摘要
针对传统谷物食用品质检测方法存在检测速度慢的问题,提出了一种基于微波口由空间测量的快速检测方法,并设计了一种新型谷物品质检测装置。采用口由空间传输反射法进行测量,借助矢量网络分析仪获取小麦粉与滑石粉的介电常数。分析其特征谱分析结果表明,根据两种粉末介电常数的差异对其进行区分是可行的.检测装置基于两次外差混频电路结构。采用自主设计的微波天线探测腔实现微波信号耦合及外部电磁屏蔽,能够有效减少检测误差。系统以STM32单片机为控制核心,通过串口与上位机进行实时通信,即时显示检测结果,以小麦粉和滑石粉为样本进行试验,分析其频谱变化曲线并采用反向传播(BP)神经网络对谷物粉进行鉴别分类,模型对样本的鉴别正确率为97.1%。该设计可推广用于区分不同种类与品质的谷物。
Aiming at the problem of slow detection speed in traditional methods of grain edible quality detection,a fast detection method based on microwave free space measurement is proposed,and a new type of grain quality detection device is designed.The free-space transmission reflection method is used for measurement.The permittivities of wheat flour and talcum powder are obtained by means of vector network analyzer,and their characteristic spectra are analyzed.The results show that it is feasible to distinguish the two powders according to the difference of their permittivities.Based on the double heterodyne mixing circuit structure,the detection device adopts the self-designed microwave antenna detection cavity to achieve microwave signal coupling and external electromagnetic shielding,which can effectively reduce the detection error.The system takes STM32 MCU as the control core,communicates with the host computer in real time through serial port,and displays the test results in real time.Wheat flour and talcum powder are taken as samples for experiment,and the frequency spectrum curve of wheat flour is analyzed,and back propagation(BP)neural network is used to identify and classify grain flour,and the detection accuracy of the model for samples is 97.1%.The design can be promoted to distinguish differrnt types and qualities of grain.
作者
王康
徐雷钧
项厚友
葛星梅
WANG Kang;XU Leijun;XIANG Houyou;GE Xingmei(School of Electrical Information Engineering,Jiangsu University,Zhenjiang 212013,China)
出处
《自动化仪表》
CAS
2021年第10期17-21,26,共6页
Process Automation Instrumentation
基金
国家自然科学基金资助项目(61874050)
江苏省农业科技自主创新基金资助项目[CX(17)3001]。
关键词
微波
谷物品质
自由空间法
介电常数
神经网络
单片机
上位机
检测
Microwave
Grain quality
Free space method
Permittivity
Neural network
Single chip
Host computer
Detection