摘要
为了提高人口预测精度,提出了基于多项式神经网络模型与递推最小二乘法的人口预测方法.方法完全避免了人为假设条件,充分利用我国六次人口普查数据来建立基于多项式神经网络模型的人口预测模型,并使用递推最小二乘算法递推计算多项式神经网络模型的加权系数.方法能有效预测中长期人口数据及其变化趋势.研究结果表明,中国将在2016年达到人口高峰1385亿.
In order to improve the prediction accuracy about the population,this paper proposed the population's prediction method based on the polynomial neural-network model and least squares algorithm.This method makes full use of our six census data to establish the population prediction model based on polynomial neural-network model without any assumptions,and compute recursively the weighted coefficients of the polynomial neural-network model using the recursive least squares(RLS).The approach can effectively predict the mid long-term demographic data and its change trend.The research results showed that the population of China will reach the peak:1385 million in 2016.
出处
《数学的实践与认识》
北大核心
2016年第18期152-158,共7页
Mathematics in Practice and Theory
基金
湖南省自然科学基金项目(2015JJ6043)
湖南科技学院重点学科(电路与系统)建设项目资助
关键词
神经网络
多项式
人口预测
递推最小二乘法
残差校正
neural-network
polynomial
population prediction
recursive least squares(RLS)
residual error correction