Will Poland fulfill its coal commitment by 2030? An answer based on a novel time series prediction method
Yan Li, Hongchao Zhang, Yingqing Kang
Abstract
Coal accounted for around 80 percent of power production in Poland in 2018. Facing the serious climate problem and pressure from all sides, Poland has laid out a long-term energy strategy: reducing the proportion of coal in electricity production to about 60% by 2030. However, considering the history of Poland’s dependence on coal and blank in the field of forecasting Polish coal consumption, many researchers and politicians have shown worry about this commitment. Therefore, we proposed a novel time series prediction method ‘ASR’ which combined Adaptive Boosting, Simulated Annealing and Relevance Vector Machine. ASR inherited the advantages of the three algorithms and was applied to predict carbon consumption in Poland. To confirm the effectiveness of ASR, we compared ASR with several previous methods by two experiments using the Polish coal consumption data from 1965 to 2018. The experimental results showed that the mean absolute error of ASR was 1.2% in executing single point prediction and 5.8% in long-term prediction, which were much lower than previous methods. After verifying ASR, finally, we trained a final ASR model to predict the coal consumption of Poland with 95% confidence interval from 2019 to 2030. Results showed Poland is likely to fulfill its promises under current policy.