用人工神经网络预测HL-2A等离子体放电破裂

使用人工神经网络(ANN)对HL-2A装置破裂放电进行了离线预测研究。采用了两种方法训练网络,一种方法是采用原始实验数据作为网络输入训练网络,另一种是把训练样本中的Mirnov原始实验信号进行预处理,目的是突出Mirnov原始信号隐含的破裂信息。比较这两种方法,结果表明第二种方法获得的网络对破裂放电能够做出更加准确的预测。     

HL-2Aװ�õ�����������ʵʱԤ��ϵͳ

HL-2A装置MHD不稳定性实时预测破裂系统采用了一种简单有效的方法来预测MHD不稳定性导致的等离子体大破裂。利用Mirnov线圈探测MHD信号,根据信号的振幅或频率特点设定计算方法,来预测等离子体破裂先兆,然后用激光吹气注入杂质来缓解等离子体破裂。研究结果表明,该系统能够实时预测破裂先兆,按量注入杂质后,可达到破裂缓解目的。     

基于小波变换的MHD信号的提取

混有干扰的Mirnov信号是一类非平稳,和Fourier变换的方法进行MHD信号的提取不能取得满意的效果,因而提出采用小波变换的方法对Mirnov信号进行处理。理论分析与实验结果表明,利用小波变换的时频分析特怀对Mirrnov信号进行滤波处理,可以有效地去除干扰,提取出有用的MHD信号。     

通过软X射线信号研究芯部等离子体的结构

介绍了如何从软X射线原始信号上分析等离子体芯部磁场结构的方法. 在HT-7托卡马克上,通过分析一炮典型的放电数据,直接从软X射线原始信号上分析了磁岛的位置和旋转方向. 通过软X射线图像反演的结果和Mirnov信号上观察到的m=2模的走向验证了磁岛旋转的方向和位置. 关键词： 软X射线 MHD不稳定性 磁岛     

基于ELMAN网络的非线性混沌微弱信号检测

为了快速和实时地从具有强噪声的较低信噪比的原始信号中检测出有用信息,设计了一种混沌相空间重构理论和ELMAN神经网络的信号检测方法;首先,描述了采用混沌相空间重构理论对原始信号进行重构的原理和方法,在获取重构的时间序列的基础上,采用ELMAN网络来近似表示用于检测信号的函数型,然后,设计了ELMAN网络中各层之间连接权值的计算方式,并提出了采用ELMAN网络进行信号检测的具体过程,最后给出了采用混沌相空间重构理论和ELMAN网络的信号检测模型;对Lorenz混沌系统模型进行仿真实验,结果证明了文章方法能有效地对瞬时信号和周期性信息进行检测,在具有高斯白噪声的情况下,仍然具有降噪效果好的优点,是一种用于信号检测的可行性方法。     

EAST上由垂直不稳定性引发破裂的分析与预测

从EAST 装置2016 年的放电实验中，选取了119 次等离子体破裂放电数据，分析诱发等离子体破裂的原因，发现约60%的破裂是由垂直不稳定性直接引起的，其破裂后将会产生更大的晕电流，从而产生更大的电磁应力损坏装置。对由垂直不稳定性引起的破裂(简称为VID)(72 次放电)进行了研究，建立了分别基于单变量(垂直位移)和两维变量(垂直位移、垂直位移增长率)的预测模型用于对VID 破裂的预测。离线测试表明，基于两维变量的预测模型可以在破裂发生前20ms 给出破裂预警信号，预测成功率达93%。     

基于局部线性嵌入的特征融合方法在岩石破裂 状态分类的应用*

为了获取岩石破裂过程有效的声发射信号特征,更好的对岩石破裂状态进行分类,提出一种基于流形学习算法的LLE特征融合方法进行数据降维。以红砂岩为研究对象设计室内单轴压缩实验采集信号,然后对原始声发射信号预处理并对信号进行特征提取,以时域、频域下的特征向量重新组合成一组新的多维特征向量,采用线性主元(PCA)和流形学习LLE算法分别进行降维。比较两种算法降维后融合特征的聚类效果二维和三维分布图,使用LLE算法降维后,四种状态分布相对更近,呈一条水平线趋势,且各状态交叉混叠数目较少,第一状态没有一个样本错判,且四个状态相比于PCA降维后的聚类效果更集中。再比较两种算法降维后融合特征的敏感度之和,LLE算法融合特征敏感度之和远大于PCA算法,说明经过LLE算法降维后得到的融合特征更多地表征了原始信号包含的局部信息同时证明了LLE算法相比PCA算法具有更好的聚类效果。最后经LLE特征融合下的砂岩破裂状态分类实验验证,融合特征后的识别率相对单一的时域特征识别提高了6%。表明该方法能显著提高岩石破裂状态分类的识别率,降维性能相对突出。     

基于规范化样本拆分的轴承缺陷检测

表面缺陷对轴承的性能和寿命存在严重影响。近年来,深度学习在缺陷检测中发挥了重要的作用,然而对于轴承检测而言,缺陷样本的采集耗时耗力。选择轴承内径作为研究对象,根据轴承的对称性特性提出一种规范化样本拆分方法,可有效扩充轴承样本数据集。分别采用不同的样本处理方法,而后利用ResNet网络训练轴承缺陷检测模型,进行多组对比实验,实验结果表明:直接采用原始图像进行网络训练,检测效果较差,模型的AUC (area under the curve)仅为0.558 0;对原始图像进行样本拆分,训练出的模型检测效果有所提升,其模型AUC提升为0.632 6;将原始图像进行4点透视变换校正后再进行网络训练,检测效果同样有所提升,其模型AUC提升为0.661 3;将原始图像进行透视变换校正且规范化样本拆分后,检测效果最好,其模型AUC增加为0.849 6。     

EAST装置的磁探针设计

介绍了EAST装置中磁探针设计中的结构、安装位置、匝面积的标定、幅频响应,并给出了该磁探针的标定误差和Mirnov线圈幅频响应特征图.两轮EAST放电试验表明,电磁测量的信号满足装置运行和等离子体控制的需要.     

HL-1M装置上MHD不稳定性磁扰动的探测和分析

给出了HL－1M装置上的Mirnov磁探针系统的布置及MHD不稳定性磁扰动模式的探测方法，探讨了实验中存在的MHD不稳定性现象。用空间傅立叶变换分析了HL－1M装置上的磁扰动模式，给出了在存在不同极向扰动模式放电条件下的各磁探针信号之间的相位比对关系结果，从而确认实验中所使用的方法的正确性和实用性。     

HL-1M装置锁模不稳定性研究

研究了HL－1M装置上锁模不稳定性和密度极限破裂。锁模不稳定性通常出现在低密度的实验中,在软X射线和Mirnov磁扰动上都观察到了频率相同的先兆振荡,其时间尺度为10ms。边3比温度的然下降是产生模不稳定性的重要标志。锁械出现几毫秒后时常发生大破裂。提出了抑制锁械的稳定性的建议。     

Mitigation and prediction of disruption on the HL-2A Tokamak

Injection of high-Z impurities into plasma has been proved to be able to reduce the localized thermal load and mechanical forces on the in-vessel components and the vacuum vessel,caused by disruptions in Tokamaks.An advanced prediction and mitigation system of disruption is implemented in HL-2A to safely shut down plasmas by using the laser ablation of high-Z impurities with a perturbation real-time measuring and processing system.The injection is usually triggered by the amplitude and frequency of the MHD perturbation field which is detected with a Mirnov coil and leads to the onset of a mitigated disruption within a few milliseconds.It could be a simple and potential approach to significantly reducing the plasma thermal energy and magnetic energy before a disruption,thereby achieving safe plasma termination.The plasma response to impurity injection,a mechanism for improving plasma thermal and current quench in major disruptions,the design of the disruption prediction warner,and an evaluation of the mitigation success rate are discussed in the present paper.     

Analysis of tokamak plasma confinement modes using the fast Fourier transformation

The Fourier analysis is a satisfactory technique for detecting plasma confinement modes in tokamaks. The confinement mode of tokamak plasma was analysed using the fast Fourier transformation (FFT). For this purpose, we used the data of Mirnov coils that is one of the identifying tools in the IR-T1 tokamak, with and without external field (electric biasing), and then compared it with each other. After the Fourier analysis of Mirnov coil data, the diagram of power spectrum density was depicted in different angles of Mirnov coils in the ‘presence of external field’ as well as in the ‘absence of external field’. The power spectrum density (PSD) interprets the manner of power distribution of a signal with frequency. In this article, the number of plasma modes and the safety factor q were obtained by using the mode number of q = m /n (m is the mode number). The maximum MHD activity was obtained in 30–35 kHz frequency, using the density of the energy spectrum. In addition, the number of different modes across 0–35 ms time was compared with each other in the presence and absence of the external field.     

Identification and Analysis of Magnetic Structures on HL-2A

1 Introduction Tearing modes in tokamak plasmas have been studied for many years In HL-2A tokamak, MHD instabilities are investigated by means of the Mirnov probes. The mode number can be determined by the methods of phase comparison analysis or correlation analysis from the experimental data of Mirnov probes, but the analysis of complicated mode structures is difficult. An identification and analysis method of magnetic islands is presented basing on simulation of the perturbation current and magnetic field in plasmas.     

Scrape-off layer turbulence modulated by Mirnov oscillations

We investigate plasma turbulence at the scrape-off layer of TCABR tokamak. We apply a power spectral analysis to the magnetic Mirnov oscillations and electrostatic fluctuations, to quantify statistical properties and to estimate the turbulence-driven radial-particle flux. A distinctive peculiarity is the modulation of electrostatic turbulence by the Mirnov oscillations shown by the partial superposition of the frequency power spectra of these two oscillations. This characteristic allows us to investigate any possible influence of the Mirnov oscillations on particle transport. In fact, a significant part of this transport occurs at the Mirnov frequencies. The effect of this modulation is also analyzed for discharges modified by external perturbations, a DC biased electrode or an ergodic magnetic limiter.Presented at the Workshop Electric Fields Structures and Relaxation in Edge Plasmas, Nice, France, October 26–27, 2004.     

The string prediction models as invariants of time series in the forex market

In this paper we apply a new approach of string theory to the real financial market. The models are constructed with an idea of prediction models based on the string invariants (PMBSI). The performance of PMBSI is compared to support vector machines (SVM) and artificial neural networks (ANN) on an artificial and a financial time series. A brief overview of the results and analysis is given. The first model is based on the correlation function as invariant and the second one is an application based on the deviations from the closed string/pattern form (PMBCS). We found the difference between these two approaches. The first model cannot predict the behavior of the forex market with good efficiency in comparison with the second one which is, in addition, able to make relevant profit per year. The presented string models could be useful for portfolio creation and financial risk management in the banking sector as well as for a nonlinear statistical approach to data optimization.     

Generation of 20-GHz picosecond pulse trains in the normal and anomalous dispersion regimes of optical fibers

In this paper, we numerically and experimentally study two methods to generate 20-GHz pulse trains at 1550 nm from a dual-frequency beat-signal. The first method is based on the multiple four-wave mixing temporal compression occurring in the anomalous dispersion regime of a standard optical fiber (SMF). In the second original method, the initial sinusoidal signal is first converted into a parabolic pulses train through nonlinear propagation in a normally dispersive fiber. A subsequent linear compression in an anomalous dispersive fiber leads to well-separated picosecond pulses.     

基于神经网络的高压输电线路单端测距方法

针对单端行波故障测距第二个行波波头性质辨识问题,提出一种将小波模极大值方法和神经网络算法相结合的测距方法。采集故障波头时间差和极性等信息作为样本,利用神经网络的非线性拟合能力对样本进行训练、测试,从而建立相应的故障测距神经网络模型。将故障信息代入神经网络模型得到初步测距结果,根据初测结果和波头极性、时间差等性质的关系,对第二个行波波头进行正确辨识,从而得到优化的测距结果。经Matlab/Simulink仿真验证,该方法有较高的可靠性和精确性。