Analysis of China's provincial carbon peak path based on LSTM neural network

As the world's largest carbon emitter and the second largest economy, China has pledged that its carbon dioxide emissions will peak around 2030, when the intensity of carbon dioxide emissions will be lower than in 2005. However, there is great heterogeneity among provinces in China, and their carbon peak paths cannot be cut off in one size fits all. Based on the "14th Five-Year Plan" of each province, this paper designs three scenarios: baseline, green development, and high-speed development. The LSTM neural network is used to dynamically predict the carbon peak paths of China and its provinces from 2020 to 2040, and the appropriate peak paths are analyzed based on the three factors of carbon emission intensity, cumulative carbon emissions, and peak time of each province. The results show that: China will achieve the carbon peak target before 2030 under different scenarios, with a peak level of 10884-11792 million tons; 24 provinces and regions can achieve the carbon peak target before 2030 under at least one scenario, and most provinces and regions show the characteristics of early peak time and low peak value under low-speed scenario, and late peak time and high peak value under high-speed scenario; Beijing, Shanghai, Fujian, Zhejiang and other provinces and regions can achieve negative carbon after 2035. The research results have important reference value for China to reasonably formulate carbon peak path measures in 2030 and coordinate the allocation of emission reduction tasks.