一种基于弛豫动态光谱技术的苹果损伤检测方法研究  

Study of Apple Slight Damage Detecting Method Based on Dynamic Relaxation Spectrum

作　　者：宁景苑 叶海芬 孙雨玘 熊思怡 梅正昊 蒋晨豪 黄科涛 张苏婕 朱哲琛 李昱权 惠国华 易晓梅 郜园园 吴鹏 NING Jingyuan;YE Haifen;SUN Yuqi;XIONG Siyi;MEI Zhenghao;JIANG Chenhao;HUANG Ketao;ZHANG Sujie;ZHU Zhechen;LI Yuquan;HUI Guohua;YI Xiaomei;GAO Yuanyuan;WU Peng(Key Laboratory of Forestry Sensing Technology and Intelligent Equipment of the Forestry and Grassland Anthority,School of Information Engineering,Zhejiang A&F University,Hangzhou Zhejiang 311300,China;Key Laboratory of Forestry Intelligent Monitoring and Information Technology of Zhejiang Province,School of Information Engineering,Zhejiang A&F University,Hangzhou Zhejiang 311300,China)

机构地区：[1]浙江农林大学信息工程学院,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 [2]浙江农林大学信息工程学院,浙江省林业智能监测重点实验室,浙江杭州311300

出　　处：《传感技术学报》2022年第8期1150-1156,共7页Chinese Journal of Sensors and Actuators

基　　金：浙江省基础公益研究项目(2019C02075);国家级大学生创新项目(202110341027);浙江省新苗计划项目(2021R412031,2021R412032);浙江农林大学学生科研项目。

摘　　要：在运输过程中,苹果的质量不断下降,导致苹果滞销严重,造成了很大的经济损失。现有的苹果品质检测技术虽然能够精准判断苹果是否发生损伤,但是价格高昂,体积庞大,难以推广使用。针对以上情况,探索了一种基于弛豫光谱技术和BP神经网络算法的苹果损伤检测方法。文中以红富士苹果为实验对象,通过自主搭建的弛豫光谱采集系统采集光谱信号,使用SNV算法优化光谱数据,基于BP神经网络算法构建SNV-BP-RS苹果损伤检测模型,该模型准确率为91.48%,检测用时为0.291 s。文中同时基于传统光谱建立了SNV-BP-CS苹果损伤检测模型,该模型准确率为86.39%,检测用时为0.454 s。经过对比,所提出的检测方法大大降低了光谱检测系统对检测波段的需求,检测系统具有体积小、价格低等优势,检测模型准确率高,稳定性良好,为水果高效、低成本判伤提供一种新的思路。The decline of apple quality during transportation leads to serious unsalable apples and huge economic losses.Although whether the apple is damaged can be accurately judged through the existing apple quality detection technology,yet the equipment is expensive and difficult to promote.In view of the above situation,an apple damage detection method based on relaxation spectroscopy and BP neural network algorithm is explored.Red Fuji apple is taken as the experimental object,the spectral signals are collected through the self-built relaxation spectrum acquisition system,the spectral data are optimized by SNV algorithm,and the apple damage detection model using relaxation spectrum is constructed based on BP neural network algorithm.The accuracy of the model is 91.48%,and the detection time is 0.291 s.At the same time,another pple damage detection model using BP traditional algorithm is also established based on the traditional spectrum.The accuracy of this model is 86.39%,and the detection time is 0.454 s.The results indicate that the detecting accuracy is greatly improved by using the proposed method.The detecting system has the advantages of small volume,low price,high accuracy,and good stability.The proposed method provides a new way for fruit slight damage detection.

关 键 词：光谱 弛豫 无损检测 苹果 神经网络 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]