Search

A sensor array is a key component of an electronic nose (E-nose). However, the practical applications of the E-nose are often inhibited by its size and energy consumption arising from the number of gas sensors. Achieving a high-performance E-nose with a minimum number of sensors is key and challenging for its practical applications. In this study, different machine learning models have been studied and compared to optimize the performance of the E-nose. The results show that the convolutional neural network (CNN) is the best-performing model, which has an accuracy of 0.986 for classification, and an R-square score of 0.979 for concentration prediction, outperforming the gated recurrent unit, long short-term memory, multi-layer perceptron neural network, and support vector machine. The performance of the E-nose experiences only a minor decrease when the number of sensors that participated in pattern recognition is reduced from eight to four, where the CNN model can yield an accuracy of 0.905 for classification and an R-square of 0.972 for concentration prediction. To further quantify the pros and cons of array optimization, a cost-effective metric is designed to reveal the suitable array size under different scenarios. This work can provide valuable guidance in the design of portable E-nose devices with a smaller size and optimal performance. • Different deep learning models are developed and adapted to take in various sizes of the sensor array. • With half size of the original sensor array, no significant decline of performance is observed. • The cost-effectiveness ratio is designed to reveal the most gainful array size under different practical scenarios. [ABSTRACT FROM AUTHOR]

In exploring pattern recognition for electronic noses via deep neural networks, traditional networks encounter key challenges, such as low training efficiency, and neglect of spatial-temporal attributes of gas sensor response sequences. In this study, an unmanned gas-sensing test system is used to generate a large dataset to ensure robust model training. The Gramian angular field-Markov transition field is utilized to convert time sequences into images. Using advanced image processing tools, the images are then subsequently compressed with data augmentation. This method fully preserves temporal and spatial features within the sequences, thus enhancing model performance. Furthermore, the proposed multi-task learning (MTL) framework competently performs simultaneous classification and regression tasks. The primary component of the MTL network, majorly constituted of convolutional neural networks, emphasizes the spatial features of the sequences. The integration of a long short-term memory layer ensures the preservation of temporal feature analysis of the input data, thereby enhancing predictive performance. When images are compressed to only 3.9 % of the original data, substantial information can still be preserved. Subsequently, the model trained by such compressed images attains an accuracy of 95.31 % and an R 2 score of 0.9510 for classification and regression tasks, respectively. Our work reveals the remarkable potential of integrating temporal and spatial features in pattern recognition, promoting the potency of multi-tasking deep learning networks in electronic nose technology. • The GASF, MTF, GADF are used for transforming time series into images, retaining spatial-temporal traits of time series. • This study enhances training, prediction with efficient image compression, preserving vital spatial-temporal details. • A multi-task learning (MTL) model is introduced for gas analysis, predicting gas type and concentration simultaneously. • The LSTM is integrated into CNN layers, enhancing spatial analysis with temporal feature handling in the model. • A large-scale gas-sensing dataset is employed for thorough model training and validation of capabilities. [ABSTRACT FROM AUTHOR]

Recent years have witnessed the splendid performance of deep learning methods used in gas recognition for electronic noses (E-nose). In addition to effective feature extraction, the architecture of the deep neural network plays a vital role. Currently, most applied network structures are hand-crafted by human experts, which is time-consuming and problem-dependent, making it necessary to design the structures of neural networks according to specific demands. In this work, a genetic algorithm with particle swarm optimization (GA-PSO), which possesses promising optimization capabilities, is applied to search for effective deep convolutional neural networks (CNNs) for gas classification based on E-nose technology. A novel image transformation strategy using Gramian angular field and a hybrid cost-saving method is employed in the search process, enabling adaptive and efficient CNN search on gas datasets. With the proposed methods, we can achieve an average classification accuracy of over 90 % on two public gas datasets, while also significantly reducing the model size compared to state-of-the-art CNNs. By using these novel strategies, our approach surpasses random search and basic PSO algorithm in achieving the global optimal solution, higher and more stable accuracy, and faster convergence in pattern recognition using E-nose. Our work suggests that the proposed method can quickly identify excellent CNN structures for E-nose applications. • A new Gramian Angular Field method transforms 1D time series into 2D images. • Kernel decomposition and depthwise separable convolution optimize CNN training. • A fast and robust hybrid algorithm searches for gas-identification models. [ABSTRACT FROM AUTHOR]

• Mesoporous CuO materials with different pore size and specific surface area were synthesized by one-pot hydrothermal method. • The effect of sensing film thickness on the gas-sensing performance was studied. • The correlation between pore size, specific surface area and gas sensing performance was discussed in detail. • The corresponding gas sensing mechanism was proposed. Mesoporous CuO materials with adjustable pore size and high specific surface area were successfully synthesized by a simple and economical one-pot hydrothermal method. The structure and morphology of mesoporous CuO materials were characterized by X-ray powder diffraction, Scanning electron microscope, Electron energy spectrometer, Nitrogen adsorption and desorption, Transmission electron microscope, Fourier infrared spectrometer. The effect of thickness of the sensing film on gas-sensing performance was studied, and it was found that the best sensing performance was shown when the film thickness was about 71 μm. Under the optimal film thickness, the influence of the pore size and specific surface area of the mesoporous CuO materials on the gas-sensing performance was further studied in detail. Analysis of gas sensing results reveled that the mesoporous CuO material with higher specific surface area and suitable pore size distribution showed enhanced acetic acid sensing performance. [ABSTRACT FROM AUTHOR]

The portable electronic nose (E-nose) systems are suffering from the limited computing ability of microcontrollers and can only adopt simple pattern recognition algorithms. The heavy coupling between the sensors greatly limits the anti-fault ability of the system. Herein, a novel ensemble learning framework based on independent artificial neural networks (ANN) is proposed in a portable E-nose system to recognize volatile organic compounds (VOCs). Edge computing is applied to complete data processing and analysis on an ARM Cortex-M3 based microcontroller in an E-nose system. Each sensing unit can recognize VOCs by training independent ANN models. Then an ensemble learning method further organizes the diverse ANN models into an onboard sensor swarm with a significantly enhanced performance. The accuracy of the type classification can reach 81.1%, which is improved by more than 20% compared with the best individual classifier. The R2 score of concentration prediction can reach 84.1%, which is improved by more than 25% compared with the best regressor in the ensemble group. For both type identification and concentration prediction, the tolerance to the sensor faults of swarm-based E-nose is 10 and 18 times greater than that of the conventional array-based sensors, respectively. Our work has designed an elastic architecture based on sensor swarms, providing a novel avenue for the development of E-nose systems with high fault tolerance. • Edge computing is used in E-nose to enable artificial neural networks running in a microcontroller. • Multiple independent recognition models are trained with feature data from each sensor separately. • Ensemble learning is used to organize independent models, significantly improving overall performance. • The swarm-based E-nose can increase the fault tolerance to more than ten times compared to the array-based E-nose. [ABSTRACT FROM AUTHOR]

Pattern recognition is the core component of the electronic nose (E-nose). Traditional machine learning algorithms highly rely on the feature data selected manually for model training and testing. A complete experiment must be performed before the data can be further processed. To realize the automatic extraction of response features and simplify the model's training and application process, a multi-task convolutional neural network (MTL-CNN) with a dual-block knowledge-sharing structure is designed to train a model for the E-nose system. This model can simultaneously perform three different classification tasks, for the purposes of target discrimination, concentration prediction, and state judgment. Only a few consecutive seconds of response data are needed to be input into the trained model to obtain various information about the E-nose. With the utilization of an unmanned gas-sensing test system, large-scale measurements of the E-nose can be carried out automatically. A baseline tracking algorithm (BTA) is designed based on the relative changes of short-term data, reducing the impact of long-term shifts. Over thousands of gas response processes and more than 10 million sensing data have participated in the training of the deep learning model. The 5-fold cross-validation method shows that the fully trained model has an outstanding generalization performance. After the baseline is tracked automatically, the accuracy of three tasks towards 12 kinds of volatile organic compounds (VOCs) is about 95% (type recognition: 95.2%, concentration prediction: 92.1%, status judgment: 97.3%) using only 4 s of sensing data during the response status of the E-nose. Our work shows the distinct advantages of combining "big data" and "deep learning" in the gas-sensing field and further proves that the employment of MTL-CNN can significantly improve the training and application efficiency of the E-nose. • Multitask deep learning is applied to realize target discrimination, concentration prediction, and status judgment. • The accuracy of three classification tasks for 12 kinds of VOCs is about 95% using 4 s of sensing data at any time. • Large-scale measurements with more than ten million data points are used to train a deep learning model for electronic nose. [ABSTRACT FROM AUTHOR]

A novel quantification technique is presented for electronic nose (E-nose), which is based on a double-step strategy combined with hierarchical classifier (HC) and partial least squares regression (PLSR). With the tree structure of HC, the complexity of classifier training process can be reduced in the case of unbalanced samples. For each level of the class hierarchy, the extreme learning machine-based artificial neural network (ELM-ANN) is applied for classification. In order to improve the classification performance of ELM-ANN, the multiple time-domain features are selected as training inputs, and a novel optimization method of the number of hidden layer neurons is given. To validate the effectiveness of the presented quantification technique, an E-nose system is designed to quantify the gases including six toxic gases (hydrogen sulfide, carbon monoxide, ammonia, toluene, formaldehyde, acetone) and three kinds of binary gas mixtures. This presented hierarchical classifier has demonstrated outstanding performance for the identification of target gases, such as the macro-averaged precision for unlabeled data is improved from 80% to 92% compared with non-hierarchical classifiers. Furthermore, an excellent performance of concentration estimation is obtained utilizing PLSR, where average values of the coefficient of determination for training and test samples are equal to 0.957 and 0.927, respectively. Overall, our work demonstrates that the proposed approach is applicable in E-nose-based odor quantification. • An electronic nose (E-nose) system is designed for quantification analysis of multiple gases in single or binary mixtures. • We apply modified hierarchical classifier to improve the classification accuracy of multiple toxic gases for E-nose system. • Six kinds of indoor air contaminants and their binary gas mixtures can be identified with high classification accuracy. • Concentration estimation is obtained accurately with a high coefficient of determination equal to 0.957 utilizing PLSR. [ABSTRACT FROM AUTHOR]

Two-dimensional (2D) transition metal dichalcogenides (TMDs) have received emerging attention due to their rich active sites, layered structure, and suitable electronic structure for gas sensing at room temperature (25 °C). However, the sensitivity, recoverability, stability and selectivity of TMDs-based gas sensors have remained a grand challenge. Herein, a novel 2D/0D molybdenum disulfide (MoS 2)/zinc sulfide (ZnS) based gas sensor was rationally designed via a two-step solution process, enabling high-performance detection of nitrogen dioxide (NO 2) gas at room temperature. Specifically, the MoS 2 /ZnS heterostructures manifest a greatly enhanced response performance (S = 7.2) to 5 ppm NO 2 in comparison with the bare MoS 2 nanosheets. The gas sensor can be completely recovered after re-exposing to the air for 4.6 min at room temperature without using any external auxiliary treatments. Our optimized gas sensor not only exhibits an ultra-low theoretical limit of detection (14 ppb), but also shows an ultrahigh selectivity and outstanding stability. We further showed that the greatly enhanced gas sensing performance of 2D/0D MoS 2 /ZnS can be ascribed to the unique structure as well as the synergistic effects between MoS 2 nanosheets and ZnS nanoparticles from the geometrical, charge transfer, equivalent resistance, and chemical aspects. This work provides a promising approach to the fabrication of high-performance gas sensing materials for environmental monitoring at room temperature. • MoS 2 /ZnS heterostructures were synthesized through wet chemical growth of ZnS nanoparticles on MoS 2 nanosheets. • MoS 2 /ZnS-based gas sensor shows high response of 7.2 towards 5 ppm NO 2 and excellent recoverability at room temperature. • The MoS 2 /ZnS-based gas sensor demonstrates an ultra-low theoretical limit (14 ppb), superb selectivity and stability. • The mechanisms are discussed from the aspects of geometrical, charge transfer, equivalent resistance, and chemical effects. [ABSTRACT FROM AUTHOR]

[Display omitted] • The uniformly dispersed ZnS NPs hinder the accumulation of rGO flakes and enlarge the specific surface area, while the heterojunction barrier effect realizes the high-efficiency response recovery of NO 2 at room temperature. • Flexible sensor shows low power consumption and good anti-bending and anti-fatigue performance. • Developed a watch-type wearable room temperature NO 2 detection device, which realizes real-time monitoring of target gas by smart terminals through Bluetooth wireless transmission. Wearable electronics consists of mechanically flexible sensor networks and the corresponding circuit system can provide real-time monitoring and recording of toxic gases. In this work, we present a convenient watch-type wearable nitrogen dioxide (NO 2) sensor based on ZnS nanoparticles/nitrogen-doped reduced graphene oxide (ZnS NPs/N-rGO) as a flexible device. The power consumption of the flexible sensor is as low as 0.52 μW. The response value of the device for 10 ppm NO 2 is 2.2; while the recovery time is 724 s, with a low theoretical limit of detection at 69 ppb, with the device maintaining stable mechanical durability. By reading and analyzing the signal data through wireless bluetooth transmission, the device can provide timely monitoring and early warning for NO 2 leakage. Furthermore, the impressive room-temperature sensing performance of ZnS NPs/N-rGO is attributed to factors of geometrical and electronic effects. More specifically, ZnS NP uniformly dispersed on rGO greatly expands the effective contact area with the target gas molecules, and the built-in electric field formed by the strong charge interaction between them also effectively improves the detection sensitivity. The successful development of this type of wireless wearable sensor provides an effective reference for the future application and development of high-performance, low-power, and lightweight sensing equipment. [ABSTRACT FROM AUTHOR]

• A strategy to design different CuO morphologies via a simple self-assemble method by the similar nanorods is proposed. • The effect of morphologies on gas sensing performance of CuO was studied. • The sensing mechanism was discussed from a new perspective. Materials with special microstructures provide excellent properties. In this study, three types of CuO microstructures with shrimp, flower, and urchin-like morphologies were synthesized via self-assembly of nanorods with a diameter of 30-50 nm using one-step hydrothermal synthesis method simply by adjusting the experimental parameters. At the same time, structural properties of the obtained materials were characterized using SEM, XRD, TEM, XPS, TGA and N 2 adsorption-desorption isotherms. The self-assembly mechanism was studied in detail, and a corresponding self-assembly strategy was proposed, which could guide the synthesis of other materials and may improve the potential of self-assembled crystalline materials in practical applications. Meanwhile, the gas sensing characteristics of three different gas sensors for volatile organic gases (methanol, ethanol, isopropanol, formaldehyde, ethyl acetate, acetone and triethylamine) were studied. The shrimp-like CuO sensor showed a higher sensitivity, shorter response time (21 s) and recovery time (19 s) than other sensors due to its higher specific surface area and special surface structure. At 200 °C, its response was upto 8.9 for 100 ppm triethylamine. This material has good application in the field of sensing. [ABSTRACT FROM AUTHOR]