Vapor identification system using quartz resonator sensor array and support vector machine

It has been developed a vapor identification system using gas sensor array and Support Vector Machine (SVM) pattern recognition. Sensor array consists of several quartz resonator sensors coated with different polymer materials in order to have a specific pattern to the vapor. In this study, the Field Programmable Gate Array was used as counters and other functions to interface the sensor array with a computer. Frequency change was measured by a counter with a period of one second. Vapors used in the experiment were kerosene, methanol, gasoline and alcohol. The data analysis was taken from the frequency changes after vapor injection. Sensors were cleaned to get the initial condition using nitrogen gas. For vapor data collecting, the measurements were performed eight times for each sample. The set of digital data was then stored as a database. Principle Component Analysis was used to visualize the performance of the sensor array to discriminate each vapor. The set of vapor pattern obtained by the sensor array was then identified by SVM algorithm. Experiment results showed that the SVM could identify each vapor with a success rate of 97.2%. The results of this study will be used for further research to detect the low concentrations of vapors contained in human breath for medical diagnoses.