基于Richards-BP神经网络的交通运输碳排放量组合预测模型  

A combined model of Richards model and BP neural network to predict transportation carbon emission

作　　者：高洁[1,2] 王建伟[1] 李琳娜[1] 

机构地区：[1]长安大学经济与管理学院,陕西西安710064 [2]江苏大学管理学院,江苏镇江212013

出　　处：《长安大学学报（自然科学版）》2013年第4期99-104,109,共7页Journal of Chang’an University(Natural Science Edition)

基　　金：教育部人文社科基金项目(10YJA790184);教育部博士点基金项目(20100205110006);中央高校基本科研业务费专项资金重点项目(Z1101)

摘　　要：为提高交通运输碳排放量的预测精度,根据交通运输碳排放量时间曲线具有的非线性饱和增长及随机性波动特点,建立基于Richards模型和BP神经网络的组合预测模型;以1985～2010年中国交通运输碳排放量数据为样本对模型进行了拟合和检验,并将Richards-BP神经网络组合模型预测结果与单项Logistic模型、GM(1,1)模型、Richards模型、BP神经网络及Logistic-BP神经网络组合模型、GM(1,1)-BP神经网络组合模型进行了误差对比分析。研究结果表明:3种组合模型的预测误差明显小于单一模型的预测误差,通过BP神经网络对单一预测模型进行误差修正可显著提高交通运输碳排放量预测精度;Richards-BP神经网络组合模型预测结果的平均绝对误差、平均绝对百分比误差及标准差值分别达到118.439×104 t、0.254%及136.915×104 t,比Lo-gistic-BP神经网络组合模型及GM(1,1)-BP神经网络组合模型精度提高了近5倍;以Richards模型的拟合误差作为BP神经网络输入效果要优于其他模型,Richards-BP神经网络组合模型具有更高的预测精度。In order to improve the prediction accuracy of transportation carbon emission, a com- bined prediction model based on Richards model and BP neural network was established according to the characteristics of nonlinear and saturated growth in transportation carbon emission time se-ries. The time series of transportation carbon emission from 1985 to 2010 in China was used to for fitting and validation of the combined model. The results of Richards-BP network model were compared with that of the individual Logistic model, GM （1,1） model, Richards model, BP neu-ral network model, Logistic-BP network model and GM （1,1）-BP network model respectively. The results show that the prediction errors of three combined prediction models are significantly less than the errors of one single model. It is indicated that the BP neural network can improve the accuracy of the single model by error correction. In addition, the prediction error indexes such as average absolute error, average absolute percentage error and standard deviation of the combined Richards-BP network model are respectively 118. 439 × 10^4 t, 0. 254% and 136. 915 × 10^4 t, which means about five times of the accuracy of the Logistic BP network model and GM （1,1）-BP network model are improved. The Richards model is better than the other single models as an input for BP network study. The combined Richards-BP network model is a reasonable method of higher accuracy for prediction of transportation carbon emission. 5 tabs, 2 figs, 14 refs.

关 键 词：交通工程 交通运输 碳排放量 组合预测 Richards模型 BP神经网络 

分 类 号：U491[交通运输工程—交通运输规划与管理] F224[交通运输工程—道路与铁道工程]