Gene expression programming as a basis for new generation of electricity demand prediction models

Highlights

• Gene expression programming is presented for the estimation of the electricity demand.

• Annual data collected through years 1986–2009 in Thailand are used to develop the model.

• GEP significantly outperform the existing electricity demand prediction methods.

Abstract

This study proposes a new gene expression programming (GEP) approach for the prediction of electricity demand. The annual population, gross domestic product, stock index, and total revenue from exporting industrial products were used to predict the electricity demand of the same year in Thailand. Several statistical criteria were used to verify the validity of the model. Further, the contributions of the influencing variables to the prediction of the electricity demand were analyzed. Correlation coefficient, root mean squared error and mean absolute percent error were used to evaluate the performance of the model. In addition to its high accuracy, the derived model outperforms regression and other soft computing-based models.

Introduction

Providing precise estimation of the electricity demand has been always an important concern for energy management. Several methods have been developed to cope with this issue such as autoregressive integrated moving average (ARIMA), multiple linear regression (MLR) and machine learning techniques (Azadeh et al., 2008b, Kandananond, 2011, Mostafavi, Mostafavi et al., 2013). ARIMA is mostly employed for the modeling of stochastic disturbances in time series analysis (Box & Jenkins, 1970). This technique has been applied to different energy problem such as prediction of the customer short term load and domestic electric energy consumption (Abdel-Aal and Al-Garni, 1997, Cho et al., 1995, Ediger and Akar, 2007). Moreover, several researchers used double seasonal ARIMA for electricity demand forecasting (Mohamed et al., 2010, Suhartono and Endharta, 2009, Taylor, 2003, Taylor, 2010). Besides, conventional approaches such as MLR are widely used for the energy demand estimation (Kandananond, 2011). However, ARIMA, MLR and other statistical analyses are based on defining the linear or nonlinear structure of the model in advance, which is not always true (Alavi and Gandomi, 2011b, Mostafavi, Mostafavi et al., 2013).

Soft computing techniques are considered as alternative to traditional methods for tackling real world problems. They automatically learn from experience and extract various discriminators (Mitchell, 1997). Artificial neural networks (ANNs) are one of the widely used branches of soft computing. ANNs and other soft computing techniques have been successfully applied to different real world problems including industrial engineering and energy management (e.g., Abdel-Aal et al., 1997, Miranda et al., 1998, Pino et al., 2000, Tien Pao, 2007, Pino et al., 2008, Abdel-Aal, 2008, Alavi and Gandomi, 2011a, Chiang and Roy, 2012, Currie, 1992, Daim et al., 2010, Moghrabi and Eid, 1998, Sharma and Srinivasan, 2013, Srinivasan, 2008, Tsujimura et al., 1997). ANNs have been used to predict the electricity demand in different countries such as Taiwan (Hsu & Chen, 2003), Ireland (Ringwood, Bofelli, & Murray, 2001), Spain (Catalao et al., 2007, González-Romera et al., 2007), Saudi Arabia (Abdel-Aal, 2008), and Iran (Azadeh, Ghaderi et al., 2007, Azadeh et al., 2008a, Azadeh et al., 2008b, Azadeh, Ghaderi et al., 2007, Azadeh, Saberi et al., 2008c, Kheirkhah et al., 2013). Support vector machine (SVM) is another soft computing technique that has been successfully applied to predict the electric consumption (Fan and Chen, 2006, Hong, 2009). Despite their good performance, ANNs and SVM are considered as black-box models. That is, they are not capable of generating practical prediction equations. The structure of ANNs should be defined in advance, which limits their practicability (Alavi and Gandomi, 2011b, Gandomi and Alavi, 2012, Mostafavi, Mostafavi et al., 2013, Mostafavi, Saeedi, et al., 2013, Yang et al., 2012).

In order to cope with the limitations of the existing methods, a robust soft computing approach, namely genetic programming (GP) is introduced (Koza, 1992). In fact, GP uses the principle of Darwinian natural selection to generate computer programs for solving a problem. GP has several advantages over the conventional and ANN techniques. A notable feature of GP is that it can produce practical prediction equations without a need to pre-define the form of the existing relationship (Alavi, Ameri, et al., 2011, Alavi and Gandomi, 2012, Alavi, Gandomi et al., 2011, Can and Heavey, 2011, Can and Heavey, 2012, Gandomi and Alavi, 2011, Gandomi and Alavi, 2013a, Mostafavi, Saeedi, et al., 2013, Tay and Ho, 2008). GP and its variants have been shown to be powerful tools for the electricity demand prediction (Bhattacharya et al., 2001, Lee et al., 1997). Gene expression programming (GEP) (Ferreira, 2001) is a recent extension to GP. GEP creates computer programs of different sizes and shapes encoded in linear chromosomes of fixed length. The numerical experiments proved the superiority of GEP to similar techniques (Gandomi and Alavi, 2013b, Oltean and Grosşan, 2003). In contrast with classical GP and ANNs, application of GEP in the field of energy conversion and management is new and original.

This study presents GEP as a new approach to develop a new generation of electricity demand prediction models. The data for total electricity demand in Thailand from 1986 to 2009 were used for the model development. The results provided by the developed GEP model were further compared with those obtained by other existing methods. The paper is organized as follows: Section 2 presents brief descriptions of the GEP technique. Section 3 outlines the model development using GEP and reviews the results. The detailed performance analysis of the proposed model is discussed in Section 4. The results of the sensitivity analysis are given in Section 5. Finally, concluding remarks are outlined in Section 6.