Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence
Statistics, Probability, Computer Sciences, Electrical and electronic engineering, Energy and fuels
The aim of this research was to characterise domestic electricity patterns of use on a diurnal, intra-daily and seasonal basis as a function of customer characteristics. This was done in order to produce a library of representative electricity demand load profiles that are characteristic of how households consume electricity. In so doing, a household’s electricity demand can be completely characterised based solely on their individual customer characteristics.
A number of different approaches were investigated as to their ability to characterise domestic electricity use. A statistical regression approach was evaluated which had the advantage of identifying key dwelling, occupant and appliance characteristics that influence electricity use within the home. An autoregressive Markov chain method was applied which proved to be effective at characterising the magnitude component to electricity use within the home but failed to adequately characterise the temporal properties sufficiently. Further time series techniques were investigated: Fourier transforms, Gaussian processes, Neural networks, Fuzzy logic, and Wavelets, with the former two being evaluated fully. Each method provided disparate results but proved to be complimentary to each other in terms of their ability to characterise different patterns of electricity use. Both approaches were able to sufficiently characterise the temporal characteristics satisfactorily, however, were unable to adequately associate customer characteristics to the load profile shape.
Finally clustering based approaches such as: k-means, k-medoid and Self Organising Maps (SOM) were investigated. SOM showed the greatest potential and when combined with statistical and regression techniques proved to be an effective way to completely characterise electricity use within the home and their associated customer characteristics. A library of domestic electricity demand load profiles representing common patterns of electricity use on a diurnal, intra-daily and seasonal basis within the home in Ireland and their associated household characteristics are then finally presented.
McLoughlin, F. (2013) Characterising Domestic Electricity Demand for Customer Load Profile Segmentation. Doctoral Thesis. Technological University Dublin. doi:10.21427/D7PK7X