枸杞果柄结合力影响因素分析及模型预测  

Analysis and Model Prediction of Factors Affecting the Binding Force of Fruit Stalks of Wolfberry

作　　者：陈运 刘全程 姜鑫娜 魏雨青 王凡 闫磊[1] 赵健[1] 曹兴达 邢红 CHEN Yun;LIU Quancheng;JIANG Xinna;WEI Yuqing;WANG Fan;YAN Lei;ZHAO Jian;CAO Xingda;XING Hong(School of Technology,Beijing Forestry University,State Forestry Administration Key Laboratory of Forestry Equipment and Automation,Key Laboratory of National Forestry and Grassland Administration on Forestry Equipment and Automation,Beijing 100089,China;Chinese Academy of Forestry,Beijing 100091,China)

机构地区：[1]北京林业大学工学院,国家林业局林业装备与自动化重点实验室,国家林业和草原局林业装备与自动化重点实验室,北京100089 [2]中国林业科学院,北京100091

出　　处：《园艺学报》2024年第11期2645-2656,共12页Acta Horticulturae Sinica

基　　金：青海省重点研发与转化计划项目(2022-HK-116)。

摘　　要：分析枸杞果柄结合力及其相关物理特性参数之间的内在联系与影响规律,发掘重要果实参数,建立果柄结合力预测模型,以期为提高枸杞机械采摘的效率和精确度提供依据。以青海省主栽枸杞品种‘宁杞1号’和‘宁杞7号’成熟期果实为研究对象,通过图像处理方法获得不同成熟度枸杞果实的RGB颜色模型分量,分析了不同品种、成熟度、采摘期温度、湿度及果实在枝条上的位置等对果柄结合力的影响,以及果柄参数、果实参数等与果柄结合力的相关性,构建果柄结合力预测模型。结果显示,可通过RGB颜色模型区分成熟度,但不同成熟度果柄结合力重叠部分较多,不能实现精确采摘。双因素方差分析结果显示品种和成熟度均对结合力有显著影响;果柄参数、果实参数和果实硬度与果柄结合力显著相关;正交试验表明果实在枝条上的位置、采摘时的温度和湿度对结合力没有显著影响。对偏最小二乘回归(PLSR)、随机森林(RF)和反向传播神经网络(BPNN)等建模方法的参数进行优化,建立果柄结合力的预测模型,其中BPNN表现出了优秀的预测性能(R_(p)^(2)=0.888,RMSEP=0.440,RPD=2.756),能够实现对果柄结合力的有效预测。To analyse the intrinsic connection and influence law between the stalk binding force of wolfberry and its related physical property parameters,to discover the important fruit parameters,and to establish a prediction model of stalk binding force,with a view to providing a basis for improving the efficiency and accuracy of mechanical picking of wolfberry.Taking the ripening fruit of‘Ningqi 1’and‘Ningqi 7’,the main cultivars of wolfberry in Qinghai Province,as the research object,the RGB colour model components of wolfberry fruits with different maturity were obtained by image processing methods,and different cultivars,maturity were analysed,temperature,humidity and the position of the fruit on the branch on the impact of the stalk binding force,as well as the correlation between the stalk parameters,fruit parameters and the stalk binding force,to construct a prediction model of the stalk binding force.The results showed that the maturity could be distinguished by the RGB colour model,but the overlapping part of the stalk binding force of different maturity levels was large,which could not be achieved for precise picking.The results of two-way ANOVA showed that both cultivar and maturity had a significant effect on the binding force;stalk parameters,fruit parameters and fruit hardness were significantly correlated with the stalk binding force;orthogonal tests showed that the position of the fruit on the branch,temperature and humidity at the time of picking did not have a significant effect on the binding force.The parameters of partial least squares regression(PLSR),random forest(RF)and back-propagation neural network(BPNN)modelling methods were optimised to establish a prediction model for fruit stalk binding,in which the BPNN showed excellent prediction performance(R_(p)^(2)=0.888,RMSEP=0.440,RPD=2.756),which was able to achieve effective prediction of fruit stalk binding.

关 键 词：枸杞 机械采收 果柄 结合力 形态学 影响因素 预测模型 

分 类 号：S567[农业科学—中草药栽培]