基于GM-ANN模型的生态足迹时间序列预测分析——以甘肃省为例

A prediction analysis of the ecological footprint time series of Gansu province based on the GM-ANN model

基于生态足迹、生态承载力及其变化时间动态特征的分析,应用灰色系统与神经网络理论与方法,组建了GM-ANN预测模型,以甘肃省为例进行了评价与预测分析.结果表明:1991-2009年期间,甘肃省的生态经济发展一直处于不可持续状态,人均生态足迹为1.517 hm^2/人,人均生态承载力为1.077 hm^2/人,人均生态冗余为-0.44hm^2/人.预测结果显示,到2015年和2020年,甘肃省人均生态足迹将分别达到2.503 hm^2/人和2.870 hm^2/人,而人均生态承载力将分别降至0.985 hm^2/人和0.930 hm^2/人,人均生态冗余则分别为-1.518hm^2/人和-1.940 hm^2/人.这说明未来该省生态经济仍处于不可持续状态,急需调整经济结构与消费模式,以降低其人均生态足迹和增加生态冗余值.另外,通过内插拟合能力检验分析证明,新建立的GM-ANN模型与常用的GM(1,1)模型相比,可使预测精度提高1.7%,在分析和预测不确定系统中有明显的优越性.因此,GM-ANN模型在研究生态足迹动态的过程中,有着较为广泛的应用前景.

Based on an analysis of the dynamic characteristics concerning the ecological footprint, ecological capacity and their changes, the theory and methods of the gray system and the neural network was applied, in combination with the respective merits of these two methods, to establish the GM-ANN forecasting model for the purpose of evaluating and predicting the sustainability of Gansu province. The results show that, during the period from 1991 to 2009, the development of Gansu＇s ecological economy was in a non-sustainable state. On average, its per capita ecological footprint （EFP） was 1.517 hm2, per capita ecological capacity （ECP） was1.077 hm2, and per capita ecological remainder （ERP） was -0.44 hm2. The prediction shows that, in 2015 and 2020, the EFP of Gansu province will be increased to 2.503 hm2, and 2.870 hm2; the ECP will be reduced to 0.985 hm2 and 0.930 hm2; the ERP will be -1.518 hm2 and -1.940 hm2. These results reveal that the future of the Gansu＇s ,e.cological economy will continue to be Uns^stalnable and adjustment of the economic structure and consumption patterns should be urgently done in order to reduce the per capita ecological footprint and increase the ecological remainder. In addition, the test analysis of the interpolated fitting ability proved that, compared with the single gray GM （1, 1） model, the new GM-ANN model could ificrease the forecasting accuracy by 1.7 percentage points. It has obvious advantages in analysis and prediction of an uncertainty system. Therefore the GM-ANN model has a relatively wide range of applications in studying the dynamic processes of ecological footprint.