<link rel="stylesheet" href="styles.cf390c6bfbe7555f.css">

Data-enabled approaches for planning and operation of smart and sustainable power distribution system

Loading...
Thumbnail Image

Date

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

The increasing penetration of renewable energy and digital monitoring technologies is transforming the power distribution grid into a more complex, data-rich environment. While these advancements promise improved sustainability and efficiency, they also introduce new challenges in planning, operation, and real-time decision-making. This dissertation presents a unified data-driven framework that addresses two fundamental challenges in modern power distribution systems: optimal integration of photovoltaic (PV) and battery storage systems, and enhanced situational awareness through advanced data analytics.

In the first part of this dissertation, a statistical and optimization-based methodology is developed for determining the optimal sizing of PV and battery storage systems in rural power networks. The proposed approach captures the inherent uncertainties of solar irradiance and electricity demand through Monte Carlo simulation and time-series clustering. By evaluating generated scenarios, the framework identifies PV–battery configurations that satisfy a specified peak-demand reduction requirement while maximizing financial benefit. A probabilistic formulation based on Gamma distributions and Kolmogorov–Smirnov goodness-of-fit testing is introduced to estimate the 95th-percentile peak-demand threshold, enabling statistical evaluation of system sizing under uncertainty. The resulting methodology provides utilities and planners with practical guidelines for cost-effective PV–battery deployment.

The second part of the dissertation focuses on improving situational awareness in distribution networks through electricity demand forecasting and distribution system state estimation (DSSE) using advanced metering infrastructure (AMI) data. A deep learning forecasting model, Seasonal Periodic Decomposition Network (SPDNet), is developed to capture the seasonal, periodic, and short-term variations in residential electricity demand. The proposed model combines seasonal-trend decomposition with periodic decomposition to capture temporal dependencies at different scales and generate forecast-based pseudo-measurements for state estimation. To address the limited availability of near-real-time AMI measurements, Bayesian Weighted Least Squares (BWLS) and Variational Bayesian Matrix Completion (VBMC) are employed to estimate the voltage states of the distribution network. The resulting estimates are combined using calibrated Maximum Ratio Combining (MRC), which accounts for the bias and uncertainty associated with each estimator. The results demonstrate that the calibrated fusion approach improves voltage-estimation accuracy compared with the individual estimators and conventional MRC. Furthermore, a comparison between individual and transformer-aggregated load forecasting shows that reducing the number of forecasting profiles from 646 individual loads to 58 aggregated groups provides nearly identical BWLS state-estimation accuracy while reducing the number of forecasting tasks.

Collectively, the proposed frameworks contribute to the planning and operation of modern power distribution systems by combining statistical modeling, optimization, machine learning, and state estimation. The results demonstrate the effectiveness of probabilistic PV–battery sizing for peak-demand reduction and economic benefit, as well as the potential of forecasting-aided and calibrated state estimation to improve situational awareness under limited measurement availability.

Description

Keywords

Smart power distribution systems, Photovoltaic and battery energy storage systems, Electricity demand forecasting, Advanced metering infrastructure, Distribution system state estimation, Data-driven methods

Graduation Month

December

Degree

Doctor of Philosophy

Department

Department of Electrical and Computer Engineering

Major Professor

Anil Pahwa

Date

Type

Dissertation

Citation

Endorsement

Review

Supplemented By

Referenced By