Use this URL to cite or link to this record in EThOS:
Title: Load pattern categorisation and dynamic pricing based demand response in smart grid
Author: Luo, X.
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
The Smart Grid is widely regarded as the next generation of the power grid in power system reform. It is the application of digital processing and communications to the power grid, making data flow and information management central to the grid. Demand response (DR) is an essential characteristic of the smart grid and it plays an important role in energy efficiency improvement and wastage reduction by providing encouraging energy-aware consumption. However, efficient DR management involves a variety of challenges including categorisation of load patterns, accurate real-time price (RTP) forecasting, effective DR program designing, etc. This thesis extends around these related challenges in the smart grid and presents significant outcomes. Load pattern categorisation (LPC) plays an important role in DR. However, how to determine a precise cluster number and choose an appropriate clustering algorithm are critical in LPC and remain challenging. In this thesis, as the first contribution, a novel parametric bootstrap (PB) algorithm is proposed, incorporated with a compatible clustering technique to address the cluster number determination problem as well as clustering the load data simultaneously. The PB algorithm is more robust against dimensionality of data and hence applicable to load demand data which is usually of high dimensionality. It is also general and independent of data type, resulting in a more appropriate cluster number determination result than existing methods with little fluctuation. The evaluation results indicate the feasibility and superiority of the proposed approach over others previously published in the literature. The RTP tariff has become a trend in the smart grid and it is usually utilised as an input control signal to enable efficient load shifting in DR. As the second contribution of the thesis, a hybrid RTP forecasting model considering deterministic and stochastic features of input data is proposed to forecast short-term electricity prices. The evaluation results clearly demonstrate that the proposed approach is effective in RTP forecasting with a higher accuracy compared with existing models from the literature. An effective DR strategy is the core of DR. As the third contribution, a number of DR strategies assisted by electric vehicles (EVs) are proposed. Innovative EV assisted DR strategies with the EV as an auxiliary power supply (EV-APS) model and a neighbour energy sharing (NES) model are proposed, to jointly optimise the load distribution for both a single household and multi-household network via vehicle to home (V2H) and vehicle to neighbour (V2N) connections, respectively. The proposed DR strategies take account of the comprehensive impacts of EVs' charging behaviors, user preferences, distributed energy, and load scheduling priorities. The effectiveness of the DR strategies are verified by numerical results in terms of load balancing and cost reduction, and the proposed DR strategies show better performance compared with previously published DR approaches.
Supervisor: Lim, Eng Gee ; Zhu, Xu Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral