An enhanced moth flame optimization extreme learning machines hybrid model for predicting CO₂ emissions.

Area of Science:

• Environmental Science
• Computer Science
• Optimization Algorithms

Background:

• Accurate prediction of carbon dioxide (CO2) emissions is crucial for effective environmental policy and sustainable development.
• Existing prediction models often struggle with accuracy and adaptability to complex emission patterns.
• Developing robust hybrid models can significantly enhance predictive capabilities for CO2 emissions.

Purpose of the Study:

• To introduce and evaluate a novel hybrid model, GMSMFO-ELM, for precise CO2 emissions prediction.
• To demonstrate the effectiveness of the Gaussian mutation and shrink mechanism-based moth flame optimization (GMSMFO) algorithm in optimizing machine learning parameters.
• To provide a reliable framework for informed decision-making in global sustainability initiatives.

Main Methods:

• Integration of the GMSMFO algorithm with the extreme learning machine (ELM) for a hybrid predictive model.
• Utilizing Gaussian mutation (GM) for enhanced population diversity and the shrink mechanism (SM) for improved exploration-exploitation balance within GMSMFO.
• Validation of GMSMFO on the CEC2020 benchmark suite and application to fine-tune ELM for CO2 emissions prediction.

Main Results:

• The GMSMFO algorithm showed superior performance over other optimization techniques on benchmark datasets.
• The GMSMFO-ELM model achieved a high coefficient of determination (R2) of 96.5% for CO2 emissions prediction.
• The model outperformed existing hybrid approaches in key metrics like RMSE, NRMSE, MAE, and MSE.
• Economic growth, foreign direct investment, and renewable energy were identified as significant predictors of CO2 emissions.

Conclusions:

• The GMSMFO-ELM model demonstrates robust and adaptable performance for accurate CO2 emissions prediction.
• This hybrid approach offers a reliable tool for advancing global sustainability objectives.
• The study underscores the potential of advanced optimization algorithms in enhancing environmental modeling and policy support.