| Kiwifruit is widely loved by consumers because of its tender,juicy,delicious taste and rich nutrition.In order to improve the yield and single fruit weight of kiwifruits,fruit growers often dip the young fruit of kiwifruits in exogenous plant growth regulators(commonly known as expansion agent).However,the quality of kiwifruits treated with expansion agent was poor,the taste became worse and the rotten fruit rate was higher.In addition,the potential effect of expansion agent on human health is not yet clear.In order to make kiwifruit industry develop healthily,many local governments have stipulated that it is forbidden to use expansion agent in kiwifruit production process.However,due to the absence of corresponding detection methods and devices for kiwifruits treated with expansion agent,the actual production process is still large-scale illegal application of expansion agent.In order to provide a non-destructive,fast and inexpensive non-destructive detector for identifying kiwifruits treated with expansion agent,in this paper,the common forchlorfenuron was used as expansion agent,and the untreated kiwifruits of ’Huayou’,’Xuxiang’,’Hayward’ and the kiwifruits treated with forchlorfenuron were taken as research objects.Based on the near-infrared(NIR)diffuse reflectance spectra of kiwifruits,the characteristic wavelengths of treated and untreated kiwifruits were selected.Then the hardware system and software of the portable detector were designed by ARM controller.Finally,the detection accuracy of the detector was tested and verified.The main contents and conclusions are as follows:(1)Based on the NIR diffuse reflectance spectra of kiwifruits,the characteristic wavelengths for identifying the normal and expanded kiwifruits were selected.The NIR diffuse reflectance spectra of normal and expanded kiwifruits of ’Huayou’,’Xuxiang’,’Hayward’ were collected from 898 nm to 1719 nm.The original spectra were preprocessed by standard normal variate,and 7,6,7 characteristic wavelengths were selected from the full spectra using successive projection algorithm respectively.(2)Design of hardware system of portable and nondestructive detector on expanded kiwifruits.With the ARM Cortex-M3 core STM32F103C8T6 chip as the control processor,the crystal clock,hardware reset and power circuit were designed.A feature wavelength extraction device was designed with five narrowband filters and micro halogen lamp with wavelength of 340~2400 nm.Using photodiodes to measure characteristic wavelengths signal,and a signal detection module was designed by using OPA277 high precision operational amplifier.The circuit schematics,PCB boards and the detector shells were completed.(3)Design of software system of portable and nondestructive detector on expanded kiwifruits.Using C language in Keil MDK 5.13 software to complete the program design,to achieve kiwifruits characteristic wavelengths signal measurement,detector calibration,menu function,interruption program,prediction model call,test results output and other functions.GPIO_PA0~PA4 of STM32F103C8T6 were triggered by external interruption to collect external input signals at the same time.(4)Models for predicting expanded kiwifruits were established and the prediction accuracy of the detector was tested.The characteristic wavelength data of kiwifruits were collected by the detector,and the PLS-DA prediction models of single and mixed expanded kiwifruits were established respectively.The result shows that the accuracy rates of the prediction models for ’Huayou’,’Xuxiang’,’Hayward’,and the mixed varieties of kiwifruits were 92.7%,85.5%,71.2% and 75.8%,respectively.The prediction model was written into the software program of the detector,and the detector recognition rates were 92.9%,85.3%,67.9% and 76.3%,respectively.The result shows that the recognition rate of single variety of kiwifruit was better than that of single variety of kiwifruit,but when multiple varieties were mixed together,the recognition rate of ‘Huayou’ was the highest,while that of ‘Hayward’was the lowest. |