Abstract
The drift of electronic nose (e-nose) sensors is the most challenging problem. An effective calibration method for e-nose is proposed in this article, which is based on projection on to convex sets (POCS) and extreme learning machine (ELM) method, including an automatic sampling platform and an e-nose sensor calibration model. It realizes a unified and universal calibration for both recognition and regression application to solve the electronic nose drift problem under long-term working conditions. Isopropanol and acetone gases that respond to most sensors were chosen as training gases for the model. Experiments show the effectiveness of the proposed e-nose calibration method.
Funding Information
  • Natural Science Foundation of China (62171066)
  • Chongqing Science and Technology Bureau (cstc2018jscx-mszdX0077, cstc2019jscx-gksb0481)

This publication has 28 references indexed in Scilit: