Evaluation of peach quality indices using an electronic nose by MLR, QPST and BP network

In this paper, responses of a gas sensor array were employed to establish a quality indices model evaluating the peach quality indices. The relationship between sensor signals and the firmness, the content of sugar (CS) and acidity of “Dabai” peach were developed using multiple linear regressions with stepwise procedure, quadratic polynomial step regression (QPST) and back-propagation network. The results showed that the multiple linear regression model represented good ability in predicting of quality indices, with high correlation coefficients (R² = 0.87 for penetrating force CF; R² = 0.79 for content of sugar CS; R² = 0.81 for pH) and relatively low average percent errors (ERR) (9.66%, 7.68% and 3.6% for CF, CS and pH, respectively). The quadratic polynomial step regression provides an accurate quality indices model, with high correlation (R² = 0.92, 0.87, 0.83 for CF, CS and pH, respectively) between predicted and measured values and a relatively low error (5.47%, 3.45%, 2.57% for CF, CS and pH, respectively) for prediction. The feed-forward neural network also provides an accurate quality indices model with a high correlation (R² = 0.90, 0.81, 0.87 for CF, CS and pH, respectively) between predicted and measured values and a relatively low average percent error (6.39%, 6.21%, 3.13% for CF, CS and pH, respectively) for prediction. These results prove that the electronic nose has the potential of becoming a reliable instrument to assess the peach quality indices.     

Simultaneous estimation of odor classes and concentrations using an electronic nose with function approximation model ensembles

This paper sets up a practical electronic nose for simultaneously estimating many kinds of odor classes and concentrations. Mathematically, such simultaneous estimation problems can be regarded as multi-input/multi-output (MIMO) function approximation problems. After decomposing an MIMO approximation task into multiple many-to-one tasks, we can use multiple many-to-one approximation model ensembles to implement them one after another. A single approximation model may be a multivariate logarithmic regression, a quadratic multivariate logarithmic regression, a multilayer perceptron, or a support vector machine. An ensemble is made up of the above four models, represents a special kind of odor, and realizes the relationship between sensor array responses and the represented odor concentrations. Naturally, all members in the ensemble are trained only by the samples from the represented odor. The real outputs of ensembles are the average predicted concentrations and the relative standard deviations (R.S.D.s). The ensemble with the smallest R.S.D. finally gives the label and concentration of an odor sample, which can be looked upon as the use of the average and the minimum combination rules. The predicted results for four kinds of fragrant materials, ethanol, ethyl acetate, ethyl caproate, and ethyl lactate, 21 concentrations in total, show that the proposed approximation model ensembles and combination strategies with the electronic nose are effective for simultaneously estimating many kinds of odor classes and concentrations.     

Fuzzy rough granular neural networks, fuzzy granules, and classification

We introduce a fuzzy rough granular neural network (FRGNN) model based on the multilayer perceptron using a back-propagation algorithm for the fuzzy classification of patterns. We provide the development strategy of the network mainly based upon the input vector, initial connection weights determined by fuzzy rough set theoretic concepts, and the target vector. While the input vector is described in terms of fuzzy granules, the target vector is defined in terms of fuzzy class membership values and zeros. Crude domain knowledge about the initial data is represented in the form of a decision table, which is divided into subtables corresponding to different classes. The data in each decision table is converted into granular form. The syntax of these decision tables automatically determines the appropriate number of hidden nodes, while the dependency factors from all the decision tables are used as initial weights. The dependency factor of each attribute and the average degree of the dependency factor of all the attributes with respect to decision classes are considered as initial connection weights between the nodes of the input layer and the hidden layer, and the hidden layer and the output layer, respectively. The effectiveness of the proposed FRGNN is demonstrated on several real-life data sets.     

Fuzzy neural network-based rescheduling decision mechanism for semiconductor manufacturing

Most semiconductor manufacturing systems (SMS) operate in a highly dynamic and unpredictable environment. The production rescheduling strategy addresses uncertainty and improves SMS performance. The rescheduling framework of SMS is presented as layered scheduling strategies with an optimization rescheduling decision mechanism. A fuzzy neural network (FNN) based rescheduling decision model is implemented which can rapidly choose an optimized rescheduling strategy to schedule the semiconductor wafer fabrication lines according to current system disturbances. The mapping between the input of FNN, such as disturbances, system state parameters, and the output of FNN, optimal rescheduling strategies, is constructed. An example of a semiconductor fabrication line in Shanghai is given. The experimental results demonstrate the effectiveness of proposed FNN-based rescheduling decision mechanism approach over the alternatives such as back-propagation neural network (BPNN) and multivariate regression (MR).     

Fuzzy neural networks for obstacle pattern recognition and collision avoidance of fish robots

The problems of detection and pattern recognition of obstacles are the most important concerns for fish robots’ path planning to make natural and smooth movements as well as to avoid collision. We can get better control results of fish robot trajectories if we obtain more information in detail about obstacle shapes. The method employing only simple distance measuring IR sensors without cameras and image processing is proposed. The capability of a fish robot to recognize the features of an obstacle to avoid collision is improved using neuro-fuzzy inferences. Approaching angles of the fish robot to an obstacle as well as the evident features such as obstacles’ sizes and shape angles are obtained through neural network training algorithms based on the scanned data. Experimental results show the successful path control of the fish robot without hitting on obstacles.     

神经网络与模糊技术的结合与发展

在神经网络与模糊技术不断发展的同时,作为两者结合的神经模糊技术和模糊神经网络已经兴起并发展直来,在对神经网络和模糊技术进行分析和比较的基础上着重论述了两者结合的原因、形式,以及模糊神经网络的理论和应用,指出了神经模糊技术未来的发展和展望。     

A novel humid electronic nose combined with an electronic tongue for assessing deterioration of wine

We report herein the use of a combined system for the analysis of the spoilage of wine when in contact with air. The system consists of a potentiometric electronic tongue and a humid electronic nose. The potentiometric electronic tongue was built with thick-film serigraphic techniques using commercially available resistances and conductors for hybrid electronic circuits; i.e. Ag, Au, Cu, Ru, AgCl, and C. The humid electronic nose was designed in order to detect vapours that emanate from the wine and are apprehended by a moist environment. The humid nose was constructed using a piece of thin cloth sewn, damped with distilled water, forming five hollows of the right size to introduce the electrodes. In this particular case four electrodes were used for the humid electronic nose: a glass electrode, aluminium (Al), graphite and platinum (Pt) wires and an Ag-AgCl reference electrode. The humid electronic nose together with the potentiometric electronic tongue were used for the evaluation of the evolution in the course of time of wine samples. Additionally to the analysis performed by the tongue and nose, the spoilage of the wines was followed via a simple determination of the titratable (total) acidity.     

Automated tuning of an electronic circuit board using the artificial neural network approach

Manufacturing of electronic circuits for microwave communication boards often requires tuning of different circuit characteristics by manual adjustment of several trimmer components, including the trimmer's resistance and capacitance. This manual tuning process was automated by applying the artificial neural network modeling approach. In the considered tuning process, which required manual adjustment of a set of trimmers, multiple specification criteria had to be satisfied by several trimmer rotations. The tuning process was described in terms of three independent steps: the circuit output measurement, trimmer selection, and trimmer rotation. The trimmer selection was performed by a semi-supervised neural network, which learned the patterns of circuit characteristics and the deviations between the ideal and practical outputs. Another network was developed for determination of trimmer rotation rate. The results, based on computer simulation of the tuning process, showed that the developed system improved performance of the tuning process, allowing for automation of the microwave circuit board tuning task in a real manufacturing environment.     

Nonsmooth training of fuzzy neural networks

The integration of fuzzy methods and neural networks often leads to nonsmoothness of the neural network and, consequently, to a nonsmooth training problem. It is shown, that smooth training methods as e.g. backpropagation fail to converge in this case. Thus a method – based on so called bundle-methods – for training of nonsmooth neural network is presented. Numerical results obtained from a character recognition problem show, that this method still converges where backpropagation fails.     

模糊人工神经网络方法在QSAR研究中的应用

使用模糊神经网络提取易于理解的“IF－THEN”模糊规则，并用于亚苄丙二腈类衍生物活性的预测，结果较好。     

A generic fuzzy aggregation operator: rules extraction from and insertion into artificial neural networks

Multilayered feedforward artificial neural networks (ANNs) are black boxes. Several methods have been published to extract a fuzzy system from a network, where the input–output mapping of the fuzzy system is equivalent to the mapping of the ANN. These methods are generalized by means of a new fuzzy aggregation operator. It is defined by using the activation function of a network. This fact lets to choose among several standard aggregation operators. A method to extract fuzzy rules from ANNs is presented by using this new operator. The insertion of fuzzy knowledge with linguistic hedges into an ANN is also defined thanks to this operator.     

神经网络控制的现状与展望

对神经网络在控制中的应用进行了综述,特别对现阶段几种较重要的神经（网络）控制的现状进行了评述,并对神经控制的发展作了展望,最后对神经网络用于控制中存在的几个问题进行了探讨。     

基于有序神经网络的神经模糊预测控制

本文使用有序神经网络和改进的模糊控制器构成了一种新型的神经模糊预测控制方法，有序网络学习速度快，所需神经数目少，用事先训练好的有序网络代替传统的预测模型，以期增强输出预测的准确性；同时，用一种改进的模糊控制器原有的PID控制器，增强系统的鲁棒性。仿真结果表明，所提出的神经模糊预测控制方法可以获得理想的控制效果。     

电子鼻鉴别有机磷农药的一种特征提取方法

由于有机磷农药的挥发性与稳定性易受环境的影响,降低了电子鼻在分类鉴别中的精确度与可比性。在定位传感器响应起始点的基础上,截取有效数据,以保证分析结果的可靠性;通过对平均微分法、积分值2种特征提取方法分析,提出了一种面积斜率比值法的特征提取方法;运用主成分分析（PCA）和BP神经网络对3种方法2种农药进行分类计算,结果表明：面积斜率比值法的分类效果最好。     

Nonlinear Systems Identification via Two Types of Recurrent Fuzzy CMAC

Normal fuzzy CMAC neural network performs well for nonlinear systems identification because of its fast learning speed and local generalization capability for approximating nonlinear functions. However, it requires huge memory and the dimension increases exponentially with the number of inputs. It is difficult to model dynamic systems with static fuzzy CMACs. In this paper, we use two types of recurrent techniques for fuzzy CMAC to overcome the above problems. The new CMAC neural networks are named recurrent fuzzy CMAC (RFCMAC) which add feedback connections in the inner layers (local feedback) or the output layer (global feedback). The corresponding learning algorithms have time-varying learning rates, the stabilities of the neural identifications are proven.     

Aging fingerprint characterization of beer using electronic nose

In this work, attempts were made in order to characterize the change of aroma of alcoholic and non alcoholic beers during the aging process by use of a metal oxide semiconductor based electronic nose. The aged beer samples were statistically characterized in several classes. Linear techniques as principal component analysis (PCA) and Linear Discriminant Analaysis (LDA) were performed over the data that revealed non alcoholic beer classes are separated except a partial overlapping between zones corresponding to two specified classes of the aged beers. A clear discrimination was not found among the alcoholic beer classes showing the more stability of such type of beer compared with non alcoholic beer. In this research, to classify the classes, two types of artificial neural networks were used: Probabilistic Neural Networks (PNN) with Radial Basis Functions (RBF) and FeedForward Networks with Backpropagation (BP) learning method. The classification success was found to be 90% and 100% for alcoholic and non alcoholic beers, respectively. Application of PNN showed the classification accuracy of 83% and 100%, respectively for the aged alcoholic and non alcoholic beer classes as well. Finally, this study showed the capability of the electronic nose system for the evaluation of the aroma fingerprint changes in beer during the aging process.     

Numerical solution of a system of fuzzy polynomials by fuzzy neural network

In this paper, a new approach for solving systems of fuzzy polynomials based on fuzzy neural network (FNN) is presented. This method can also lead to improve numerical methods. In this work, an architecture of fuzzy neural networks is also proposed to find a real root of a system of fuzzy polynomials (if exists) by introducing a learning algorithm. Finally, we illustrate our approach by numerical examples.     

基于FAM的模糊神经控制器的研究

根据模糊联想记忆(FAM)理论, 提出了预解模糊FAM原理, 给出了预解模糊FAM和一般FAM的等价性的构造性证明. 为了提高FAM推理过程的自适应能力, 将神经网络应用于预解模糊FAM推理, 提出了一种新的智能控制器——FAM神经控制器(FAMNC), 以小车倒立摆为控制对象进行了仿真研究, 表明了所提方法的可行性.     

一种基于无源性理论的模糊Hopfield神经网络学习律设计方法

本文研究了一类模糊Hopfield神经网络系统的稳定性问题.首先,基于无源性理论,设计了一种新的权重学习律,并通过构造的模糊Lyapunov函数证明了系统从输入到输出是无源的.在此基础上,证明了系统在该学习律下是输入到状态稳定的.相比于传统的公共Lypaunov函数,本文所提的模糊Lyapunov函数能保证系统具有更好的性能.最后,通过数值仿真验证了所提方法的有效性.