A Fuzzy-Neural Network Control of Nonlinear Dynamic Systems

In this paper, an adaptive dynamic control scheme based on a fuzzy neural network is presented, that presents utilizes both feed-forward and feedback controller elements. The former of the two elements comprises a neural network with both identification and control role, and the latter is a fuzzy neural algorithm, which is introduced to provide additional control enhancement. The feedforward controller provides only coarse control, whereas the feedback controller can generate on-line conditional proposition rule automatically to improve the overall control action. These properties make the design very versatile and applicable to a range of industrial applications.

A Fuzzy Neural Network Model of Linguistic Dynamic Systems Based on Computing with Words

Linguistic dynamic systems(LDS)are dynamic processes involving computing with words(CW)for modeling and analysis of complex systems.In this paper,a fuzzy neural network(FNN)structure of LDS was proposed.In addition,an improved nonlinear particle swarm optimization was employed for training FNN.The experiment results on logistics formulation demonstrates the feasibility and the efficiency of this FNN model.

Dynamic Bandwidth Allocation Technique in ATM Networks Based on Fuzzy Neural Networks and Genetic Algorithm

ＤｙｎａｍｉｃＢａｎｄｗｉｄｔｈＡｌｏｃａｔｉｏｎＴｅｃｈｎｉｑｕｅｉｎＡＴＭＮｅｔｗｏｒｋｓＢａｓｅｄｏｎＦｕｚｙＮｅｕｒａｌＮｅｔｗｏｒｋｓａｎｄＧｅｎｅｔｉｃＡｌｇｏｒｉｔｈｍ①ＺｈａｎｇＬｉａｎｇｊｉｅＬｉＹａｎｄａＷａｎｇＰｕ（Ｄｅｐｔｏｆ...

Fault Diagnosing System of Steam Generator for Nuclear Power Plant Based on Fuzzy Neural Networks

All kinds of reasons are analysed in theory and a fault repository combined with local expert experiences is establishedaccording to the structure and the operation characteristic of steam generator in this paper. At the same time, Kohonen algo-rithm is used for fault diagnoses system based on fuzzy neural networks. Fuzzy arithmetic is inducted into neural networks tosolve uncertain diagnosis induced by uncertain knowledge. According to its self-association in the course of default diagnosis. thesystem is provided with non-supervise, self-organizing, self-learning, and has strong cluster ability and fast cluster velocity.

Improving Generalization of Fuzzy Neural Network

Explores the generalization error of fuzzy neural network, analyzes the reason for occurrence and presents the equation of calculating error by the confidence interval approach. In addition, a generalization error transfering(GET) method of improving the generalization error is proposed. The simulation experimental results of heating furnance show that the GET scheme is efficient.

ReinforcementBased Fuzzy Neural Network Control with Automatic Rule Generation

A reinforcemen based fuzzy neural network control with automatic rule generation (RBFNNC) is proposed. A set of optimized fuzzy control rules can be automatically generated through reinforcement learning based on the state variables of object system. RBFNNC was applied to a cart pole balancing system and simulation result shows significant improvements on the rule generation.

Rotation Angle Control Strategy for Telescopic Flexible Manipulator Based on a Combination of Fuzzy Adjustment and RBF Neural Network

The length of fexible manipulators with a telescopic arm alters during movement.The dynamic parameters of telescopic fexible manipulators exhibit signifcant time-varying characteristics owing to variations in length.With an increase in the manipulators’length,the nonlinear terms caused by fexibility in the manipulators’dynamic equations cannot be ignored.The time-varying characteristics and nonlinear terms of telescopic fexible manipulators cause fuctuations in rotation angles,which afect the operation accuracy of end-efectors.In this study,a control strategy based on a combination of fuzzy adjustment and an RBF neural network is utilized to improve the control accuracy of fexible telescopic manipulators.First,the dynamic equation of the manipulators is established using the assumed mode method and Lagrange’s principle,and the infuence of nonlinear terms is analyzed.Subsequently,a combined control strategy is proposed to suppress the fuctuation of the rotation angle in telescopic fexible manipulators.The variation ranges of the feedforward PD controller parameters are determined by the pole placement strategy and length of the manipulators.Fuzzy rules are utilized to adjust the controller parameters in real-time.The RBF neural network is utilized to identify and compensate the uncertain part of the dynamic model of the fexible manipulators.The uncertain part comprises time-varying parameters and nonlinear terms.Finally,numerical simulations and prototype experiments prove the efectiveness of the combined control strategy.The results prove that the proposed control strategy has a smaller standard deviation of errors.Therefore,the combined control strategy is more suitable for telescopic fexible manipulators,which can efectively improve the control accuracy of rotation angles.

FAULT DIAGNOSIS OF ROTATING MACHINERY USING KNOWLEDGE-BASED FUZZY NEURAL NETWORK

A novel knowledge-based fuzzy neural network （KBFNN） for fault diagnosis is presented. Crude rules were extracted and the corresponding dependent factors and antecedent coverage factors were calculated firstly from the diagnostic sample based on rough sets theory. Then the number of rules was used to construct partially the structure of a fuzzy neural network and those factors were implemented as initial weights, with fuzzy output parameters being optimized by genetic algorithm. Such fuzzy neural network was called KBFNN. This KBFNN was utilized to identify typical faults of rotating machinery. Diagnostic results show that it has those merits of shorter training time and higher right diagnostic level compared to general fuzzy neural networks.

The Fuzzy Modeling Algorithm for Complex Systems Based on Stochastic Neural Network

A fuzzy modeling method for complex systems is studied. The notation of general stochastic neural network (GSNN) is presented and a new modeling method is given based on the combination of the modified Takagi and Sugeno's (MTS) fuzzy model and one-order GSNN. Using expectation-maximization(EM) algorithm, parameter estimation and model selection procedures are given. It avoids the shortcomings brought by other methods such as BP algorithm, when the number of parameters is large, BP algorithm is still difficult to apply directly without fine tuning and subjective tinkering. Finally, the simulated example demonstrates the effectiveness.

无师General Fuzzy Min-Max人工神经网络

针对一般模糊极小极大(generalfuzzymin max,GFMM)神经网络不能够完全无师聚类和自适应在线学习的问题,提出了一种无师训练的一般模糊极小极大(generalfuzzymin max,GFMM)人工神经网络。它继承了GFMM网络的优点,可以输入n维模糊量,尤其是新增加了无师学习的功能,弥补了GFMM网络不能自适应在线学习新类的缺陷。实验测试结果与分析表明,该网络在自动目标识别的实际应用中具有广泛的适用性。

Flatness Control Based on Dynamic Effective Matrix for Cold Strip Mills

Steel strips are the main of steel products and flatness is an important quality indicator of steel strips. Flatness control is the key and highly difficult technique of strip mills. The bottle-neck restricting the improvement of flatness control techniques is that the research on flatness theories and control mathematic models is not in accordance with the requirement of technique developments. To build a simple, rapid and accurate explicit formulation control model has become an urgent need for the development of flatness control technique. This paper puts forward the conception of dynamic effective matrix based on the effective matrix method for flatness control proposed by the authors under the consideration of the influence of the change of parameters in roiling processes on the effective matrix, and the concept is validated by industrial productions. Three methods of the effective matrix generation are induced： the calculation method based on the flatness prediction model; the calculation method based on the data excavation in rolling processes and the direct calculation method based on the network model. A fuzzy neural network effective matrix model is built based on the clusters, and then the network structure is optimized and the high-speed-calculation problem of the dynamic effective matrix is solved. The flatness control scheme for cold strip mills is proposed based on the dynamic effective matrix. On stand 5 of the 1 220 mm five-stand 4-high cold strip tandem mill, the industrial experiment with the control methods of tilting roll and bending roll is done by the control scheme of the static effective matrix and the dynamic effective matrix, respectively. The experiment result proves that the control effect of the dynamic effective matrix is much better than that of the static effective matrix. This paper proposes a new idea and method for the dynamic flatness control in the rolling processes of cold strip mills and develops the theory and model of the flatness control effective matrix method.

Research on the Evolution of Thinking Dynamics Attractor

In this paper we introduce the evolution law of thinking-neural network attractor in the field of thinking dynamics,a new idea about the attractor evolution using evolutional mapping transformation is given on the bases of generalized isologous concept. The idea is connected with the intuitive thinking.This paper lays a foundation for the further studying of the brain thinking process.

基于VMD-FE-CNN-BiLSTM的短期光伏发电功率预测

为提高光伏功率的预测精度,提出一种变分模态分解(VMD)、模糊熵(FE)、卷积神经网络(CNN)和双向长短期记忆神经网络(BiLSTM)的光伏功率组合预测模型。该方法首先采用VMD将原始光伏序列数据分解成多个子序列,从而减少随机波动分量和噪声干扰对预测模型的影响,通过FE对每个子序列进行重组,使用一维CNN的局部连接及权值共享提取不同分量的特征,将CNN输出的特征融合并输入到BiLSTM模型中;利用BiLSTM模型建立历史数据之间的时间特征关系,得到光伏发电功率预测结果。与BiLSTM、CNN-BiLSTM、EEMD-CNN-BiLSTM、VMD-CNN-BiLSTM这4种模型进行比较,该文提出的VMD-FE-CNN-BiLSTM模型在光伏发电功率预测中具有较高的精确度和稳定性,满足光伏发电短期预测的要求。

变工况下动态卷积域对抗图神经网络故障诊断

针对基于无监督域自适应故障诊断方法忽略了域间数据结构信息、传统域对齐平均最大差异法全局泛化能力差等问题,本文提出一种基于无监督域自适应理论的动态卷积域对抗图神经网络故障诊断模型,分别通过对数据的类别标签、域标签和数据结构信息进行建模。通过分类器和域鉴别器分别建模类别标签和域标签,通过图神经网络将数据结构信息嵌入到实例图节点中,利用高斯Wasserstein距离来度量不同领域的实例图之间的差异。本文对比了不同工况下共14种迁移任务在各模型下故障识别的准确率。实验结果表明:基于动态卷积的域对抗图神经网络模型在变工况下的故障诊断效果均优于其他对比模型,且模型性能稳定。

低附着路况条件下车辆横向稳定性控制

针对在冰雪环境下车辆横向稳定控制,为解决在低附着、分布不均的路面情况下车辆对参考轨迹的稳定跟踪问题,设计了基于神经网络调节的模糊PID(Proportional-Integral-Differential)制器,以及基于线性化车辆模型的模型预测控制(MPC:Model Predictive Controll)。以路面附着系数及车辆速度作为输入构建BP(Back-Propagation)神经网络,输出调节系数优化模糊PID控制器控制性能;设计了十自由度模型表征车辆在冰雪环境下的动力学特性,使用MPC实现车辆横向稳定控制。使用CarSim/Simulink进行联合仿真实验,结果表明该控制器能显著提高车辆轨迹跟踪性能。

A Fuzzy Neural Network Based Dynamic Data Allocation Model on Heterogeneous Multi-GPUs for Large-scale Computations

The parallel computation capabilities of modern graphics processing units （GPUs） have attracted increasing attention from researchers and engineers who have been conducting high computational throughput studies. However, current single GPU based engineering solutions are often struggling to fulfill their real-time requirements. Thus, the multi-GPU-based approach has become a popular and cost-effective choice for tackling the demands. In those cases, the computational load balancing over multiple GPU ＂nodes＂ is often the key and bottleneck that affect the quality and performance of the real=time system. The existing load balancing approaches are mainly based on the assumption that all GPU nodes in the same computer framework are of equal computational performance, which is often not the case due to cluster design and other legacy issues. This paper presents a novel dynamic load balancing （DLB） model for rapid data division and allocation on heterogeneous GPU nodes based on an innovative fuzzy neural network （FNN）. In this research, a 5-state parameter feedback mechanism defining the overall cluster and node performance is proposed. The corresponding FNN-based DLB model will be capable of monitoring and predicting individual node performance under different workload scenarios. A real=time adaptive scheduler has been devised to reorganize the data inputs to each node when necessary to maintain their runtime computational performance. The devised model has been implemented on two dimensional （2D） discrete wavelet transform （DWT） applications for evaluation. Experiment results show that this DLB model enables a high computational throughput while ensuring real=time and precision requirements from complex computational tasks.

基于GD-FNN的金融股指预测模型

针对股票市场内部结构复杂性和外部因素多变性,构建一种基于椭圆基函数且能够动态调整网络结构的广义动态模糊神经网络模型对金融股指进行预测。以上证指数为例,在价格和成交量的基础上,将与股票市场密切相关的宏观经济指标引入模型预测指标体系。通过滑动时间窗对数据集进行处理,提高了模型预测准确性并降低了运算时间。与其他神经网络模型预测效果进行比较,结果表明提出的模型具有较好的预测效果。

基于GD-FNN的特高压直流输电暂态稳定控制

特高压直流输电输电容量大、控制灵活迅速,在我国"西电东送"战略中扮演了重要角色,其对受端交流系统暂态稳定性的影响也随输电容量的增大而变大。提出采用广义动态模糊神经网络(GD-FNN)控制来提高系统暂态稳定性,通过附加控制信号动态调节输送功率从而给系统提供足够阻尼。根据系统选择合适的调制信号以及控制维数,对GD-FNN系统训练后通过系统误差和模糊规则的ε-完备性作为判据来优化系统结构,同时对隶属度函数的参数进行修正,从而保证了控制器具有紧凑的结构和良好的泛化能力。仿真结果表明,所设计的暂态稳定控制器在保持系统稳定方面具有优越的性能,并且鲁棒性较好,可有效保障机组和电网的安全稳定运行。

基于GD-FNN的微生物发酵过程软测量建模

针对微生物发酵过程关键生物参数(如菌体浓度、基质浓度、产物浓度等)难以直接在线测量的问题,提出了一种基于广义动态模糊神经网络(GD-FNN)的软测量建模方法。GD-FNN算法基于椭圆基函数(EBF),以模糊-完备性作为在线参数分配机制。该方法学习时参数调整和结构辨识同时进行,并能自动地确定模糊规则从而达到系统的特定性能。文中以青霉素发酵过程为研究对象,应用一致关联度法确定软测量模型的辅助变量后,建立了GD-FNN软测量模型。仿真结果表明,基于GD-FNN的软测量建模比基于径向基(RBF)神经网络的软测量建模运算速度快、预测精度高、泛化能力强。

全局和局部感知的交通速度预测模型

面对日益严峻的交通堵塞问题,智能交通系统获得飞速发展和广泛应用,作为基石工作的交通速度预测因此备受关注.近些年来,深度学习被广泛用于交通速度预测的研究工作,并且研究方向也从单一的建模时间相关性迁移到复杂的时空相关性,图神经网络由于契合交通路网的图结构数据这一本质属性,成为建模空间相关性的主流方法 .目前,大多数的研究工作已经注意到动态的空间相关性对交通速度预测任务的重要性.然而,基于这一发现所提出的建模思路主要预定义矩阵或自适应矩阵,属于静态矩阵,并不足以应对空间相关性的复杂和动态的特性.同时通过对真实交通速度数据集的分析,本文发现交通节点间依赖的局部波动相比交通路网的全局影响具有更强的动态性,这表明空间相关性可以从全局和局部的角度分开建模,因此本文提出了一个端到端全局和局部融合的动态图神经网络模型来进行交通速度预测.首先,交通速度流被自分解层分解为静态分量和动态分量,随后动态图生成模块为动态分量构造实时的动态图以匹配其动态性.基于构造的动态图和输入的预定义图,本文借助图卷积操作来学习这两类空间相关性的高阶表达.除此之外,本文在时间模块使用空洞因果卷积捕获交通数据中时间相关性.最后,残差连接被用来聚合时空相关性并输送给输出层完成最终的速度预测.在两个高速公路数据集和一个城市路数据集上的实验结果表明本文提出的模型相比主流模型在平均绝对误差和均方根误差两个预测指标上均优于主流模型.