A novel grey prediction model with a feedforward neural network based on a carbon emission dynamic evolution system...

Area of Science:

• Environmental Science
• Data Science
• Energy Policy

Background:

• Accurate carbon dioxide (CO2) emission prediction is crucial for effective governmental energy policies.
• Existing models may not fully capture the dynamic evolution and external influencing factors of carbon emissions.

Purpose of the Study:

• To develop and validate a novel grey model integrated with neural networks for predicting carbon dioxide emissions.
• To enhance the prediction accuracy of dynamic carbon emission systems by incorporating external factors.

Main Methods:

• Established a grey model for carbon emissions, expanding its structure.
• Integrated the mechanism of a feedforward neural network with external influencing factors.
• Optimized model parameters and derived modeling steps.
• Validated the model using Beijing's carbon emission data (2009-2018).

Main Results:

• The developed model achieved simulation and prediction errors below 10% across four test cases.
• Case 1 yielded the best results with errors of 1.56% and 2.07%.
• The model accurately predicted Beijing's CO2 emissions for the next five years, aligning with actual trends.

Conclusions:

• The proposed neural network-based grey model is effective and feasible for predicting carbon dioxide emissions.
• The model's accuracy supports its application in informing energy policy and planning.