基于组合模型的中国氢气产量动态预测

Dynamic prediction of hydrogen production in China based on combination model

基于灰色GM(0,N)模型、BP神经网络、偏最小二乘法回归,采用方差倒数法建立产量动态预测组合模型,以2012—2021年样本期间,动态预测2022—2025年氢气产量,结果表明组合模型2012—2021年的平均相对误差0.91%,2022—2025年的平均相对误差1.84%,均低于单项模型平均相对误差,验证了组合模型的有效性和准确性,为氢气产量预测提供借鉴和参考。受工艺技术复杂和高成本的影响,我国氢气产量整体增幅较缓慢,建议从降低氢气制储运加成本、推动氢气规模化应用和加大氢能技术创新力度方面助力氢能产业高质量发展。

Based on the grey GM(0,N)model,BP neural network,partial least squares regression,a dynamic prediction of production combination model was established by using the inverse variance method.During the sample period from 2012 to 2021,the hydrogen production from 2022 to 2025 was dynamically predicted.The results showed that the average relative error of the combined model from 2012 to 2021 was 0.91%,and the average relative error of the single model from 2022 to 2025 was 1.84%,both lower than the average relative error of the single model.The effectiveness and accuracy of the combined model were verified,and it could provide reference for hydrogen production prediction.Affected by the complexity of process technology and high cost,the overall growth of hydrogen production in China is relatively slow.It is suggested to help the high-quality development of hydrogen energy industry by reducing the cost of hydrogen production,storage,transportation and processing,promoting the large-scale application of hydrogen and increasing the innovation of hydrogen energy technology.