewable resources has created new challenges in the transmission system operations.

Energy storage system is well poised to mitigate uncertainties of renewable generation outputs. However, there are several challenges to the widespread deployment of energy storage. As identified in the U.S. Department of Energy report, the most crucial hurdle to storage adoption is how to ensure energy storage are cost competitive with other energy resources. To overcome this hurdle my research group developed a comprehensive optimization and valuation model (ESVOT) which allows energy storage to provide multiple electricity market products simultaneously.

Energy Storage Optimization and Valuation Tool (ESVOT)

ESVOT allows the user to conduct a comprehensive stochastic valuation of energy storage systems. In addition, ESVOT identifies the optimal energy storage integration location, size and technology for each customer.