Hard magnetization direction and its relation with magnetic permeability of highly grain-oriented electrical steel

The magnetic properties of highly grain-oriented electrical steel vary along different directions. In order to investigate these properties, standard Epstein samples were cut at different angles to the rolling direction. The hard magnetization direction was found at an angle of 60° to the rolling direction. To compare the measured and fitting curves, when the magnetic field intensity is higher than 7000 A/m, it is appropriate to simulate the relation of magnetic permeability and magnetization angle using the conventional elliptical model. When the magnetic field intensity is less than 3000 A/m, parabolic fitting models should be used; but when the magnetic field intensity is between 3000 and 7000 A/m, hybrid models with high accuracy, as proposed in this paper, should be applied. Piecewise relation models of magnetic permeability and magnetization angle are significant for improving the accuracy of electromagnetic engineering calculations of electrical steel, and these new models could be applied in further industrial applications.

Bi-Directional Giant Magneto Impedance Sensor

A thin film giant magneto impedance (GMI) based on magnetic field sensor has been developed using electrodeposited Ni-Fe permalloy. Chemical composition, surface morphology, and magnetic properties of Ni-Fe permalloy were char-acterized as a function of plateup parameters, and process conditions were established to deposit a Ni-Fe thin film with a high permeability (~1000) and a low coercivity (0.6 Oersted). Conventional GMI sensors are uni-directional and are several millimeters long. In this work, a spiral-shaped sensor using electroplated Ni-Fe permalloy to detect bi-directional magnetic field is reported. Excellent bi-directional magnetic field sensing has been demonstrated using the 1 mm2 compact double-spiral structure.

基于Inception-BiLSTM和迁移学习的结构损伤识别

针对传统卷积神经网络(convolutional neural network,CNN)方法在时空特征提取存在不足,提出了一种改进的Inception与双向长短期记忆(bi-directional long short-term memory,BiLSTM)联合模型,以全面学习振动信号中的空间和时序信息。首先,构建具有多尺度感受野的Inception模块,自适应地提取不同尺度下的空间特征;其次,BiLSTM序列化处理时间特征,以深度挖掘时间相关性;最后,通过全局平均池化和Softmax分类器来实现钢框架结构的损伤识别。为评估该模型对噪声的鲁棒性,引入高斯白噪声作为干扰。此外,采用迁移学习策略来评估模型在不同强度激励和小样本下的泛化能力,确保适用于不同的损伤识别任务。结果表明,与传统的CNN方法相比,该模型在无噪声条件下及信噪比超过25 dB时保持了100%的识别精度。该方法解决了土木工程应用中样本量不足和不同强度激励的实际挑战。通过微调预训练模型的参数,实现了在不同强度激励和小样本情况下的知识迁移与泛化,从而增强了模型的实际适用性。

基于BO-BiGRU-Attention短期电力负荷预测

电力系统的可靠供应对于工业、商业和居民的生活至关重要。为了满足电力需求并维持电力系统的稳定运行,提高短期电力负荷预测的准确性和可靠性尤为关键;针对负荷数据存在复杂的非线性特性,该文提出一种基于贝叶斯优化算法的双向门控循环单元和注意力机制(BO-BiGRU-Attention)的混合预测模型对短期电力负荷进行精准预测。首先,使用Min-Max Normalization方法对负荷数据进行归一化处理。其次,利用BiGRU网络捕获序列中的长期依赖关系和上下文信息,结合注意力机制,通过在输入序列的不同部分给予不同的权重,从而突出关键特征。最后,针对BiGRU-Attention模型的超参数难以选取最优解的问题,引入贝叶斯优化算法对BiGRU-Attention模型的超参数进行寻优,完成短期电力负荷的预测。采用印度北部某地区的电力负荷数据进行预测分析,仿真结果表明,BO-BiGRU-Attention网络表现优于其他模型,各误差评价指标最小,其中MAE、RMSE和MAPE分别为56.67,73.49和1.16%,预测精度达到了99.47%。

基于聚类的HPO-BILSTM光伏功率短期预测

考虑到光伏发电功率在不同天气类型下的波动性和不确定性,对此提出一种基于模糊C均值聚类算法(FCM)和猎食者优化算法(HPO)优化双向长短期记忆网络(BILSTM)的光伏发电短期功率预测模型。首先对光伏发电数据进行处理和分析,再进行主成分分析(PCA)降维和FCM聚类算法将数据按天气类型分为阴、晴、雨;最后通过HPO筛选得出BILSTM神经网络的最佳超参数,避免因超参数设置不佳对实验带来的影响,进一步提高实验的准确性和模型的泛化能力。最后通过预测和对比实验进行分析,验证所提方法的优越性。

基于CNN-BiGRU-Attention的短期电力负荷预测

针对目前电力负荷数据随机性强,影响因素复杂,传统单一预测模型精度低的问题,结合卷积神经网络(Convolutional neural network,CNN)、双向门控循环单元(Bi-directional gated recurrent unit,BiGRU)以及注意力机制(Attention)在短期电力负荷预测上的不同优点,提出一种基于CNN-BiGRU-Attention的混合预测模型。该方法首先通过CNN对历史负荷和气象数据进行初步特征提取,然后利用BiGRU进一步挖掘特征数据间时序关联,再引入注意力机制,对BiGRU输出状态给与不同权重,强化关键特征,最后完成负荷预测。试验结果表明,该模型的平均绝对百分比误差(Mean absolute percentage error,MAPE)、均方根误差(Root mean square error,RMSE)、判定系数(R-square,R~2)分别为0.167%、0.057%、0.993,三项指标明显优于其他模型,具有更高的预测精度和稳定性,验证了模型在短期负荷预测中的优势。

基于Bo-BiLSTM网络的IGBT老化失效预测方法

针对绝缘栅双极型晶体管(IGBT)受热应力冲击后对其进行老化失效预测精度不高的情况,提出一种基于贝叶斯优化(Bo)-双向长短期记忆(BiLSTM)网络的IGBT老化失效预测方法。首先分析IGBT模块老化失效原理,然后基于NASA老化实验数据集建立失效特征数据库,最后利用Matlab软件构造Bo-BiLSTM网络预测失效特征参数数据。选取常用回归预测性能评估指标将长短期记忆(LSTM)网络模型、BiLSTM网络模型与Bo-BiLSTM网络模型的预测结果进行对比分析。结果表明,Bo-BiLSTM网络的模型拟合精度更高,基于Bo-BiLSTM网络的IGBT老化失效预测方法具有较好的预测效果,能够应用于IGBT的失效预测。

基于TCN-BiLSTM网络的电力电缆故障诊断

为了提升电力电缆故障诊断技术的准确率,解决传统电力电缆诊断过程中操作复杂、可靠性低和精准度不够等问题,提出了一种基于TCN和BiLSTM的电力电缆故障诊断方法。该方法的核心是使用Matlab/Simulink搭建三相电缆的仿真模型,按照电缆的实际参数设置模型,然后提取出电缆的四种短路故障:单相接地短路、双相接地短路、双相相间短路以及三相短路的电压信号。构建电缆故障样本集,搭建TCN和BiLSTM网络对电缆故障信号进行特征提取和序列捕捉,通过与TCN网络和CNN-BiLSTM网络进行实验对比,以及对从淮南某煤矿采集到的数据进行验证,证明该方法对电缆故障诊断具有良好的性能。

基于MRSDAE-KPCA结合Bi-LST的滚动轴承剩余使用寿命预测

针对现有滚动轴承剩余使用寿命预测方法在提取数据特征时没有充分考虑数据的内部分布,且在构建健康因子时还需要专家经验进行人工提取等问题,提出一种基于流形正则化堆栈去噪自编码器、核主成分分析并结合双向长短时记忆网络的滚动轴承剩余使用寿命预测方法。首先采用无监督的堆栈去噪自编码器网络对原始振动数据进行深层特征提取,并使用核主成分分析法进一步降维,以提高健康因子的指标稳定性;然后在堆栈去噪自编码器中加入流形正则化,最大程度保留编码器隐藏层内部的数据分布结构,提高模型提取数据特征的有效性。最后使用双向长短时记忆网络预测轴承的剩余使用寿命,并采用AdaMax优化算法对网络模型的超参数进行自适应寻优。分析结果表明,提出的滚动轴承剩余使用寿命预测方法具有更高的精度。

基于EMD-PSO-Bi LSTM组合模型的短期风电功率预测

风电功率预测对风电并网的稳定运行具有重要意义。为了解决风电功率预测中的精度和模型稳定性问题,引入了EMD-PSO-BiLSTM模型。通过经验模态分解技术将原始风电功率序列分解为一系列固有模态函数,以有效捕捉数据中的多尺度特征,并为每个模态序列建立了各自的预测模型。鉴于双向长短时记忆神经网络良好的泛化能力,建立了基于BiLSTM的各模态预测模型。进一步采用粒子群算法优化了BiLSTM参数,解决了模型非线性、高维、多模态等问题,获得了各模态分量的最优模型,并通过汇总各模态分量的结果得到了风电功率预测值。最后,以湖南省某风电场的实际运行数据为例,验证了EMD-PSO-BiLSTM模型可以有效提高风电功率短期预测精度。

THE ANISOTROPY OF LOESS MAGNETIC SUSCEPTIBILITY IN THE NORTHEASTERN FRINGE OF QINGHAI-XIZANG PLATEAU AS AN INDICATOR OF PALAEOWIND DIRECTION

ＴＨＥＡＮＩＳＯＴＲＯＰＹＯＦＬＯＥＳＳＭＡＧＮＥＴＩＣＳＵＳＣＥＰＴＩＢＩＬＩＴＹＩＮＴＨＥＮＯＲＴＨＥＡＳＴＥＲＮＦＲＩＮＧＥＯＦＱＩＮＧＨＡＩＸＩＺＡＮＧＰＬＡＴＥＡＵＡＳＡＮＩＮＤＩＣＡＴＯＲＯＦＰＡＬＡＥＯＷＩＮＤＤＩＲＥＣＴＩＯＮＷａｎ...

Effect of power parameter and induction coil on magnetic field in cold crucible during continuous melting and directional solidification

Bottomless electromagnetic cold crucible is a new apparatus for continuous melting and directional solidification;however,improving its power efficiency and optimizing the configuration are important for experiment and production.In this study,a 3-D finite element (FE) method based on experimental verification was applied to calculate the magnetic flux density (Bz).The effects of the power parameters and the induction coil on the magnetic field distribution in the cold crucible were investigated.The results show that higher current intensity and lower frequency are beneficial to the increase of Bz at both the segment midpoint and the slit location.The induction coil with racetrack section can induce greater Bz,and a larger gap between the induction coil and the shield ring increases Bz.The mechanism for this effect is also discussed.

Research into magnetic guidance technology for directional drilling in SAGD horizontal wells

SAGD horizontal wells are used to enhance oil recovery from heavy oil reservoirs.This technology requires precise separation between the production well and the injection well to ensure the efficient drainage of the reservoir.By studying the attitude of the downhole probe tube and the production well trajectory,an algorithm is proposed for eliminating ferromagnetic interference while drilling injection wells.A high accuracy filter circuit has been designed to correct the detected magnetic signals,which are ultra-weak,frequency-instable,and narrow-band.The directional drilling magnetic guidance system（DD-MGS） has been developed by integrating these advanced techniques.It contains a sub-system for the ranging calculation software,a magnetic source,a downhole probe tube and a sub-system for collecting ＆ processing the detected signals.The DD-MGS has succeeded in oilfield applications.It can guide the directional drilling trajectory not only in the horizontal section but also in the build section of horizontal injection wells.This new technology has broad potential applications.

基于ATT-CNN-BiLSTM的虚拟编组列车时空轨迹预测

保障虚拟编组平稳追踪运行的关键问题是实现对列车运行状态的精准预测。针对列车运行过程多变的特点,提出基于融合注意力机制的卷积双向长短期记忆神经网络(ATT-CNN-BiLSTM)的时空轨迹预测方法。针对列车历史运行数据中非正常运行场景稀少产生的数据非均衡问题,利用卷积神经网络和双向长短期记忆网络提取列车运行数据维度之间的特征关联,并增加注意力机制提升泛化能力。同时引入运行时验证方法在线监控预测结果,降低由预测错误造成的行车风险。以成都地铁8号线真实数据为例进行实验,设计5种评价指标,通过基线模型与消融实验对所提ATT-CNN-BiLSTM进行评价,该模型对于异常场景的预测误差至少减小9.626%。

Influence of Tb on easy magnetization direction and magnetostriction of PrFe1.9 alloy

The crystal structures, magnetization, and spontaneous magnetostriction of ferromagnetic Laves phase Pr1-xTbxFe1.9 compounds are investigated in a temperature range between 5 K and 300 K. High resolution synchrotron x-ray diffraction(XRD) analysis shows that different proportions of Tb in Pr1-xTbxFe1.9 alloys can result in different easy magnetization directions(EMD) below 70 K, i.e., [100] with x = 0.0, and [111] with x ≥ 0.1. This indicates Tb substitution can lead the EMD to change from [100] to [111] with x rising from 0.0 up to 0.1. The Tb substitution for Pr reduces the saturation magnetization Ms and the magnetostriction to their minimum value when x = 0.6, but it can increase low-field(0 ≤ H ≤9 kOe, the unit 1 Oe = 79.5775 A·m-1) magnetostriction when x = 0.8 and 1.0 at 5 K. This can be attributed to the larger magnetostriction of PrFe1.9 than that of TbFe1.9, as well as the decrease of the resulting anisotropy due to Tb substitution at low temperatures.

THE BI DIRECTIONAL REGULATION OF FILAMIN ON THE ATPase ACTIVITY OF SMOOTH MUSCLE MYOSIN

The aim of this study is to investigate the functional relationship between filamin, a known actin binding protein, and myosin and the effects of filamin on the interaction between myosin and actin. Methods.Ultra centrifugation method was used to investigate the binding of filamin to both phosphorylated and unphosphorylated myosins. Mg ATPase activities of both phosphorylated and unphosphorylated myosins in the presence and absence of actin were measured to observe the effects resulted from filamin actin and filamin myosin interactions. Results. It was found that filamin is also a myosin binding protein. Filamin inhibited the actin activated Mg ATPase activity of phosphorylated myosin and stimulated Mg ATPase of phosphorylated myosin in the absence of actin; in addition, filamin stimulated Mg ATPase activity of unphosphorylated myosin in both the presence or absence of actin. Conclusion. The results suggest that the effects of filamin on the myosin Mg ATPase activities are bi directional, i.e., stimulatory via binding to myosin and inhibitory via binding to actin.

模型误差影响下基于CNN+BiLSTM神经网络的非圆信号目标直接跟踪算法

针对运动观测阵列状态误差与接收频率抖动同时影响下的非圆信号无源跟踪问题,提出了一种基于卷积神经网络(Convolutional Neural Network,CNN)+双向长短时记忆网络(Bi-directional Long Short-Term Memory,BiL⁃STM)的直接跟踪算法.该算法首先利用多运动观测阵列信号各频带间的相关性与辐射源信号的非圆特性,建立模型误差影响下的扩展多站观测矢量;接着利用多个观测时隙内扩展多站观测矢量的信号子空间构造空时特征输入序列;然后设计基于CNN与BiLSTM混合神经网络的直接跟踪模型,通过训练实现对非圆目标的轨迹矢量直接估计.本文算法是从原始数据信号子空间中估计轨迹矢量的直接跟踪模式,相比传统“观测参数估计+滤波轨迹跟踪”的两步估计模式,具有更高的估计精度.由于本文算法在神经网络训练过程中学习到模型误差的信息,因此能够实现对多种误差的校正.仿真结果表明,本文算法较传统两步跟踪算法与现有直接跟踪算法均具有更高的轨迹估计精度,能够明显提升模型误差影响下多站协同跟踪的鲁棒性.

Study on Application of Bi-directional Combination Technology Integrating Residue Hydrotreating with Catalytic Cracking RICP

After analysing the disadvantages of the traditional residue hydrotreating-catalytic cracking combination process, RIPP has proposed a bi-directional combination technology integrating residue hydrotreating with catalytic cracking called RICP which does not further recycles the FCC heavy cycle oil (HCO) inside the FCC unit and delivers HCO to the residue hydrotreating unit as a diluting oil for the residue that is concurrently subjected to hydrotreating prior to being used as the FCC feed oil. The RICP technology can stimulate residue hydrotreating reactions through utilization of HCO along with an in- creased yield of FCC light distillate, resulting in enhanced petroleum utilization and economic benefits of the refinery.

A High-Speed Bi-Directional Visible Light Communication System Based on RGB-LED

In this paper,we propose and experimentally demonstrate a bi-directional indoor communication system based on visible light RGB-LED.Spectrally efficient modulation formats(QAM-OFDM),advanced digital signal processing,pre- and postequalization are adopted to compensate the severe frequency response of indoor channel.In this system,we utilize red-green-blue Light emitting diodes(LEDs),of which each color can be used to carry different signals.For downlink,the low frequencies of each color are used while for uplink,the high frequencies are used.The overall data rate of downlink and uplink are 1.15-Gb/s and 300-Mb/s.The bit error ratios(BERs) for all channels after0.7 m indoor delivery are below pre-forwarderror-correction(pre-FEC) threshold of3.8×l0^(-3) To the best of our knowledge,this is the highest data rate in bi-directional visible light communication system.

Construction and analysis of a plant transformation binary vector pBDGG harboring a bi-directional promoter fusing dual visible reporter genes

The constitutive promoter of cauliflower mosaic virus 35S （CaMV 35S） is a polar unidirectional promoter and is widely used in plant genetic engineering. In the present study, the unidirectional CaMV 35S promoter has been modified to a bi-directional promoter by fusing its minimal promoter element to the 5＇ end of CaMV 35S promoter in the opposite orientation. To qualitatively and quantitatively estimate its bi-directional transcriptional function and activity, two visible reporter genes, gusA （13-glucuronidase, GUS） and gfp （green fluorescent protein, GFP）, were fused to the two ends of the promoter in bi-orientations ending with NOS terminator sequences, respectively. Stable expression of gusA and gfp genes in transgenic tobacco （Nicotiana tabacum L.） was visulized by histochemically staining for GUS and fluorescence microscopic observation under UV for GFP in transgenic plants. The expression of two reporter genes showed that the constructed bi-directional promoter did have the bi-directional transcriptional function in both expected orientations. The quantitative estimation of GUS and GFP were determined on a HITACHI F1000 Fluorescence Spectrophotometer with various wavelengths of excitation and emission. The GUS activity varied from g to 250 pmol 4-MU/min/mg protein and the GFP content varied from 0.9 to 1.8 μg/ mg protein in various lines of transgenic tobacco plants. Higher GUS activity generally coupled with lower GFP content, and vice versa.