欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
传统GMDH算法在进行多变量非线性建模时耗时较长,一定程度上限制了它的应用范围.针对这个问题,提出了一种改进的GMDH算法,扩大了每一个初始输入元素的信息含量,采用随机分组建立中间模型的方式代替原算法枚举出所有两两组合中间模型的方式,减少了中间模型的数量,提高了建模效率.将改进算法应用于中国GDP的趋势预测,结果表明与传统GMDH算法相比,在不牺牲预测精度的情况下,改进算法效率更高.
英文摘要：
GMDH algorithm takes much time to obtain the final solutions when building nonlinear models with many variables, so its application is limited. Aimed at this problem, some improvements are implemented on the transfer functions of classical GMDH algorithm. More information is added to every initial input, and a stochastic way instead of the emumerating way is used to create the next middle layer models in order to decrease the number of middle layer models and enhance the modeling efficiency. Finally, through building a prediction model both with the classical GMDH algorithm and improved GMDH algorithm to forecast Chinese GDP, it is proved that the latter can be more efficient.
同期刊论文项目
同项目期刊论文
期刊信息
位置：