Forecasting electricity demand in the industrial sector based on disaggregate data

The objective of this work was to investigate the use of disaggregation in the development of models for forecasting the electrical energy demand in the industrial sector. A variety of modelling techniques were examined and models were developed based on past demand data and external variables.<br/><br/>A number of different types of disaggregation were investigated and it was found that grouping based on the standard industrial classification codes resulted in the optimum models. The combination of the predictions from these models resulted in more accurate forecasts than the aggregate model for the total industrial sector for short lead times. The response of electrical demand to external variables in the different industries was investigated and when included in the model no significant increase in accuracy was found. However when these inputs were included for the modelling of individual consumers, a significant improvement was obtained. This was associated with the quality of external data available for individual consumers.<br/><br/>The identification of outliers and the use of intervention analysis showed significant improvements in the models developed for all levels of disaggregation and was found to deal adequately with the opening and closure of large industries.<br/><br/>Forecasting on a half hour basis was also examined and at this level intervention analysis was again found to be most appropriate. Based on half hour forecasting techniques, a novel design of a maximum demand controller was proposed and the performance assessed under typical conditions.

Name dc:type.qualificationname

Doctor of Philosophy

Level dc:type.qualificationlevel

Doctoral Thesis

Grantor dc:publisher.institution

Queen's University Belfast

Year dc:date.issued

1991

Author dc:creator

• McCafferty, Peter

Advisor dc:contributor.advisor

• Beattie, W.C.

Language dc:language

eng

Identifier

oai:pure.qub.ac.uk/portal:studenttheses/6c5f595e-9983-4b75-a753-61f8c3fd4c80

OAI identifier oai:identifier

oai:pure.qub.ac.uk/portal:studenttheses/6c5f595e-9983-4b75-a753-61f8c3fd4c80