Background Interference Elimination in Wound Infection Detection by Electronic Nose Based on Reference Vector-based Independent Component Analysis

Calibration Update and Drift Correction for Electronic Noses and Tongues

One of the obstacles to the wider practical use of the multisensor systems for gas and liquid analysis-electronic noses and tongues, is the limited temporal validity of the multivariate calibration models. Frequent recalibration of multisensor systems is often excessively costly and time consuming due to the large number of necessary reference sample and their limited availability. There are several circumstances that can invalidate multivariate calibration model. The most common problem in the case of sensor systems is temporarily drift or gradual change of sensor characteristics occurring during sensor exploitation. Another common situation is a change in the composition of the analyzed samples that also alters sensor response due to the matrix effects. Finally, a necessity to replace sensors in the array or to transfer calibration model from one sensor set or one type of sensors to the other can arise. As an alternative to the recalibration of the sensor system using full set of calibration samples, drift correction and calibration update has been proposed. The main approaches can be summarized as follows: Drift correction that consists in modeling sensor temporarily drift or drift direction using a series of measurements and then using it for correcting new data.Calibration standardization that aims to correct new measured data by eliminating new variation. For this purpose, a relationship between two experimental conditions is established using a reduced set of samples measured at both conditions (standardization subset).Calibration update that consists in incorporation of new sources of variance in the calibration model by recalculating it using initial calibration samples and reduced set of samples measured at new conditions. The latter can be either standard or unknown samples. This paper presents an overview of different methods reported for the drift correction and calibration update of the electronic noses and tongue and discussion of the practical aspects of their implementation.

Study on Interference Suppression Algorithms for Electronic Noses: A Review

Electronic noses (e-nose) are composed of an appropriate pattern recognition system and a gas sensor array with a certain degree of specificity and broad spectrum characteristics. The gas sensors have their own shortcomings of being highly sensitive to interferences which has an impact on the detection of target gases. When there are interferences, the performance of the e-nose will deteriorate. Therefore, it is urgent to study interference suppression techniques for e-noses. This paper summarizes the sources of interferences and reviews the advances made in recent years in interference suppression for e-noses. According to the factors which cause interference, interferences can be classified into two types: interference caused by changes of operating conditions and interference caused by hardware failures. The existing suppression methods were summarized and analyzed from these two aspects. Since the interferences of e-noses are uncertain and unstable, it can be found that some nonlinear methods have good effects for interference suppression, such as methods based on transfer learning, adaptive methods, etc.

A correlated information removing based interference suppression technique in electronic nose for detection of bacteria

A sensor array with 30 gas sensors is used in the electronic nose (e-nose) for bacteria detection in wound infection. However, the interference is an urgent problem in e-nose, since it would impact on the detection of target due to the cross-sensitivity of gas sensors, especially the background interference caused by carrier gas. The related methods to suppress the background interference are independent component analysis and orthogonal signal correction algorithm which are unreasonable, because it is difficult to obtain the so-called reference vector in complex real-world scenario. Consider that the sampling process of pump suction is divided into three parts: baseline collecting, sample collecting and system purging. In the case of stabilized carrier gas, the information in baseline can be fully used to suppress the interference in sampling stage. Thus a novel and effective correlated information removing based interference suppression (CIRIS) method is proposed. Specifically, the principle of this method is to suppress the interference of the sampling stage by removing the information correlated with baseline samples. Experimental results show that the proposed method (CIRIS with principal component analysis used to calculate the projection matrix) is significantly effective for interference suppression in e-nose.

Suppression of Strong Background Interference on E-Nose Sensors in an Open Country Environment

The feature extraction technique for an electronic nose (e-nose) applied in tobacco smell detection in an open country/outdoor environment with periodic background strong interference is studied in this paper. Principal component analysis (PCA), Independent component analysis (ICA), re-filtering and a priori knowledge are combined to separate and suppress background interference on the e-nose. By the coefficient of multiple correlation (CMC), it can be verified that a better separation of environmental temperature, humidity, and atmospheric pressure variation related background interference factors can be obtained with ICA. By re-filtering according to the on-site interference characteristics a composite smell curve was obtained which is more related to true smell information based on the tobacco curer’s experience.