Design And Experiment Research Of Bionic Olfactory Detection And Analysis Platform Using Purging Method

Bionic olfactory detection designed for mimicking biological olfaction is a kind of intelligent sensor technology,which has a wide range of applications in food,medicine and agriculture.However,the detection instruments used by researchers mainly come from abroad,there are fewer commercial instruments coming from China.Although domestic researchers had developed some olfactory detection devices for laboratory testing,the main emphasis of these research was focused on the application of the device,there were few research reports on the uniformity and stability of the flow field in the olfactory sensor and optimization of bionic olfactory control parameters.At the same time,it was rarely involved in the application of bionic olfactory technology to the detection of citrus pests.Therefore,the bionic olfactory detection and analysis platform using purging method was designed from the hardware and software in this paper,the Fluent software was used to simulate the interior fluid field of the olfactory detection chamber with different structures to select the detection chamber structure suitable for the platform;The effects of sensor heating voltage,gas humidity and flow on olfactory detection performance were analyzed to select the better test control parameters;Finally,the sugar orange with different pest infestation degree was detected by the detection and analysis platform and identified by the pattern recognition algorithm to verify the feasibility and reliability of the whole platform in the optimization of control parameters.The main conclusions of this study were as follows:1)The hardware system was designed from the gas sensor array,gas transmission pipeline and bionic olfaction measurement and control unit.The Lab VIEW graphical programming environment was taken as the development platform of this software system to write the human-computer interaction control interface of this platform,the main functions of the system included system parameter setting,real-time data acquisition and storage,feature parameters extraction.The pattern recognition was off-line processed by MATLAB software.2)The uniformity and stability of the linear arrangement,parallel arrangement and multi-nasal duct detection chamber was analyzed by using Fluent software,it was concluded that the structure of multi-nasal duct detection chamber had advantages in velocity uniformity and stability,the speed deviation ratio and coefficient of non-uniformity ranged from 0.712 to 0.831 and from 0.253 to 0.292,respectively.The airflow velocity had a smooth area in the axial direction of the pipeline 0.035-0.049 m,which was the ideal location for installing sensor.3)The optimized bionic olfactory control parameters:the heating voltage of sensor was 5.0 V,the relative humidity of gas was 30%,the flow was 100 mL/min,at this time,the platform could get a better performance,the sensibility of the sensor array was maximum,which ranged from 2.853 to 7.559,the response and recovery time were minimum,which ranged from 35 to 45 s and from 30 to 50 s,respectively.4)The different treatments had significant effects on volatile compounds in citrus fruits(P<0.05),the classification and prediction effects of the model established by relative stability value were better than that of integral value.The correct recognition rate of LDA prediction model was 96.43%for citrus with different pest infestation time;After sensor array was optimized by Loadings method,the correct recognition rate for citrus with different pest infestation time was 98.21%by the LDA prediction model,which was 1.78%higher than that of the non-optimized sensor array,and,the non-infestation,the invasion and incubation period of citrus could be distinguished well by LDA model,there was no overlap between sample data points;According to the volatile components of citrus tested by GC-MS,the relative contents of the 3-thujene,?-pinene,sabinene,caprinaldehyde,?-terpineol,trans-d-Limonene oxide decreased with increase in pest infestation degree,while the relative contents of other olefins,aldehydes and alcohols increased with increase in pest infestation degree,the results of GC-MS further confirmed the availability of the bionic olfactory detection and analysis platform for pest-infested samples detection.In summary,the developed bionic olfactory detection and analysis platform can basically meet the test requirements in function,and,the platform has a certain feasibility in the detection of citrus pests,which can provide certain data reference and technical support for the further research on nondestructive testing of agricultural products after pest infestation.