Research on Carbon Emission Prediction Method Considering Data Preprocessing

Abstract

Achieving accurate carbon emission forecasts is an important prerequisite for formulating and evaluating low-carbon transformation and upgrading strategies. However, the existing carbon emission forecasting methods are vulnerable to abnormal data and social factors. Aiming at the above problems, this paper studies a carbon emission prediction model considering the data preprocessing method. First, establish the calculation model of 'converting carbon by electricity', and calculate the conversion coefficient of 'carbon-electricity'; Secondly, data preprocessing methods such as isolated forest algorithm and exponential smoothing method are used to eliminate abnormal data, and then the processed 'carbon-electricity' conversion coefficient is input into the established BA-LSTM prediction model, and the prediction result of carbon emissions is obtained by combining with electricity consumption calculation. Finally, a verification analysis is carried out with the data of the whole society's electricity consumption and energy consumption in a certain province in China. The results show that, compared with the traditional carbon emission prediction method, the calculation model combined with the data preprocessing method proposed in this paper has higher prediction accuracy.