首页 | 简介 | 作者 | 编者 | 读者 | Ei(光盘版)收录本刊数据 | 网络预印版 | 点击排行前100篇 | English
邹小波,张俊俊,黄晓玮,郑开逸,吴胜斌,石吉勇.基于音频和近红外光谱融合技术的西瓜成熟度判别[J].农业工程学报,2019,35(9):301-307.DOI:10.11975/j.issn.1002-6819.2019.09.036
基于音频和近红外光谱融合技术的西瓜成熟度判别
投稿时间:2018-12-19  修订日期:2019-05-10
中文关键词:  声学  近红外光谱  模型  西瓜成熟度  可溶性固形物
基金项目:国家重点研发计划重点专项(2018YFD0400803);国家自然科学基金项目(31601543,31671844,31801631);江苏省自然基金项目(BK20160506; BK20180865)。
作者单位
邹小波 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
张俊俊 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
黄晓玮 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
郑开逸 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
吴胜斌 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
石吉勇 江苏大学食品与生物工程学院农产品加工及贮藏工程实验室镇江 212013 
摘要点击次数: 427
全文下载次数: 180
中文摘要:为了满足西瓜成熟度的快速无损检测需求,该研究主要利用声学技术、近红外光谱技术结合K最近邻法(k-nearest neighbor,KNN)、线性判别分析(linear discriminant analysis,LDA)和反向传播人工神经网络(back propagation artificial neural network,BP-ANN)3种化学计量学方法对不同成熟度的西瓜进行定性判别;同时采用联合区间偏最小二乘筛选法(synergy interval partial least squares,Si-PLS)分别建立声学技术、近红外光谱技术、融合技术的西瓜可溶性固形物预测模型。结果表明融合技术处理结果均优于单一信号,其LDA模型数据的西瓜成熟度模型识别率较佳,校正集和预测集的识别率分别为100.00%和91.67%。同时,基于融合技术所建立的西瓜可溶性固形物预测模型效果较佳,其校正集的均方差根误差(root mean squared error of the calibration set,RMSECV)为 0.601%,预测集的均方差误差(root mean squared error of the prediction set,RMSEP)为0.725%,相比的单独音频信号其均方根误差分别降低了0.081、0.068个百分点。研究结果可为高精度的西瓜品质快速鉴别提供参考。
Zou Xiaobo,Zhang Junjun,Huang Xiaowei,Zheng Kaiyi,Wu Shengbin,Shi Jiyong.Distinguishing watermelon maturity based on acoustic characteristics and near infrared spectroscopy fusion technology[J].Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE),2019,35(9):301-307.DOI:10.11975/j.issn.1002-6819.2019.09.036
Distinguishing watermelon maturity based on acoustic characteristics and near infrared spectroscopy fusion technology
Author NameAffiliation
Zou Xiaobo Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Zhang Junjun Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Huang Xiaowei Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Zheng Kaiyi Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Wu Shengbin Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Shi Jiyong Agricultural Product Processing and Storage Lab., School of Food and Biological Engineering, Jiangsu University, Zhenjiang 212013, China 
Key words:acoustics  near-infrared spectroscopy  models  watermelon maturity  soluble solids content
Abstract:Rapid non-destructive detection internal quality of fruit can improve the marketable value of watermelon. Non-destructive technology discrimination and quality assessment of watermelon in real time is a hot topic. More and more studies have been used for watermelon quality detection, such as near infrared spectroscopy, acoustic methods, electrical and magnetic methods etc. Compared with non-destructive methods such as near-infrared spectroscopy and nuclear magnetic resonance, acoustic signal detection has the advantages of simple equipment and low price. So it is widely used to detect the internal quality information of watermelon. Near-infrared (NIR) spectroscopy is a suitable method to characterize organic compounds which have been successfully applied for the detection of soluble solids content (SSC). However, the accuracy of single detection method is not very high. In this study, we aimed to combine acoustic technology with near-infrared spectroscopy of discrimination and watermelon quality at three maturity level. For the acoustic signal of watermelons, a simple single wire pendulum platform was built for the collection of acoustic. Then the acoustic signal was converted into a spectrum signal. The spectrum signal was modeled after extracting feature variable by (Principal Component Analysis) PCA method. Linear discriminant analysis (LDA), K-near neighbor (KNN) and back-propagation artificial neural network (BP-ANN) were successfully discriminated the watermelon based on the maturity stages. Meanwhile, SSC was the most important evaluation index of watermelon. Therefore synergy interval partial least squares (Si-PLS) was used to establish the SSC prediction of watermelon multivariate model of acoustic signal, spectral signal and conjunction information. The LDA of concatenated technique was the best calibration model and the recognition rate of the calibration set and prediction set get highest were 100.00% and 83.33%, respectively. The rc was 0.846 9 and rp was 0.723 3 of Si-PLS model, respectively. For the near-infrared spectroscopy of watermelon, the study used a portable near-infrared spectrometer to collect information about the quality of watermelons at the equator. Then the spectral signal was collected by portable NIR spectrometer for characteristic wavelength screening by GA (Genetic Algorithm). Finally, the 126 variables were obtained to the next modeling analysis. BP-ANN model showed the highest accuracy while Si-PLS model presented higher prediction (rp was 0.834 8) and calibration (rc was 0.855 9). For the acoustic signal merged the near-infrared spectral signature signal, maximum and minimum normalization of the acoustic signal and the spectral signal were obtained. The precision of quantitative and qualitative models were improved. LDA of concatenated technique showed the best calibration model and the recognition rate of the calibration set and prediction set were 100.00% and 91.67%. And the established calibration model had the best effect. The root mean squared error of the calibration set (RMSECV) was 0.601%, and the root mean squared error of the prediction set (RMSEP) was 0.725%. The rc was 0.901 5 and rp was 0.850 6. Compared with the individual audio signals, the root mean square error is reduced by 0.081 and 0.068 percentage points, respectively. Thus, it evidenced that the acoustic signal merged the near-infrared spectral signature signal can improve the accuracy of the qualitative judgment of watermelon maturity. Based on portable NIR spectroscopy and acoustic signal, the rapid prediction of soluble solids in watermelon can be achieved.
查看全文   下载PDF阅读器

京ICP备06025802号-3

主办单位:中国农业工程学会 单位地址:北京朝阳区麦子店街41号

服务热线:010-59197076、59197077 传真: 邮编:100125 Email:tcsae@tcsae.org
本系统由北京勤云科技发展有限公司设计

京公网安备 11010502031390号