Let F be a number field and p be a prime. In the successive approximation theorem, we prove that, for each integer n ≥ 1, finitely many candidates for the Galois group of the nth stage of the p-class tower over F are...Let F be a number field and p be a prime. In the successive approximation theorem, we prove that, for each integer n ≥ 1, finitely many candidates for the Galois group of the nth stage of the p-class tower over F are determined by abelian type invariants of p-class groups C1pE of unramified extensions E/F with degree [E : F] = pn-1. Illustrated by the most extensive numerical results available currently, the transfer kernels (TE, F) of the p-class extensions TE, F : C1pF → C1pE from F to unramified cyclic degree-p extensions E/F are shown to be capable of narrowing down the number of contestants significantly. By determining the isomorphism type of the maximal subgroups S G of all 3-groups G with coclass cc(G) = 1, and establishing a general theorem on the connection between the p-class towers of a number field F and of an unramified abelian p-extension E/F, we are able to provide a theoretical proof of the realization of certain 3-groups S with maximal class by 3-tower groups of dihedral fields E with degree 6, which could not be realized up to now.展开更多
为监测分布式驱动电动汽车中轮毂电机运行状态,确保整车运行安全,提出一种基于改进的多类支持向量数据描述(multi-class support vector data description,简称MCSVDD)的轮毂电机故障诊断方法。首先,针对MCSVDD算法的改进,基于近邻传播(...为监测分布式驱动电动汽车中轮毂电机运行状态,确保整车运行安全,提出一种基于改进的多类支持向量数据描述(multi-class support vector data description,简称MCSVDD)的轮毂电机故障诊断方法。首先,针对MCSVDD算法的改进,基于近邻传播(affinity propagation,简称AP)聚类算法提出了MCSVDD以“距离类内簇中心最小”的类别判断法则,并基于Weibull函数构造了Weibull核函数,用于优化数据描述模型;其次,针对轮毂电机运行状态的多维特征参数组,提出一种基于最小距离传播鉴别投影(minimum-distance propagation discriminant projection,简称MPDP)的降维法,提高了不同工况下轮毂电机故障状态的可分性;最后,定制带有典型轴承故障的轮毂电机,采集7种工况下的振动信号,验证所提出方法的有效性。结果表明:基于MPDP降维后的轮毂电机运行状态观测样本的可分性优于线性判别分析(linear discriminant analysis,简称LDA)、局部保持投影(locality preserving projection,简称LPP)及最小距离鉴别投影(minimum-distance discriminant projection,简称MDP)方法,基于Weibull核函数的MCSVDD状态识别系统的识别精度整体高于基于多项式和高斯核函数的MCSVDD系统。展开更多
针对电力通信网等工业互联网中非受控终端不能通过安装代理软件进行异常行为监测的问题,采用非侵入式网络监听手段,采集各终端设备进网流量、出网流量、IP组播流量、IP广播流量、会话总数等数据,提出一种基于长短时记忆网络的自适应动...针对电力通信网等工业互联网中非受控终端不能通过安装代理软件进行异常行为监测的问题,采用非侵入式网络监听手段,采集各终端设备进网流量、出网流量、IP组播流量、IP广播流量、会话总数等数据,提出一种基于长短时记忆网络的自适应动态多核单类支持向量机方法(Long ShortTerm Memory Adaptive Dynamic Multiple Kernel One Class Support Vector Machine,LSTM-ADMK-OCSVM),精确刻画各类非受控终端正常工作行为模态,构建异常行为描述和监测模型,实现对非受控终端设备非设定异常行为安全监测。通过电力信息内网非受控终端实际系统实验,得出所提方法可有效对非受控终端异常行为进行监测,精度达到95.36%,满足实际系统应用要求。展开更多
文摘Let F be a number field and p be a prime. In the successive approximation theorem, we prove that, for each integer n ≥ 1, finitely many candidates for the Galois group of the nth stage of the p-class tower over F are determined by abelian type invariants of p-class groups C1pE of unramified extensions E/F with degree [E : F] = pn-1. Illustrated by the most extensive numerical results available currently, the transfer kernels (TE, F) of the p-class extensions TE, F : C1pF → C1pE from F to unramified cyclic degree-p extensions E/F are shown to be capable of narrowing down the number of contestants significantly. By determining the isomorphism type of the maximal subgroups S G of all 3-groups G with coclass cc(G) = 1, and establishing a general theorem on the connection between the p-class towers of a number field F and of an unramified abelian p-extension E/F, we are able to provide a theoretical proof of the realization of certain 3-groups S with maximal class by 3-tower groups of dihedral fields E with degree 6, which could not be realized up to now.
文摘针对电力通信网等工业互联网中非受控终端不能通过安装代理软件进行异常行为监测的问题,采用非侵入式网络监听手段,采集各终端设备进网流量、出网流量、IP组播流量、IP广播流量、会话总数等数据,提出一种基于长短时记忆网络的自适应动态多核单类支持向量机方法(Long ShortTerm Memory Adaptive Dynamic Multiple Kernel One Class Support Vector Machine,LSTM-ADMK-OCSVM),精确刻画各类非受控终端正常工作行为模态,构建异常行为描述和监测模型,实现对非受控终端设备非设定异常行为安全监测。通过电力信息内网非受控终端实际系统实验,得出所提方法可有效对非受控终端异常行为进行监测,精度达到95.36%,满足实际系统应用要求。
