Electronic Noses Using Quantitative Artificial Neural Networ

The present paper covers a new type of electronic nose(e-nose) with a four-sensor array,which has been applied to detecting gases quantitatively in the presence of interference. This e-nose has adapted fundamental aspects of relative error(RE) in changing quantitative analysis into the artificial neural network (ANN).. Thus, both the quantitative and the qualitative requirements for ANN in implementing e-nose can be satisfied. In addition, the e-nose uses only 4 sensors in the sensor array, and can be designed for different usages simply by changing one or two sensor(s). Various gases were tested by this kind of e-nose, including alcohol vapor, CO, liquefied-petrol-gas and CO2. Satisfactory quantitative results were obtained and no qualitative mistake in prediction was observed for the samples being mixed with interference gases.     

用于测定空气中甲醛的电子鼻

制作了可定量检测空气中甲醛的便携式电子鼻.该电子鼻由传感器阵列、信号调理电路、模式识别系统以及显示系统等4个部分组成,其中传感器阵列为4个半导体金属氧化物传感器.模式识别系统采用模糊神经网络算法.便携式甲醛电子鼻对甲醛气体响应专一,抗干扰能力强,且定量结果精确,可用于甲醛气体的现场检测.对于0.001～0.25mg/L浓度范围内的甲醛气体,电子鼻定量测报的正确率达到81.3%;对于干扰气体存在下的甲醛气体,未出现错误测报.     

Virtual sensor array as a tool for classifying air pollution

This work examined the acquisition of information about gases using a virtual sensor array and classification. We were particularly interested in the approach in which classes are defined in a qualitative–quantitative manner, that is, by identifying the gas and concentration range. This type of information will be of interest for air pollution assessment purposes. In this field of application, it is often not necessary to provide very precise information. The idea of the virtual sensor array exploits the dependence of a gas sensor’s response on operating conditions. Originally it was developed as a means to improve the selectivity of an electronic nose when energy consumption by this device was a serious limitation. If the response of one sensor is measured in n time points, and each time point is characterised by different controlled exposure conditions, the sensor becomes analogous to an n-dimensional virtual sensor array. Compared with conventional approaches, a virtual sensor array based on a single sensor offers low power consumption, low volume, and low cost, which opens up new markets for wide application of portable and handheld devices. In this article, we show that a virtual sensor array and classification may serve as a reliable source of qualitative–quantitative information about gases. Twenty-six classes (five substances, each at five concentration ranges, and pure air) were recognised with a true positive rate higher than 99.14 ± 0.49% and a true negative rate higher than 99.21 ± 0.52%. As demonstrated, the basis for recognition could be a virtual sensor array associated with a low-power consuming sensor (210–280 mW). The complexity of the applied classifier could be adjusted depending on the choice of sensor operating conditions. For a complex classifier like support vector machine, dynamic exposure was sufficient to obtain high classification performance. A simpler classifier like k-nearest neighbours required more information, that is, information associated with static as well as dynamic exposure.     

Gas sensor array based on metal-decorated carbon nanotubes

Here we demonstrate design, fabrication, and testing of electronic sensor array based on single-walled carbon nanotubes (SWNTs). Multiple sensor elements consisting of isolated networks of SWNTs were integrated into Si chips by chemical vapor deposition (CVD) and photolithography processes. For chemical selectivity, SWNTs were decorated with metal nanoparticles. The differences in catalytic activity of 18 catalytic metals for detection of H(2), CH(4), CO, and H(2)S gases were observed. Furthermore, a sensor array was fabricated by site-selective electroplating of Pd, Pt, Rh, and Au metals on isolated SWNT networks located on a single chip. The resulting electronic sensor array, which was comprised of several functional SWNT network sensors, was exposed to a randomized series of toxic/combustible gases. Electronic responses of all sensor elements were recorded and the sensor array data was analyzed using pattern-recognition analysis tools. Applications of these small-size, low-power, electronic sensor arrays are in the detection and identification of toxic/combustible gases for personal safety and air pollution monitoring.     

VOCs分子的半导体型传感器识别检测研究进展

具有体积小、功耗低、灵敏度高、硅工艺兼容性好等优点的金属氧化物半导体(MOS)气体传感器现已广泛地应用于军事、科研和国民经济的各个领域。然而MOS传感器的低选择性阻碍了其在物联网(IoT)时代的应用前景。为此,本文综述了解决MOS传感器选择性的研究进展,主要介绍了敏感材料性能提升、电子鼻和热调制三种改善MOS传感器选择性的技术方法,阐述了三种方法目前所存在的问题及其未来的发展趋势。同时,本文还对比介绍了机器嗅觉领域主流的主成分分析(PCA)、线性判别分析(LDA)和神经网络(NN)模式识别/机器学习算法。最后,本综述展望了具有数据降维、特征提取和鲁棒性识别分类性能的卷积神经网络(CNN)深度学习算法在气体识别领域的应用前景。基于敏感材料性能的提升、多种调制手段与阵列技术的结合以及人工智能(AI)领域深度学习算法的最新进展,将会极大地增强非选择性MOS传感器的挥发性有机化合物(VOCs)分子识别能力。     

Quantitative fuzzy neural network for analytical determination

A quantitative fuzzy neural network (Q-FNN) for pattern recognition in analytical determination is reported in this paper. The fuzzy neural network (FNN) combines a fuzzy logic system with an artificial neural network (ANN) so that it has both advantages of a high training speed and strong anti-interference. Importantly, the analytical concept of relative error (RE) in quantitative determination has been integrated into FNN so that the Q-FNN provides a very good quantitative capability in chemical analysis, and prevents the system from an over-fitting problem. The logarithm curve with noise in terms of analytical response versus concentration is calibrated by trained FNN and a close approximation to the ideal one without noise is obtained. The Q-FNN has been applied to the concentration determination of freon in the presence of interference gases. The prediction error for a test set in quantification is less than 10% while no qualitative mistake is observed, implying that the quantitative FNN has sustained the feature of pattern recognition. The results indicate that the Q-FNN has obvious advantages not only in converging speed, but also in the quantitative accuracy over the ANN.     

Quantitative determination of Freon gas using an electronic nose

This paper reports a quantitative electronic nose (enose) for the quantitative determination of Freon gas within the concentration range 0-1000 ppm in the presence of interfering gases such as water, lubricant and petrol vapours. This quantitative enose is a new type of Freon detection system, composed of an array of four sensors. The artificial neural network (ANN) and fuzzy logic type of ANN (FNN), in combination with the relative error concept in analytical chemistry, are integrated for both quantification and discrimination. The predicted results are satisfied with a pass rate of > 80% within the permitted relative errors. The results show that the Freon enose developed in this study is reliable for both the qualitative and quantitative determination of Freon gas and exhibits the merits of high sensitivity, anti-interference and accuracy.     

Quantitative Analysis of Active Pharmaceutical Ingredients (APIs) Using a Potentiometric Electronic Tongue in a SIA Flow System

An advanced potentiometric electronic tongue and Sequential Injection Analysis (SIA) measurement system was applied for the quantitative analysis of mixtures containing three active pharmaceutical ingredients (APIs): acetaminophen, ascorbic acid and acetylsalicylic acid, in the presence of various amounts of caffeine as interferent. The flow‐through sensor array was composed of miniaturized classical ion‐selective electrodes based on plasticized PVC membranes containing only ion exchangers. Partial Least Squares (PLS) analysis of the steady‐state sensor array responses, measured in API mixtures prepared by the SIA system permitted a correct quantitative analysis of acetylsalicylic acid and ascorbic acid. Further optimization using multiway PLS fed by dynamic responses without additional feature extraction did not improve significantly the resolution of acetaminophen. Lastly, the chemometric treatment, involving the extraction of dynamic components of the transient response employing the Wavelet transform, the removal of less‐significant coefficients by means of Causal Index pruning and training of an Artificial Neural Network (ANN) with the selected coefficients, allowed the simultaneous determination of all the three studied APIs, while counterbalancing any interference due to caffeine.     

Development of an analytical microsystem for organic gas detection based on surface acoustic wave resonators

Modified Surface Acoustic Wave (SAW) devices can be used as highly sensitive detectors for gases: coated with a selective sorptive layer, they serve as a frequency determining element of an oscillator circuit. To perform a quantitative analysis of organic gas mixtures with polymer coatings as sensitive membranes, it is necessary to use an array of several SAW sensors with sensitivities towards various gaseous components and a subsequent pattern recognition of the sensor responses. An automatic working, self purging and compact analytical microsystem has been developed for organic gas detection. As the central component it contains an array of nine commercially available SAW resonators working at frequency of 433.92 MHz. Eight sensor oscillators, one common reference oscillator for temperature compensation and circuits for frequency mixing are mounted within a compact radial housing (diameter 108 mm). A selection of polymers has been made to detect a broad variety of gaseous organic analytes. Of particular interest have been hydrocarbons, halogenated hydrocarbons and aromatic compounds. Initial experiments with these analytes have been carried out to proof the selection and to optimize the combination of polymer coatings. With the compact sensor configuration a very fast response time in the range of seconds is obtained for qualitative determination of the analytes.Dedicated to Professor Dr. Dr. h.c. mult. J. F. K. Huber on the occasion of his 70th birthday     

Colorimetric sensor array–smartphone–remote server coupling system for rapid detection of saccharides in beverages

A kind of “colorimetric sensor array–smartphone–remote server” coupling system was constructed for rapid on-site testing of saccharides. First, the binding capacity between saccharides and boric acid compounds (boric acid, phenylboronic acid and 3-nitrophenylboronic acid) was studied. The binding capacity of 3-nitrophenylboronic acid was found to be the highest, followed by phenylboronic acid and boric acid. Then a small-scale colorimetric sensor array (2 × 2) of pH indicator based on affinity interaction between 3-nitrophenylboronic acid and saccharides was developed to detect 19 kinds of saccharides. A camera phone was used to acquire the array images before and after reaction, then the self-developed color discrimination software in smartphone was applied to process pictures in order to obtain the color difference image and data of analytes. The color difference data were analyzed by several methods, including principal component analysis, hierarchical cluster analysis and linear discriminant analysis. The analysis results showed that the sensor array (2 × 2) established in this paper has great discriminative capability for 19 kinds of saccharides, and the classification accuracy is as high as 100%. Nineteen different quantitative models of saccharides that showed high accuracy and precision were established based on partial least-square method. Finally, the smartphone was connected to a remote server on which the qualitative analysis and quantitative analysis models for analytes had been established. The color difference data obtained by the smartphone were uploaded to the remote server for the qualitative analysis and quantitative analysis of saccharides. The effectiveness of the “colorimetric sensor array–smartphone–remote server” coupling system in rapid on-site detection of saccharides was further verified by the spike and recovery experiments. The qualitative analysis results showed that this coupling system could distinguish all of the analytes without a mistake, and the quantitative analysis results showed that the predicted values for saccharides were close to the real values.     

E-Nose Discrimination of Abnormal Fermentations in Spanish-Style Green Olives

Current legislation in Spain indicates that table olives must be free of off-odors and off-flavors and without symptoms of ongoing alteration or abnormal fermentations. In this regard, the International Olive Council (IOC) has developed a protocol for the sensory classification of table olives according to the intensity of the predominantly perceived defect (PPD). An electronic nose (e-nose) was used to assess the abnormal fermentation defects of Spanish-style table olives that were previously classified by a tasting panel according to the IOC protocol, namely zapateria, butyric, putrid, and musty or humidity. When olives with different defects were mixed, the putrid defect had the greatest sensory impact on the others, while the butyric defect had the least sensory dominance. A total of 49 volatile compounds were identified by gas chromatography, and each defect was characterized by a specific profile. The e-nose data were analyzed using principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA). The different defects were clearly separated from each other and from the control treatment, independently of PPD intensity. Moreover, the e-nose differentiated control olives from table olives with combined sensory defects despite the dilution effect resulting from the combination. These results demonstrate that e-nose can be used as an olfactory sensor for the organoleptic classification of table olives and can successfully support the tasting panel.     

Voltammetric Electronic Tongue Based on Carbon Paste Electrodes Modified with Biochar for Phenolic Compounds Stripping Detection

Biochar is a charcoal produced from the biomass pyrolysis process that presents a highly porous and functionalized surface. In the present work an array of carbon paste electrodes (CPE) made of different forms of carbon (graphite, carbon nanotubes and activated biochar) was evaluated in the development of an electronic tongue for discrimination and stripping voltammetric determination of catechol (CAT), 4‐ethylcatechol (4‐EC) and 4‐ethylguaiacol (4‐EG) phenolic compounds. Morphological characterization of carbon materials and electrodes surfaces was performed by scanning electron microscopy (SEM) and semi‐quantitative elemental composition by energy dispersive spectroscopy (EDS). Electrochemical Impedance Spectroscopy (EIS) measurements were used for electrochemical characterization of electrodes. Cyclic voltammetry measurements were performed for the phenolic compounds evaluated using different concentrations. Principal component analysis (PCA) was performed to evaluate the qualitative analysis. Quantitative data modeling was done using artificial neural networks (ANN). The proposed sensor array presented analytical potentiality allowing the distinction and determination of CAT, 4‐EC and 4‐EG by using chemometric processing. The method showed sensibility, reproducibility and a good linearity (R²>0.9940) for three compounds evaluated. Spontaneous preconcentration of three compounds was possible using all three sensors, which can allow the application of these as passive samplers for remote determinations of phenolic compounds in wine and food samples.     

A miniature electronic nose system based on an MWNT–polymer microsensor array and a low-power signal-processing chip

This article introduces a power-efficient, miniature electronic nose (e-nose) system. The e-nose system primarily comprises two self-developed chips, a multiple-walled carbon nanotube (MWNT)–polymer based microsensor array, and a low-power signal-processing chip. The microsensor array was fabricated on a silicon wafer by using standard photolithography technology. The microsensor array comprised eight interdigitated electrodes surrounded by SU-8 “walls,” which restrained the material–solvent liquid in a defined area of 650?×?760 μm². To achieve a reliable sensor-manufacturing process, we used a two-layer deposition method, coating the MWNTs and polymer film as the first and second layers, respectively. The low-power signal-processing chip included array data acquisition circuits and a signal-processing core. The MWNT–polymer microsensor array can directly connect with array data acquisition circuits, which comprise sensor interface circuitry and an analog-to-digital converter; the signal-processing core consists of memory and a microprocessor. The core executes the program, classifying the odor data received from the array data acquisition circuits. The low-power signal-processing chip was designed and fabricated using the Taiwan Semiconductor Manufacturing Company 0.18-μm 1P6M standard complementary metal oxide semiconductor process. The chip consumes only 1.05 mW of power at supply voltages of 1 and 1.8 V for the array data acquisition circuits and the signal-processing core, respectively. The miniature e-nose system, which used a microsensor array, a low-power signal-processing chip, and an embedded k-nearest-neighbor-based pattern recognition algorithm, was developed as a prototype that successfully recognized the complex odors of tincture, sorghum wine, sake, whisky, and vodka.

Application of an artificial neural network (ANN) and piezoelectric chemical sensor array for identification of volatile organic compounds

A piezoelectric chemical sensor array was developed using four quartz crystals. Gas chromatographic stationary phases were used as sensing materials and the array was connected to an artificial neural network (ANN). The application of the ANN method proved to be particularly advantageous if the measured property (mass, concentration, etc.) should not be connected exactly to the signal of the transducers of the piezoelectric sensor. The optimum structure of neural network was determined by a trial and error method. Different structures were tried with several neurons in the hidden layer and the total error was calculated. The optimum values of primary weight factors, learning rate (η=0.15), momentum term (μ=0.9), and the sigmoid parameter (β=1) were determined. Finally, three hidden neurons and 900 training cycles were applied. After the teaching process the network was used for identification of taught analytes (acetone, benzene, chloroform, pentane). Mixtures of organic compounds were also analysed and the ANN method proved to be a reliable way of differentiating the sensing materials and identifying the volatile compounds.     

Application of an artificial neural network (ANN) and piezoelectric chemical sensor array for identification of volatile organic compounds

A piezoelectric chemical sensor array was developed using four quartz crystals. Gas chromatographic stationary phases were used as sensing materials and the array was connected to an artificial neural network (ANN). The application of the ANN method proved to be particularly advantageous if the measured property (mass, concentration, etc.) should not be connected exactly to the signal of the transducers of the piezoelectric sensor. The optimum structure of neural network was determined by a trial and error method. Different structures were tried with several neurons in the hidden layer and the total error was calculated. The optimum values of primary weight factors, learning rate (η=0.15), momentum term (μ=0.9), and the sigmoid parameter (β=1) were determined. Finally, three hidden neurons and 900 training cycles were applied. After the teaching process the network was used for identification of taught analytes (acetone, benzene, chloroform, pentane). Mixtures of organic compounds were also analysed and the ANN method proved to be a reliable way of differentiating the sensing materials and identifying the volatile compounds.     

提取多维催化发光信号鉴别有害气体的传感器

基于2种催化敏感材料, 设计了通过提取多维发光信号鉴别有害气体的催化发光(CTL)传感器. 在最佳检测条件下, 18种有害气体依次经过纳米Al₂O₃(或MgO)表面进行CTL反应, 产生的CTL 响应信号组成其特征图谱. 通过主成分分析法(PCA)鉴别了各种气体. 采用线性判别分析(LDA)对浓度分别为100, 300及500 mL/m³的18种气体的识别正确率均为100%. 该传感器可以同时测量气体的浓度, 检出限均低于我国工作场所有害因素职业接触限值(GBZ2.1-2007), 空气中几种常见的污染物共存不影响这些有害气体的检测. 该传感器具有传感元件少、 稳定性好、 操作方便、 信息丰富和识别能力强等优点, 可用于发展微型实用的传感器.     

A comparison of different chemometrics approaches for the robust classification of electronic nose data

Accurate detection of certain chemical vapours is important, as these may be diagnostic for the presence of weapons, drugs of misuse or disease. In order to achieve this, chemical sensors could be deployed remotely. However, the readout from such sensors is a multivariate pattern, and this needs to be interpreted robustly using powerful supervised learning methods. Therefore, in this study, we compared the classification accuracy of four pattern recognition algorithms which include linear discriminant analysis (LDA), partial least squares-discriminant analysis (PLS-DA), random forests (RF) and support vector machines (SVM) which employed four different kernels. For this purpose, we have used electronic nose (e-nose) sensor data (Wedge et al., Sensors Actuators B Chem 143:365–372, 2009). In order to allow direct comparison between our four different algorithms, we employed two model validation procedures based on either 10-fold cross-validation or bootstrapping. The results show that LDA (91.56 % accuracy) and SVM with a polynomial kernel (91.66 % accuracy) were very effective at analysing these e-nose data. These two models gave superior prediction accuracy, sensitivity and specificity in comparison to the other techniques employed. With respect to the e-nose sensor data studied here, our findings recommend that SVM with a polynomial kernel should be favoured as a classification method over the other statistical models that we assessed. SVM with non-linear kernels have the advantage that they can be used for classifying non-linear as well as linear mapping from analytical data space to multi-group classifications and would thus be a suitable algorithm for the analysis of most e-nose sensor data.     

烃类混合气体的神经网络模型检测

八十年代末科学家模仿生物鼻研制一种传感器阵列与计算机模式识别的气体检测系统．传感器阵列相当于生物鼻的嗅觉细胞，计算机模式识别系统相当于嗅泡和大脑「‘］．传感器阵列对气体的响应是一个多维空间的响应模式，这种响应模式经过一定的数学处理后可以实现气体的种类识别或浓度检测［’－‘j．传感器的响应和混合气体浓度之间呈非线性关系，这一特性给定量检测多组分气体混合物造成很大的限制．应用人工神经元网络技术（ANN）可以克服这一缺陷，并使检测气体的选择性大大提高．本工作运用ANN中的反向传播（BP）算法识别由16个不同…     

Bi‐channel Surface Acoustic Wave Gas Sensor for Carbon Disulfide and Methanol Vapors in Polymer Plants

A C₆₀‐polyphenylacetylene (C₆₀‐PPA) and polyvinylpyrrolidone (PVP) coated two‐channel surface acoustic wave (SAW) crystal gas sensor with a homemade computer interface for data acquisition and data processing was developed and employed to detect carbon disulfide (CS₂) and methanol (CH₃OH) vapors in polymer plants. The frequency of surface acoustic wave oscillator decreases due to the adsorption of gas molecules on the coated materials of the SAW sensor. Six coating materials (C₆₀‐PPA, nafion, PPA, crytand [2,2], polyethene glycol and PVP) were used to adsorb and detect carbon disulfide and methanol gases. Adsorption of all the six coating materials to CS₂ and CH₃OH was found to be physical adsorption. The C₆₀‐PPA coated SAW detector exhibited more sensitive to CS₂ than the other coating materials. In contrast, the PVP coated SAW detector was more sensitive to CH₃OH than the other coating materials. With the two‐channel SAW sensor, the C₆₀‐PPA coated SAW showed a good detection limit of 0.4 ppm and good reproducibility with RSD of 3.37 % (n=10) for CS₂. Similarly, the PVP coated SAW also showed a good detection limit of 0.05 ppm and good reproducibility, with RSD of 0.86 % (n=10) for CH₃OH. The interference effect of other organic molecules on the SAW detection system was negligible, except for the irreversible adsorption of C₆₀‐PPA to propylamine. The frequency signals from the two‐channel SAW sensor array C₆₀‐PPA and PVP coatings were processed by a back‐propagation artificial neural network (BPN) and multiple regression analysis (MRA). Thus a two‐channel SAW sensor array with BPN and MRA has been successfully applied for the qualitative and quantitative analyses of CS₂ and CH₃OH in mixtures.