A new method for predicting carbon emissions in energy industry based on CEEMD wavelet denoising and hybrid machine learning

As an important source of carbon emissions in contemporary society, the energy industry is currently affected by the redundant impact of massive data, which generates significant noise and reduces prediction accuracy. To address this, a new method for predicting carbon emissions in the energy industry is proposed, based on Complementary Ensemble Empirical Mode Decomposition (CEEMD), wavelet denoising, and hybrid machine learning. By collecting historical carbon emission data from the energy industry as the training set, test set, and validation set for the prediction model, the data is denoised using a wavelet threshold denoising method based on CEEMD. The dataset is decomposed using the CEEMD algorithm to obtain several noisy Intrinsic Mode Function (IMF) components. The wavelet denoising method is then applied to each IMF component, and the denoised IMF components along with the residual component are combined and reconstructed to obtain the denoised carbon emission dataset. A hybrid machine learning model is constructed using a Nonlinear Autoregressive (NAR) neural network and a deep extreme learning machine to achieve carbon emission prediction for the energy industry. The two prediction models are combined through optimal weighting, and the denoised carbon emission dataset is used in the training and testing phases of the combined prediction model. Finally, the trained model is applied to predict carbon emissions. Experimental analysis demonstrates that this method achieves higher accuracy, exhibits good stability, effectively reduces error fluctuations, and performs excellently in carbon emission prediction.