欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
基于构造型神经网络运算复杂度低、构造直观方便、学习速度快、可解释性强的特点，提出一种跳频信号动态检测方法．首先从滑动窗口的数据流模型入手，采用适合增量学习的覆盖算法动态聚类，聚合相似度大的样本，分离相似度小的样本，降低了聚类复杂度，并减轻了噪声的影响，实现了覆盖簇的动态维护．然后在不同的覆盖簇中提取信号数据概要，构造数据结构数组，运用时频关联方法，排除数组中的定频信号、突发信号、随机噪声信号等，分离出了其中的跳频信号，实现了跳频信号的动态检测．实验结果表明该方法能快速、准确地动态检测跳频信号．
英文摘要：
In this paper, the method for FI-I signals dynamic detection is presented based on the Constructive Neural Network which has the advantages of low computation complexity and convenient construction. Firstly, based on the sliding windows data stream models 0 the method takes advantage of Covering Algorithm to cluster data stream, aggregates the similar samples, separates the small similarity samples, reduces the clustering complexity and realizes the dynamic maintenance of the covering clusters, then reduces the influence of noises. Secondly, the data synopses are abstracted from each covering cluster and the data structure arrays are construc- ted, the time-frequency correlation method is used, the fixed-frequency signals, burst signals, random noise signals are eliminated, then the FH signals is dynamic detected. The experimental results have shown that our method can detect dynamically with high speed and accuracy.
同期刊论文项目
同项目期刊论文
期刊信息
位置：