基于环境气体信息的BP神经网络苹果贮藏品质预测  

Prediction of apple storage quality using BP neural network based on environmental gas information

作　　者：张永超[1] 赵录怀[2] 王昊 张宇航 ZHANG Yong-chao;ZHAO Lu-huai;WANG Hao;ZHANG Yu-hang(City College,Xi'an Jiaotong University,Xi'an 710018,China;School of Electrical Engineering,Xi'an Jiaotong University,Xi'an 710049,China)

机构地区：[1]西安交通大学城市学院,陕西西安710018 [2]西安交通大学电气学院,陕西西安710049

出　　处：《江苏农业学报》2020年第1期194-198,共5页Jiangsu Journal of Agricultural Sciences

摘　　要：针对苹果贮藏品质预测复杂、精度低的问题,设计了基于环境气体信息的BP神经网络苹果贮藏品质预测。首先分析了贮藏环境中温度、气体体积比与苹果理化特性指标的相关性,再通过对苹果的贮藏温度、气体(氧气、二氧化碳)体积比和理化特性指标(硬度、可溶性固形物含量、总酸含量、水分含量)进行检测,将16组温度和气体体积比数据作为BP神经网络的输入,理化特性指标分别作为BP神经网络的输出,对建立的BP神经网络进行训练。训练后用5组非训练样本进行试验验证,结果表明用BP神经网络模型预测苹果贮藏品质的预测值与实测值相对误差在5%以下,可以满足苹果贮藏品质预测的精度要求。In order to solve the problem of complex and low accuracy of apple storage quality prediction,the BP neural network based on environmental gas information was designed.The correlation between temperature,gas volume ratio and physical and chemical properties of apple in storage environment was analyzed,and then the storage temperature,gas(oxygen,carbon dioxide)volume ratio and physical and chemical properties(hardness,soluble solid content,total acid content,moisture content)were measured.Sixteen groups of temperature and gas volume ratio data were used as the input of BP neural network,and the physical and chemical properties were used as the output of BP neural network to train the established BP neural network.After training,five groups of non-training samples were used to verify the results.The results showed that the relative error between the predicted value and the measured value of apple storage quality using BP neural network model was less than 5%,which could meet the accuracy requirements of apple storage quality prediction.

关 键 词：苹果 贮藏温度 气体信息 贮藏品质预测 BP神经网络 

分 类 号：TS255.3[轻工技术与工程—农产品加工及贮藏工程]