Optimization Method Of Gas Sensor Array For Tea Aroma Based On Electronic Nose

Electronic nose is an advanced instrument for detecting odor.It can be used to make a judgment on one or more odor through the gas sensor.At present,It has been widely used in process monitoring and quality identification of foodstuff and drugs.Because of the complex components of tea aroma,although many people have made a lot of attempts at the algorithm,the detection effect is still unsatisfactory.The key to solve the problem is determining the experimental parameters and to optimize the gas sensor array for the detection of tea aroma.In this paper,determined the method of rejecting redundant gas sensors based on correlation coefficient and distinguishing performance.And proposed the method of optimizing gas sensor array based on clustering analysis and distinguishing performance.Gas sensor arrays were optimized for green tea(fried green tea,roasted green tea),black tea and oolong tea.STM32 is the core processor of the electronic nose,and the pattern recognition algorithm of PCA and LDA is fused in the software.The median-value filter is used to improve the accuracy of detecting data.In addition,determined the optimum experimental parameters for detecting tea aroma.Based on the components of tea aroma,15 metal oxide gas sensors were selected,and the redundant sensors were selected by correlation analysis.One of the redundant sensor is eliminated,whose distinguishing performance is poor.The method of optimizing sensor array based on clustering analysis and discriminating performance is proposed.The independence of each sensor is determined by clustering analysis.When we choose gas sensor arrays,start by selecting the sensor with the highest distinguishing performance.And select gas sensors with large independence step by step.Finally,the best gas sensor array is optimized.Based on eliminating redundant gas sensors and optimizing best sensor array,gas sensor arrays for green tea,black tea,oolong tea,fried green tea and baked green tea were screened.Though the method of random forest and the basic pattern recognition,proved that using gas sensor array to detect different tea aroma,the accuracy can reach more than 98%.Besides,the interactive experiment proved the gas sensor array with higher matching degree with the detection object,the detection accuracy is better.