基于小波变换-DGM(2,1)模型的海水养殖品产量预测

Prediction of Mariculture Output Based on Wavelet Transform-DGM(2,1)

由于我国海水养殖品产量数据呈非单调、不光滑分布特征,为了提高预测精度,采用小波变换和DGM(2,1)模型相结合的方法对其进行预测。首先运用小波变换去除原始信号中的噪声,然后以去噪后的信号为基础,利用DGM(2,1)模型进行预测,取得了较好的效果。模型的平均预测误差为2.0857%,比传统灰色GM(1,1)模型的2.8747%减小了27.4463%,比传统DGM(2,1)模型的2.2773%减小了8.4135%。由模型预测得到2021年我国海水养殖品产量为2058.865万吨。

Due to the non-monotonic and non-smooth distribution of mariculture production data in China,the method of combining wavelet transform and DGM(2,1)model was used to predict the output of mariculture products in China,in order to improve the prediction accuracy.Firstly,wavelet transform was used to remove the noise in the original signal.Then,based on the de-noised signal,DGM(2,1)model was used for prediction,and good results were achieved.The average prediction error of the model was 2.0857%,which is 27.4463% less than 2.8747% of the traditional grey GM(1,1)model and 8.4135% less than 2.2773% of the traditional DGM(2,1)model.According to the prediction of the present model,the output of mariculture in China will be 20.58865 million tons in 2021.