基于加速遗传算法的区域生态足迹预测一般回归模型

 为选择区域生态足迹的预测模型结构的合理形式，提出把一般回归模型(GRA)作为区域生态足迹预测模型结构，用加速遗传算法(AGA)进行高精度GRA统一建模的新方案，从而建立了基于AGA的区域生态足迹一般回归模型(AGA-GRA)。AGA-GRA的实证结果说明：在2005-2020年期间，安徽省人均生态足迹将由1.724 6升至2.148 6 hm²，平均年增长率为1.48%；人均生态承载力由0.415 8降至0.338 8 hm²，平均年减少率为1.36%；人均生态赤字由1.323 7升至1.876 0 hm²，平均年增长率为2.35%；万元GDP生态足迹由1.836 7降至0.395 9 hm²/万元，平均年减少率为9.73%。安徽省的人均生态赤字仍在不断加剧，说明现有的社会经济发展模式是不可持续的。AGA-GRA克服了普通时间序列分析和回归分析模型中预先确定简单函数曲线则预测误差大而预先确定复杂曲线则模型求解困难之间的矛盾，提高了模型拟合和预测能力。

Accelerating Genetic Algorithm Based General Regression Analysis Model of Predicting Regional Ecological Footprint

In order to choose rational structure of prediction model, a general regression analysis model was suggested as the prediction model structure of regional ecological footprint, and accelerating genetic algorithm was used to optimiz the model parameters, and then a scheme of accelerating genetic algorithm based general regression analysis model for predicting dynamic change of regional ecological footprint, named AGA GRA for short, was established. The AGA GRA was applied to forecasting dynamic change of the ecological footprint of Anhui province.The results show that the ecological footprint per capita will increase from 1.7246 hm² in 2005 to 2.1486 hm² in 2020 with average annual increase rate of 1.48%;the ecological capacity per capita will decrease from 0.4158 hm² in 2005 to 0.3388 hm² in 2020 with average annual decrease rate of 1.36%; the ecological deficit per capita will increase from 1.3237 hm² in 2005 to 1.8760 hm² in 2020 with average annual increase rate of 2.35%, the ecological footprint per 10 thousand Yuan GDP will decrease from 1.8367 hm² in 2005 to 0.3959 hm² in 2020 with average annual decrease rate of 9.73%. The average personal ecological deficit of the province is increasing year by year, and the present developing model of economy and society is not sustainable. The conflict between predefined simple function curve inducing big error and complex function curve inducing modeling difficulty can be overcome, the fitting and forecasting power of model can be advanced by using AGA GRA.