课程文献中心 更多>>

关键词： 变压器   BP神经网络   主成分分析   故障诊断  

摘要： BP神经网络在变压器故障诊断方面己经得到了应用,但仍存有收敛速度慢、易陷入局部极小等不足。为解决以上问题,采用主成分分析法优化BP神经网络诊断模型,达到提高诊断精度的目的。通过采用主成分分析对网络学习样本进行降维分析,使得神经网络的结构简化及神经网络的不足得到改善。最后基于该模型对变压器故障进行实例分析,并对比其他变压器诊断方法,结果表明主成分-BP神经网络判别模型具有更高的准确性和可靠性。

关键词： 分布式光伏   概率预测   分层关联建模   深层时空关联性  

摘要： 准确的区域分布式光伏功率概率预测可为有源配电网优化运行提供更全面的信息支撑。当缺乏气象测量或预报数据时,对分布式光伏时空相关信息的挖掘利用可以有效提升功率预测精度,然而,现有研究或难以针对性挖掘时空关联信息,或在建模过程中丢失大量有效信息。为此,提出了一种基于分层关联建模的区域分布式光伏功率超短期概率预测方法。首先,采用基于深度一致性的聚类方法对分布式光伏集群进行子区域划分,支撑对子区域内部时空关联的针对性建模;在此基础上,构建分层图结构以同步建模子域内与子域间时空关联关系,有效利用不同层级间关联信息;然后,提出了基于分层图卷积神经网络的概率预测模型,挖掘光伏电站之间的深度时空关联特征,提升区域分布式光伏超短期功率概率预测精度。最后,利用实际分布式光伏功率数据集验证了该方法的有效性。

关键词： SF_(6)变压器   气体冷却循环风机   叶片断裂   智慧化管理   运行状态评估   故障诊断  

摘要： 气体冷却循环风机(气泵)被广泛使用在气体绝缘电力变压器的冷却系统内,是冷却系统的核心部件,因此,气体冷却循环风机的稳定性是变压器运行稳定的关键条件之一。然而随着设备使用年限的增加,不可避免的会产生各类机械故障,其中叶片断裂是最常见也是最严重的故障之一。为了确保风机的安全运行,需要实时监测和有效干预风机的运行状态,以防止故障发生。同时,出现故障时,要能准确诊断并采取应对措施。基于SF_(6)气体冷却循环风机的高效和安全需求,提出了SF_(6)变压器气体冷却循环风机智能监测系统的功能并加以实施。这项研究为SF_(6)变压器气体冷却循环系统的安全可靠运行提供了有力的支持。

关键词： 解耦控制   LCL滤波器   复矢量   主动阻尼  

摘要： 采用复矢量对同步旋转坐标系(SRF)下的LCL型电机逆变器进行建模,深入研究不同控制策略下系统的整体相位裕度与耦合性能影响,揭示了逆变器不仅在dq轴间有耦合关系,主动阻尼反馈支路同样会加剧耦合现象。由此提出一种复矢量解耦策略,依次设计主动阻尼反馈支路和电流环系统的解耦环节,分别对反馈支路与PI控制器输出电流施加解耦环节,不仅可以有效消除动态耦合而且使系统具有良好的动态性能。通过仿真和试验与传统解耦控制效果作对比,验证了该控制策略的有效性。

关键词： Semiconductor device modeling   Predictive models   MOSFET   Uncertainty   Data models   Long short term memory   Hidden Markov models   Power electronics   Mathematical models   Logic gates   Bayesian neural network (BNN)   machine learning   model uncertainty   power metal-oxide-semiconductor field-effect transistor (MOSFET)   precursor prediction   reliability  

摘要： As the core component of power electronic systems, health monitoring of metal-oxide-semiconductor field-effect transistor (MOSFET) is extremely crucial. In this article, a hybrid failure precursor prediction model based on machine learning techniques is proposed. It consists of an isolation forest method and a long short-term memory (LSTM) network. The proposed model extracts information from different aspects of the input data to make predictions and can be sensitive to abnormal data behavior. By detecting the abnormality in the curve and predicting its future behavior, the model can give early warning of the power MOSFET failure and help avoid unexpected accidents. Besides, the model uncertainty is discussed. Two main factors that affect the model uncertainty of the proposed model are evaluated. To reduce the model uncertainty, a Bayesian neural network (BNN) is used to quantify the uncertainty of the proposed model with different parameters. The performance of the proposed model is verified based on the power MOSFET data collected from the accelerated life tests (ALTs). The experimental results indicate satisfying performances of the proposed model, because it can not only give early warning of MOSFET failures but also provide more stable prediction results with less model uncertainty compared with other existing models.

关键词： Meminductor   Memristor   Emulator   Chaotic oscillator   Pinched hysteresis loop   Chua's circuit  

摘要： In this paper, a grounded meminductor emulator has been proposed using six MOSFETs (3NMOS and 3PMOS) only. Three transconductance stages and two capacitors are generally required to realize a meminductor emulator. Meminductor is considered as an inductor having memory. In the proposed design of meminductor emulator, the gyrator circuit is realized using common-source and common-gate amplifiers with one capacitor. The first capacitor is used to obtain the behaviour of inductance while the second capacitor is used for charge storage which acts as a memory element. In the proposed design of meminductor emulator, the gate-to-source capacitor of MOSFET is utilized in place of external capacitor for memory element. Therefore, the proposed design is free from the requirement of passive components as the required capacitor is formed by MOSFET. Also, the frequency characteristics of proposed meminductor are found to be satisfactory up to 100 MHz. The other essential characteristics such as non-volatility test, temperature analysis, tunability, Monte Carlo, and corner analysis are also found to be satisfactory. The proposed design of meminductor emulator is compared with other existing meminductor emulators. The performance of the emulator is successfully verified through the realization of chaotic oscillator circuit.

关键词： Total ionizing dose   Single event gate rupture   Synergistic effect   VDMOS with stacked layers   TCAD simulation  

摘要： The synergistic effect of total ionizing dose on single event gate rupture (SEGR) was simulated in the vertical double diffusion metal oxide semiconductor device with SiO2-Si3N4 stacked gate layer. In comparison to the device with a single SiO2 gate layer, the synergistic effect was revealed to be suppressed in the device with SiO2-Si3N4 stacked layer. The mechanism is that the oxide layer is a sensitive area of the SEGR effect. Compared with the single SiO2 layer, the superposition of the additional electric field formed by the trapped holes in the sensitive area of the stacked layer leads to a decrease in the sensitivity of the synergistic effect, which is more obvious with increasing the volume of the Si3N4 layer.

关键词： SiC MOSFET   Gate oxide degradation   DM EMI   Amplitude-frequency characteristics  

摘要： This paper investigates gate oxide degradation and its influence on Differential Mode(DM) EMI signals. The study reveals that changes in the parasitic capacitance within the chip, resulting from gate oxide degradation, can modify the amplitude-frequency characteristics of DM EMI signals, leading to unexpected transmission outcomes. This paper subjects SiC MOSFET modules to periodic high-temperature gate bias stress, extracts the spectrum characteristics of DM EMI, and assesses the impact of junction temperature. The analysis explores the impact of gate oxide degradation on low and high frequency DM EMI signals. Experimental results reveal distinct sensitivities of amplitude-frequency characteristics in the DM EMI signals to temperature variations and responses to gate oxide degradation at various frequency bands. The extrapolation of the correlation between the frequency-domain characteristics of the DM EMI signals and the degree of gate oxide degradation introduces a novel approach to evaluate the lifespan of power electronic devices.

关键词： Power MOSFETs   Silicon-germanium (SiGe)   Heterojunction   Reverse recovery   Reverse conduction  

摘要： -A novel power MOSFET structure with Drain-side N-Si/N-SiGe Heterojunctions (DH-MOS) is proposed by introducing a N SiGe region which is sandwiched between the N-drift region and N+ substrate on the drain side. The operation mechanism and simulation verification of DH-MOS are presented. Due to the difference in valence band of N Si/N SiGe heterojunction, holes can be more easily expelled from the N-drift region into N SiGe region during the reverse conduction of DHMOS, resulting in a significant reduction in hole density inside the N-drift region, thus improving the reverse recovery performance. Simulation results show that the reverse recovery charge (Qrr) of DH-MOS is 2.08 mu C/cm2, approximately 66.34 % lower than the 6.18 mu C/cm2 of conventional MOS. The introduction of new DH-MOS structure with N-Si/N-SiGe heterojunctions does not result in a significant increase in reverse conduction voltage (VF). In addition, it does not sacrifice any forward conduction and blocking characteristics. The proposed DH-MOS introduces changes on the drain side, unlike existing methods which focus on the source side or drift region. And it can be compatible with existing improving measures to further improve the reverse recovery performance.

关键词： Logic gates   Degradation   Silicon carbide   Junctions   Temperature measurement   Monitoring   MOSFET   Gate-oxide degradation   junction temperature   monitoring   silicon carbide (SiC) metal-oxide-semiconductor field-effect transistors (MOSFETs)   temperature-sensitive electrical parameters (TSEPs)  

摘要： The online accurate junction temperature monitoring improves the reliability of power converters throughout the lifetime of silicon carbide (SiC) metal-oxide-semiconductor field-effect transistor (MOSFETs). Without the correction, the methods based on the temperature-sensitive electrical parameters (TSEPs) immune to packaging degradation have low accuracy for SiC MOSFETs with the gate-oxide degradation. Without shutting down the power converters, an online correction method is proposed by measuring the turn-on delay time and the specific charge time of input capacitance. The turn-on delay time is used to monitor the junction temperature. The specific charge time of input capacitance is used to characterize the gate-oxide degradation state. The junction temperature of the device under test (DUT) is monitored by the relationship between the turn-on delay time and the junction temperature of twin device (TD) in the same degradation state as the gate oxide of the DUT. The junction temperatures of SiC MOSFETs with different degradation states of gate oxide are accurately monitored online based on the buck converter. The monitoring error of junction temperature is one order of magnitude smaller than the uncorrected one. The accurate online monitoring of junction temperature for degraded SiC MOSFETs can be achieved.