Temperature‐dependent models of low‐noise microwave transistors based on neural networks

Neural networks are proposed for efficient temperature‐dependent modeling of small‐signal and noise performances of low‐noise microwave transistors over a wide temperature range. The proposed models can be based either on neural networks only or on a combination of neural networks and empirical transistor models. © 2005 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2005.     

Small‐signal and noise modeling of class of HEMTs using knowledge‐based artificial neural networks

In this article, the neural network approach is exploited for development of bias‐dependent small‐signal and noise models of a class of microwave field effect transistor (FETs) made in the same technology but differing in the gate width. The prior knowledge neural approach is applied. Introducing gate width at the input of proposed neural networks, as well as the S/noise parameters of a device that belongs to the same class as the modeled device representing the prior knowledge, leads to very accurate scattering and noise parameters' modeling, as exemplified by modeling of class of pseudomorphic high electron mobility transistor (pHEMT) devices. © 2012 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2013.     

Application of data mining methods to efficient microwave active device modeling

In this work, the signal and noise behaviors of a microwave transistor within its operation domain (voltage drain to source–V_DS, current of drain to source—I_DS, frequency—f) are modeled by data mining techniques (DMT) without using any information on the microwave circuit theory. The device is modeled by a black box whose small signal (S) and noise parameters are evaluated through data mining techniques, based on the fitting of both of these parameters for multiple bias and configuration. It has been shown that DMT have a high potential of faithful and efficient device modeling. © 2012 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2013.     

Cost‐effective global surrogate modeling of planar microwave filters using multi‐fidelity bayesian support vector regression

A computationally efficient method is presented for setting up accurate Bayesian support vector regression (BSVR) models of the highly nonlinear |S₂₁| responses of planar microstrip filters using substantially reduced finely discretized training data (compared to traditional design of experiments techniques). Inexpensive coarse‐discretization full‐wave simulations are exploited in conjunction with the sparseness property of BSVR to identify the regions of the input space requiring denser sampling. The proposed technique allows for substantial reduction (by up to 51%) of the computational expense necessary to collect the finely discretized training data, with negligible loss in predictive accuracy. The accuracy of the reduced‐data BSVR models is confirmed by their use within a space mapping optimization algorithm. © 2013 Wiley Periodicals, Inc. Int J RF and Microwave CAE 24:11–17, 2014.     

Hybrid modeling of microwave devices using multi‐kernel support vector regression with prior knowledge

This article proposes a support‐vector hybrid modeling method of microwave devices when only a small number of measurements are available. In this method, a hybrid model of microwave device has been obtained by combining a coarse model and a support‐vector model, where the coarse model is complemented by a support‐vector model capable of correcting the difference between the measurements and the coarse model. The support‐vector model was developed using a novel algorithm. In the algorithm, multi‐kernel and prior knowledge from a calibrated simulator were incorporated into the framework of the linear programming support vector regression by utilizing multiple feature spaces and modifying the optimization formulation. The experimental results from two microwave devices show that the hybrid modeling can enhance the physical meaning of the support‐vector model and improve the modeling accuracy for a small dataset, and that the proposed algorithm shows great potential in some applications where sufficient experimental data is difficult and costly to obtain, but the prior knowledge from a simulation model is available. The hybrid modeling is suited to a microwave computer‐aided design tool or an automatic tuning robot. © 2014 Wiley Periodicals, Inc. Int J RF and Microwave CAE 25:219–228, 2015.     

Gain gradients applied to optimization of distributed‐parameter matching circuits for a microwave transistor subject to its potential performance

In this article, a rigorous design procedure is carried out for a microwave amplifier by employing the Feasible Design Space and simple analytical gain gradients of the matching circuits. Physical lengths and characteristic impedances of the transmission lines used in the matching circuits are chosen as the design variables and their lower and upper limits are bounded by the limits of the planar transmission line technology so that resulted microwave amplifier can be realized by this technology. Feasible Design Target Space is determined by the compatible performance [noise (F), input VSWR (V_i), gain (G_T)] triplets and their source Z_S(ω_i) and load Z_L(ω_i) terminations resulted from the performance characterization of the active device. These triplets take into account the physical limitations of the device and realization conditions so that F_req ≥ F_min, V_ireq ≥ 1, G_{T min} ≤ G_{T req} ≤ G_{T max}; and Z_S(ω_i) and Z_L(ω_i) terminations be taken place within the “Unconditionally Stable Working Area”. Design of the amplifier for the compatible performance triplets is reduced to the design of the Z_S(ω_i) and Z_L(ω_i), i = 1…N terminations, which is achieved by the gain optimization of the two passive, reciprocal matching two‐ports using the Darlington theorem. Analytical expressions of the gain gradients of the matching circuits are obtained by the two different methods: (i) chain sensitivity matrix approach; (ii) adjoint network approach. Gain gradients of the L‐, T‐, and Π‐types of distributed‐parameter matching circuits are obtained as worked examples. Then typical design examples are given with together the synthesized, target, simulated characteristics. © 2008 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2008.     

基于人工免疫算法的最小二乘支持向量机参数优化算法*

针对最小二乘支持向量机（LSSVM）处理大数据集时确定最优模型参数耗时长、占内存大的问题,提出了一种基于人工免疫算法的参数寻优方法。通过分析LSSVM模型参数对分类准确率的影响发现,存在多种参数组合,使得分类准确率相同;当其中一个参数固定,另外一个参数在某些范围内变化取值时,它们的组合并不影响分类的准确率。将LSSVM模型参数作为抗体的基因设计了抗体的编码方案,利用人工免疫算法对LSSVM参数优化搜索。仿真结果表明,与使用交叉验证和网格搜索方法相比,提出的LSSVM参数优化算法在不降低分类准确率的前提下,寻优效率大大提高。     

Modeling and optimization of microwave filter by ADS‐based KBNN

When microwave devices are designed by knowledge‐based neural network (KBNN), the empirical formula is always used as priori knowledge. However, it is difficult to derive the corresponding formulas for the most electromagnetic problems, especially for complex electromagnetic problems, the formula derivation is almost impossible. In this article, they combine neural network with simulation software and use results of Agilent ADS as priori knowledge and HFSS as teaching signal to train the neural network by particle swarm optimization (PSO), which solves the difficulty in obtaining priori knowledge and effectively reduces the complexity of the neural network structure. Based on the KBNN, the microwave filters are designed. The results of optimization satisfy the required specifications which show the effectiveness and superiority of the method.     

Trust region‐based optimization of PKI neural models for RF/microwave devices

This article proposes a new trust region‐based optimization technique for Radio Frequency (RF)/microwave devices. The proposed approach is apt for modeling scenarios, where standard ANN multilayer perceptron (MLP) and Prior Knowledge Input (PKI) models fail to deliver a satisfactory model. This approach feeds output of standard ANN model as knowledge input to PKI model. The ANN model and the PKI model form a symbiotic pair to yield accurate results. In this paper, the dogleg routine is exploited in the process of optimization to obtain valid trust region steps. The proposed method is compared with sensitivity technique via several RF/microwave components. © 2012 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2013.     

基于遗传算法对支持向量机模型中参数优化

支持向量机是基于统计学习理论的结构风险最小化原理基础上提出来的一种学习算法,其在理论上保证了模型的最大泛化能力.针对支持向量机结构参数的选取在没有理论支持,选取又比较困难的情况下,对影响模型分类能力的相关参数进行了研究,提出了一种基于遗传算法和十折交叉检验相结合的遗传支持向量机(GA-SVM)算法,利用遗传算法的全局搜索特性得到支持向量机(SVM)的最优参数值,并用算例表明了此算法有效提高了分类的精度和效率.     

一种基于GA和支持向量机的车牌字符识别方法

以高斯核为其核函数的支持向量机在实际应用中表现出优良的学习性能,被广泛应用于模式分类中。支持向量机的识别性能对参数的选取是敏感的,惩罚因子C和核函数参数σ对支持向量机性能会产生重要的影响。针对高斯核支持向量机在车牌字符识别问题中的应用,提出了一种基于遗传算法的参数选择方法。首先确定合适的遗传算法适应度函数,然后利用遗传算法对支持向量机的参数进行优化,最后在各个识别子网中分别采用参数优化后的支持向量机对车牌字符进行识别。实验结果表明,该方法取得了令人满意的识别率。     

基于支持向量机的蒙脱土悬浮液电动电位预测

应用多元线性回归、人工神经网络、支持向量机3种方法,对加入聚乙二醇、十二烷基苯磺酸钠、石油磺酸盐和部分水解聚丙烯酰胺四种处理剂的蒙脱土悬浮液的电动电位进行预测。在模型训练中,分别采用了神经网络集成和非启发式参数优化来提高人工神经网络和支持向量机模型的泛化能力。检验结果表明,参数优化的支持向量机模型预测精度最高,其平均误差率为3.88%,最大误差率为7.55%。     

一种并行协同粒子群优化的支持向量机预测模型

转炉提钒过程是一个非常复杂的多元非线性反应过程,从统计学和反应机理等角度出发,难以建立终点控制静态模型．针对这样的问题,提出了并行协同粒子群优化的支持向量机预测模型,不仅克服了支持向量机偏差ε和折中参数C选择的随机性,而且较好地解决了大数据集的快速并行计算,缩短了计算时间,从而有利于连续生产操作．试验表明,用该模型预测转炉提钒的冷却剂加入量和吹氧时间,结果的误差减小,满足了终点命中率在90%以上的指标,具有工程实用性．     

Fuzzy neural‐based approaches for efficient RF/microwave transistor modeling

In today's RF and microwave circuits, there is an ever‐increasing demand for higher level of system integration that leads to massive computational tasks during simulation, optimization, and statistical analyses, requiring efficient modeling methods so that the whole process can be achieved reliably. Since active devices such as transistors are the core of modern RF/microwave systems, the way they are modeled in terms of accuracy and flexibility will critically influence the system design, and thus, the overall system performance. In this article, the authors present neural‐ and fuzzy neural‐based computer‐aided design techniques that can efficiently characterize and model RF/microwave transistors such as field‐effect transistors and heterojunction bipolar transistors. The proposed techniques based on multilayer perceptrons neural networks and c‐means clustering algorithms are demonstrated through examples. © 2008 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2009.     

Support‐vector modeling of electromechanical coupling for microwave filter tuning

This article presents a support‐vector modeling method for electromechanical coupling of microwave filter tuning in the case of the scarcity of experimental data available. This has been done for the purpose of establishing an accurate coupling model which can be used in an automatic tuning device of volume‐producing filters. In the method, a coupling model that reveals the effect of mechanical structure on the filter electrical performance is established by using a proposed algorithm which can incorporate multi‐kernel and prior knowledge into linear programming support vector regression (LPSVR). Some experiments from three microwave filters have been performed, and the results confirm the effectiveness of the support‐vector modeling method. Moreover, the comparative results also show that the proposed multi‐kernel prior knowledge LPSVR can improve the data‐driven modeling accuracy of small dataset. The proposed algorithm show great potential in some problems where a sufficient experimental data is difficult and costly to obtain, but some prior knowledge data from a simulation model can be easily obtained. © 2012 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2013.     

Neural model of microwave cylindrical cavity loaded with arbitrary‐raised dielectric slab

A combined knowledge‐based neural‐multilayer perceptron (KBN‐MLP) model to account for a loading effect of arbitrary raised dielectric slab in a microwave cylindrical metallic cavity is presented. Existing partial knowledge about the resonant frequency behavior of loaded cavity is incorporated in the KBN part of suggested model. In comparison with the model based on classical MLP network, more accurate and efficient resonant frequencies calculation is achieved. © 2008 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2009.     

Neural network modeling of microwave FETs based on third‐order distortion characterization

A new method for characterization of HEMT distortion parameters, which extracts the coefficents of a Taylor series expansion of I_ds(V_gs, V_ds), including all cross‐terms, is developed from low‐frequency harmonic measurements. The extracted parameters will be used either in a Volterra series model around a fixed bias point for 3^rd‐order characterization of small‐signal I_ds nonlinearity, or in a large‐signal model of I_ds characteristic, where its partial derivatives are locally characterized up to the 3^rd order in the whole bias region, using a novel neural‐network representation. The two models are verified by one‐tone and two‐tone intermodulation distortion (IMD) tests on a PHEMT device. © 2006 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2006.     

Prediction of slot‐position and slot‐size of a microstrip antenna using support vector regression

This article presents a synthesis modeling scheme of rectangular microstrip antenna with support vector regression (SVR) scheme. Here, radiating patch and ground surface is loaded with two asymmetrical slots and two symmetrical slots, respectively. The position of the slots on the radiating patch as well as the size of the slots on the ground surface are predicted using SVR model and artificial neural network (ANN) model. A good convergence rate has been addressed in synthesis model by employing the adaptive step‐size. A comparison between SVR model and ANN model is presented where SVR is more accurate and faster than ANN. The suggested SVR approach is also validated by fabricating and characterizing a prototype of microstrip antenna. A very good agreement is observed in measured, simulated, and predicted results. The predicted microstrip antenna has displayed quite good agreement between measured and simulated performance parameters.     

改进型支持向量回归预测模型的轧机轧制力预测

对轧机轧制力预测模型进行研究.使用人工鱼群优化算法对支持向量回归(SVR)参数选取进行最优的参数组合,将粒子群优化算法引入到常规人工鱼群算法中,并对其进行改进,提高了人工鱼群算法的性能.研究结果表明:Ekelund模型的轧制力计算结果误差较大,超过了10％,常规SVR预测模型的轧制力预测精度低于10％,而本文研究的改进SVR预测模型得到的轧制力误差低于5％,说明通过人工鱼群算法优化SVR算法模型的参数能够提高预测模型的预测精度,并且预测消耗时间在3种预测模型中是最短的.     

Modeling power and intermodulation behavior of microwave transistors with unified small‐signal/large‐signal neural network models

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 I_ds as well as gate and drain charge Q_g and Q_d with respect to intrinsic voltages V_gs and V_ds 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.