Ya-Qi Jing

Electronic nose with a new feature reduction method and a multi-linear classifier for Chinese liquor classification

An electronic nose (e-nose) was designed to classify Chinese liquors of the same aroma style. A new method of feature reduction which combined feature selection with feature extraction was proposed. Feature selection method used 8 feature-selection algorithms based on information theory and reduced the dimension of the feature space to 41. Kernel entropy component analysis was introduced into the e-nose system as a feature extraction method and the dimension of feature space was reduced to 12. Classification of Chinese liquors was performed by using back propagation artificial neural network (BP-ANN), linear discrimination analysis (LDA), and a multi-linear classifier. The classification rate of the multi-linear classifier was 97.22%, which was higher than LDA and BP-ANN. Finally the classification of Chinese liquors according to their raw materials and geographical origins was performed using the proposed multi-linear classifier and classification rate was 98.75% and 100%, respectively.

A New Method Combining KECA-LDA With ELM for Classification of Chinese Liquors Using Electronic Nose

We proposed a hybrid algorithm by combining kernel entropy component analysis (KECA) with linear discriminant analysis (LDA), namely, KECA-LDA for feature reduction in electronic-nose systems. It combined the advantages of KECA and LDA. Then, the data extracted by KECA-LDA were inputted to extreme learning machine (ELM) for classification. In order to examine the performance of the proposed method, eight types of strong-flavor Chinese liquors were tested using an electronic nose (e-nose) system designed by ourselves, and the results after cross validation showed that features extracted by KECA-LDA were more beneficial to classification than KECA, and the performance of ELM was better than that of backpropagation neural network. The highest classification rate by KECA-LDA-ELM was 100%. In conclusion, an e-nose combined with KECA-LDA and ELM is a feasible method to classify Chinese liquors.

Distributed Sequential Location Estimation of a Gas Source via Convex Combination in WSNs

Localization of the hazardous gas source plays an important role in the protection of public security, since it can save a lot of time for subsequent rescue works. For gas source localization (GSL), a large number of gas sensor nodes can be rapidly deployed to construct a wireless sensor network (WSN) and cover the whole concerned area. Although least-squares (LS) methods can solve the problem of GSL in WSNs regardless of the distribution of measurement noises, centralized LS methods are not power efficient and robust since they require the gathering and processing of large-scale measurements on a central node. In this paper, we propose a novel distributed method for GSL in WSNs, which is performed on a sequence of sensor nodes successively. Each sensor node in the sequence conducts an individual estimation and a convex combination. The individual estimation is inspired by the LS formulation of the problem of GSL in WSNs. The proposed method is fully distributed and computationally efficient, and it does not rely on the absolute location of the sensor nodes. Extensive simulation results and a set of experimental results demonstrate that the success rate and localization accuracy of the proposed method are generally higher than those of the trust-region-reflective method.

A Bio-Inspired Breathing Sampling Electronic Nose for Rapid Detection of Chinese Liquors

The rapid detection of Chinese liquors is one of the most demanded techniques of the China light industry and electronic noses (e-noses) have been used to solve this problem. However, these e-noses using classic dynamic sampling (CDS) method consume quite long sampling time. In this paper, a novel bio-inspired breathing sampling (BBS) method is proposed to shorten the sampling time. The proposed BBS is a bionic multi-stage sampling method that imitates the biological process of continuous short breathings. Based on the CDS and BBS methods, the same data processing procedures are employed to process the sampled data. Compared with the CDS, the sampling time of the BBS is much shorter and the extracted features of the BBS are simpler. To test the performance of the two sampling methods, three classification methods are employed to identify seven kinds of Chinese liquors using the self-designed electronic nose. The experimental results indicate that the BBS method achieves not only rapider detection but also higher classification accuracy.

Combining PSO-KECA with ELM in an electronic nose for classification of Chinese liquors

We designed an electronic nose to classify different Chinese liquors. Kernel entropy component analysis (KECA) was applied to reduce the dimensionality of data sets. In order to avoid the blindness of parameter setting, particle swarm optimization (PSO) algorithm was employed to optimize parameters in KECA. At last, we adopted extreme learning machine (ELM) as a classifier to classify eight kinds of strong-flavor Chinese liquors. The results indicate that ELM has a better performance for classification of Chinese liquors with different brands than back propagation neural network (BPNN). The highest classification rate by ELM is 97.5%.

A Portable Odor-Tracing Instrument

In this paper, we present the design of a portable odor-tracing instrument and propose a new method for odor tracing using the instrument. The odor-tracing instrument comprises three gas sensors, a microprocessor, and a liquid crystal display. The gas sensors are placed in a regular triangle. When odor molecules reach the surface of a gas sensor, the output of the sensor changes. To obtain more features from the sensor outputs, we use a scale-space method to obtain multitime-scale information. A time series for the sensor data can then be calculated by feature-point matching. The direction of the odor source can be estimated using a graphical statistical method. The method uses relative output information from the sensors to eliminate the influence of sensor drift. Indoor and outdoor experiments showed that the instrument could indicate the odor source precisely in a windy environment. According to the guidance of the system, users can locate an odor source precisely.

A portable E-nose system for classification of Chinese liquor

In this paper, a portable electronic nose (e-nose), being highlighted with high speed and portability, was developed to classify Chinese liquor. It consisted of sampling module, gas chamber and ARM processor. First, two cascaded chambers wrapped with thermostat (Silicone heating band) were designed, one for rapid evaporation, and the other for reaction of the liquor samples. Then a novel bio-inspired segmented breathing sampling method was proposed, which can notably enrich the available dynamic information in limited time. Consequently, by taking the advantage of fast online algorithm, one test process was limited in 4.5 minutes, including sample evaporation, sequential sampling, data output and cleanout. Besides, the developed e-nose was easily carried (1.9 Kg including battery). To examine the performance of the e-nose, six kinds of Chinese liquors were tested (each for 20 times) using optimized support vector machine (SVM), and the accuracy rate was 90.8%.

Learning to Rapidly Re-Contact the Lost Plume in Chemical Plume Tracing

Maintaining contact between the robot and plume is significant in chemical plume tracing (CPT). In the time immediately following the loss of chemical detection during the process of CPT, Track-Out activities bias the robot heading relative to the upwind direction, expecting to rapidly re-contact the plume. To determine the bias angle used in the Track-Out activity, we propose an online instance-based reinforcement learning method, namely virtual trail following (VTF). In VTF, action-value is generalized from recently stored instances of successful Track-Out activities. We also propose a collaborative VTF (cVTF) method, in which multiple robots store their own instances, and learn from the stored instances, in the same database. The proposed VTF and cVTF methods are compared with biased upwind surge (BUS) method, in which all Track-Out activities utilize an offline optimized universal bias angle, in an indoor environment with three different airflow fields. With respect to our experimental conditions, VTF and cVTF show stronger adaptability to different airflow environments than BUS, and furthermore, cVTF yields higher success rates and time-efficiencies than VTF.

Rapid detection of Chinese liquors using a portable e-nose based on C-SVM

We designed a portable electronic nose (e-nose) for rapid detection of Chinese liquors. The e-nose had an evaporation chamber wrapped with silicone heating band, which could vaporize the liquor samples in 1 minute. The sequential sampling and clean-out lasted 100 seconds and 3 minutes, respectively. Hence, one fast detection process could be finished within 3 minutes without regard to clean-out. In order to acquire enough information from arrays sampling curves for classification, we chose 60 features (including 6 features per sensor) according to the response curves, the derivative of the response curves and the relative change rate curves of conductivity. Then the dimension of features was reduced by principle component analysis (PCA). Finally, classification of Chinese liquors was performed using C-Support Vector Machine (C-SVM) and the classification rate was 96.7%.