摘要
【目的】增强农产品价格预测准确度,为农产品价格的有效预测提供参考。【方法】以河南省白条猪每周平均批发价格为研究对象,提出一种基于序列分解、主成分分析和神经网络(CEEMDAN-PCA-CNN-LSTM)的白条猪价格预测方法。首先,使用自适应白噪声完全集合模态分解方法(CEEMDAN)对白条猪价格序列进行分解;其次,选用皮尔逊相关系数筛选影响价格波动的相关因素;再次,利用主成分分析(PCA)对影响因素及分解得到的子序列降维处理并作为原始价格序列的特征值,并行输入到作为编码器的卷积神经网络(CNN)中进行特征提取;最后,引入长短期记忆网络(LSTM)作为解码器输出得到预测结果。将该方法应用于河南省白条猪每周平均价格数据,与LSTM、门控循环单元(GRU)、CNN、基于卷积的长短期记忆网络(ConvLSTM)模型进行比较。【结果】CEEMDAN-PCA-CNN-LSTM组合模型预测方法得到的平均绝对误差分别降低了44.95%、27.30%、28.13%、43.17%。【结论】CEEMDAN-PCA-CNN-LSTM模型对于河南省白条猪市场价格的预测性能更优,有助于相关部门针对河南省白条猪价格波动做出科学决策。
【Objective】This study aims to enhance the accuracy of agricultural price prediction and provide reference for the effective prediction of agricultural product prices.【Method】Taking the average weekly wholesale pork price in Henan province as the research object,a price forecasting method based on sequence decomposition,principal component analysis and neural network(CEEMDAN-PCA-CNN-LSTM)was proposed.Firstly,the price seriesofpork was decomposed by using the complete ensemble mode decomposition with adaptive white noise(CEEMDAN).Secondly,Pearson correlation coefficient was used to screen the relevant factors affecting price volatility.Thirdly,principal component analysis(PCA)was used to reduce the dimension of the influencing factors and the subsequence obtained by decomposition,which were used as the feature values of the original price sequence,and were input into the convolutional neural network(CNN)as the encoder for feature extraction.Finally,long short-term memory network(LSTM)was introduced as the decoder output to obtain the prediction results.The proposed method was applied to the weekly average price data of pork in Henan Province,and the prediction results were compared with those by LSTM,gated recurrent unit(GRU),CNN and convolution-based long short-term memory network(ConvLSTM)models.【Result】The mean absolute error obtained by the CEEMDAN-PCA-CNN-LSTM combination model prediction method is reduced by 44.95%,27.30%,28.13%and 43.17%,respectively.【Conclusion】The CEEMDAN-PCA-CNN-LSTM model has better prediction performance for the market price of pork in Henan Province,which is helpful for relevant departments to make scientific countermeasures against the price fluctuations of pork in Henan Province.
作者
刘合兵
华梦迪
席磊
尚俊平
LIU Hebing;HUA Mengdi;XI Lei;SHANG Junping(College of Information and Management Science,Henan Agricultural University,Zhengzhou 450046,China)
出处
《河南农业大学学报》
CAS
CSCD
北大核心
2024年第1期123-131,共9页
Journal of Henan Agricultural University
基金
河南省研究生教育改革与质量提升工程项目(YJS2023AL046)
河南省科技攻关项目(212102110204)。
关键词
价格预测
自适应白噪声完全集合模态分解
主成分分析
神经网络
组合模型
price forecasting
complementary ensemble empirical mode decomposition with adaptive noise
principal component analysis
neural network
combination model
