基于气体传感器阵列的Tent-SSA-BP苹果鲜度预测模型  

Tent-SSA-BP Apple Freshness Prediction Model Based on Gas Sensor Array

作　　者：杨伟震 王伟[1] YANG Weizhen;WANG Wei(College of Information and Communication Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学信息与通信工程学院,山西太原030051

出　　处：《山西农业科学》2023年第3期333-339,共7页Journal of Shanxi Agricultural Sciences

基　　金：中北大学重点实验室开放研究基金项目(DXMBJJ2018-08)。

摘　　要：为了能够准确预测苹果的鲜度等级、有效降低苹果的储藏风险、避免经济损失,基于一种气体传感器阵列和无线传输模块的电子鼻系统,采集苹果散发的气味信息。在实验室现有的苹果鲜度识别系统的研究基础上,以室内条件下储存的富士苹果为研究对象,建立基于混沌序列(Tent)改进的麻雀搜索算法(SSA)优化反向传播(BP)神经网络的预测模型,通过预测苹果挥发的气味信息完成对苹果鲜度的预测。结果表明,基于气体传感器阵列的Tent-SSA-BP预测模型的决定系数为0.942 03,均方误差为0.000 4,准确性总体高于BP神经网络预测模型的0.800 57,且具有更高的预测稳定性。该预测模型解决了当前基于传统动力学模型的预测准确率低、检测效率低、且对苹果造成破坏的问题;同时相较于红外光谱等其他预测手段,该模型具有更低的经济成本,且具有操作简便、移动便携等优点。综上所述,通过对苹果挥发的气味信息进行预测以实现对苹果鲜度的预测,具有操作简便、成本低、结果可靠的特点,避免了预测鲜度的过程中对苹果造成破坏,实现了无损检测。In order to accurately predict apple freshness degree, effectively reduce the storage risk of apples, and avoid economic losses, in this study, an electronic nose system based on a gas sensor array and wireless transmission module was proposed to acquire smell information from apples. On the basis of our lab’s existing apple freshness recognition systems, Fuji apples stored in indoor conditions were taken as the research object, a prediction model based on back propagation(BP) neural network optimized by sparrow search algorithm(SSA) improved by chaotic sequence(Tent) was established to predict apple freshness by predicting apple smell information. The results showed that the determination coefficient R2 of the Tent-SSA-BP prediction model using gas sensor array was 0.942 03 with a mean square error of 0.000 4, which was higher than that of the BP neural network prediction model of 0.800 57 and had higher prediction stability. The Tent-SSA-BP prediction made up for the current problems of low prediction accuracy, low detection efficiency, and damage to apples of traditional dynamic models.Simultaneously, compared with other prediction methods such as infrared spectroscopy, the model had the advantages of lower economic cost, easier operation, and portability. In conclusion, prediction of the apple freshness achieved by prediction of smell information volatilized from apples had the characteristics of simple operation, low cost, and reliable results, avoided damage to apples in the process of prediction of freshness, and achieved non-destructive testing.

关 键 词：苹果鲜度 预测算法 气味识别 传感器阵列 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]