Truncated-Newton training algorithm for neurocomputational viscoplastic model

We present an estimate approach to compute the viscoplastic behavior of a polymer matrix composite under different thermomechanical environments. This investigation incorporates computational neural network as the tool for determining the creep behavior of the composite. We propose a new second-order learning algorithm for training the multilayer networks. Training in the neural network is generally specified as the minimization of an appropriate error function with respect to parameters of the network (weights and learning rates) corresponding to excitory and inhibitory connections. We propose here a technique for error minimization based on the use of the truncated Newton (TN) large-scale unconstrained minimization technique with quadratic convergence rate. This technique offers a more sophisticated use of the gradient information compared to simple steepest descent or conjugate gradient methods. In this work we briefly specify the necessary details for implementing the TN method for training the neural networks that predicts the viscoplastic behavior of the polymeric composite. We provide comparative experimental results and explicit model results to verify the effectiveness of the neural networks-based model. These results verify the superiority of the present approach compared to the explicit modeling scheme. Moreover, the present study demonstrates for the first time the feasibility of introducing the TN method, with quadratic convergence rate, to the field of neural networks.     

应用前馈神经网络的大柔性机翼阵风响应分析

大柔性飞行器因结构重量低、柔性大使得机翼等部件在受载时产生较大的弹性变形,呈现显著的几何非线性效应,因此准确的结构大变形建模方法对于几何非线性气动弹性分析至关重要,而神经网络对非线性系统具有强大的拟合能力,可通过将神经网络应用于非线性结构建模,构造适用于结构大变形的前馈神经网络预测模型,在样本特征和数据结构相对较优的条件下结合曲面涡格法,搭建非线性气动弹性分析框架,对某机翼模型进行阵风响应计算;结果表明神经网络模型能准确预测大柔性机翼结构大变形,应用到气动弹性分析后能进行准确的阵风响应计算,验证了将神经网络应用到结构大变形预测的可行性,为以后机器学习技术与气动弹性分析结合的研究提供思路和方法。     

神经网络在热电厂对象建模中的应用

针对一个375 MW热电厂的锅炉-汽轮机系统仿真模型,采用多层前向神经网络进行离线建模;讨论了网络结构设计、训练算法等神经网络建模问题;采用相同的固定负荷数据分别建立了线性ARX模型和局部神经网络模型并做多步预测比较;通过对基于一层隐层的全局神经网络模型的训练和仿真,结果证实了神经网络在非线性系统建模和辨识上的有效性.     

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.     

Numerical prediction of elastic properties for carbon nanotubes reinforced composites using a multi-scale method

The main aim of this research was to investigate longitudinal elastic and effective modulus of composites reinforced with zigzag and armchair single-walled (CNT) and multi-walled carbon nanotubes (MWCNT) with different volume fractions and aspect ratios via finite element simulation. A three-phased volume element was adopted for the modeling of nanocomposite behavior and nonlinear spring elements were used to model interphase part joints and the effective force between nanotubes and resin were determined based on Lennard-Jones potential. After the evaluation and validation of the model, elastic modulus and Poisson’s ratio of composites reinforced with zigzag and armchair CNTs with different volume fractions and aspect ratios were extracted. It was found that by increasing volume fraction and aspect ratio, elastic modulus of representative volume element of composite was increased and its Poisson’s ratio was decreased. At similar aspect ratio and volume fraction, the elastic modulus of composites reinforced with armchair CNTs and Poisson’s ratio of those reinforced with zigzag CNTs were higher. Also, results showed that elastic modulus of composite was independent from elastic modulus of interphase.

     

A Neural-Network-Based Model for the Dynamic Simulation of the Tire/Suspension System While Traversing Road Irregularities

This paper deals with the simulation of the tire/suspension dynamics by using recurrent neural networks (RNNs). RNNs are derived from the multilayer feedforward neural networks, by adding feedback connections between output and input layers. The optimal network architecture derives from a parametric analysis based on the optimal tradeoff between network accuracy and size. The neural network can be trained with experimental data obtained in the laboratory from simulated road profiles (cleats). The results obtained from the neural network demonstrate good agreement with the experimental results over a wide range of operation conditions. The NN model can be effectively applied as a part of vehicle system model to accurately predict elastic bushings and tire dynamics behavior. Although the neural network model, as a black-box model, does not provide a good insight of the physical behavior of the tire/suspension system, it is a useful tool for assessing vehicle ride and noise, vibration, harshness (NVH) performance due to its good computational efficiency and accuracy.      

Nonlinear control structures based on embedded neural system models

This paper investigates in detail the possible application of neural networks to the modeling and adaptive control of nonlinear systems. Nonlinear neural-network-based plant modeling is first discussed, based on the approximation capabilities of the multilayer perceptron. A structure is then proposed to utilize feedforward networks within a direct model reference adaptive control strategy. The difficulties involved in training this network, embedded within the closed-loop are discussed and a novel neural-network-based sensitivity modeling approach proposed to allow for the backpropagation of errors through the plant to the neural controller. Finally, a novel nonlinear internal model control (IMC) strategy is suggested, that utilizes a nonlinear neural model of the plant to generate parameter estimates over the nonlinear operating region for an adaptive linear internal model, without the problems associated with recursive parameter identification algorithms. Unlike other neural IMC approaches the linear control law can then be readily designed. A continuous stirred tank reactor was chosen as a realistic nonlinear case study for the techniques discussed in the paper.     

Postbuckling of multilayer cylindrical and spherical shell panels reinforced with graphene platelet by isogeometric analysis

The present work fills a gap on the postbuckling behavior of multilayer functionally graded graphene platelet reinforced composite (FG-GPLRC) cylindrical and spherical shell panels resting on elastic foundations subjected to central pinching forces and pressure loadings. Based on a higher-order shear deformation theory and the von Kármán’s nonlinear strain–displacement relations, the governing equations of the FG-GPLRC cylindrical and spherical shell panels are established by the principle of virtual work. The non-uniform rational B-spline (NURBS) based isogeometric analysis (IGA), the modified arc-length method and the Newton’s iteration method are employed synthetically to obtain nonlinear load–deflection curves for the panels numerically. Several comparative examples are performed to test reliability and accuracy of IGA and arc-length method in present formulation and programming implementation. Parametric investigations are carried out to illustrate the effects of dispersion type of the graphene platelet (GPL), weight fraction of the GPL, thickness of the panel, radius of the panel and parameters of elastic foundation on the load–deflection curves of the FG-GPLRC shell panels. Some complex load–deflection curves of the FG-GPLRC cylindrical and spherical shell panels resting on elastic foundations may be useful for future references.

     

基于多层局部回归神经网络的多变量非线性系统预测控制

以罐式搅拌反应器为例,针对复杂多变量系统的强耦合性、非线性、时变性等问题,研究了多变量非线性系统的预测控制及改善控制性能的方法,采用多层局部回归神经网络离线建立预测模型,以偏差补偿和模型修正相结合的方式对预测模型进行误差补偿,以要线校正用于预测控制,通过对性能指标中的偏差项负指数加权,进一步改善预测控制性能,住址结果表明了控制算法的有效性。     

Two-dimensional modeling of material failure in reinforced concrete by means of a continuum strong discontinuity approach

The paper presents a new methodology to model material failure, in two-dimensional reinforced concrete members, using the Continuum Strong Discontinuity Approach (CSDA). The mixture theory is used as the methodological approach to model reinforced concrete as a composite material, constituted by a plain concrete matrix reinforced with two embedded orthogonal long fiber bundles (rebars). Matrix failure is modeled on the basis of a continuum damage model, equipped with strain softening, whereas the rebars effects are modeled by means of phenomenological constitutive models devised to reproduce the axial non-linear behavior, as well as the bond-slip and dowel effects. The proposed methodology extends the fundamental ingredients of the standard Strong Discontinuity Approach, and the embedded discontinuity finite element formulations, in homogeneous materials, to matrix/fiber composite materials, as reinforced concrete. The specific aspects of the material failure modeling for those composites are also addressed. A number of available experimental tests are reproduced in order to illustrate the feasibility of the proposed methodology.     

Elastic-plastic analysis of composite beams with incomplete interaction by finite element method

This paper presents a nonlinear finite element analysis of composite beams with incomplete interaction. A simplified nonlinear model is assumed in this approach. This is applied to the elastic-plastic analysis of reinforced concrete beams and composite beams with incomplete interaction. The numerical results are compared with the test results and existing values based on other numerical methods, and found to be in good agreement. The elastic-plastic behavior of partial composite beams without shear connectors in the negative bending moment region is discussed by the proposed method.     

NEURAL PREDICTIVE FORCE CONTROL FOR A HYDRAULIC ACTUATOR: SIMULATION AND EXPERIMENT

This paper proposes a neural-based predictive control algorithm for online control of a force-acting industrial hydraulic actuator. In the algorithm, a multilayer feedforward neural network is employed to modeling the highly nonlinear hydraulic actuator. The nonlinear neural model is instantaneously linearized at each sampling point. Estimated parameters from the linearized model are used in the generalized predictive control (GPC) algorithm to control the contact force. Simulation and experimental results show that the neural-based predictive controller can adapt to different environments and keep the contact force in a desired value despite high nonlinearity and uncertainty in the hydraulic actuator system.     

Neural network ensemble strategies for financial decision applications

Considerable research effort has been expended to identify more accurate models for decision support systems in financial decision domains including credit scoring and bankruptcy prediction. The focus of this earlier work has been to identify the “single best” prediction model from a collection that includes simple parametric models, nonparametric models that directly estimate data densities, and nonlinear pattern recognition models such as neural networks. Recent theories suggest this work may be misguided in that ensembles of predictors provide more accurate generalization than the reliance on a single model. This paper investigates three recent ensemble strategies: crossvalidation, bagging, and boosting. We employ the multilayer perceptron neural network as a base classifier. The generalization ability of the neural network ensemble is found to be superior to the single best model for three real world financial decision applications.     

A combination of linear and nonlinear activation functions in neural networks for modeling a de-superheater

This paper deals with modeling a power plant component with mild nonlinear characteristics using a modified neural network structure. The hidden layer of the proposed neural network has a combination of neurons with linear and nonlinear activation functions. This approach is particularly suitable for nonlinear system with a low grade of nonlinearity, which can not be modeled satisfactorily by neural networks with purely nonlinear hidden layers or by the method of least square of errors (the ideal modeling method of linear systems). In this approach, two channels are installed in a hidden layer of the neural network to cover both linear and nonlinear behavior of systems. If the nonlinear characteristics of the system (i.e. de-superheater) are not negligible, then the nonlinear channel of the neural network is activated; that is, after training, the connections in nonlinear channel get considerable weights. The approach was applied to a de-superheater of a 325 MW power generating plant. The actual plant response, obtained from field experiments, is compared with the response of the proposed model and the responses of linear and neuro-fuzzy models as well as a neural network with purely nonlinear hidden layer. A better accuracy is observed using the proposed approach.     

Application of the recurrent multilayer perceptron in modelingcomplex process dynamics

A nonlinear dynamic model is developed for a process system, namely a heat exchanger, using the recurrent multilayer perceptron network as the underlying model structure. The perceptron is a dynamic neural network, which appears effective in the input-output modeling of complex process systems. Dynamic gradient descent learning is used to train the recurrent multilayer perceptron, resulting in an order of magnitude improvement in convergence speed over a static learning algorithm used to train the same network. In developing the empirical process model the effects of actuator, process, and sensor noise on the training and testing sets are investigated. Learning and prediction both appear very effective, despite the presence of training and testing set noise, respectively. The recurrent multilayer perceptron appears to learn the deterministic part of a stochastic training set, and it predicts approximately a moving average response of various testing sets. Extensive model validation studies with signals that are encountered in the operation of the process system modeled, that is steps and ramps, indicate that the empirical model can substantially generalize operational transients, including accurate prediction of instabilities not in the training set. However, the accuracy of the model beyond these operational transients has not been investigated. Furthermore, online learning is necessary during some transients and for tracking slowly varying process dynamics. Neural networks based empirical models in some cases appear to provide a serious alternative to first principles models.     

Modeling of a proton exchange membrane fuel cell based on the hybrid particle swarm optimization with Levenberg–Marquardt neural network

This paper presents a nonlinear modeling approach of a proton exchange membrane fuel cell (PEMFC) based on the hybrid particle swarm optimization with Levenberg–Marquardt algorithm neural network (PSO-LM NN). The PSO algorithm converges rapidly during the initial stages of a global search, while it becomes extremely slow around the global optimum. On the contrary, the LM algorithm can achieve faster convergent speed around the global optimum, while it is prone to being trapped in the local minimum. Therefore the hybrid algorithm with a transition from PSO search to LM training is proposed to train the weights and thresholds of neural network, which aims to exploit the advantage of the both algorithms. An accurate mathematical model is an extremely useful tool for the fuel cell design, and neural network is an excellent optional tool for complex nonlinear dynamic system modeling such as PEMFC. In the paper, firstly a highly reduced PEMFC dynamic physical model is established to generate the data for the PSO-LM NN model training and validation, and then the neural network nonlinear autoregressive model based on the PSO-LM algorithm is applied in modeling PEMFC voltage and temperature model, and finally the validation test result demonstrates that the trained PSO-LM NN model can efficiently approach the dynamic behavior of a PEMFC.     

A plastic damage approach for confined concrete

There are many situations in which it is necessary to increase the capacity of structures in use. This need maybe either for a change of use or because the structures have suffered some damage or have shown little resistance in case of extreme loads such as earthquakes. The most common methods for repair and retrofit of reinforced concrete columns are concrete jacketing, steel jacketing and fiber wrapping. This last type of reinforcement has many advantages as it offers a high-strength, low-weight and corrosion-resistant jacket with easy and rapid installation. The reinforcement with composite materials improves shear and compression strength and ductility as a result of concrete core confinement. The present analytical and numerical ability to quantify the efficiency of fiber confinement is rather limited, especially with respect to ductility.A constitutive model that approximately reproduces the behavior of structural concrete elements under confinement is developed in this paper. The model allows the assessment of concrete columns and bridge piles repaired and/or reinforced with fiber reinforced composites (FRP). The model presented is a modification of an existing coupled plastic damage model. A new definition for the plastic hardening variable and a new yielding surface with curved meridians are proposed. Both improvements enable the adequate reproduction of concrete behavior in high confinement conditions.The comparison of numerical and experimental results shows the model capacity to simulate concrete behavior under triaxial compression conditions like the ones present in concrete columns confined with fiber reinforced composites.     

化工过程建模中的一类复合型模糊神经网络

针对化工非线性过程建模问题,本文提出了一类由函数逼近和规则推理网络构成的复合型模糊神经网络,其规则网络基于过程先验知识用于对操作区间的划分,而函数网络采用改进型模糊神经网络结构完成非线性函数逼近。该技术已成功地用于某工业尿素ＣＯ２汽提塔液位建模。