Development Of Cloud Electronic Nose Platform And Its Application In The Detection Of Rice Wine Marked Age

As a rapid and non-destructive technology,electronic nose(E-nose)has been widely applied in the quality detection of agricultural products and foods.To meet different needs,E-nose system is developed in a direction of portability and networking.Related researches were mostly based on embedded technology and wireless communication technology to develop E-nose system,which had deficiencies in operability,versatility,and extensibility on account of the limitations in MCU performance and system architecture.In this work,an E-nose system based on smartphone and cloud server was developed.The main work contents were as follows:(1)An cloud E-nose platform was designed,which includes E-nose device,E-nose client and E-nose server.The E-nose device was responsible for data acquisition and transmission,and E-nose client provided user interaction interface,and the E-nose cloud server took charge to complete data storage and calculation.(2)E-nose device was developed on the basis of modular design,including main-control module,network communication module,sensor array,sampling circuit,air chamber,gas path,display,button and power module.Gas circuit structure was optimized by separating sampling gas path and cleaning gas path to make sure dynamic steady state of gas circularly flowing through the air chamber.Moreover,the manual detection and automatic monitoring were both achieved.(3)E-nose cloud server was developed.E-nose database was designed based on the analysis of data models in E-nose technic field.Software on cloud server was developed by PHP for logging in,inquiring and uploading.The data processing were accomplished by Python.(4)An Android App was developed as E-nose client,by which the E-nose device could be searched,connected,real-time controlled and displayed in LAN.Also,by interacting with E-nose Cloud Server,the E-nose Client App could provide functions such as creating experiment,inquiring,uploading and model computing.(5)Experiments were carried out with rice wine.The original features in both sampling and cleaning phase were extracted.Stability and variance analysis were conducted to select features subset.Logistic regression and support vector machine were applied to build classification models to identify the marked age of rice wine.These models were saved in E-nose cloud server to accomplish rapid and real-time detection for unknown samples.