基于自适应局部线性化的环境时间序列数据预测

The Data Forecasting of Environmental Time Series Based on the Self-adaption of Local Linearization

将自适应局部线性化方法应用于环境时间序列数据的预测。对于当前时刻,首先使用自适应局部线性化方法找出合理的嵌入维数,然后在此基础上计算出局部线性化模型的估计参数值,最后利用估计参数值计算出下一个时刻的值。在实际采集的环境时间序列数据上的实验结果表明:相对于标准的局部线性化方法和BP方法,本方法的平均均方根误差和平均绝对误差显著低于以上两种方法的预测结果。

This study applies the method of self-adaption of local linearization to the data forecasting of environmental time series.Firstly,the method of self-adaption of local linearization was used to find out reasonable embedded dimension,and the estimated parameters were counted,then the parameter values of next moment were counted as well b,ased on the previous data.Results show that compared with the methods of standard local linearization and BP,the root-mean-square error and the absolute error of the method of selfadaption of local linearization can be lower in data forecasting.