Embedded neural network for fire classification using an array of gas sensors

2008 ◽  
Author(s):  
Shishir Bashyal ◽  
Ganesh Kumar Venayagamoorthy ◽  
Bandana Paudel
Keyword(s):  
Neural Network ◽  
Gas Sensors

scholarly journals Electronic Nose Technologies in Monitoring Black Tea Manufacturing Process

2020 ◽  
Vol 2020 ◽  
pp. 1-8
Author(s):  
Tharaga Sharmilan ◽  
Iresha Premarathne ◽  
Indika Wanniarachchi ◽  
Sandya Kumari ◽  
Dakshika Wanniarachchi
Keyword(s):  
Neural Network ◽  
Electronic Nose ◽  
Gas Sensors ◽  
Black Tea ◽  
Quality Parameters ◽  
Polyphenolic Compounds ◽  
Metal Oxide Semiconductor ◽  
Oxide Semiconductor ◽  
Optimum Level ◽  
Fermentation Stage

“Tea” is a beverage which has a unique taste and aroma. The conventional method of tea manufacturing involves several stages. These are plucking, withering, rolling, fermentation, and finally firing. The quality parameters of tea (color, taste, and aroma) are developed during the fermentation stage where polyphenolic compounds are oxidized when exposed to air. Thus, controlling the fermentation stage will result in more consistent production of quality tea. The level of fermentation is often detected by humans as “first” and “second” noses as two distinct smell peaks appear during fermentation. The detection of the “second” aroma peak at the optimum fermentation is less consistent when decided by humans. Thus, an electronic nose is introduced to find the optimum level of fermentation detecting the variation in the aroma level. In this review, it is found that the systems developed are capable of detecting variation of the aroma level using an array of metal oxide semiconductor (MOS) gas sensors using different statistical and neural network techniques (SVD, 2-NM, MDM, PCA, SVM, RBF, SOM, PNN, and Recurrent Elman) successfully.

Methods to Improve the Accuracy of Urease Respiratory Diagnosis of Helicobacter Pylori

2015 ◽  
Vol 22 (3) ◽  
pp. 86-92
Author(s):  
Ясовеев ◽  
V. Yasoveev ◽  
Матанцев ◽  
A. Matantsev ◽  
Уразбахтина ◽  
...  
Keyword(s):  
Neural Network ◽  
Gas Sensors ◽  
Gas Mixture ◽  
Network Technology ◽  
Electronic Database ◽  
Cross Sensitivity ◽  
Diagnosis Method ◽  
H Pylori ◽  
Mixed Gases ◽  
Neural Network Technology

This article describes existing methods of H. pylori diseases diagnosis, procedures of interpreting acquired results and the diagnosis method with the use of semi-conductor catalytic gas sensors combined into the system. Gas sensors have cross-sensitivity to various gases in addition for which they are designed. The use of several sensors allows to reduce the influence of mixed gases. This method is especially useful within medical entities, where air inside can contain alcohol or chloramine vapors. Sensors are selected in the way to overlap main sensor´s cross-sensitivity zone to the maximum extent possible. This is how mixed gases´ influence on the main sensor is compensated. The proposed system uses methods of artificial neural network technology, which allows to enhance system´s stability in changing gas mixture. Due to microcontroller driven calculations, the system can automatically provide data processing. The proposed system can reduce the influence of factors that contribute uncertainty to the measurement result. These results can be transmitted to PC, which one can use to create electronic database or to hold case history.

scholarly journals Electronic Nose-Based Technique for Rapid Detection and Recognition of Moldy Apples

2019 ◽  
Author(s):  
Wenshen Jia ◽  
Gang Liang ◽  
Hui Tian ◽  
Jing Sun ◽  
Cihui Wan
Keyword(s):  
Neural Network ◽  
Pattern Recognition ◽  
Electronic Nose ◽  
Gas Sensors ◽  
Penicillium Expansum ◽  
Support Vector ◽  
Linear Discriminant ◽  
Pattern Recognition Methods ◽  
Group A ◽  
Group B

In this paper, PEN3 electronic nose was used to detect and recognize fresh and moldy apples (inoculated with Penicillium expansum and Aspergillusniger) taken Golden Delicious apples as model subject. Firstly, the apples were divided into two groups: apples only inoculated with different molds (Group A) and mixed apples of inoculated apples with fresh apples (Group B). Then the characteristic gas sensors of the PEN3 electronic nose that were most closely correlated with the flavor information of the moldy apples were optimized and determined, which can simplify the analysis process and improve the accuracy of results. Four pattern recognition methods, including linear discriminant analysis (LDA), backpropagation neural network (BPNN), support vector machines (SVM) and radial basis function neural network (RBFNN), were then applied to analyze the data obtained from the characteristic sensors, respectively, aiming at establishing the prediction model of flavor information and fresh/moldy apples. The results showed that only the gas sensors of W1S, W2S, W5S, W1W and W2W in the PEN3 electronic nose exhibited strong signal response to the flavor information, indicating were most closely correlated with the characteristic flavor of apples and thus the data obtained from these characteristic sensors was used for modeling. The results of the four pattern recognition methods showed that BPNN presented the best prediction performance for the training and validation sets for both the Group A and Group B, with prediction accuracies of 96.29% and 90.00% (Group A), 77.70% and 72.00% (Group B), respectively. Therefore, it first demonstrated that PEN3 electronic nose can not only effectively detect and recognize the fresh and moldy apples, but also can distinguish apples inoculated with different molds.

Sign in / Sign up

Export Citation Format

Share Document