This paper is concerned with the stochastic bounded consensus tracking problems of leader-follower multi-agent systems,where the control input of an agent can only use the information measured at the sampling instants from its neighbours or the virtual leader with a time-varying reference state,and the measurements are corrupted by random noises.The probability limit theory and the algebra graph theory are employed to derive the necessary and sufficient conditions guaranteeing the mean square bounded consensus tracking.It is shown that the maximum allowable upper boundary of the sampling period simultaneously depends on the constant feedback gains and the network topology. Furthermore,the effects of the sampling period on the tracking performance are analysed.It turns out that from the view point of the sampling period,there is a trade-off between the tracking speed and the static tracking error. Simulations are provided to demonstrate the effectiveness of the theoretical results.
近年来,针对说话人识别算法普遍受到信道因素的干扰问题,研究者提出使用总变化因子分析的识别方法对语音信道进行补偿得到了很不错的效果,其中概率线性判别分析(Probabilistic Linear Discriminant Analysis,PLDA)因其表现优异而受到学者们的关注。然而,高斯PLDA模型中I-Vector并非符合标准正态分布。因此论文在特征域利用特征弯折算法对梅尔倒谱系数(MFCC)进行处理,以消除背景噪声以及线性信道的影响。然后在模型域对I-Vector进行非线性转换使其分布更适合用PLDA模型区分说话人,以提高说话人识别系统的识别率。实验结果表明,使用传统GMM/UBM的系统,在NIST SRE-2010评估数据集上使用所提议的技术获得了很好的效果。