ID 原文 译文
6014 通过引入神经网络中的正切Sigmoid激励函数和终端吸引子函数,构造了一种非线性跟踪微分器。 By introducing a tangent Sigmoid excitation function of neural network and terminal attractor function, to construct a nonlinear tracking differentiator.
6015 采用完备的稳定性理论证明了所设计跟踪微分器的渐近收敛性,并通过扫频测试分析其频域特性,总结出参数整定规则。 Using the stability of the complete theory proved that the designed asymptotical convergence of tracking differentiator, and the frequency domain characteristics through the analysis of the frequency sweep test, summed up the parameter setting rules.
6016 和之前的文献相比,进一步优化了跟踪微分器的结构,整合了功能重叠的参数使得更易调参。 Compared with the previous literature, further optimize the structure of the tracking differentiator, integrates the functions of overlapping parameters makes more adjustable parameters.
6017 同时,加入了终端吸引子函数降低高频信号引起的颤振使得噪声抑制能力更强。 Joined the terminal attractor function at the same time, reduce the high frequency signal caused by flutter has stronger noise suppression ability.
6018 仿真结果表明,所设计的跟踪微分器具备对包含一定噪声干扰的正弦信号的滤波和求导能力,且响应速度快、精度高,具备对方波信号的快速稳定跟踪和广义导数逼近的能力,且具有更强的噪声抑制能力。 The simulation results show that the design of tracking differentiator have to contain certain derivation of sine signal filtering and noise interference ability, and fast response speed, high accuracy, with the other wave signal fast track and generalized derivative approximation ability, and has stronger ability of noise suppression.
6019 基于脉冲描述字进行雷达信号分选时,传统聚类算法需要预先人工设定聚类中心和聚类数目。 Based on pulse description for radar signal sorting, traditional clustering algorithms need to manually set beforehand clustering center and clustering number.
6020 针对该问题,提出一种基于数据场理论联合脉冲重复间隔(pulse repetition interval,PRI)变换与聚类的雷达信号分选新方法。 Aiming at the problem, put forward a kind of based on data field theory combined pulse repetition interval (pulse repetition interval, PRI) transform and clustering, a new method of radar signal sorting.
6021 首先,依据数据场理论,基于势值大小实现干扰点剔除; First of all, based on the theory of data field, based on the size of the potential value to achieve noise points out;
6022 而后利用PRI变换算法进行PRI估计,依据PRI估计值将归一化脉冲描述字数据预分类; and then using the PRI transform algorithm is used to estimate the PRI, according the PRI estimation described the normalized pulse data classification;
6023 进而以各类数据集中心间的欧氏距离小于辐射因子为准则进行类别合并,自动得到初始聚类中心和聚类数目; and then to the heart of all kinds of data set of Euclidean distance is less than the radiation factor as a criterion for the category to merge, automatically get the initial clustering center and clustering number;