Development of a PCA-Based Drift Compensation Algorithm for a Gas Sensor Array

• Main Authors: Jhang, Jia Yin, 張家寅
• Other Authors: Tang, Kea Tiong
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/64832845547330327312

碩士 === 國立清華大學 === 電機工程學系 === 104 === Recently, gas sensing system has been widely used in many applications, including environmental monitoring, food quality management, and medical diagnosis. However, gas sensing systems are still not popular for these applications, restricted by sensor drift problem usually caused by environmental factor, such as humidity, temperature, air pollution, and so on. Therefore, it is necessary to develop a drift calibration method in gas sensing system. In this thesis, we develop a simple and low-cost gas sensing system with humidity and temperature control. We can control the environmental condition by this system for most portable e-nose system. In this system, the range of relative humidity (RH) is controlled from 10% to 80%； the range of temperature is controlled from room temperature up to 100°C. With this environment controlled system, we can test the environment parameters for most gas sensing applications. By observing the response of humidity and gas sensor, we find the humidity factor of gas sensor arrays. By using a linear calibration method, we developed a PCA-based drift calibration algorithm for the drift effect caused by humidity variation of the methanol and ethanol gas data collected under the different environment conditions. Our calibration result performs better than Linear Discriminant Analysis (LDA) algorithm in both classification accuracy and Fisher Class Separability. Besides, our algorithm also has a better grouping result than the traditional calibration algorithm such as Independent Component Analysis and Neural Network.