基于EWT-ARIMA组合模型的银杏液流预测与因子关联分析  

Ginkgo sap flow prediction based on EWT-ARIMA model and factor correlation analysis

作　　者：王子祥 李颜娥 武斌[1] 徐达宇[1] 吴斌 Wang Zixiang;Li Yan’e;Wu Bin;Xu Dayu;Wu Bin(College of Mathematics and Computer Science,Zhejiang A&F University,Hangzhou 311300,China;Key Laboratory of Forestry Intelligent Monitoring and Information Technology of Zhejiang Province,Hangzhou 311300,China;China Key Laboratory of State Forestry and Grassland Administration on Forestry Sensing Technology and Intelligent Equipment,Hangzhou 311300,China)

机构地区：[1]浙江农林大学数学与计算机学院,浙江杭州311300 [2]浙江省林业智能监测与信息技术实验室,浙江杭州311300 [3]林业感知技术与智能装备国家林业局重点实验室,浙江杭州311300

出　　处：《电子技术应用》2023年第10期89-95,共7页Application of Electronic Technique

基　　金：国家自然科学基金(72001190);教育部人文社科基金(20YJC630173);浙江省科技重点项目(2022C02009,2022C02044,2022C02020);浙江省基本公益项目(GN21F020001)。

摘　　要：树干液流速率由于受到外在环境因子与内在生长机理的综合作用,往往呈现出非线性与高随机的特点,单一的预测方法往往难以对其做出较为准确的预测。对此,提出引入经验小波变换(EWT)方法,对非线性、高随机的银杏液流数据进行分解,得到两组多分辨率分析分量,分别对分量采用统计模型ARIMA进行预测。根据真实数据实验结果验证,提出了EWT-ARIMA组合模型能够较为准确地预测树干液流的变化趋势,模型评价指标MSE、MAE、MAPE、R2分别为11.05、2.488、0.1640、0.9599,相较单一ARIMA模型各项评价指标均有较大提升。此外,还利用传递熵(EWT),无模型假设地对时滞内环境因子与银杏液流之间的因果关系进行了探讨。Due to the comprehensive effect of environmental factors and growth mechanism,the sap flow often presents the characteristics of nonlinearity and high randomness,and it is often difficult to predict it accurately by a single prediction method.This paper proposes to introduce the empirical wavelet transform(EWT)method to decompose the nonlinear and highly random ginkgo sap flow data to obtain two sets of multi-resolution components,and the ARIMA model is used to predict the components respectively.According to the results,it is proposed that the EWT-ARIMA model can accurately predict the change trend of sap flow,and the model evaluation indicators MSE,MAE,MAPE,R2 are 11.05,0.1640,0.9599 and 0.9598,respectively,which are greatly improved compared with the single ARIMA model.In this paper,transfer entropy(TE)is also used to explore the causal reflection between environmental factors in time delay and ginkgo sap flow without model assumptions.

关 键 词：银杏液流预测 经验小波变换 ARIMA模型 传递熵 因果分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]