Domestic load forecasting using neural network and its use for missing data analysis

Ai Songpu,Kolhe, M.L.,Lei Jiao,Qi Zhang
DOI: https://doi.org/10.1109/ATEE.2015.7133866
2015-01-01
Abstract:Domestic demand prediction is very important for home energy management system and also for peak reduction in power system network. In this work, active and reactive power consumption prediction model is developed and analysed for a typical Southern Norwegian house for hourly power (active and reactive) consumptions and time information as inputs. In the proposed model, a neural network is adopted as a main technique and historical domestic load data of around 2 years are used as input. The available data has some measurement errors and missing segments. Before using the data for training purpose, missing and inaccurate data are considered and then it is used for testing the model. It is observed that the possible reasons of prediction errors may be due to local external parameters (e.g. ambient temperature, moisture, solar radiation etc.). It may be required to include analysis of these external parameters on domestic demand prediction model with peak prediction and timing and this will be carried out in our further work.
What problem does this paper attempt to address?