Search

This paper describes an efficient simulation technique for field‐effect transistor devices exhibiting significant frequency dispersive behavior. The device model is formulated such that a single energy function completely describes the channel displacement current, thus imposing charge and energy conservations principles from the outset. It is shown that adherence to energy conservation necessarily implies a reciprocal intrinsic displacement network, which is typically too rigid to follow measured device data. This problem is overcome by considering a frequency dispersive time‐delay parameter, which has the effect of allowing nonreciprocal susceptance behavior to be captured by the overall device model. This dispersive time‐delay parameter is included within the nonlinear device simulation via an efficient discrete‐time convolution. [ABSTRACT FROM AUTHOR]

The electrolysis of water using a polymer electrolyte membrane (PEM) electrolyzer is a very vital and efficient method of producing hydrogen (H). The performance of this method can be significantly improved if a photovoltaic (PV) array, with maximum-power-point (MPP) tracker, is utilized as an energy source for the electrolyzer. This paper suggests a stand-alone PV/PEM electrolyzer system to produce pure hydrogen. The paper also develops the different mathematical models for each constituent subsystem. Moreover, the paper develops the suitable maximum-power-point tracking (MPPT) algorithm that is based on utilizing the neural network. This algorithm is utilized together with the action of the PI controller to improve the performance of the suggested stand-alone PV/PEM electrolyzer system through maximizing the hydrogen production rate for every instant. Finally, the suggested hydrogen production system is simulated using the Matlab/Simulink and neural network toolbox. The simulation results of the system indicate the improved relative performance of the suggested hydrogen production system compared with the traditional case of direct connection between the PV array and the PEM electrolyzer. Copyright © 2010 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Inverse modeling of microwave components plays an important role in microwave design and diagnosis or tuning. Since the analytical function or formula of the inverse input‐output relationship does not exist and is difficult to obtain, artificial neural network (ANN) becomes an efficient tool to develop inverse models for microwave components. This paper provides an overview of recent advances in neural network‐based inverse modeling techniques for microwave applications. We review two different shallow neural network‐based inverse modeling techniques, including the comprehensive neural network inverse modeling methodology and the multivalued neural network inverse modeling technique. Both techniques address the problem of nonuniqueness in inverse modeling. We also provide an overview of recently developed hybrid deep neural network modeling technique and the application to inverse modeling. For the inverse modeling problem with high‐dimensional inputs, the relationship between the inputs and the outputs of the inverse model will become more complicated and the inverse modeling problem will become harder. The deep neural network becomes a practical choice. The hybrid deep neural network structure is presented. The recently proposed activation function, specifically for microwave application, and a three‐stage deep learning algorithm for training the hybrid deep neural network are reviewed. [ABSTRACT FROM AUTHOR]

Artificial neural networks‐based module for fast direction‐of‐arrival (DoA) estimation of the signal received by multi‐element textile wearable antenna array (TWAA) is proposed in this paper. The developed multilayer perceptron (MLP) DoA module considers the effects of changing the gain of the antenna elements, the distance between the antenna elements and their resonant frequencies during the movement of the TWAA wearer and the crumpling of the textile. The inputs of the MLP DoA module are the elements of the spatial correlation matrix of the signal sampled by the TWAA and the output is the angular position of the signal source in the azimuthal plane. The influence of the number of TWAA elements on the accuracy of the MLP DoA module in DoA estimation under textile crumple conditions is investigated. The performances of the MLP DoA module under increased noise conditions are investigated, as well as its behavior under different degrees of textile crumpling. A comparison was made between the proposed module and the corresponding root MUSIC DoA module in terms of accuracy and speed of DoA estimation. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, the equivalent series resistance (ESR) and capacitance (ESC) of piezoelectric energy harvesting devices have been modeled accurately over a wide frequency range for the first time. It is shown that the impedance modeling of the ESR and ESC is demonstrated as an important factor for the design of the Bias Flip (or the parallel synchronized switching harvesting on an inductor) interface circuit, which affects the performance of energy harvesting seriously. The conventional Butterworth‐Van Dyke (BVD) model and the modified BVD model are analyzed to state their imprecise modeling performance. To address this problem, an artificial neural network (ANN) technique with prior knowledge input method is employed to improve the model accuracy of the ESR and ESC. Also, a least‐squares curve fitting method is presented to compare with the ANN model. A good matching has been obtained between the measured and ANN‐predicted ESR and ESC in the frequency range of 2600 to 3600 Hz. The excellent performance on the accuracy of the proposed method is further illustrated through the comparison on the average relative error and maximum relative error with the conventional BVD, modified BVD model, and the curve fitting method. [ABSTRACT FROM AUTHOR]

This paper proposes a hybrid technique to detect and classify the transmission line faults in the power system. The proposed approach is the joint execution of linear discriminant analysis (LDA) and cuttlefish optimizer (CFO) learning process‐based random forest algorithm (RFA), ie, named as LDA‐CFRFA technique. Here, two modules are utilized for fault analysis in power system: fault detection and fault classification. The first procedure of the proposed method is the power system transmission line parameters in normal and abnormal condition dataset preparation by using LDA. LDA‐based dataset preparation process consists of feature extraction of power flow parameters and defines the nature of signals occurred by the system. The extracted dataset is assessed by CFO‐based RFA technique for classifying the fault type occurred in the transmission system. By then, the proposed model is executed in MATLAB/Simulink working stage, and the execution is assessed with the existing techniques such as CFO, RFA, Feed forward neural network (FNN), and artificial neural network (ANN). In our research, the faults implemented on test system are phase A, phase B, phase C, phase A to ground (AG), phase B to ground (BG), phase C to ground (CG), phase AB, phase AC and phase BC, phase AB to ground (ABG), phase AC to ground (ACG), and phase BC to ground (BCG). From the results, the proposed technique guarantees the system with less complexity and less consumption time for the detection and classification of the fault, and hence, the accuracy of the system is increased. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to provide a comprehensive overview of the field‐effect transistor (FET) small‐signal modeling using artificial neural networks (ANNs). To gain an in‐depth insight into how to effectively develop an ANN model, we present a comparative study on the application of the ANNs for modeling the scattering (S‐) parameters of a variety of FET technologies versus bias point, ambient temperature, and geometrical dimensions. As will be shown, the main challenge consists of identifying the most appropriate ANN model for the specific case under study. This is because the performance of an ANN‐based model can vary significantly, depending especially on the choice of the model structure and the size and parameters of the chosen ANN. In addition, the choice of the model is related directly to the behavior of the FET characteristics, which might greatly depend on the selected device technology and operating conditions. The analysis of the present comparative study allows understanding how to properly construct ANN models to perform at their best for a successful FET modeling. [ABSTRACT FROM AUTHOR]

ABSTRACT This paper proposes a new technique to train neural network (NN); with the result, we can solve some real-world application problems such as microwave components modeling and optimization. Its major advance is achieved in avoiding the testing error falling into local minimum. After the generalization, the ability of three-layer and four-layer NN is also checked; our investigations show that four-layer NN trained by the proposed training method can map the electromagnetic simulation of microwave components better than its counterpart. Besides, the modeling of microwave circuits and slotted patch antennas is examined to demonstrate the validity of this technique. Copyright © 2012 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A non-coherent serial acquisition scheme for direct sequence spread spectrum communication systems is analyzed and discussed in this paper. The adaptive thresholding based on constant false alarm rate and multilayer perceptron neural network (MLP-NN) techniques are combined to improve the performance of code division multiple access systems. One of the most important problems in code acquisition of pseudo-noise sequences for multiuser detection is the presence of interferences caused by the multiple access technique and multipath replicas. To solve this problem, an MLP-NN is trained and adapted to work as a constant false alarm rate detector using the error back propagation gradient descendent algorithm. It is named MLP-NN adaptive processor. The performance of this proposed algorithm is presented using the serial search acquisition system, which is chosen because of its simple hardware implementation. The performance of the MLP-NN adaptive processor algorithm in homogeneous and non-homogenous environments for additive white Gaussian noise and Rayleigh fading channels is evaluated via computer simulations. The obtained results are compared to other serial acquisition schemes using the cell-averaging adaptive processor, the order statistics adaptive processor, and the automatic censoring adaptive processor algorithms. [ABSTRACT FROM AUTHOR]

This paper presents an advanced analytical neuro-space mapping (neuro-SM) technique for accurate and efficient modeling of transistor devices. This is an improvement over the existing neuro-SM, which aims to use neural networks to map a given approximate device model towards an accurate model. The proposed neuro-SM retains the ability of the existing neuro-SM in modifying the voltage relationship between the given approximate device model and the accurate model. The proposed technique can also map the current relationship between the given model and the accurate model. In this way, the proposed neuro-SM can produce improved accuracy over the existing neuro-SM. In addition, analytical formulas of mapping and sensitivities of the direct current, small-signal S parameter, and large-signal harmonic of the proposed neuro-SM model with respect to mapping parameters and coarse-model parameters are also derived. The sensitivity analysis can be used with a gradient-based training technique to improve the model training efficiency. The validity and efficiency of the proposed approach are verified through 2 transistor modeling examples and use of the proposed neuro-SM models in a large-signal behavior analysis of an amplifier. [ABSTRACT FROM AUTHOR]

In this paper, the bias-dependent current-voltage (I-V) characteristics and their high-order derivatives of GaAs pseudomorphic high electron mobility transistors (pHEMTs) have been modeled over a wide temperature range. To simulate these characteristics at different temperatures, the model is developed considering the dependence on the ambient temperature. It is the first time that the temperature-dependent high-order derivatives of I-V characteristics of pHEMT are predicted, which can guarantee their accuracy under different bias conditions. The artificial neural networks are employed with the temperature as one of the input variables. The validity of this model has been demonstrated by comparing the measured and modeled Ids and its derivatives ( gm, gm2 and gm3, derived from the I-V characteristics numerically) of a GaAs pHEMT at different temperature range (250-400 K, with step of 50 K). The results show that the proposed model has a better agreement of high-order derivatives than the popularly used Angelov model, especially for the third-order derivative. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

The wave approach has appeared as a very efficient tool for modeling as well as for measurements of noise parameters of microwave transistors. Having in mind the attractiveness of transistors in gallium nitride technology in modern communication systems, where it is very important to keep the noise on a low level and, thus, to have accurate transistor noise models, in this paper, the wave approach is applied for the noise modeling of high electron-mobility transistor in gallium nitride technology. The noise wave representation of the transistor intrinsic circuit noise is used, where the noise wave parameters are modeled by exploiting the artificial neural networks. The modeling results, compared with the measured data and with those obtained by the conventional noise equivalent circuit model, provide a verification of the developed model accuracy. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A novel approach for parameter extraction of microwave transistor noise models based on artificial neural networks is proposed in this work. Neural networks are applied to determine parameters of the noise model directly from the measured noise and small-signal scattering parameters without any optimization procedure. Moreover, unlike the similar existing procedures, development of the extraction procedure does not require any measured data or optimizations in a circuit simulator, making the procedure more efficient, as described in detail in the paper. The approach has been applied to extraction of the Pospieszalski's noise model parameters for a specific pseudomorphic high-electron-mobility transistor (pHEMT) device working under different temperatures. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

The FinFET architecture is currently attracting increasing attention to enable further downscaling of the complementary metal-oxide-semiconductor (CMOS) technology. The interest towards the FinFET technology for microwave applications is not only limited to transistors but extended also to varactors. Therefore, there is a need for efficient and accurate varactor models in the high-frequency range. In this paper, an artificial neural networkbased behavioral model of varactors fabricated in advanced FinFET technology is proposed. The model is developed and verified by comparing measured and simulated scattering parameters up to 50 GHz. The extracted model can reproduce very well the measured behavior of the tested varactor before and after applying the de-embedding procedure based on open dummy structure. [ABSTRACT FROM AUTHOR]

Most of the electromagnetic devices, especially electrical machines, have the disadvantage to be exposed to high vibrations caused by magnetic forces. The aim of this study is to propose a methodology to optimize the cylindrical stators generally used in electrical machines regarding the vibration phenomena. Techniques for vibration reduction require knowledge of the proper frequencies, which depend on mechanical shapes and dimensions as well as material properties such as mass density, Young's modulus and Poisson's ratio. This paper proposes a new approach which is based on the identification of mass density (lamination stacking factor) and Young's modulus in the goal to minimize the vibratory behavior of electrical machines. In this goal, we have used artificial intelligent and finite element method (FEM) analysis to solve the magneto-mechanical inverse problem (IP). In the proposed approach, a Multilayer Perceptron Neural Network (MLPNN) is used as a forward model in order to decrease the FEM time consuming. Thus, a Genetic Algorithm (GA) is used to solve the IP in a reasonable time of running. An example study of an induction machine proves that the developed approach may be applied in both design and identification applications. Copyright © 2011 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

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 ƒmax 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. [ABSTRACT FROM AUTHOR]

This paper presents two new methods, space mapping (SM) with prior knowledge input (PKI-D) with difference and compound space mapping-based neuromodeling. Both methods combine two powerful techniques, space mapping-based neuromodeling and PKI-D with difference. The knowledge-based modeling methods in the RF/microwave literature merge the prior knowledge about the device to be modeled with neural network structures while a knowledge-based method, SP, focuses on reducing the computational burden. The main advantage of the proposed methods over these already existing knowledge-based methods are their better extrapolation capability and reduced number of training set data. The simulation results obtained reveal that both methods decrease the cost of training and improve the extrapolation capability and output performance of the SP-based neuromodeling. Copyright © 2007 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

This paper introduces two time-domain field-based optimization procedures for microwave engineering. The methods are built on the foundations of MATLAB's optimization and neural network toolboxes. The first procedure makes use of a direct connection linking MATLAB's optimization toolbox with MEFISTO-3D NOVA. In this approach the field simulator acts as an objective function server for the optimization toolbox; these two programs work cooperatively with each other to tune the structure parameters to obtain a target response. The second procedure is an indirect optimization approach that makes use of MATLAB's neural network toolbox in conjunction with MEFISTO-3D NOVA to create a neural network model to emulate the structure of interest; the resulting neural network model is then used an objective function server in a normal MATLAB optimization process. Copyright © 2006 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

This paper describes a systematic approach for extracting the lumped circuit parameters of microstrip discontinuities from their S-parameter characteristics—which are assumed to be known—by using a three- step procedure. We begin by computing the lumped equivalent circuit parameters, e.g., L, C, R and/or G, with a direct extraction procedure, which relates the S-parameters to the Z- (or Y-) parameters of the discontinuity. Next, we use these values as initial guesses for an optimization code based on the genetic algorithm, to obtain circuit parameters that are invariant over a specified frequency range. Finally, we develop an artificial neural network model for predicting the values of these circuit elements rapidly as we change the physical and electrical parameters of the discontinuity, for instance the width or height of the etch, and the complex dielectric constant of the substrate. The application of the direct extraction technique is first illustrated by considering a number of representative two-port microstrip discontinuities, viz., the chamfered bend and gap discontinuities. It is then extended to three-port structures, for instance, a T-junction. The validation is given for the case examples investigated by comparing the S-parameter calculated from the equivalent circuit with those obtained from a full-wave EM solver. We also use the chamfered bend as a case example to illustrate the application of this three-step procedure. Copyright © 2002 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A scalable model combining the advantages of the compact model and space-mapping neural network (SMNN) has been presented to characterize radio-frequency behaviors of on-chip spiral inductors. The physics-based T equivalent circuit model has been used for constructing the proposed scalable SMNN model. All values of the T model elements are fast and accurately extracted based on the mathematical formulations derived by analyzing the resonant responses. A 4-layer perceptron neural network has been applied for the space mapping of the T model and the measurement data. Compared with the conventional models of on-chip spiral inductors, the proposed SMNN model not only preserves the accuracy of measurement data but also runs as fast as an approximate compact model. The presented SMNN model has been verified by a series of octagon spiral inductors fabricated by 130-nm BiCMOS process of HHNEC. Excellent agreements are obtained between the measurement and simulation of the proposed SMNN model up to 40 GHz. [ABSTRACT FROM AUTHOR]

This article proposes a neural network-based approach to increase accuracy of two-dimensional direction of arrival (DOA) estimation of an electromagnetic signal. The proposed method combines two neural networks developed using simulated and small amount of empirical data, respectively. The output of the simulation-based neural network represents approximate information on DOAs. It is then considered as a priori knowledge for the small empirical network that is crucial for obtaining more accurate DOA estimates. The developed cascade-connected model is validated using real data from a rectangular antenna array. Improvements in terms of accuracy and reliability are obtained and compared with the MUSIC algorithm. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

In this paper, an innovative Artificial Neural Network (ANN)‐based approach to solve the bi‐dimensional Electric Capacitance Tomography inverse problem is proposed in order to assess the health state of standing trees. Using a set of examples, an ANN is trained to solve the direct problem, ie, to associate the properties of the tree trunk affected by defects to a set of capacitance values. Then, the trained ANN is inverted in order to obtain the defect position and dimensions on the basis of measured capacitance values. The potentiality of the method is presented making reference to several target examples. [ABSTRACT FROM AUTHOR]

A novel large‐signal model construction technique is proposed in this paper. High‐order current and charge sources with additional dimensions derived from pulsed small‐signal measurement data are used to describe the dispersive behaviour of FETs. Added dimensions can effectively account for trapping‐induced dispersions. Instead of look‐up tables, empirical functions and artificial neural network are adopted to implement the model into simulators, which can reduce the volume of the model and the possible oscillation caused by interpolation and provide better prediction performance beyond the measurement data range. The validity of proposed modelling technique has been verified by a 2 × 75um GaAs pHEMT. This proposed technique can be easily extended to GaN devices with the similar procedure. [ABSTRACT FROM AUTHOR]

In this paper, by using of gas flow pattern, a novel neural network‐based fault detection method is presented to detect the leakage in the gas pipeline. The pipe is divided into four segments, and each segment is modeled by using input/output pressure of the gas flow. For this purpose, the acquired practical data from the real life gas pipeline are gathered and utilized for training a neural network to model the process. Some of the data are used for training set to adjust the neural network weights, and others are used to evaluate the performance of the neural network‐based fault detection system. Gathered practical data from a real life pipeline made sure that the proposed method is prominent and applicable for practical implementations. The model was verified with the data obtained from the test in the actual pipeline and compared with leakage mode. [ABSTRACT FROM AUTHOR]

Artificial neural network modeling techniques have been recognized as important vehicles in the microwave computer‐aided design (CAD) area in addressing the growing challenges of designing next generation microwave device, circuits, and systems. This article provides an overview of recent advances in knowledge‐based neural network model generation and extrapolation techniques for microwave applications. We first introduce the unified knowledge‐based neural network structure optimization technique. Using the distinctive property for feature selection of l1 optimization, this unified modeling technique efficiently determines the type and topology of the mapping structure in a knowledge‐based model. This knowledge‐based model structure optimization technique is more flexible and systematic, and can further speed up the knowledge‐based neural model development. As a further advancement, we also discuss the advanced multi‐dimensional extrapolation technique for neural‐based microwave modeling. The purpose is to make the neural network model can be reliably used not only inside the training range but also outside the training range. Multi‐dimensional cubic polynomial extrapolation formulation and optimization over grids outside the training range are utilized to make neural models more robust and reliable when they are used outside the training range. The validity of these techniques is demonstrated by microwave modeling examples. [ABSTRACT FROM AUTHOR]

The main inconvenience in design process of modern high performance reflectarray antennas is that these designs are heavily depended on full‐wave electromagnetic simulation tools, where in most of the cases the design optimization process would be an inefficient or impractical. However, thanks to the recent advances in computer‐aided design and advanced hardware systems, artificial neural networks based modeling of microwave systems has become a popular research topic. Herein, design optimization of an alumina‐based ceramic substrate reflectarray antenna by using multilayer perceptron (MLP) and 3D printing technology had been presented. MLP‐based model of ceramic reflectarray (CRA) unit element is used as a fast, accurate, and reliable surrogated model for the prediction of reflection phase of the incoming EM wave on the CRA unit cell with respect to the variation of unit elements design parameters, operation frequency, and substrate thickness. The structural design of a reflectarray antenna with nonuniform reflector height operating in X band has been fabricated for the experimental measurement of reflectarray performance using 3D printer technology. The horn feeding based CRA antenna has a measured gain characteristic of 22 dBi. The performance of the prototyped CRA antenna is compared with the counterpart reflectarray antenna designs in the literature. [ABSTRACT FROM AUTHOR]

To facilitate the frequency regulation, here an adaptive artificial neural network (ANN) tuned proportional‐integral‐derivative (PID) controller is suggested for load frequency control (LFC) investigation in a system with distributed generation (DG) resources. The various DG resources include wind turbine generators (WTG), battery energy storage system (BESS), aqua electrolyzer (AE), diesel engine generators (DEG), and fuel cell (FC). Initially, an isolated thermal generating system is considered with DG. Then an interconnected two‐area thermal power system with DG is considered for LFC analysis. The implemented PID controller parameters are achieved using two methodologies. In the first case, the PID controller parameters are tuned by a recent optimization technique known as grasshopper optimization algorithm (GOA). In the second case, the PID controller parameters are tuned by an ANN. The dynamic behavior of the two categories of the system is inspected with GOA tuned PID controller and ANN tuned PID controller and it is established that ANN tuned PID controller exhibits superior performance as compared to GOA tuned PID controller in terms of time‐based performance evaluative factors such as minimum undershoots, settling time and maximum overshoots. Also, the robustness of the recommended ANN tuned PID controller is verified by applying random loading in the system. [ABSTRACT FROM AUTHOR]

Abstract: There are many applications to ranking including page searching, question answering, recommender systems, sentiment analysis, and collaborative filtering, to name a few. In the past several years, machine learning and information retrieval techniques have been used to develop ranking algorithms, and several listwise approaches to learning to rank have been developed. We propose a new method, which we call GeneticListMLE++ and GeneticListNet++, which builds on the original ListMLE and ListNet algorithms. Our method substantially improves on the original ListMLE and ListNet ranking approaches by incorporating genetic optimization of hyperparameters, a nonlinear neural network ranking model, and a regularization technique. [ABSTRACT FROM AUTHOR]

Artificial neural networks (ANNs) have been often used for engineering design problems. In this work, an inverse model of a reconfigurable N-shaped microstrip patch antenna which is formed by ANN is considered to find design parameters. For this task, knowledge-based response correction consists of two steps, which include generating response using multilayer perceptron as a first step and correcting this response using knowledge based methods such as source difference, prior knowledge input, and prior knowledge input with difference as a second step. The proposed antenna has four states of operation controlled by two Positive-Intrinsic-Negative (PIN) diodes with ON/OFF states. The two-step ANN models are inversely trained using the optimum of the resonant frequency parameter as the input and the physical dimensions of the proposed antenna as outputs of the multilayer perceptron. The outputs and, in some methods, the input parameters of the multilayer perceptron are sent as input to the knowledge-based models while the obtained outputs from the two steps are the results of the new physical dimensions of the redesigned reconfigurable antenna that will be compared and analyzed. This input/output complexity of the proposed reconfigurable antenna allows an accurate and fast inverse model to be developed with less training data. Users may use this antenna and its ANN models to develop new products in the market where any frequency in the operating region can be given to the input to result an appropriate form of the new reconfigurable antenna. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

In this article, a simple, accurate, fast, and reliable black-box modeling is proposed for the scattering (S)-parameters and noise (N)-parameters of microwave transistors using the general regression neural network (GRNN) with the substantially reduced measurements and computational cost. In this modeling method, GRNN is employed as a nonlinear extrapolator to generalize the S-data and N-data belonging to only a single bias voltage in the middle region into the entire device operation domain of the bias condition (VDS/VCE, IDS/IC, f) within the shortened human effort. The proposed method is implemented to the modeling of the two transistors BFP640 and ATF-551 M4 as study cases. Thus, comparisons are made with the multilayer perceptrons, trained by the two standard backward propagation algorithms, which are the Levenberg-Marquardt, Bayesian regularization and the 10 data mining methods recently published in the literature using the chosen training data sets in both ınterpolation and extrapolation types of generalization. All the comparisons are achieved using four criteria commonly used in the literature. It can be concluded that GRNN is found to be a fast and accurate modeling method that extrapolates the reduced amount of training data consisting of measured S-parameters and N-parameters at the typical currents of the middle bias voltage to the wide operating range. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A methodology based on the support vector machine (SVM) combined with a hybrid kernel function (HKF) for accurately modeling the resonant frequencies of the compact microstrip antenna (MSA) is presented and dedicated to reduce the number of samples and simplify the structure when predicting the resonant frequency of the compact MSA by artificial neural network. The parameters of the SVMs and weight coefficients of the HKF are optimized by means of particle swarm optimization algorithm. In addition, two different kernel functions (KFs), namely polynomial KF (a kind of global KF) and Cauchy KF (a kind of local KF), are employed to overcome the disadvantages of traditional KF. The proposed method is validated by the UCI database. The evaluation results show that the HKF can improve the learning ability and generalization ability of the SVM. Furthermore, the resonant frequencies of a planar inverted F-shaped antenna and an L-shaped MSA are modeled by the proposed method. Predictive results with high accuracy demonstrate that the particle swarm optimization-based SVM with the HKF can improve the prediction accuracy for a small dataset. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Because the nonlinear and time-varying characteristics of the continuously variable transmission system operated using a six-phase copper rotor induction motor are unknown, improving the control performance of the linear control design is time-consuming. To capture the nonlinear and dynamic behaviour of the six-phase copper rotor induction motor servo-driven continuously variable transmission system, a blend modified recurrent Gegenbauer orthogonal polynomial neural network (NN) control system, which has the online learning capability to return to the nonlinear time-varying system, was developed. The blend modified recurrent Gegenbauer orthogonal polynomial NN control system can perform overseer control, modified recurrent Gegenbauer orthogonal polynomial NN control, and recompensed control. Moreover, the adaptation law of online parameters in the modified recurrent Gegenbauer orthogonal polynomial NN is based on the Lyapunov stability theorem. The use of amended artificial bee colony optimization yielded two optimal learning rates for the parameters, which helped improve convergence. Finally, comparison of the experimental results of the present study with those of previous studies demonstrated the high control performance of the proposed control scheme. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

The microgrid (MG) is described as an electrical network of small modular distributed generation, energy storage devices and controllable loads. In order to maximize the output of solar arrays, maximum power point tracking (MPPT) technique is used by artificial neural network (ANN), and also, control of turbine output power in high wind speeds is proposed using pitch angle control technic by fuzzy logic. To track the maximum power point (MPP) in the photovoltaic (PV), the proposed ANN is trained by the genetic algorithm (GA). In other word, the data are optimized by GA, and then these optimum values are used in ANN. The simulation results show that the ANN-GA in comparison with the conventional algorithms with high accuracy can track the peak power point under different insolation conditions and meet the load demand with less fluctuation around the MPP; also it can increase convergence speed to achieve MPP. Moreover, pitch angle controller based on fuzzy logic with wind speed and active power as inputs that have faster responses which leads to have flatter power curves enhances the dynamic responses of wind turbine. The models are developed and applied in Matlab/Simulink. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Gallium nitride high electron-mobility transistors have gained much interest for high-power and high-temperature applications at high frequencies. Therefore, there is a need to have the dependence on the temperature included in their models. To meet this challenge, the present study presents a neural approach for extracting a multi-bias model of a gallium nitride high electron-mobility transistors including the dependence on the ambient temperature. Accuracy of the developed model is verified by comparing modeling results with measurements. Copyright © 2014 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

SUMMARY Three nonlinear reduced-order modeling approaches are compared in a case study of circuit variability analysis for deep submicron complementary metal-oxide-semiconductor technologies where variability of the electrical characteristics of a transistor can be significantly detrimental to circuit performance. The drain currents of 65 nm N-type metal-oxide-semiconductor and P-type metal-oxide-semiconductor transistors are modeled in terms of a few process parameters, terminal voltages, and temperature using Kriging-based surrogate models, neural network-based models, and support vector machine-based models. The models are analyzed with respect to their accuracy, establishment time, size, and evaluation time. It is shown that Kriging-based surrogate models and neural network-based models can be generated with sufficient accuracy that they can be used in circuit variability analysis. Numerical experiments demonstrate that for smaller circuits, Kriging-based surrogate modeling yields results faster than the neural network-based models for the same accuracy whereas for larger circuits, neural network-based models are preferred as, in all metrics, better performance is obtained. Within-die variations for an XOR circuit are analyzed, and it is shown that the nonlinear reduced-order models developed can more effectively capture the within-die variations than the traditional process corner analysis. Copyright © 2011 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

This study proposes neural network-based iterative inverse solutions for non-destructive evaluation (NDE) in which vector finite elements (VFEM) represent the forward model that closely models the physical process. The iterative algorithm can eventually estimate the material parameters. Vector finite element method global matrix is stored in a compact form using its sparsity and symmetry. The stored matrix elements are employed as the neurons weights, and preconditioning techniques are used to accelerate convergence of the neural networks (NN) algorithm. Detailed algorithm describing this new method is given to facilitate implementation. Combining vector finite elements and NNs offers several advantages over each technique alone, such as reducing memory storage requirements and the easily computed fixed weights of the NN. Various examples are solved to show the performance and usefulness of the proposed method, including lossy printed circuit board and lossy inhomogeneous cylindrical problems with ferromagnetic materials. These solutions compare very well with other published data where the maximum relative error was 5%. Copyright © 2010 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

A relevant automated electromagnetic (EM) optimization method is presented. This optimization method combines a rigorous and accurate global EM analysis of the device performed with a finite element method (FEM) and a fast analytical model deducted from its segmented EM analysis applying a neural network approximation. First we present our optimization tools, then we describe our full EM optimization method with the definition of the analytical model. Afterward, we apply it to optimize three volumetric dielectric resonator (DR) filters, considering the novel topology of the dual-mode filter. The accuracy of this automated method is demonstrated considering the good agreement between theoretical optimization results and experimental ones. Copyright © 2000 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Accurate design of miniaturized antenna is constrained by the limited well‐formulated exact mathematical expressions. Demands for smart devices with features like portability, implantability, and configurability have further placed bigger challenges in front of the antenna design engineers or scientists. As a part of the search for various solutions, many innovative approaches have been proposed by various authors in different literatures. Application of soft computing is also another design approach to accurate design of fractal antenna. Here, the authors have attempted to propose a better solution to miniaturized antenna and its design. A fractal antenna based on circular outer geometry has been proposed as a solution to the search of miniaturized antennas, and a particle swarm optimization–based selective artificial neural networks ensemble is developed, which is employed as the objective function of a bacterial foraging optimization algorithm leading to a hybridized algorithm. The developed hybrid algorithm is utilized to develop the proposed antenna at 2.45 GHz. A good agreement of the simulated, desired, and experimental results validates the proposed design approach. [ABSTRACT FROM AUTHOR]