A CVD epitaxial deposition in a vertical barrel reactor: processmodeling using cluster-based fuzzy logic models

A chemical vapor deposition (CVD) epitaxial deposition process modeling using fuzzy logic models (FLM's) has been proposed. The process modeling algorithm consists of a cluster estimation method and backpropagation algorithm to construct a number of modeling structures from the training data. A decision rule based on the multiple correlation factor is used to obtain the optimum structure of the fuzzy model using the testing data. Upon the optimum structure being reached, the gradient-descent method is used to refer the parameters of the final fuzzy model using both training and testing data. The algorithm has been applied to a nonlinear function and a vertical chemical vapor deposition process. The results demonstrate the efficiency and effectiveness of the proposed fuzzy logic model in comparison with existing fuzzy logic models and artificial neural network models     

Learning algorithms for a class of neurofuzzy network and application

A class of neurofuzzy networks and a constructive, competition-based learning procedure is introduced. Given a set of training data, the learning procedure automatically adjusts the input space portion to cover the whole space and finds membership functions parameters for each input variable. The network processes data following fuzzy reasoning principles and, due to its structure, it is dual to a rule-based fuzzy inference system. The neurofuzzy model is used to forecast seasonal streamflow, a key step to plan and operate hydroelectric power plants and to price energy. A database of average monthly inflows of three Brazilian hydroelectric plants located at different river basins was used as source of training and test data. The performance of the neurofuzzy network is compared with period regression, a standard approach used by the electric power industry to forecast streamflows. Comparisons with multilayer perceptron, radial basis network and adaptive neural-fuzzy inference system are also included. The results show that the neurofuzzy network provides better one-step-ahead streamflow forecasting, with forecasting errors significantly lower than the other approaches.     

一种利用神经网络的故障模糊诊断系统

本文提出一种神经网络与模糊逻辑相结合的故障诊断系统,该系统包括2个方面:模糊推理模块和规则学习模块。模糊推理规则记忆在网络的记忆层中,记忆节点的激活水平则反映了输入矢量与已记忆规则的匹配程度;规则学习模块通过自组织聚类过程自动生成规则。作为该诊断系统的一个应用实例,模拟了旋转主轴的故障诊断试验。     

Evolutionary fuzzy neural networks for hybrid financial prediction

In this paper, an evolutionary fuzzy neural network using fuzzy logic, neural networks (NNs), and genetic algorithms (GAs) is proposed for financial prediction with hybrid input data sets from different financial domains. A new hybrid iterative evolutionary learning algorithm initializes all parameters and weights in the five-layer fuzzy NN, then uses GA to optimize these parameters, and finally applies the gradient descent learning algorithm to continue the optimization of the parameters. Importantly, GA and the gradient descent learning algorithm are used alternatively in an iterative manner to adjust the parameters until the error is less than the required value. Unlike traditional methods, we not only consider the data of the prediction factor, but also consider the hybrid factors related to the prediction factor. Bank prime loan rate, federal funds rate and discount rate are used as hybrid factors to predict future financial values. The simulation results indicate that hybrid iterative evolutionary learning combining both GA and the gradient descent learning algorithm is more powerful than the previous separate sequential training algorithm described in.     

The use of multilayer neural networks in material synthesis

This paper demonstrates the incorporation of a multilayer neural network in semiconductor thin film deposition processes. As a first step toward neural network-based process control, we present results from neural network pattern classification and beam analysis of reflection high energy electron diffraction RHEED images of GaAs/AlGaAs crystal surfaces during molecular beam epitaxy growth. For beam analysis, we used the neural network to detect and measure the intensity of the RHEED beam spots during the growth process and, through Fourier transformation, determined the thin film deposition rate. The neural network RHEED pattern classification and intensity analysis capability allows, powerful in situ real time monitoring of epitaxial thin film deposition processes. Our results show that a three layer network with sixteen hidden neurons and three output neurons had the highest correct classification rate with a success rate of 100% during testing and training on 13 examples     

A NEURAL NETWORK BASED FAULT FUZZY DIAGNOSTIC SYSTEM

A fault fuzzy diagnostic system(FFDS) based on neural network and fuzzy logic hybrid is proposed. FFDS consists of two modes: a fuzzy inference mode and a rule learning mode. The fuzzy inference rules are stored in the memory layer. The excitation levels of the memory neurons reflect the matching degrees between the input vectors and the prototype rules. In the rule learning mode, the rules can be produced automatically through the cluster process. As an application case of this diagnostic system, the fault diagnosis experiment of the rotating axis is simulated.     

A Hybrid Neuro-Fuzzy System for Adaptive Vehicle Separation Control

The primary purpose of this paper is to develop a robust adaptive vehicle separation control in the increasingly important roles of intelligent transportation system (ITS). A hybrid neuro-fuzzy system (HNFS) is proposed for developing the adaptive vehicle separation control to minimize the distance (headway) between successive cars. This hybrid system consists of two modules: a headway identification (prediction) module and a control decision module. Each of these modules is realized with a distinct neuro-fuzzy network that upgrades hierarchical granularity and reduces the complexity in the control system. Given the current headway and relative velocity between the two consecutive cars, the headway identification module predicts the headway of the next time instant. This identified headway, together with the desired velocity are input to the control decision module whose output is the actual acceleration/deceleration control output. The HNFS encapsulates the adaptive learning capabilities of a neural network into a fuzzy logic control framework to fine-tune the fuzzy control rules. On the other hand, rules derived initially from well-defined fuzzy phase plane accelerate the training of the neural network. Simulation results are very encouraging. It is observed that the headway decreases significantly without sacrificing speed. Furthermore, both stability and robustness of HNFS are demonstrated.     

神经网络自动生成模糊系统

在模糊系统的生成过程中，最主要的任务是隶属函数和模糊规则的提取和调整，但用传统方法，其工作量往往随变量数的增长而爆炸性地增加。为了解决这一问题，本文提出了一种新颖的，利用神经网络来自动地撮模糊系统 隶属函数和规则。     

A Recurrent Fuzzy-Network-Based Inverse Modeling Method for a Temperature System Control

Temperature control by a Takagi-Sugeno-Kang (TSK)-type recurrent fuzzy network (TRFN) designed by modeling plant inverse is proposed in this paper. TRFN is a recurrent fuzzy network developed from a series of TSK-type fuzzy if--then rules, and is characterized by structure and parameter learning. In parameter learning, two types of learning algorithms, the Kalman filter and the gradient descent learning algorithms, are applied to consequent parameters depending on the learning situation. The TRFN has the following advantages when applied to temperature control problems: 1) high learning ability, which considerably reduces the controller training time; 2) no a priori knowledge of the plant order is required, which eases the design process; 3) good and robust control performance; 4) online learning ability, i.e., the TRFN can adapt itself to unpredictable plant changes. The TRFN-based direct inverse control configuration is applied to a real water bath temperature control plant, where various control conditions are experimented. The same experiments are also performed by proportional-integral (PI), fuzzy, and neural network controllers. From comparisons, the aforementioned advantages of a TRFN have been verified     

Modeling stripline discontinuities by neural network with knowledge-based neurons

A three-layer neural network with knowledge-based neurons in the hidden layer (NNKBN) is presented for modeling stripline discontinuities. In NNKBN, prior knowledge for stripline discontinuity is incorporated into each hidden neuron. With knowledge-based neurons, the learning ability and generalization of the neural network are improved. Compared with conventional multi-layer perceptron neural network, the NNKBN can map the input-output relationships with fewer hidden neurons and has higher reliability for extrapolation beyond training data range. Two examples are given to illustrate the potential power of this approach.     

Neurofuzzy classification and rule generation of modes of radiowave propagation

This paper describes, in a neurofuzzy framework, a method for the classification of different modes of radiowave propagation, followed by generation of linguistic rules justifying a decision. Weight decay during neural learning helps in imposing a structure on the network, resulting in the extraction of logical rules. Use of linguistic terms at the input enables better human interpretation of the inferred rules. The effectiveness of the system is demonstrated on radiosonde data of four different seasons in India.     

A Novel Virtual Metrology Scheme for Predicting CVD Thickness in Semiconductor Manufacturing

In an advanced semiconductor fab, online quality monitoring of wafers is required for maintaining high stability and yield of production equipment. The current practice of only measuring monitor wafers may not be able to timely detect the equipment-performance drift happening in-between the scheduled measurements. This may cause defects of production wafers and, thereby, raise the production cost. In this paper, a novel virtual metrology scheme (VMS) is proposed for overcoming this problem. The proposed VMS is capable of predicting the quality of each production wafer using parameters data from production equipment. Consequently, equipment-performance drift can be detected promptly. A radial basis function neural network is adopted to construct the virtual metrology model. Also, a model parameter coordinator is developed to effectively increase the prediction accuracy of the VMS. The chemical vapor deposition (CVD) process in semiconductor manufacturing is used to test and verify the effectiveness of the proposed VMS. Test results show that the proposed VMS demonstrates several advantages over the one based on back-propagation neural network and can achieve high prediction accuracy with mean absolute percentage error being 0.34% and maximum error being 1.15%. The proposed VMS is simple yet effective, and can be practically applied to construct the prediction models of semiconductor CVD processes.     

Evolving Fuzzy Neural Networks for Extracting Rules

ＥｖｏｌｖｉｎｇＦｕｚｚｙＮｅｕｒａｌＮｅｔｗｏｒｋｓｆｏｒＥｘｔｒａｃｔｉｎｇＲｕｌｅｓ＊＊ＴｈｉｓｗｏｒｋｗａｓｓｕｐｐｏｒｔｅｄｂｙｔｈｅＣｌｉｍｂｉｎｇＰｒｏｇｒａｍｍｅ┐ＮａｔｉｏｎａｌＫｅｙＰｒｏｊｅｃｔｆｏｒＦｕｎｄａｍｅｎｔａｌＲｅｓ...     

闭环非线性对象辨识的模糊和神经网络方法

闭环系统的辨识是近年来国内很受重视的研究课题。本文基于闭环对象的历史数据，采用模糊方法构造系统的初始模型，以克服闭环数据稀疏给系统辩识带来的困难，并基于现场数据，采用ＯＬＳ算法对初始模型进行了修正，以提高系统的辨识精度。将这种方法应用于某合成氨过程的实际数据，得到了良好的辨识效果。     

Identification of pneumatic artificial muscle manipulators by a MGA-based nonlinear NARX fuzzy model

This study investigates the technique of modeling and identification of a new dynamic NARX fuzzy model by means of genetic algorithms. In conventional identification techniques, there are difficulties such as poor knowledge of the process, inaccurate process or complexity of the resulting mathematical model. All these factors deteriorate the identification performance when dealing with dynamic nonlinear industrial processes. To overcome these difficulties, this paper proposes a novel approach by using a modified genetic algorithm (MGA) combined with the predictive capability of nonlinear ARX (NARX) model for generating the dynamic NARX Takagi–Sugeno (TS) fuzzy model. The MGA algorithm processes the experimental input–output training data from the real system and optimizes the NARX fuzzy model parameters. This is referred to as fuzzy identification, which automatically generates the appropriate fuzzy if-then rules to characterize the dynamic nonlinear features of the real plant. The prototype pneumatic artificial muscle (PAM) manipulator, being a typical nonlinear and time-varying system, is used as a test system for this novel approach. This result shows that, with this MGA-based modeling and identification, the novel NARX fuzzy model identification approach to the PAM manipulator achieved highly outstanding performance and high precision as well. The accuracy of the proposed MGA-based NARX fuzzy model proves excellent in comparison with the MGA-based TS fuzzy model and the conventional GA-based TS fuzzy model.     

基于改进BP神经网络的ATM系统信息安全评估方法

根据ATM系统3层体系结构,针对ATM系统面临的信息安全问题,提出了应用人工神经网络(ANN)技术来评估ATM系统信息安全的思想;设计了基于改进的BPANN的ATM系统3层神经网络评估模型。根据建立的BP神经网络模型,以ATM信息系统主要信息安全指标作为训练样本,通过学习和训练找出输入与输出之间的内在联系,用训练好的BP网络对ATM信息系统进行评估,并将评估结果与传统的评估方法进行比较。实验结果表明,提出的评估模型具有很强的自适应性和容错能力,适用于复杂的ATM信息系统的安全性评估。实验数据与实际ATM信息系统的运行状态相吻合。     

有记忆效应的功放实数延时模糊神经网络模型

描述了一种基于实数延时模糊神经网络的有记忆效应的功率放大器模型.该模糊神经系统即自适应模糊神经推理系统,采用模糊c类均值聚类方法来减少模型的规则数目和简化模型结构.在训练过程中,采用最小二乘和反向传播相结合的高效算法提取模型参数.在测试平台上用三载波WCDMA宽带信号对射频功率放大器进行测试,并借助矢量信号分析仪采样功率放大器输入和输出数据,成功地对模型进行了训练和验证.通过和实数延时神经网络模型(RVTDNN)比较,该模型的收敛速度远快于这些前馈结构的神经网络模型.比较和分析时域和频域结果表明模型有很好的性能,其归一化均方误差达-38dB.     

Nonlinear System Control Using Adaptive Neural Fuzzy Networks Based on a Modified Differential Evolution

This study presents an adaptive neural fuzzy network (ANFN) controller based on a modified differential evolution (MODE) for solving control problems. The proposed ANFN controller adopts a functional link neural network as the consequent part of the fuzzy rules. Thus, the consequent part of the ANFN controller is a nonlinear combination of input variables. The proposed MODE learning algorithm adopts an evolutionary learning method to optimize the controller parameters. For design optimization, a new criterion is introduced. A hardware-in-the loop control technique is developed and applied to the designed ANFN controller using the MODE learning algorithm. The proposed ANFN controller with the MODE learning algorithm (ANFN-MODE) is used in two practical applications—the planetary-train-type inverted pendulum system and the magnetic levitation system. The experiment is developed in a real-time visual simulation environment. Experimental results of this study have demonstrated the robustness and effectiveness of the proposed ANFN-MODE controller.      

模糊回声状态网络

针对基于梯度下降的模糊递归神经网络训练效率低、容易陷入局部极小的缺点,本文基于回声状态网络(Echo State Networks,ESNs)和TS模型提出一种新的模糊模型结构——模糊回声状态网络(Fuzzy Echo State Networks,FESNs).FESNs由多条TS类型的模糊规则组成,规则后件采用ES...     

基于模糊化输入和反转提高神经网络分类性能的方法

为有效提高神经网络的分类性能,首先提出了一个可处理模糊输入的模糊神经网络结构,然后由模糊输出和非模糊目标输出定义了代价函数,推导出相应的学习算法,并对该模糊神经网络进行反转,提出了模糊化的反转算法．最后,通过计算机仿真实际的模式分类问题,验证了所提出的方法的有效性．实验结果表明,所提出的方法具有学习效率高、分类准确率高、泛化能力高的优点．