Drift reduction of gas sensor by wavelet and principal component analysis

• DOI: 10.1016/s0925-4005(03)00569-0
• Publication Date: 2003-09-12T17:26:12Z
• Authors: Ding hui; Liu Jun-hua; Shen Zhong-ru
• Keywords: 0103 physical sciences; 01 natural sciences

Abstract This paper mainly deals with the sensor drift in the application of gas concentration measurement, but little has been done in previous works. The algorithm of detecting the drift of sensors presented in this paper is based on the combination of the principal component analysis (PCA) with the wavelet analysis. By this algorithm the sensor drift can be detected online sensitively. For compensating the drift of sensors, an adaptive dynamic drift compensation algorithm (ADDC) based on a drift model is also provided in this paper. From the drift model, the drift compensation factors used to compensate the drifting sensor’s data are calculated. When the drift feature changed, the drift model will be updated online adaptively. In this way, a lifelong efficient drift compensation is made possible for every sensor. The superior performance of this drift counteraction strategy is illustrated with the examples using real semiconductor sensor array data.