ID 原文 译文
58468 天线的谐振频率和带宽由 2 个寄生耦合单元的结构尺寸以及寄生单元与馈电传递单元之间的相互耦合共同确定，可以通过调整寄生耦合单元的结构参数独立调整各谐振频点. The resonant frequency and bandwidth of the antenna are determined by the structure of the meander-linepatch parasitic coupling element，the T-shaped patch parasitic coupling element，and the mutual couplingbetween the two parasitic elements and the feeding transfer element.
58469 通过计算电磁学软件分析了各结构参数对谐振频率的影响，并仿真了一种典型情况的天线，其工作频段为 0.972 ～ 0.988 GHz、2.178 ～ 2.27 GHz 和 3.293 ～ 3.356 GHz. The proposed antenna achievesmultiband resonant frequencies that can be tuned independently by adjusting individual structure parameters. The associations between the antenna resonant frequencies and the structure parameters are simulated by electromagnetic simulation software，and it is shown that the antenna can typically work in0. 972 ～ 0. 988 GHz，2. 178 ～ 2. 27 GHz and 3. 293 ～ 3. 356 GHz.
58470 该天线仿真和测试结果符合得很好，验证了可通过单馈点激励多个寄生耦合单元的方法有效地设计多频段天线. The measurement results fit very wellwith the simulation，which validates the multiband microstrip antenna design method using parasitic ele?ments excited by single feed point.
58471 提出了一种基于熵和线性关系的两级流量异常检测方法，综合考虑了流量异常检测方法的准确性和实时性要求. To improve the accuracy and real-time traffic anomaly detection methods，a two-stage trafficanomaly detection method based on entropy and linear relation was presented.
58472 该方法在时间域上设定两级动态阈值，采用基于熵的方法对异常时间点进行检测，对熵值变化程度明显的时间点可使用一级阈值检测出来，而对熵值变化程度处于一级阈值和二级阈值之间的时间点采用基于线性关系的方法再次进行检测，并通过定义的报警触发函数识别异常类型. A two-level dynamicthreshold was set and used based on entropy and linear relation to detect abnormal traffic in time domain. The type of traffic anomaly can be indentified through the alarm trigger function.
58473 仿真实验结果证明，该方法在准确性和实时性方面优于现有的方法. Simulations verify thatthis method is superior to the existing methods in the accuracy and real-time respects.
58474 在相关性快速过滤特征选择算法( FCBF) 基础上，通过最大相关系数的方式改进 FCBF 算法. Based on the correlation fast Filtering Feature selection algorithm ( FCBF) ，which is improvedby the maximum correlation coefficient.
58475 首先，通过最大相关系数和对称不确定性度量准则，计算出每个特征与标签之间的相关度量值，并按照数值大小顺序进行排序; Firstly，It calculates the correlation measure between each fea?ture and label with the‘maximum normalized information coefficient’criterion
58476 其 次，通过最大相关系数和近似马尔可夫毯原理进行无关特征和冗余特征的筛选， and‘measurement princi?ple of symmetric uncertainty’and sort these feature according to the calculated value.
58477 最终选择出最优特征子集. Finally，It filters ir?relevant features and redundant features by the ‘maximum normalized information coefficient’criterionand approximate Markov Blanket and obtain the optimal feature subset.