Mid-Long Term Load Forecasting using Multi-Model Artificial Neural Networks

Abstract

Electrical load is a major input factor in a country's economic development. To support economic growth and meet future demands for electricity, Load forecasting has become a very important task for electrical power management and planning. Several techniques have been employed to accomplish this task. One of those that is mostly used is Artificial Neural Networks (ANNs) method, which have seen the largest number of studies in the field. the load time series data are auto-correlated and are influenced by other variables, such as temperature and population, etc. Furthermore, a large data sample is needed to perform mid-long term forecasts. For the latter purpose and in order to overcome both the lack of data and the high complexity of adding exogenous variables, a multi-model method, based on neural networks, is proposed for the mid-long term load consumption forecasting. The proposed model decomposes, the initial time series into three components using the X12-ARIMA algorithm, and performs a forecast for each component using three Nonlinear Auto Regressive (NAR) neural networks, with a Feed Forward neural network combining the outputs of the NARs. The experiments that were conducted showed that the proposed model provided good performance even when using a small sample of available data when compared to benchmark models.

An adaptive backpropagation algorithm for long-term electricity load forecasting

“…In Ref. [23], a Univariate Multi-Model (UMM) based on neural networks was proposed to forecast electrical load. The purpose was to increase the performance of mid to long-term forecasting.…”

Medium term load demand forecast of Kano zone using neural network algorithms

“…Several models related to this work were developed such as grey model [5], support vector regression [6], but the main issues with the support vector machine are the choice of the kernel function parameters, extensive memory requirement and difficulty of interpretation, multi-model artificial neural networks [7], fast-learning recurrent neural network [8]; stability is the major drawback of recurrent neural network. Deep learning neural networks [9], large amount of data requirement and determination of suitable topology are main demerits of deep learning method.…”