Effects of noise and synaptic weight on propagation of subthreshold excitatory postsynaptic current signal in a feed-forward neural network

Nonlinear Dynamics - Excitatory postsynaptic current (EPSC) is a biological signal of neurons; the propagation mechanism of subthreshold EPSC signal in neural network and the effects of background...     

Chaotic dynamics in a neural network under electromagnetic radiation

In this paper, a small Hopfield neural network with three neurons is studied, in which one of the three neurons is considered to be exposed to electromagnetic radiation. The effect of electromagnetic radiation is modeled and considered as magnetic flux across membrane of the neuron, which contributes to the formation of membrane potential, and a feedback with a memristive type is used to describe coupling between magnetic flux and membrane potential. With the electromagnetic radiation being considered, the previous steady neural network can present abundant chaotic dynamics. It is found that hidden attractors can be observed in the neural network under different conditions. Moreover, periodic motion and chaotic motion appear intermittently with variations in some system parameters. Particularly, coexistence of periodic attractor, quasiperiodic attractor, and chaotic strange attractor, coexistence of bifurcation modes and transient chaos can be observed. In addition, an electric circuit of the neural network is implemented in Pspice, and the experimental results agree well with the numerical ones.     

Multistable dynamics in a Hopfield neural network under electromagnetic radiation and dual bias currents

Nonlinear Dynamics - This paper investigates a Hopfield neural network under the simulation of external electromagnetic radiation and dual bias currents, in which the fluctuation of magnetic flux...     

Chaotic resonance in Izhikevich neural network motifs under electromagnetic induction

Nonlinear Dynamics - Chaotic resonance (CR) is the response of a nonlinear system to weak signals enhanced by internal or external chaotic activity (such as the signal derived from Lorenz system)....     

Control of epileptic activities in a cortex network of multiple coupled neural populations under electromagnetic induction

Epilepsy is believed to be associated with the abnormal synchronous neuronal activity in the brain, which results from large groups or circuits of neurons. In this paper,we choose to focus on the temporal lobe epilepsy, and establish a cortex network of multiple coupled neural populations to explore the epileptic activities under electromagnetic induction. We demonstrate that the epileptic activities can be controlled and modulated by electromagnetic induction and coupling among regions. In cert...     

Dynamics explore of an improved HR neuron model under electromagnetic radiation and its applications

The influence of electromagnetic field to neuron firing rhythm is not negligible. In order to investigate the behavior mechanism, a five-dimensional neuron model based on the Faraday's law of electromagnetic induction is improved by introducing magnetic flux variables and electric field variables on the three-dimensional Hindmarsh–Rose (HR) neuron model, and then, its rich dynamics and application in image encryption are discussed. Specifically, the equilibrium point distribution is analyzed using Matcont software and it is found that there are subcritical Hopf bifurcation and coexisting mode firing first. Second, numerical simulations are performed in terms of two-parameter bifurcation, ISI bifurcation, the maximum Lyapunov exponent and firing sequences, and the experimental results show that the new model exhibits various firing rhythms. The rich dynamic behaviors make the model more suitable for application in image encryption. So in the end, a grayscale image encryption scheme containing five parts called sparse, compression calculation, forward diffusion, rank scrambling and backward diffusion is designed by combining with the compressive sensing theory. The security analysis results show that the designed encryption scheme not only has excellent compression performance and high security, but also displays faster encryption speed. That is to say, the algorithm can be applied to the field of real encryption owning to the advantages of the lower costs of data transmission and higher efficiency of encryption. It is worth mentioning that the influence of different dimensional compression methods on the encryption and reconstruction effects is analyzed for the first time. The research results of this paper provide some ideas for perfecting the neuron model, revealing the influence of electromagnetic field on biological nervous system, and the excellent performance of the new neuron model provides theoretical guidance and experimental basis for the practical application of digital image encryption.

     

On radiation of electromagnetic and electroacoustic waves in plasma

Summary The infinitely long cylindrical antenna immersed in a compressible plasma is considered as a boundary-value problem. The analysis starts with a specified voltage at a circumferential gap which is uniformly excited. It is found that the radiation pattern of the antenna is influenced only, to a slight extent, by the finite compressibility of the plasma, provided that the wire diameter is somewhat larger than a Debye length in the plasma.Equations in this Part II are continuous with those in Part I, Appl. sci. Res.B 11 (1965) 423.The research reported in this paper was sponsored by the Air Force Cambridge Research Laboratories, Bedford, Mass., under Contract PRO-62-201.     

On radiation of electromagnetic and electroacoustic waves in a plasma

Summary Radiation from current distributions in a lossless compressible plasma media is considered. A linearized theory is used such that the isotropic electron plasma is regarded as a single fluid continuum. It is found that a considerable portion of the power is radiated as an electro-acoustic type wave. Some specialized results of other investigators are recovered.The research reported in this paper was sponsored by the Air Force Cambridge Research Laboratories, Bedford, Massachusetts, under contract PRO-62-201.     

Study of Rusar-type polymer fibrous materials in uniaxial compression using the electromagnetic radiation effect

We experimentally study the behavior of polymer threads in uniaxial tension with the use of a constant load and mechanical vibrations up to 1000 Hz. The deformation of polymer threads results in the appearance of electromagnetic signals which can be registered in the radio frequency band. We propose and test an effective method for obtaining information by analyzing the electromagnetic signals for determining the pre-fracture moment of polymer materials. We show that it is principally possible to use electromagnetic radiation (EMR) effect to investigate the dynamics of polymer thread breaking.     

甲烷/煤尘复合爆炸火焰的传播特性

为揭示甲烷/煤尘复合爆炸火焰的传播机理,利用气粉两相混合爆炸实验系统,在低于甲烷爆炸下限条件下,采用高速摄影机记录火焰传播图像,通过热电偶采集火焰温度,研究了煤尘种类以及甲烷体积分数对甲烷/煤尘复合火焰传播特性的影响。结果表明:挥发分是衡量煤尘燃烧特性的主导因素;随着煤尘挥发分的升高,燃烧反应增强,火焰传播速度升高,火焰温度升高;挥发分含量差异较小时,水分含量越低,燃烧反应越剧烈;在相同条件下,焦煤的燃烧反应强度最高,其次为长焰煤,最后为褐煤;随着甲烷体积分数的增加,煤尘颗粒的燃烧可由释放挥发分的扩散燃烧转变为气相预混燃烧,燃烧反应增强,火焰传播速度和火焰温度显著升高;热辐射和热对流作用促进煤尘颗粒热解,释放挥发分进行燃烧反应,维持复合火焰的持续传播;随着混合体系中甲烷体积分数的增加,混合爆炸机制由粉尘驱动型爆炸转为气体驱动型爆炸,燃烧反应增强;甲烷/煤尘复合爆炸火焰可由未燃区、预热区、气相燃烧区、多相燃烧区和焦炭燃烧区5部分组成,湍流扰动导致燃烧介质空间分布存在差异,使得燃烧区无规则交错分布。     

BP神经网络及其在结构动力分析中的应用研究

回顾了ＢＰ神经网络在结构分析中的应用和发展过程，针对ＢＰ网络在实际应用时存在的收敛速度慢以及所需训练样本多等主要问题，提出了相应的网络快速学习算法以及训练样本正交化处理方法。同时，对网络结构设计、训练样本规范化处理、网络映射结果精度的自适应修正等具体技术问题也进行了详细讨论，并提出了相应的解决方法。文中最后给出了二个工程应用实例，结果表明：ＢＰ网络在结构动力分析中不失为一种崭新的有效方法。     

Study on the motion error of main spindle of lathe based on the harmonic wavelet neural network

Experimental Techniques -     

Conversion of longitudinal waves into electromagnetic radiation in a slowly varying plasma under W.K.B. approximation

This paper investigates the conversion of a dispersive longitudinal oscillation into reflected and transmitted electromagnetic radiation fields in slowly varying unmagnetized warm fluid plasmas, using W.K.B. approximations. The expressions for the power of the transmitted and reflected electromagnetic radiations, generated by electron acoustic waves, have also been obtained. It is shown that this conversion process becomes most efficient under certain conditions.

Nomenclature

In § 2 H magnetic field - H ₁ - u electron fluid velocity - k _t wave number of the transverse wave - k ₁ wave number of the longitudinal wave in electron fluid - m electronic mass - N ₀ number density of electrons in the unperturbed state - N perturbation in the electron number density - p perturbation in the electron fluid pressure - v _e adiabatic sound velocity of the electron fluid - K _t ² c ^{2 2}–e² - K ₁ ² v _e ^{2 2}–e² - wave frequency - _e electron plasma frequency - 1– _e ² / ² - c velocity of light in vacuum In § 3 K ₀ wave number in the 0X direction - K ₁ ² K ₁ ² –K ₀ ² - K ₂ ² K _t ² –K ₀ ² - K ₃ K ₁–K ₂ - K ₄ K ₁+K ₂ - K ₅ (K ₁ K ₂)^1/2 See Appendix A - A ₁ pressure amplitude of the reflected part of the incident wave - B ₁ pressure amplitude of the transmitted part of the incident wave - L characteristic length of variation ofN ₀ - e _x unit vector along 0X - e _z unit vector along 0Z In § 4 S _t Poynting flux of the transverse electromagnetic radiation - S _tZ ^/t Average of the transmitted part of the poynting flux along 0Z over the time period 2/ - S _tZ ^/r Average of the reflected part of the poynting flux along 0Z over the time period 2/ In § 5 S ₁ Energy flux carried by the longitudinal pressure wave - S _1Z ^/t Average of the transmitted part ofS ₁ along 0Z over the time period 2/     

基于灰色模型和RBF神经网络的MEMS陀螺温度补偿

MEMS陀螺的零偏随温度呈非线性变化,同时含有较大的随机噪声.针对传统的多项式模型难以精确表达零偏随温度变化的问题,提出了一种基于灰色模型和RBF神经网络的MEMS陀螺温度补偿方法:首先用灰色模型对数据进行预处理,以减小原始数据的噪声;然后用降噪后的样本数据对RBF神经网络进行训练.在相同的训练次数下训练误差可减小一个数量级.验证试验结果表明,采用该模型补偿后的陀螺零偏误差较传统的多项式模型减小一个数量级,较未经预处理的RBF神经网络减小2/3.     

基于神经网络的间隙机构运动副磨损预测

应用BP神经网络建立了磨损率与接触应力、滑动速度和材料硬度之间的非线性关系模型,并对该网络模型进行了验证和测试,结果表明,训练良好的神经网络模型能够准确反映样本所蕴含的内在磨损规律,且具有较好的预测效果。基于非线性弹簧阻尼模型和修正的Coulomb摩擦力模型对含间隙曲柄滑块机构进行数值仿真分析,获得间隙机构运动副的接触应力和相对滑动速度,利用训练好的神经网络磨损模型对轴套的磨损进行迭代磨损预测分析,发现随着曲柄转数的增加,轴套表面一些特定位置处的磨损越来越严重,最终导致轴套表面出现非均匀磨损现象,其原因是间隙机构运转过程在一些特定位置处产生了较大接触应力和碰撞力。     

矩形板结构损伤的分区域神经网络识别方法

通过引入LM优化算法,针对矩形薄板中对称结构的损伤识别问题,提 出了一种基于神经网络的分区域分步识别方法. 对于预测输出量比较多且对预测精度要求比 较高的问题,常会出现网络训练时收敛速度慢、网络预测精度低,并且当网络训练达到目标 误差时,输出的预测量中常有某个输出量的误差还很大的情况. 针对这些问题,利用选 取的组合输入参数,提出了基于神经网络的分区域识别方法. 通过对悬臂板结构的数值模拟 结果表明：提出的分区域识别方法对结构损伤的分区和预测是可行和有效的, 其预测精度要明显的高于只用单个网络的预测结果,并且预测子网络对损伤的位置和程度是 同步输出的,从而避免了传统分步识别理论中子网络过多的问题.     

基于遗传RBF网络的惯性测量组合模拟电路故障诊断

为解决惯性测量组合模拟电路的诊断不易定位到元件级故障的问题,提出了一种基于遗传RBF网络的智能诊断方法。该方法首先利用RBF神经网络快速准确识别故障的能力,以RBF的训练均方误差为遗传算法的适应度函数,依靠遗传算法强大的全局寻优能力实现故障特征选择。在特征选择的过程中,同时记录使训练均方误差达到最小的最优RBF网络,然后直接利用特征选择过程中训练好的最优RBF网络诊断故障,而无需利用特征选择后的训练数据对RBF网络进行再训练,简化了诊断步骤,同时增强了网络的抗干扰能力。仿真结果表明,该方法能有效去除冗余特征,准确诊断惯性测量组合模拟电路的故障,并有较好的抗噪能力,证明了该方法的有效性和可行性。