Bayesian Optimization Algorithm-Based Statistical and Machine Learning Approaches for Forecasting Short-Term Electricity Demand

Abstract

This article focuses on developing both statistical and machine learning approaches for forecasting hourly electricity demand in Ontario. The novelties of this study include (i) identifying essential factors that have a significant effect on electricity consumption, (ii) the execution of a Bayesian optimization algorithm (BOA) to optimize the model hyperparameters, (iii) hybridizing the BOA with the seasonal autoregressive integrated moving average with exogenous inputs (SARIMAX) and nonlinear autoregressive networks with exogenous input (NARX) for modeling separately short-term electricity demand for the first time, (iv) comparing the model’s performance using several performance indicators and computing efficiency, and (v) validation of the model performance using unseen data. Six features (viz., snow depth, cloud cover, precipitation, temperature, irradiance toa, and irradiance surface) were found to be significant. The Mean Absolute Percentage Error (MAPE) of five consecutive weekdays for all seasons in the hybrid BOA-NARX is obtained at about 3%, while a remarkable variation is observed in the hybrid BOA-SARIMAX. BOA-NARX provides an overall steady Relative Error (RE) in all seasons (1~6.56%), while BOA-SARIMAX provides unstable results (Fall: 0.73~2.98%; Summer: 8.41~14.44%). The coefficient of determination (R2) values for both models are >0.96. Overall results indicate that both models perform well; however, the hybrid BOA-NARX reveals a stable ability to handle the day-ahead electricity load forecasts.

Short-Term Load Forecasting of Electricity Demand for the Residential Sector Based on Modelling Techniques: A Systematic Review

“…The load of the residential sector is a typical time series, many statistical methods have been and continue to be used for STLF solutions, such as the auto-regressive integrated moving average (ARIMA) [26], or the Bayesian [12] [30] or Gaussian processes (GP) [28]. However, due to the non-linearity of the behaviour of residential energy consumption, the effects of these models are often limited, and the availability of the original time sequences is mandatory, the statistical models present in the STLF solutions are low-performance.…”

“…However, ANN suffers from a limitation of settling in the local minima and overfitting problems [46]. To avoid overfitting, the authors increased the amount of data, dropout others and trained with momentum [12]. Training a neural network consists of modifying its parameters through gradient descent optimization, which minimizes a given loss function that quantifies the network's accuracy in performing the desired task [43] reducing the training error [47].…”

“…Increasing electricity consumption at an alarming rate has become a challenge for distribution companies [5]. Energy institutions and organizations in each country need to ensure and adjust the demand for supply at competitive prices, avoid waste of resources and, safely, provide it to their economy and their citizens [12] [13].…”

“…An accurate load forecasting system can reduce the high cost of excessive contracts and subcontracts in balance markets due to load forecasting errors [14]. Electricity demand forecasting has been decisive in electricity production capacity, transmission planning and pricing [12]. In addition, it keeps energy markets efficient and provides a better understanding of the dynamics of the monitored system [14].…”

Electricity Demand Forecasting with Use of Artificial Intelligence: The Case of Gokceada Island

“…Conversely, lower load forecasts may cause the system to operate in a risky region, resulting in insufficient supply [2]. At the same time, load and demand forecasts form the basis of many decisions made in energy markets, allowing electricity markets to be planned and operated in an efficient, transparent, reliable manner and to meet the needs of the sector [3].…”