Estimating Hydrogen Price Based on Combined Machine Learning Models by 2060: Especially Comparing Regional Variations in China

Hydrogen energy’s economic efficiency is the key for China to obtain the goal of “carbon neutrality” by 2060. Different from the bottom-up methods and learning rate methods, this study estimates the hydrogen prices in China and typical regions by 2060 from the perspectives of economics and machine learning. The main factors influencing hydrogen price are determined from the perspectives of economics: hydrogen production, demand, and cost. A novel model is established based on combined machine learning models to predict hydrogen price. The hydrogen production is predicted based on the trained BP neural network model optimized by particle swarm optimization considering the uses of hydrogen. The hydrogen prices prediction model is built by applying a least squares support vector machine optimized by Bayesian optimization considering the hydrogen production, hydrogen demand, natural gas price, coal price, electricity price, and green hydrogen share. Moreover, the hydrogen prices in typical regions in China are compared with the average prices. The results show that the hydrogen price is estimated to decrease below CNY 12/kg and the hydrogen price in Northwest China will be lower than CNY 7.5/kg due to low electricity cost by 2060.