GA-SA模型预测中国能源需求

为预测中国能源需求,本文提出了一种新的遗传算法(genetic algorithm)和模拟退火法(simulated annealing)混合方法,称为GA-SA,该方法能利用各自的优势,克服彼此之间的弱点.分析了GDP、人口、固定资产投资、能源效率、能源消费结构及居民生活水平六个主要能源影响因子后,使用1981至2012共32年间各因子及能源消费的观察数据,运用GA-SA方法优化非线性二次能源评估方程,得出了该方程的权重系数.结果显示,提出的方法和选取的能源因子构建的GA-SA模型预测的能源值与观察数据值及其接近,并与其它模型进行了比较,准确性上提高了至少两个数量级.运用提出的GA-SA模型预测了中国在2015和2020年时两种不同情景下的能源需求量,并提出了一些政策性建议.

GA-SA model forecasting China's energy demand

In order to forecast China's energy demand, the study develops a genetic algorithm and simulated annealing robust hybrid algorithm, called GA-SA, which can overcome their weaknesses by acquiring others' strong points. After analyzing and using the main influencing factors, such as GDP, population, fixed assets investment, energy efficiency, energy consumption structure, living standards of individual resident, the GA-SA optimize the quadratic model based on 32-year historical data between years 1981 and 2012. Compared with other models, the results show that the predicted energy demands and the observed ones are very close, and accuracy is at least two orders of magnitude higher than other models, which shows the proposed GA-SA forecasting model has very high accuracy. The forecasting system model plans the energy demand of China in 2015 and 2020 under two different scenarios. At last, some policy proposals are put forward.