An adaptive backpropagation algorithm for long-term electricity load forecasting

Abstract

Artificial Neural Networks (ANNs) have been widely used to determine future demand for power in the short, medium, and long terms. However, research has identified that ANNs could cause inaccurate predictions of load when used for long-term forecasting. This inaccuracy is attributed to insufficient training data and increased accumulated errors, especially in long-term estimations. This study develops an improved ANN model with an Adaptive Backpropagation Algorithm (ABPA) for best practice in the forecasting long-term load demand of electricity. The ABPA includes proposing new forecasting formulations that adjust/adapt forecast values, so it takes into consideration the deviation between trained and future input datasets' different behaviours. The architecture of the Multi-Layer Perceptron (MLP) model, along with its traditional Backpropagation Algorithm (BPA), is used as a baseline for the proposed development. The forecasting formula is further improved by introducing adjustment factors to smooth out behavioural differences between the trained and new/future datasets. A computational study based on actual monthly electricity consumption inputs from 2011 to 2020, provided by the Iraqi Ministry of Electricity, is conducted to verify the proposed adaptive algorithm's performance. Different types of energy consumption and the electricity cut period (unsatisfied demand) factor are also considered in this study as vital factors. The developed ANN model, including its proposed ABPA, is then compared with traditional and popular prediction techniques such as regression and other advanced machine learning approaches, including Recurrent Neural Networks (RNNs), to justify its superiority amongst them. The results reveal that the most accurate long-term forecasts with the minimum Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE) values of (1.195.650) and (0.045), respectively, are successfully achieved by applying the proposed ABPA. It can be concluded that the proposed ABPA, including the adjustment factor, enables traditional ANN techniques to be efficiently used for long-term forecasting of electricity load demand.

Analysis of College Students’ Network Moral Behavior by the History of Ideological and Political Education under Deep Learning

“…Each layer structure is calculated according to the framework of RNN. According to the parameter sharing nature of RNN, the gradient of the weight is the sum of all layers [27]. e loss function L operates as shown in the following equation:…”

Deterioration of Electrical Load Forecasting Models in a Smart Grid Environment

“…and Al-Bazi A. [31] proposed an improved ANN along with an adaptive backpropagation algorithm (ABPA) to cover the limitations of ANN and provide enhanced forecasting accuracy. The datasets comprised nine years of data attained from the ministry of electricity in Iraq.…”

“…The ARIMA model could not detect the importance of newly introduced parameters, and the results remained more or less the same with minimal deviations. In Figure 13, (a After reviewing several proposed models for load forecasting utilizing ANN [31,33,42,44], it is concluded that ANN methods have better forecasting accuracy than statistical methods. However, in this study, we utilized ANN and other forecasting models to understand the performance and behavior of models towards feature change scenarios.…”

“…Some of the models that resulted in performance degradation are presented in Table 4. Figure 14, (a In this article, we reviewed studies that proposed LSTM models for electrical load predictions [12,21,[29][30][31][32] and observed their performance. We have utilized the configuration of multivariate LSTM to cater to the maximum number of variables.…”

“…After reviewing several proposed models for load forecasting utilizing ANN [31,33,42,44], it is concluded that ANN methods have better forecasting accuracy than statistical methods. However, in this study, we utilized ANN and other forecasting models to understand the performance and behavior of models towards feature change scenarios.…”