Short-term prediction of carbon emissions based on the EEMD-PSOBP model

A novel short-term carbon emission prediction model based on secondary decomposition method and long short-term memory network

“…Time series forecasting is a means of using historical data to predict the data in a certain period of time in the future. It emphasizes the importance of the time factor in prediction and temporarily ignores other external influential factors (Sun and Ren 2021 ). Applying the time series forecasting method effectively overcomes the above problems.…”

“…Wu and Huang ( 2009 ) proposed ensemble empirical mode decomposition (EEMD), which can significantly reduce the fluctuation of original data. Sun and Ren ( 2021 ) constructed a hybrid prediction model combining EEMD and backpropagation neural network optimized by particle swarm optimization (PSO-BP) to predict short-term carbon emissions in China. This research conclusion highlighted that EEMD is more suitable to deal with short-term carbon emission data than other primary decomposition methods.…”

“…Moreover, in contrast with traditional machine learning algorithms, LSTM does not need feature engineering in the construction process and can better mine the time relationship between data series (Sun and Huang 2020 ). And there are nonlinearity and fluctuation in the original sequence of daily carbon emissions, and the volatility of data will have a certain impact on the prediction (Sun and Ren 2021 ). When the amount of data is large and highly irregular, data decomposition technology will improve the precision of prediction to a certain extent (E et al 2021 ).…”

A novel short-term carbon emission prediction model based on secondary decomposition method and long short-term memory network

“…Moreover, in contrast with traditional machine learning algorithms, LSTM does not need feature engineering in the construction process and can better mine the time relationship between data series (Sun and Huang 2020). And there are nonlinearity and uctuation in the original sequence of daily carbon emissions, and the volatility of data will have a certain impact on the prediction (Sun and Ren 2021). When the amount of data is large and highly irregular, data decomposition technology will improve the precision of prediction to a certain extent (E et al 2021).…”

“…It illustrates that different decomposition means have signi cantly different effects on the improvement of prediction accuracy. In contrast with EMD, EEMD is more suitable for processing the short-term carbon emission data, which is the same as described by the conclusion ofSun and Ren (2021).In this part, there are nine comparison models and the error results of each model are depicted in Fig.9. It can be seen from the image that EEMD-VMD-LSTM has the highest accuracy, which suggests EEMD-VMD is superior to EEMD and EMD in improving the model prediction ability.…”