Home About us Contact
|
Home About us Contact | ||
|
|
||
Neural Network Approach (neural + network_approach)
Selected AbstractsNeural network approach to stereoscopic correspondence of three-dimensional particle tracking velocimetryIEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING, Issue 6 2008Achyut Sapkota Student Member Abstract Particle tracking velocimetry (PTV) is a reliable measurement technique for the quantitative study of fluid flows by observing the motion of the particles seeded in them and is widely used in several industrial applications. The nature of the flow can be precisely observed only if all the three components of the velocity are computed. In 3-D PTV system, particles viewed by two (or more than two) stereoscopic cameras with a parallax have to be correctly paired at every synchronized time step. This is important because the 3-D coordinates of individual particles cannot be computed without the knowledge of the correct stereo correspondence of the particles. In the present work, a neural network,based algorithm has been proposed for the stereoscopic particle pairing process. The correspondence between the particle pairs is modeled as a constrained optimization problem. The constraints are provided on the basis of the epipolar geometry of the particle images and on the basis of the uniqueness of the matched pairs. The results are tested with various standard images. Copyright © 2008 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [source] Neural network approach to firm grip in the presence of small slipsJOURNAL OF FIELD ROBOTICS (FORMERLY JOURNAL OF ROBOTIC SYSTEMS), Issue 6 2001A. M. Al-Fahed Nuseirat This paper presents a two stage method for constructing a firm grip that can tolerate small slips of the fingertips. The fingers are assumed to be of frictionless contact type. The first stage was to formulate the interaction in the gripper,object system as a linear complementarity problem (LCP). Then it was solved using a special neural network to find minimal fingers forces. The second stage was to use the obtained results in the first stage as a static mapping in training another neural network. The second neural network training included emulating the slips by random noise in the form of changes in the positions of the contact points relative to the reference coordinate system. This noisy training increased robustness against unexpected changes in fingers positions. Genetic algorithms were used in training the second neural network as global optimization techniques. The resulting neural network is a robust, reliable, and stable controller for rigid bodies that can be handled by a robot gripper. © 2001 John Wiley & Sons, Inc. [source] Neural network approach for comodeling design of multichip moduleMICROWAVE AND OPTICAL TECHNOLOGY LETTERS, Issue 7 2008M. El Zoghbi Abstract An original comodeling approach based on neural network is proposed in order to optimize multichip modules (MCM). This approach permits to characterize and to optimize millimeter-wave module behavior by taking into account electromagnetic phenomena. All the design procedure is implemented in a circuit software to reduce the simulation time. Encouraging results are obtained. © 2008 Wiley Periodicals, Inc. Microwave Opt Technol Lett 50: 1770,1774, 2008; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.23530 [source] Two variable semi-empirical and artificial neural-network-based modeling of peptide mobilities in CZE: The effect of temperature and organic modifier concentrationELECTROPHORESIS, Issue 5 2009Stefan Mittermayr Abstract This work was focused on investigating the effects of two separation influencing parameters in CZE, namely temperature and organic additive concentration upon the electrophoretic migration properties of model tripeptides. Two variable semi-empirical (TVSE) models and back-propagation artificial neural networks (ANN) were applied to predict the electrophoretic mobilities of the tripeptides with non-polar, polar, positively charged, negatively charged and aromatic R group characteristics. Previously published work on the subject did not account for the effect of temperature and buffer organic modifier concentration on peptide mobility, in spite of the fact that both were considered to be influential factors in peptide analysis. In this work, a substantial data set was generated consisting of actual electrophoretic mobilities of the model tripeptides in 30,mM phosphate buffer at pH 7.5, at 20, 25, 30, 35 and 40°C and at four different organic additive containing running buffers (0, 5, 10 and 15% MeOH) applying two electric field strengths (12 and 16,kV) to assess our mobility predicting models. Based on the Arrhenius plots of natural logarithm of mobility versus reciprocal absolute temperature of the various experimental setups, the corresponding activation energy values were derived and evaluated. Calculated mobilities by TVSE and back-propagation ANN models were compared with each other and to the experimental data, respectively. Neural network approaches were able to model the complex impact of both temperature and organic additive concentrations and resulted in considerably higher predictive power over the TVSE models, justifying that the effect of these two factors should not be neglected. [source] Recurrent Neural Networks for Uncertain Time-Dependent Structural BehaviorCOMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING, Issue 5 2010W. Graf The approach is based on recurrent neural networks trained by time-dependent measurement results. Thereby, the uncertainty of the measurement results is modeled as fuzzy processes which are considered within the recurrent neural network approach. An efficient solution for network training and prediction is developed utilizing ,-cuts and interval arithmetic. The capability of the approach is demonstrated by means of the prediction of the long-term structural behavior of a reinforced concrete plate strengthened by a textile reinforced concrete layer. [source] Enhancing Neural Network Traffic Incident-Detection Algorithms Using WaveletsCOMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING, Issue 4 2001A. Samant Researchers have presented freeway traffic incident-detection algorithms by combining the adaptive learning capability of neural networks with imprecision modeling capability of fuzzy logic. In this article it is shown that the performance of a fuzzy neural network algorithm can be improved through preprocessing of data using a wavelet-based feature-extraction model. In particular, the discrete wavelet transform (DWT) denoising and feature-extraction model proposed by Samant and Adeli (2000) is combined with the fuzzy neural network approach presented by Hsiao et al. (1994). It is shown that substantial improvement can be achieved using the data filtered by DWT. Use of the wavelet theory to denoise the traffic data increases the incident-detection rate, reduces the false-alarm rate and the incident-detection time, and improves the convergence of the neural network training algorithm substantially. [source] Neural Network Earnings per Share Forecasting Models: A Comparative Analysis of Alternative MethodsDECISION SCIENCES, Issue 2 2004Wei Zhang ABSTRACT In this paper, we present a comparative analysis of the forecasting accuracy of univariate and multivariate linear models that incorporate fundamental accounting variables (i.e., inventory, accounts receivable, and so on) with the forecast accuracy of neural network models. Unique to this study is the focus of our comparison on the multivariate models to examine whether the neural network models incorporating the fundamental accounting variables can generate more accurate forecasts of future earnings than the models assuming a linear combination of these same variables. We investigate four types of models: univariate-linear, multivariate-linear, univariate-neural network, and multivariate-neural network using a sample of 283 firms spanning 41 industries. This study shows that the application of the neural network approach incorporating fundamental accounting variables results in forecasts that are more accurate than linear forecasting models. The results also reveal limitations of the forecasting capacity of investors in the security market when compared to neural network models. [source] A neural network approach for structural identification and diagnosis of a building from seismic response dataEARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS, Issue 2 2003C. S. Huang Abstract This work presents a novel procedure for identifying the dynamic characteristics of a building and diagnosing whether the building has been damaged by earthquakes, using a back-propagation neural network approach. The dynamic characteristics are directly evaluated from the weighting matrices of the neural network trained by observed acceleration responses and input base excitations. Whether the building is damaged under a large earthquake is assessed by comparing the modal parameters and responses for this large earthquake with those for a small earthquake that has not caused this building any damage. The feasibility of the approach is demonstrated through processing the dynamic responses of a five-storey steel frame, subjected to different strengths of the Kobe earthquake, in shaking table tests. Copyright © 2002 John Wiley & Sons, Ltd. [source] A postural workload evaluation system based on a macro-postural classificationHUMAN FACTORS AND ERGONOMICS IN MANUFACTURING & SERVICE INDUSTRIES, Issue 3 2002Min K. Chung Many Korean workers are exposed to repetitive or prolonged poor working postures, which are closely related with pains or symptoms of musculoskeletal disorders. Poor working postures in Korea were reviewed and an observational method to assess the postural load was developed. A computer-based postural workload evaluation system based on a macro-postural classification scheme was developed. The macro-postural classification is based on the perceived discomforts for various joint motions. On the basis of the perceived discomfort, postural stress levels for the postures at each joint were also defined in a ratio scale to the standing neutral posture. A neural network approach was used to predict the whole-body postural stresses from the body joint motions. A computer-based postural stress evaluation system was designed to automate the procedure for analyzing postures and enhance the usability and practical applicability. © 2002 Wiley Periodicals, Inc. [source] Support vector machines-based modelling of seismic liquefaction potentialINTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS, Issue 10 2006Mahesh Pal Abstract This paper investigates the potential of support vector machines (SVM)-based classification approach to assess the liquefaction potential from actual standard penetration test (SPT) and cone penetration test (CPT) field data. SVMs are based on statistical learning theory and found to work well in comparison to neural networks in several other applications. Both CPT and SPT field data sets is used with SVMs for predicting the occurrence and non-occurrence of liquefaction based on different input parameter combination. With SPT and CPT test data sets, highest accuracy of 96 and 97%, respectively, was achieved with SVMs. This suggests that SVMs can effectively be used to model the complex relationship between different soil parameter and the liquefaction potential. Several other combinations of input variable were used to assess the influence of different input parameters on liquefaction potential. Proposed approach suggest that neither normalized cone resistance value with CPT data nor the calculation of standardized SPT value is required with SPT data. Further, SVMs required few user-defined parameters and provide better performance in comparison to neural network approach. Copyright © 2006 John Wiley & Sons, Ltd. [source] A hybrid Bayesian back-propagation neural network approach to multivariate modellingINTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS, Issue 8 2003C. G. Chua Abstract There is growing interest in the use of back-propagation neural networks to model non-linear multivariate problems in geotehnical engineering. To overcome the shortcomings of the conventional back-propagation neural network, such as overfitting, where the neural network learns the spurious details and noise in the training examples, a hybrid back-propagation algorithm has been developed. The method utilizes the genetic algorithms search technique and the Bayesian neural network methodology. The genetic algorithms enhance the stochastic search to locate the global minima for the neural network model. The Bayesian inference procedures essentially provide better generalization and a statistical approach to deal with data uncertainty in comparison with the conventional back-propagation. The uncertainty of data can be indicated using error bars. Two examples are presented to demonstrate the convergence and generalization capabilities of this hybrid algorithm. Copyright © 2003 John Wiley & Sons, Ltd. [source] Image reconstructions from two orthogonal projectionsINTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY, Issue 2 2003Yuanmei Wang Abstract A vector entropy optimization-based neural network approach is presented to handle image reconstructions from two orthogonal projections. An accurate and parallel reconstruction is attained with this method allowing parallel implementation. This is an attempt to extract the image information from two projections. It is especially meaningful for clinical applications and three-dimensional modeling of the coronary arteries. © 2003 Wiley Periodicals, Inc. Int J Imaging Syst Technol 13, 141,145, 2003; Published online in Wiley Inter-Science (www.interscience.wiley.com). DOI 10.1002/ima.10036 [source] Multicriteria maximum likelihood neural network approach to positron emission tomographyINTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY, Issue 6 2000Yuanmei Wang Abstract The emerging technology of positron emission image reconstruction is introduced in this paper as a multicriteria optimization problem. We show how selected families of objective functions may be used to reconstruct positron emission images. We develop a novel neural network approach to positron emission imaging problems. We also studied the most frequently used image reconstruction methods, namely, maximum likelihood under the framework of single performance criterion optimization. Finally, we introduced some of the results obtained by various reconstruction algorithms using computer-generated noisy projection data from a chest phantom and real positron emission tomography (PET) scanner data. Comparison of the reconstructed images indicated that the multicriteria optimization method gave the best in error, smoothness (suppression of noise), gray value resolution, and ghost-free images. © 2001 John Wiley & Sons, Inc. Int J Imaging Syst Technol 11, 361,364, 2000 [source] An efficient neural network approach for nanoscale FinFET modelling and circuit simulationINTERNATIONAL JOURNAL OF NUMERICAL MODELLING: ELECTRONIC NETWORKS, DEVICES AND FIELDS, Issue 5 2009M. S. Alam Abstract The present paper demonstrates the suitability of artificial neural network (ANN) for modelling of a FinFET in nano-circuit simulation. The FinFET used in this work is designed using careful engineering of source,drain extension, which simultaneously improves maximum frequency of oscillation ,max because of lower gate to drain capacitance, and intrinsic gain AV0,=,gm/gds, due to lower output conductance gds. The framework for the ANN-based FinFET model is a common source equivalent circuit, where the dependence of intrinsic capacitances, resistances and dc drain current Id on drain,source Vds and gate,source Vgs is derived by a simple two-layered neural network architecture. All extrinsic components of the FinFET model are treated as bias independent. The model was implemented in a circuit simulator and verified by its ability to generate accurate response to excitations not used during training. The model was used to design a low-noise amplifier. At low power (Jds,10,µA/µm) improvement was observed in both third-order-intercept IIP3 (,10,dBm) and intrinsic gain AV0 (,20,dB), compared to a comparable bulk MOSFET with similar effective channel length. This is attributed to higher ratio of first-order to third-order derivative of Id with respect to gate voltage and lower gds in FinFET compared to bulk MOSFET. Copyright © 2009 John Wiley & Sons, Ltd. [source] Modeling power and intermodulation behavior of microwave transistors with unified small-signal/large-signal neural network modelsINTERNATIONAL JOURNAL OF RF AND MICROWAVE COMPUTER-AIDED ENGINEERING, Issue 4 2003F. Giannini Abstract This article presents a detailed procedure to learn a nonlinear model and its derivatives to as many orders as desired with multilayer perceptron (MLP) neural networks. A modular neural network modeling a nonlinear function and its derivatives is introduced. The method has been used for the extraction of the large-signal model of a power MESFET device, modeling the nonlinear relationship of drain-source current Ids as well as gate and drain charge Qg and Qd with respect to intrinsic voltages Vgs and Vds over the whole operational bias region. The neural models have been implemented into a user-defined nonlinear model of a commercial microwave simulator to predict output power performance as well as intermodulation distortion. The accuracy of the device model is verified by harmonic load-pull measurements. This neural network approach has demonstrated to predict nonlinear behavior with enough accuracy even if based only on first-order derivative information. © 2003 Wiley Periodicals, Inc. Int J RF and Microwave CAE 13: 276,284, 2003. [source] A hybrid integral equation and neural network approach for fast extraction of frequency dependent parameters of multiconductor transmission linesINTERNATIONAL JOURNAL OF RF AND MICROWAVE COMPUTER-AIDED ENGINEERING, Issue 1 2002G. Pan Abstract Multiconductor transmission lines (MTL) have been modeled by the distributed parameters R, L, C, and G in many commercial CAD packages, where most of the parameters are assumed to be frequency independent or at most . At gigahertz frequencies, such assumptions may introduce significantly large errors in the waveform simulation and timing. In this article, we present a new and fast technique based on a combination of neural network techniques and the integral equation method (IEM) to evaluate frequency dependences accurately, while dramatically reducing the computation time. © 2002 John Wiley & Sons, Inc. Int J RF and Microwave CAE 12: 37,50, 2002. [source] Hybrid modeling of inulinase bio-production processJOURNAL OF CHEMICAL TECHNOLOGY & BIOTECHNOLOGY, Issue 4 2010Marcio A. Mazutti Abstract BACKGROUND: A potential application of inulinase in the food industry is the production of fructooligosaccharides (FOS) through transfructosilation of sucrose. Besides their ability to increase the shelf-life and flavor of many products, FOS have many interesting functional properties. The use of an industrial medium may represent a good, cost-effective alternative to produce inulinase, since the activity of the enzyme produced may be improved or at least remain the same compared with that obtained using a synthetic medium. Thus, inulinase production for use in FOS synthesis is of considerable scientific and technological appeal, as is the development of a reliable mathematical model of the process. This paper describes a hybrid neural network approach to model inulinase production in a batch bioreactor using agroindustrial residues as substrate. The hybrid modeling makes use of a series artificial neural network to estimate the kinetic parameters of the process and the mass balance as constitutive equations. RESULTS: The proposed model was shown to be capable of describing the complex behavior of inulinase production employing agroindustrial residues as substrate, so that the mathematical framework developed is a useful tool for simulation of this process. CONCLUSION: The hybrid neural network model developed was shown to be an interesting alternative to estimate model parameters since complete elucidation of the phenomena and mechanisms involved in the fermentation is not required owing to the black-box nature of the ANN used as parameter estimator. Copyright © 2010 Society of Chemical Industry [source] Improving the prediction of liquid back-mixing in trickle-bed reactors using a neural network approachJOURNAL OF CHEMICAL TECHNOLOGY & BIOTECHNOLOGY, Issue 9 2002Simon Piché Abstract Current correlations aimed at estimating the extent of liquid back-mixing, via an axial dispersion coefficient, in trickle-bed reactors continue to draw doubts on their ability to conveniently represent this important macroscopic parameter. A comprehensive database containing 973 liquid axial dispersion coefficient measurements (DAX) for trickle-bed operation reported in 22 publications between 1958 and 2001 was thus used to assess the convenience of the few available correlations. It was shown that none of the literature correlations was efficient at providing satisfactory predictions of the liquid axial dispersion coefficients. In response, artificial neural network modeling is proposed to improve the broadness and accuracy in predicting the DAX, whether the Piston,Dispersion (PD), Piston,Dispersion,Exchange (PDE) or PDE with intra-particle diffusion model is employed to extract the DAX. A combination of six dimensionless groups and a discrimination code input representing the residence-time distribution models are used to predict the Bodenstein number. The inputs are the liquid Reynolds, Galileo and Eötvos numbers, the gas Galileo number, a wall factor and a mixed Reynolds number involving the gas flow rate effect. The correlation yields an absolute average error (AARE) of 22% for the whole database with a standard deviation on the AARE of 24% and remains in accordance with parametric influences reported in the literature. © 2002 Society of Chemical Industry [source] Critical factors in consumer supermarket shopping behaviour: a neural network approachJOURNAL OF CONSUMER BEHAVIOUR, Issue 1 2001Luiz A. Moutinho Abstract This paper studies UK supermarket shopping behaviour, by analysing the antecedent variables of three critical factors: overall levels of customer satisfaction, number of trips to the supermarket, and amount spent. A neural network approach predicts these factors using ten input variables and three hidden nodes. Results show that the most satisfied and high-spending customers tend to be those who have the income to take full advantage of the choice and quality offered. Other customers are more concerned with prices being reasonable and discounts available, but the satisfaction of these shoppers is also linked with store atmosphere. Copyright © 2001 Henry Stewart Publications. [source] EXPERIMENTAL AND NEURAL NETWORK PREDICTION OF THE PERFORMANCE OF A SOLAR TUNNEL DRIER FOR DRYING JACKFRUIT BULBS AND LEATHERJOURNAL OF FOOD PROCESS ENGINEERING, Issue 6 2005B.K. BALA ABSTRACT This article presents the field performance of a solar tunnel drier for drying jackfruit bulbs and leather. The drier consists of a transparent plastic-covered flat-plate collector and a drying tunnel connected in series to supply hot air directly into the drying tunnel using two direct-current fans operated by a photovoltaic module. The drier has a loading capacity of 120,150 kg of fruits. Sixteen experimental runs were conducted for drying jackfruit bulbs and leather (eight runs each). The use of a solar tunnel drier led to a considerable reduction in drying time and dried products of better quality in comparison to products dried under the sun. A multilayered neural network approach was used to predict the performance of the solar tunnel drier. Using solar drying data of jackfruit bulbs and leather, the model has been trained using backpropagation algorithm. The prediction of the performance of the drier was found to be excellent after it was adequately trained. It can be used to predict the potential of the drier for different locations, and can also be used in a predictive optimal control algorithm. [source] Stimulatory Effect of Procaine on the Growth of Several Microalgae and CyanobacteriaJOURNAL OF PHARMACY AND PHARMACOLOGY: AN INTERNATI ONAL JOURNAL OF PHARMACEUTICAL SCIENCE, Issue 2 2000TAKAHIRO SUZUKI Procaine has been used to stimulate plant growth and it has been noted that it also promotes growth of microorganisms. The effect of procaine hydrochloride concentration on the growth rates of several species of microalgae and cyanobacteria was studied under both photoautotropic and heterotrophic growth conditions. Procaine hydrochloride was added to cultures at concentrations over the range 0.01,1000 mg L,1. A stimulating effect of procaine hydrochloride on photoautotrophic growth was observed for the cyanobacteria Anabaena cylindrica and Anabaena variabilis, and for the salt-tolerant green algae Dunaliella primolecta and Dunaliella parva. During active growth in batch culture an increase in growth rate (compared with control culture without procaine hydrochloride) of about 25% was observed at 0.1 mg L,1 of procaine hydrochloride for A. cylindrica. However, procaine hydrochloride was toxic at concentrations of > 10 mg L,1. Simultaneous administration of hydrolysis products of procaine, p -amino-benzoic acid and diethyl aminoethanol, in lieu of procaine hydrochloride, was as effective as procaine in stimulating growth of A. cylindrica. Heterotrophic growth of Chlorella ellipsoidea and Prototheca zopfii was not stimulated by procaine hydrochloride over the concentration range studied (0.1,10 mg L,1). The combined effects of procaine hydrochloride concentration and four other environmental factors (temperature, light intensity, CO2 concentration in the flushing gas and NaCl concentration) on growth rate of D. primolecta was modelled using both a neural network approach and a response surface method. These results indicate that procaine hydrochloride exerts different effects on the growth of microalgal and cyanobacterial cells as functions of dosage, species and culture conditions. [source] Neural Network Modeling of Constrained Spatial Interaction Flows: Design, Estimation, and Performance IssuesJOURNAL OF REGIONAL SCIENCE, Issue 1 2003Manfred M Fischer In this paper a novel modular product unit neural network architecture is presented to model singly constrained spatial interaction flows. The efficacy of the model approach is demonstrated for the origin constrained case of spatial interaction using Austrian interregional telecommunication traffic data. The model requires a global search procedure for parameter estimation, such as the Alopex procedure. A benchmark comparison against the standard origin constrained gravity model and the two,stage neural network approach, suggested by Openshaw (1998), illustrates the superiority of the proposed model in terms of the generalization performance measured by ARV and SRMSE. [source] Humidity estimation using neural network and optical fiber sensorMICROWAVE AND OPTICAL TECHNOLOGY LETTERS, Issue 3 2009Kirthi L. Sreenivasan Abstract An optical fiber (OF) sensor with enhanced dynamic range and high sensitivity is realized to detect varying range of relative humidity (RH). For precise and independent estimation/prediction of RH from its nonlinear response, feed forward artificial neural network (ANN) model is developed. Preprocessed stable sensor outputs and corresponding commercial sensor outputs were used to train ANN. Results show that ANN has been effective in successfully predicting RH from the response of OF sensor. Overall, neural network approach showed better performance in comparison to alternative calibration method. © 2009 Wiley Periodicals, Inc. Microwave Opt Technol Lett 51: 641,645, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.24132 [source] Characterization of monolithic spiral inductor using neural networksMICROWAVE AND OPTICAL TECHNOLOGY LETTERS, Issue 4 2002A. Ouchar Abstract The characterization of a monolithic spiral inductor (MSI) by using a multilayer neural network approach is presented in this Letter. The inductance, physical, and geometrical parameters are extracted in order to perform a full characterization of MSI. A three-layer neural network was used for accurate representation. The results obtained by using neural networks are compared with measured S parameters of typical MSI. © 2002 Wiley Periodicals, Inc. Microwave Opt Technol Lett 34: 299,302, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.10442 [source] Neuro-fuzzy structural classification of proteins for improved protein secondary structure predictionPROTEINS: STRUCTURE, FUNCTION AND BIOINFORMATICS, Issue 8 2003Joachim A. Hering Abstract Fourier transform infrared (FTIR) spectroscopy is a very flexible technique for characterization of protein secondary structure. Measurements can be carried out rapidly in a number of different environments based on only small quantities of proteins. For this technique to become more widely used for protein secondary structure characterization, however, further developments in methods to accurately quantify protein secondary structure are necessary. Here we propose a structural classification of proteins (SCOP) class specialized neural networks architecture combining an adaptive neuro-fuzzy inference system (ANFIS) with SCOP class specialized backpropagation neural networks for improved protein secondary structure prediction. Our study shows that proteins can be accurately classified into two main classes "all alpha proteins" and "all beta proteins" merely based on the amide I band maximum position of their FTIR spectra. ANFIS is employed to perform the classification task to demonstrate the potential of this architecture with moderately complex problems. Based on studies using a reference set of 17 proteins and an evaluation set of 4 proteins, improved predictions were achieved compared to a conventional neural network approach, where structure specialized neural networks are trained based on protein spectra of both "all alpha" and "all beta" proteins. The standard errors of prediction (SEPs) in % structure were improved by 4.05% for helix structure, by 5.91% for sheet structure, by 2.68% for turn structure, and by 2.15% for bend structure. For other structure, an increase of SEP by 2.43% was observed. Those results were confirmed by a "leave-one-out" run with the combined set of 21 FTIR spectra of proteins. [source] DYNAMIC MODELING OF RETORT PROCESSING USING NEURAL NETWORKSJOURNAL OF FOOD PROCESSING AND PRESERVATION, Issue 2 2002C. R. CHEN ABSTRACT Two neural network approaches , a moving-window and hybrid neural network , which combine neural network with polynomial regression models, were used for modeling F(t) and Qv(t) dynamic functions under constant retort temperature processing. The dynamic functions involved six variables: retort temperature (116,132C), thermal diffusivity (1.5,2.3 × 10,7m2/s), can radius (40,61 mm), can height (40,61 mm), and quality kinetic parameters z (15,39C) and D (150,250 min). A computer simulation designed for process calculations of food thermal processing systems was used to provide the fundamental data for training and generalization of ANN models. Training data and testing data were constructed by both second order central composite design and orthogonal array, respectively. The optimal configurations of ANN models were obtained by varying the number of hidden layers, number of neurons in hidden layer and learning runs, and a combination of learning rules and transfer function. Results demonstrated that both neural network models well described the F(t) and Qv(t) dynamic functions, but moving-window network had better modeling performance than the hybrid ANN models. By comparison of the configuration parameters, moving-window ANN models required more neurons in the hidden layer and more learning runs for training than the hybrid ANN models. [source] Online identification of nonlinear multivariable processes using self-generating RBF neural networksASIAN JOURNAL OF CONTROL, Issue 5 2010Karim Salahshoor Abstract This paper addresses the problem of online model identification for multivariable processes with nonlinear and time-varying dynamic characteristics. For this purpose, two online multivariable identification approaches with self-organizing neural network model structures will be presented. The two adaptive radial basis function (RBF) neural networks are called as the growing and pruning radial basis function (GAP-RBF) and minimal resource allocation network (MRAN). The resulting identification algorithms start without a predefined model structure and the dynamic model is generated autonomously using the sequential input-output data pairs in real-time applications. The extended Kalman filter (EKF) learning algorithm has been extended for both of the adaptive RBF-based neural network approaches to estimate the free parameters of the identified multivariable model. The unscented Kalman filter (UKF) has been proposed as an alternative learning algorithm to enhance the accuracy and robustness of nonlinear multivariable processes in both the GAP-RBF and MRAN based approaches. In addition, this paper intends to study comparatively the general applicability of the particle filter (PF)-based approaches for the case of non-Gaussian noisy environments. For this purpose, the Unscented Particle Filter (UPF) is employed to be used as alternative to the EKF and UKF for online parameter estimation of self-generating RBF neural networks. The performance of the proposed online identification approaches is evaluated on a highly nonlinear time-varying multivariable non-isothermal continuous stirred tank reactor (CSTR) benchmark problem. Simulation results demonstrate the good performances of all identification approaches, especially the GAP-RBF approach incorporated with the UKF and UPF learning algorithms. Copyright © 2010 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society [source] | ||||||||||||||||||||||||||||||||||