基于神经网络的口腔非线性动力学系统建模

神经网络方法是用于非线性动力学系统建模的有效方法，本文针对多层神经网络结构，运用递推预报误差（ＲＰＥ）算法对离散非线性动力学系统进行了辨识研究，并对口腔非线性动力学系统进行了神经网络动态建模，为了解语音和实用化的语音识别提供了良好的基础。     

基于神经网络的机械磨损故障光谱定位诊断法

在分析常用光谱定位诊断方法的基础上提出了基于神经网络的光谱定位诊断法；将机械摩擦副材质的元素含量作为神经网络输入，将材质所对应的部件作为神经网络输出，建立了相应的神经网络训练样本；通过整理训练样本和训练神经网络，利用神经网络超强的非线性映射能力和容错性实现了磨损故障部位诊断；通过算例分析验证了所提出的诊断方法的可行性和准确性．结果表明，所建立的方法简洁有效，并具有很高的诊断精度．     

具有不同激活特性神经元的前向神经网络在捷联惯导系统对准中的应用

提出了一种基于H∞滤波的动态神经网络。该神经网络不仅运用H∞滤波理论在践调整网络权值，而且采用与调整权值相同的算法，对神经网络各层神经元的不同非线性激活函数的参数进行在线调整，替代传统神经网络中各层神经元相同的激活特性，并将其用于捷联惯性导航系统(简称SINS)对准的误差状态模型中。仿真结果表明：与采用相同激活特性的方法相比，这种方法不仅提高了系统状态估值运算的实时性，而且也提高了精度。     

确定非线性隔振装置参数的一种方案

对单自由度刚度可调非线性隔振模型进行数学建模仿真和实验，对于待训练的神经网络模型，将数值仿真结果作为学习样本，将实验结果作为检验样本，训练成功的神经网络模型具有较好的内延能力和一定的外推能力，在设定非线性隔振装置要求的隔振效能参数后，可以给出隔振装置对应的刚度和阻尼参数，从而为设计非线性隔振装置提供了一种方案。     

摩擦表面边界膜温度特性的神经网络模型

采用非线性变换单元组成的多层前馈神经网络建立了丝杆螺母磨擦副表面边界膜温度特性的磨损自补偿教学模型,该模型可用于准确地预测边界膜对摩擦学特性的影响。采用Ｌ－Ｍ规则进行神经网络学习训练使网络收敛快且误差小,所得网络输出结果与实验结果有较好的吻合性。该神经网络可为工程设计人员进行摩擦学设计提供有效的计算工具。     

基于人工神经网络的结构振动系统重分析模型

本文利用人工神经网络的非线性映射功能，提出了一种基于人工神经网络技术的，具有全局意义的结构振动系统重分析模型。在Ｋｏｌｍｏｇｏｒｏｖ多层神经网络映射存在定理基础上，从理论上证明了一个结构振动系统的设计变量与其动态特性参数之间的映射关系可由一个三层神经网络模型来精确实现夷出了具体的建模算法，算例结果初步证实了本文方法的有效性。     

神经网络在陀螺漂移误差模型辨识中的应用

神经网络具有很强的自学习、自适应能力及非线性变换特性，为模型的辨识提供了一条十分有效的途径。本文基于反向传播（Back-Propagation）网络的研究，将神经网络应用于陀螺漂移误差模型辨识，通过陀螺的实际测试数据对神经网络的加权进行训练，得到了较为满意的结果。     

基于复数神经网络的粘弹性材料动态模量拟合方法

固体火箭发动机药柱粘弹性材料除具有弹塑性特性,还具有粘滞性,这一特性使得材料变形具有明显的时间效应,本构关系复杂,进行动态力学分析时,动态模量难以有效拟合．本文提出了一种基于(Levenberg-Marquardt, L-M)算法的复数神经网络拟合粘弹性材料动态模量的方法．通过广义Maxwell模型推导得到材料的动态模量表达式,以此构造未定网络参数为复数的神经网络,从而提供了一种输入、输出样本均为复数的神经网络解决方法．将实数L-M训练算法进行改进,衍生到复数领域,提出复数L-M训练算法．通过粘弹性材料实验,将实验数据时温等效转换,获得复数神经网络的训练及测试样本．通过对神经网络进行训练,实现粘弹性材料动态模量的高精度拟合．数值算例表明,与传统神经网络拟合方法相比,所提方法在训练速度和泛化能力方面都有其优越性．     

基于深度学习和近场动力学的层合板冲击工况识别

在冲击荷载作用下复合材料会产生断裂和分层等损伤。基于损伤数据对冲击工况进行识别，对改善复合材料的设计和确保其安全使用具有重要意义。基于此，本文提出一种基于深度学习和近场动力学（PD）理论的层合板冲击工况识别方法。首先使用改进的表面修正系数PD理论建立复合材料层合板刚体冲击损伤演化分析PD模型，PD模型数值模拟结果结合噪声数据增强技术构建层合板的冲击工况数据库；基于深度学习-卷积神经网络（CNN），对不同工况下的冲击损伤演化数据进行训练，实现对未知冲击工况的识别。结果显示，对于钢球冲击速度和角度的识别准确率均高于90%。     

基于经验小波变换和卷积神经网络的水泵水轮机无叶区流态特征提取及识别研究

监测和识别原型水泵水轮机无叶区的流动状态，对于保证抽水蓄能电站的运行安全性和稳定性有非常重要的意义。本文提出了一种基于经验小波变换、散布熵和卷积神经网络原理的流态特征提取和识别方法，首先使用经验小波变换对压力脉动信号进行分解，然后通过计算各分量的散布熵提取流态相关特征，最后通过利用特征–标签对训练卷积神经网络得到的智能识别模型，实现了无叶区流态识别。利用从国内某水泵水轮机采集到的发电、抽水和空转工况下实测压力脉动信号对该方法进行了测试，测试的平均准确率达到了94.84%，证明了该方法的有效性。     

Application of neural networks to identification of nonlinear characteristics in cushioning packaging

The structural neural network method is applied to the identification of nonlinear characteristics of cushioning liners in cushioning packaging. The simulated results on the two typical cushioning liner models show that the nonlinear characteristics can be identified perfectly.     

Advances in identification of nonlinear characteristics of packaging based on computational intelligence

Cushioning packaging is the most important subject in theoretical studies and real world applications of packaging dynamics. One of the fundamental steps in the design of cushioning packaging is to choose appropriate cushioning liner materials so as to control the peak value of acceleration transfered to the product and avoid the damage of the product resulting from the excessive internal stress owing to a large inertia force. It is necessary to determine the nonlinear behaviours of cushioning liners for designing reasonably packaging items and improving the cushioning effect of the packaging of products. It gives a new way of providing theoretical basis for judging, demonstrating and designing cushioning packaging that the artificial neural network methods, fuzzy adaptive control techniques and evolutionary computations are used to deal with the identification problems of cushioning packaging. This paper outlines several advances in the identification of nonlinear characteristics of packaging using computational intelligence and discusses the existent problems and the relevant further efforts.     

结构试验机的神经网络内模自适应跟踪控制系统

针对结构试验系统的非线性和不确定性特性，研究了一种基于神经网络的非线性内模自适应加载控制方法。引入的神经网络内模可跟踪学习对象的时变动力学，控制器的设计较少依赖于对象的先验知识，控制参数的调整是基于被控过程的测量信息，利用导出的神经网络算法来实现的。实验结果证明该系统具有良好的控制效果。     

模糊—神经网络控制算法及其在离心力—振动复合环境试验系统中的应用

针对离心－振动复合环境试验系统所存在的耦合性、非线性和不确定性提出了一种模糊－神经网络控制算法,利用被控对象输入输出信息离线、在线相结合学习系统的动态特性,对时变、非线性系统进行跟踪控制,并研究了该算法在系统中的实现方法。实现表明了控制系统具有良好的跟踪能力。该算法也适用于快速变化这类系统的实时控制。     

猎雷声纳基阵姿态稳定系统自适应反步法控制

针对带非线性摩擦力矩和负载扰动的高精度猎雷声纳基阵姿态稳定系统,提出了一种基于神经网络的自适应反步法控制方法。其中神经网络用于估计未知非线性摩擦力矩,进而设计反步法控制器和参数自适应律来对神经网络估计误差和负载扰动进行补偿。最后应用Lyapunov方法证明了所提出的自适应控制器能保证闭环系统的稳定性,并且可以通过选择适当的控制器参数来调整收敛率。仿真结果表明,基于神经网络的自适应反步法控制方法与PID控制相比,系统的动、静态性能指标及鲁棒性得到了全面的改善,与双闭环PID控制相比,跟踪精度提高了3倍多。     

Fuzzy relaxed-finite step size method to enhance the instability of the fuzzy first-order reliability method using conjugate discrete map

Fuzzy reliability analysis can be implemented using two discrete optimization maps in the processes of reliability and fuzzy analysis. Actually, the efficiency and robustness of the iterative reliability methods are two main factors in the fuzzy-based reliability analysis due to the huge computational burdens and unstable results. In the structural fuzzy reliability analysis, the first-order reliability method (FORM) using discrete nonlinear map can provide a C membership function. In this paper, a discrete nonlinear conjugate map is proposed using a relaxed-finite step size method for fuzzy structural reliability analysis, namely Fuzzy conjugate relaxed-finite step size method fuzzy CRS. A discrete conjugate map is stabilized using two adaptive factors to compute the relaxed factor and step size in FORM. The framework of the proposed fuzzy structural reliability method is established using two linked iterative discrete maps as an outer loop, which constructs the membership function of the response using alpha level set optimization based on genetic operator, and the inner loop, implemented for reliability analysis using proposed conjugate relaxed-finite step size method. The fuzzy CRS and fuzzy HL-RF methods are compared to evaluate the membership functions of five structural problems with highly nonlinear limit state functions. Results demonstrated that the fuzzy CRS method is computationally more efficient and is strongly more robust than the HL-RF for fuzzy-based reliability analysis of the nonlinear structural reliability problems.     

Nonlinear adaptive neural network control for a model-scaled unmanned helicopter

In this paper, a nonlinear adaptive neural network control is proposed for trajectory tracking of a model-scaled helicopter. The purpose of this research is to reduce the ultimate bounds of tracking errors resulted from small coupling forces (or small parasitic body forces) and aerodynamic uncertainties. The proposed control is designed under backstepping framework, with neural network compensators being added. Updating laws of neural networks are designed through projection algorithm, so that adaptive parameters are bounded. Derivatives of virtual controls are obtained through command filters. It is proved that, by using neural network compensators, tracking errors of the closed-loop system can be restricted within very small ultimate bounds. Superiority of the proposed nonlinear adaptive neural network control over a backstepping control is demonstrated by simulation results.     

NEURO FUZZY MODEL FOR PREDICTING THE DYNAMIC CHARACTERISTICS OF BEAMS

An adaptive neuro-fuzzy inference system （ANFIS） is introduced to predict the dynamic behavior of beams. The effects of axial forces and large displacements are considered in the analysis. A database of tests for the dynamic characteristics of beams is developed from the experimental tests. The responses of nonlinear vibration force for the single and multiple-stepped beams are calculated from the finite element method （FEM）, experimental tests and neuro-fuzzy model for comparison. The neuro-fuzzy model provides a general framework for the combination of neural networks and fuzzy logic. It is more flexible with more options of incorporating the fuzzy nature of the real-world system and is an useful estimation tool for the dynamic characteristics of beams. Therefore, ANFIS can be a useful tool for dynamic behaviour analysis of multiple-stepped beams subjected to axial loads and large displacement.     

基于模糊神经网络的惯性导航系统预热控制

针对惯性导航系统预热中存在的非线性问题,采用了基于模糊神经网络的控制系统。讨论了模糊系统与神经网络的优缺点,给出了适于非线性时滞、基于BP网络的控制方案,并通过仿真说明方案的可行性。     

Fuzzy adaptive unscented Kalman filter control of epileptiform spikes in a class of neural mass models

A new closed-loop control method based on the fuzzy adaptive unscented Kalman filter (FAUKF) is proposed to suppress epileptiform spikes in a class of neural mass models with uncertain measurement noise. The FAUKF is used to estimate the nonlinear system states of the underlying models and amend measurement noise adaptively. The control law is constructed via the estimated states. Numerical simulations illustrate the efficiency of the proposed method.