Data-driven demand response optimisation considering uncertain residential demand flexibility

Abstract

Towards the national and city-wide net-zero targets, integrating distributed energy resources (DER), including renewable energy generation and electric vehicles, is critically important. However, the inherent variability and complexity of energy patterns pose significant challenges to their seamless integration. Demand-side management (DSM) emerges as a key solution to address these challenges, necessitating a comprehensive analysis of consumption and DER profiles for optimisation. This thesis examines representative features and metrics of DER to identify DSM potential. It utilised the data-driven unsupervised clustering methodologies. Residential customers are segmented into representative clusters based on time series data, facilitating the implementation of DSM applications. To effectively coordinate integrated DSM and account for the uncertainty associated with residential demand flexibility, a data-driven two-stage distributionally robust optimisation (DRO) model is constructed based on the residential area integrated demand response to promote efficient utilisation of grid resources and renewable energy. The ambiguity set of the probability distribution is formulated using Kullback-Leibler divergence. The proposed method is validated through simulations using real consumption and DER data from Hong Kong. Results confirm that the proposed DRO model appropriately balances the trade-off between economical operation and robustness while showcasing its adaptability. Additionally, our approach demonstrates economic viability with a lowered rate of renewable power curtailment, computational efficiency, and practical feasibility.

Date of Award	2023
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology
Supervisor	Li QIU (Supervisor) & Jian Liang (Supervisor)