An improved ASM-HEMT model for kink effect on GaN devices

With the analysis of experiment and theory on GaN HEMT devices under DC sweep,an improved model for kink effect based on advanced SPICE model for high electron mobility transistors(ASM-HEMT)is pro⁃posed,considering the relationship between the drain/gate-source voltage and kink effect.The improved model can not only accurately describe the trend of the drain-source current with the current collapse and kink effect,but also precisely fit different values of drain-source voltages at which the kink effect occurs under different gatesource voltages.Furthermore,it well characterizes the DC characteristics of GaN devices in the full operating range,with the fitting error less than 3%.To further verify the accuracy and convergence of the improved model,a load-pull system is built in ADS.The simulated result shows that although both the original ASM-HEMT and the improved model predict the output power for the maximum power matching of GaN devices well,the im⁃proved model predicts the power-added efficiency for the maximum efficiency matching more accurately,with 4%improved.

基于动态加权符号互信息与k均值聚类的帕金森病患者静息脑电关联状态识别

目前,帕金森病发生率在逐渐上升,严重影响着患者生活质量,社会诊疗负担不断加重。然而,该病的早期监测手段有限,很难及时干预。为了发现其标志物,本文对服药前后帕金森病患者及健康人的32通道静息态脑电数据进行分频段研究。首先利用动态加权符号互信息计算各通道脑电信号间相关性,再通过k均值聚类实现信号关联矩阵的分类,最后得到脑电信号关联状态。通过统计分析发现,在Beta频段(P=0.034)与Gamma频段(P=0.010)各有一个脑电信号关联状态可显著区分未服药帕金森病患者与健康人。这表明未服药帕金森病患者与健康人的静息态脑电各通道信号相关性差异有统计学意义。而在服药与未服药帕金森病患者、服药帕金森病患者与健康人之间,其关联状态差异均没有统计学意义。这可为帕金森病的临床诊断提供一种参考。