{Home} {Publications and Codes} {Research} {Projects} {Professional Activities} {People}

Machine Olfaction, Signal Processing and Pattern Recognition

[Temperature Modulation in Electronic Nose]

Paper:

Xin Yin, Lei Zhang*, Fengchun Tian, and David Zhang, Temperature Modulated Gas Sensing E-Nose System for Low-Cost and Fast Detection, IEEE Sensors Journal, vol. 16, no. 2, pp. 464-474, 2016. [paper]

Abstract:

Constant sensor-heating voltage is commonly used in electronic nose, such that multiple sensors should be integrated as a sensor array in order to differentiate multiple odor analytes. However, supplying a constant heating voltage for each sensor cannot provide rich pattern information, resulting in high cost and weak capability of an E-nose in detection. To address this issue, this paper aims at introducing an optimal temperature modulation technique of gas sensors for achieving low-cost, fast, and accurate detection. The contributions of this paper include: 1) the temperature modulated gas sensing system proposed in this paper operates in a linearly dynamical region by generating a linear control signal waveform of sensors’ heating voltage; 2) with a highly efficient extreme learning machine that follows a random projection-based learning mechanism, the proposed system is developed for simultaneous gas classification and gas concentration prediction; and 3) the optimal heating voltage analysis of gas sensors is explored by using machine learning methods for providing some perspective and insight for optimal heating voltage selection. The experimental results and comparisons in terms of gas classification accuracy, concentration prediction error, system cost, and power consumption significantly demonstrate the high precision and efficacy of our proposed E-nose system.

Approach:

Framework and Platform

Temperature-modulated sensor response curves for [0.03, 0.56, 1.08, 1.20, 1.6, 1.68] ppm formaldehyde (a), [0.05, 0.35, 0.91, 1.25, 1.47, 2.0] ppm Nitrogen dioxide (b) and [0.06, 8, 27, 46, 63, 76] ppm Carbon monoxide (c) gases in one experimental cycle (i.e, 6 samples/1800s).

Experimental Results

Performance analysis with different heating voltage for each sensor based on three algorithms. (a) NN. (b) SVR. (c) ELM.

References:

[1] R. Gosangi and R. Gutierrez-Osuna, “Active temperature modulation of metal-oxide sensors for quantitative analysis of gas mixtures,” Sens. Actuators B, Chem., vol. 185, pp. 201–210, Aug. 2013.

[2] E. Martinelli, D. Polese, A. Catini, A. D’Amico, and C. D. Natale, “Self-adapted temperature modulation in metal-oxide semiconductor gas sensors,” Sens. Actuators B, Chem., vol. 161, no. 1, pp. 534–541, Jan. 2012.

[3] G.-B. Huang, “An insight into extreme learning machines: Random neurons, random features and kernels,” Cognit. Comput., vol. 6, no. 3, pp. 376–390, Sep. 2014.

[4] L. Zhang, F. Tian, C. Kadri, G. Pei, H. Li, and L. Pan, “Gases concentration estimation using heuristics and bio-inspired optimization models for experimental chemical electronic nose,” Sens. Actuators B, Chem., vol. 160, no. 1, pp. 760–770, Dec. 2011.