ID 原文 译文
2643 仿真实验结果表明，模型和算法有效且可行。 Finally, the experiments demonstrate that the model and method pro-posed in this paper are valid.
2644 建立了两种碳化硅(SiC)器件 JFET 和 MOSFET 的失效模型。 The failure models of SiC JFET and SiC MOSFET have been developed.
2645 失效模型是在传统的电路模型的基础上引入了额外附加的泄漏电流， Based on the conventional circuitmodels of SiC JFET and SiC MOSFET, the additional leakage currents between the electrodes are introduced.
2646 其中，SiC JFET 是在漏源极引入了泄漏电流，SiC MOSFET 是在漏源极和栅极引入了泄漏电流; For SiC JFET, the leakage current between the drain and the source is considered. For SiC MOSFET, two leakage currents are considered, one is the current between the drain and the source, another is the additional leakage current of the gate.
2647 同时，为了体现温度和电场强度与失效的关系，用与温度和电场强度相关的沟道载流子迁移率代替了传统电路模型所采用的常数迁移率。 Furthermore, themobility dependent on the temperature and the electric-field strength replaces the constant mobility in conventional circuit models.
2648 有关文献的实验结果和半导体器件的计算机模拟(Technology Computer Aided Design，TCAD)验证了两种 SiC 器件失效模型的准确性。 The results from other experimental works and TCAD simulations verify the failure models of SiC JFET and SiCMOSFET.
2649 所建立的失效模型能够对比 SiC JFET 和 SiC MOSFET 的短路特性。 The developed failure models can be used to compare the short-circuit performances of SiC JFET and SiC MOS-FET.
2650 目前众多的研究者通常直接将标签置信度矩阵作为先验知识直接加入到分类模型中，并没有考虑未标注先验知识对标签集质量的影响。 At present, many researchers usually directly add the label confidence matrix as a priori knowledge to the classification model, and do not consider the influence of non-equilibrium prior knowledge on the quality of the label set.
2651 基于此，引入非平衡参数的方法，将先验知识获得的基础置信度矩阵进行非平衡化，从而提出一种非平衡化的标签补全的核极限学习机多标签学习算法(KELM-NeLC): Based on this, the method of non-equilibrium parameters is introduced, and the basis confidence matrix obtained from the prior knowledge is non-equilibrium. Therefore, a multi-label learning algorithm is proposed, which uses kernel extreme learn-ing machine with non-equilibrium label completion (KELM-NeLC).
2652 首先使用信息熵计算标签之间的相关关系得到标签置信度矩阵， Firstly, information entropy is used to measure the cor-relation between labels which gets the basic label confidence matrix.