基于BP人工神经网络的北京城市人口预测

人口系统是一个非常复杂的非线性系统,对人口数量的准确预测能为应急避险规划等提供重要依据。通过建立一维时间序列的BP人工神经网络模型,根据1980-2005年的人口统计数据,借助Matlab5.3软件进行了北京城近郊区的户籍人口数量预测。检验结果显示,2003、2004和2005年预测值与实际样本的差值在6～7万人之间,其与实际样本的比值约为1%,预测精度完全符合要求。最终预测得到2006、2007、2008、2009、2010年的人口分别为703.2、708.5、720.9、733.8、734.6万人。

Forecasting the Population of Beijing Urban Based on BP Artificial Neural Network

The population system is a very complicated nonlinear system and an accurate prediction of population will be helpful for the planning of emergency adapting.Through the time series model of BP Artificial Neural Network,according to the statistics on amounts of population from 1980 to 2005 and with the help of Matlab5.3 software,the amounts of registered permanent residence population of city and outskirts in Beijing was predicted.The result showed that difference value was 6~7 ten thousand between prediction value and actual value in 2003、2004、2005 year.The ratio is 1 percent compared with actual value.The forecasting result was more precise and effective.Finally,the prediction result showed that amounts of population was 703.2 、 708.5、720.9、733.8、734.6 ten thousand in 2006、2007、2008、2009、2010 year,respectively.