Next-day MV/LV substation load forecaster using time series method

•This paper deals with the electricity load forecast of MV/LV substations (several dozen kW), which is rarely the subject of other load forecasting papers.•With a linear structure, the proposed time series model is easy for application, quick in computation, and provides good precision.•Advanced sig...

Full description

Saved in:
Bibliographic Details
Published in:Electric power systems research 2015-02, Vol.119, p.345-354
Main Authors: Ding, Ni, Bésanger, Yvon, Wurtz, Frédéric
Format: Article
Language:English
Subjects:
Distribution system
Electric power
Engineering Sciences
Load forecast
Smart grid
Spectral analysis
Time series
Weather uncertainty
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•This paper deals with the electricity load forecast of MV/LV substations (several dozen kW), which is rarely the subject of other load forecasting papers.•With a linear structure, the proposed time series model is easy for application, quick in computation, and provides good precision.•Advanced signal processing methods are incorporated to build and adjust the parameters of the model.•Results are validated with the real measurements collected in the French distribution networks, reducing the error from 50% to less than 20%. With smart-grid development, distribution networks have attracted more and more attention. As an important infrastructure of smart-grid projects, smart meters collect individual consumption information based on which forecasting models are to be built. A time series method is applied to answer the need for next day Medium Voltage/Low Voltage (MV/LV) substation load forecasts. An additive time series method divides the analysis into three parts: a trend, a cyclic, and a random error component. The first two parts are deterministic parts and are modeled separately with dummy variable regression models. The final forecasting results are obtained by adding the forecasting results of these two models. The random error part is looked through in detail to ensure the wellbeing of the parametric model. Advanced signal processing methods are incorporated in the method such as ANalysis Of VAriance (ANOVA) nullity test and smoothed periodogram. The results of the method are encouraging compared with a naive model. Weather uncertainty as an important aspect to the load forecast is also discussed at the end of the paper. The method has been successfully applied to French distribution networks.
ISSN:0378-7796
1873-2046
DOI:10.1016/j.epsr.2014.10.003