基于电子舌结合VMD-IMG组合模型的清香型白酒掺假检测
Adulteration detection of mild aromatic Chinese spirits on electronic tongue combined with VMD-IMG composite model
白雪瑞 1李鑫 1孙涛 1王彦荣 1曾琬晴 1王志强1
作者信息
- 1. 山东理工大学 计算机科学与技术学院,山东淄博 255049
- 折叠
摘要
针对传统白酒掺假检测方法分析流程长、仪器操作复杂等问题,提出一种基于电子舌结合变分模态分解和改进多通道GhostNet网络组合模型,实现对白酒掺假快速检测的方法.采用贝叶斯优化后的VMD对信号进行自适应模态分解,将原始信号分解为多个本征模态函数.在传统GhostNet网络的基础上,引入瓶颈注意力模块和空洞卷积,并扩展为多通道GhostNet,进而构建为VMD-IMG组合模型,以实现电子舌信号特征的有效提取和分类识别.试验结果表明,VMD-IMG组合模型具有较好的白酒掺假分辨性能,其测试的准确率、精确率、F1分数分别为98.33%,98.43%,98.34%.研究为白酒掺假检测提供一种快速、低成本的检测方法,并可为其他酒类或饮品掺假检测提供参考.
Abstract
In view of the problems of long analysis process and complex instrument operation with the traditional liquor adulteration detection methods,a method based on the electronic tongue combined with Variational mode decomposition(VMD)and Improved Multichannel Ghostnet(IMG)was proposed to realize the rapid detection of liquor adulteration.The Bayesian optimized VMD was used to perform adaptive mode decomposition of the signal,and the original signal was decomposed into multiple intrinsic mode functions(IMFs).Based on the traditional GhostNet network,the bottleneck attention module and dilated convolution were introduced and extended to multi-channel GhostNet,which was then constructed as a combined VMD-IMG model to realize the effective extraction of electronic tongue signal features and classification recognition.The experimental results show that the VMD-IMG combined model proposed has a better performance in liquor adulteration discrimination,and its tested accuracy,precision,and F1-Score are 98.33%,98.43%,and 98.34%,respectively.This study provides a fast and low-cost detection method for liquor adulteration detection,and can provide a new technical means for other liquor or beverage adulteration detection.
关键词
白酒掺假/电子舌/变分模态分解/贝叶斯优化/多通道GhostNet网络Key words
liquor adulteration/electronic tongue/variational modal decomposition/Bayesian optimization/multi-channel GhostNet network引用本文复制引用
基金项目
山东省自然科学基金(ZR2022MF330)
教育部科技发展中心产学研创新基金(2018A02010)
出版年
2024
国家科技期刊平台
NSTL
万方数据