An optimized gene expression programming model for forecasting the national CO2 emissions in 2030 using the metaheuristic algorithms

Abstract

To cope with the approaching POST-2020 scenario, the national CO2 emission in the building sector, which accounts for 25.5% of the total CO2 emissions, should be managed effectively and efficiently. To do this, it is essential to forecast the national CO2 emissions in the building sector by region. As the South Korean government does not currently do this by region, regional characteristics are rarely taken into consideration when managing the national CO2 emissions in the building sector. Towards this end, this study developed an optimized gene expression programming model for forecasting the national CO2 emissions in 2030 using the metaheuristic algorithms. Compared to the forecasting performance of the gene expression programming model, the forecasting performance of the optimized gene expression programming – harmony search optimization model has improved by 7.11, 2.05, and 2.06% in terms of the mean absolute error, root mean square error, and mean absolute percentage error, respectively. Various national CO2 emissions scenarios in the building sector were established in order to better analyze the variation range of the national CO2 emissions in the building sector. Compared to the national CO2 emissions in 2016 (i.e., scenario 1: 41,337 ktCO2; scenario 2: 45,373 ktCO2; scenario 3: 46,024 ktCO2) in multi-family housing complexes, the national CO2 emissions in 2030 (i.e., scenario 1: 37,579 ktCO2; scenario 2: 37,736 ktCO2; scenario 3: 37,754 ktCO2) in multi-family housing complexes are forecasted to increase by 10.00–21.91%. The developed optimized gene expression programming – harmony search optimization model will potentially be able to assist policymakers in central and local governments forecast the national CO2 emissions in 2030. Through this, national CO2 emission management that more closely reflects the characteristics at the regional or national level can be supported.