A New Compensation Method for Sensor-Drift Effect Based on the Cross-Correntropy Concept
Namyong Kim, Hyung‐Gi Byun, Krishna C. Persaud · Sensor Letters · 2011
Cross-correntropy which has the characteristics of correlation and entropy can be a generalized similarity measure of two probability density functions (PDFs). In this paper, we apply the cross-correntropy concept to sensor-signal processing for sensor drift compensation by maximizing the correntropy between the Parzen PDF of test-phase system output and the target pattern. The gas sensing system uses a conducting polymer sensor array and signal processing unit equipped with radial basis function network (RBFN). The RBFN weights are readjusted in the test phase by the proposed method for sensor drift compensation. Copyright © 2011 American Scientific Publishers. All rights reserved.