Robust Output Feedback Controller Design for Time‐Delayed Teleoperation: Experimental Results

In this paper, a robust output feedback control strategy is proposed for a nonlinear teleoperation system which can deal with stability as well as transparency despite the variable time‐delay and uncertain dynamics. The proposed approach is composed of two steps. First, local Lyapunov based adaptive controllers are applied to both master and slave sides in order to suppress the nonlinearities in the system dynamics. Afterwards, a new observer‐based controller scheme is proposed to achieve stability and performance (transparency) of the teleoperation system. Using the Lyapunov techniques, stability and performance objectives are cast as some linear matrix inequality (LMI) feasibility conditions. To evaluate the performance of the proposed controller, a set of simulations and experiments are performed. Through simulation results, it is demonstrated that the proposed approach significantly outperforms the existing methodologies reported in the literature.     

Convergence Results of the Biconjugate Residual Algorithm for Solving Generalized Sylvester Matrix Equation

In this article, we investigate a variant of the biconjugate residual (BCR) algorithm to solve the generalized Sylvester matrix equation which includes the well‐known Lyapunov, Stein and Sylvester matrix equations. We show that the BCR algorithm with any (special) initial matrix pair can smoothly compute the (least Frobenius norm) solution pair of the generalized Sylvester matrix equation within a finite number of iterations in the absence of round‐off errors. Finally the accuracy and effectiveness of the BCR algorithm in comparison to some existing algorithms are demonstrated by two numerical examples.     

Robust Stochastic Stability and Control for Uncertain Singular Markovian Jump Systems with Multiplicative Noise

This paper focuses on the problems of robust stability and stabilization and robust control for uncertain singular Markovian jump systems with (x,v)‐dependent noise. The parameter uncertainties appearing in state, input, disturbance as well as diffusion terms are assumed to be time‐varying but norm‐bounded. Based on the approach of generalized quadratic stability, the memoryless state feedback controller is designed for the robust stabilization problem, which ensures that the resulting closed‐loop system has an impulse‐free solution and is asymptotically stable in the mean square. Furthermore, the results of robust control problem are derived. The desired state feedback controller is presented, which not only meets the requirement of robust stabilization but also satisfies a prescribed performance level. The obtained results are formulated in terms of strict LMIs. What we have obtained can be viewed as corresponding extensions of existing results on uncertain singular systems. A numerical example is finally given to demonstrate the application of the proposed method.     

基于Zadeh推理的故障空调噪音识别

通过声音来实现故障诊断在工业领域应用比较广泛。本文的主要目标是给出故障空调在运行状态下的声音,使用Zadeh推理的方法对故障空调的声音进行识别,进而推断出空调内部具体哪个部件或零件发生故障。本文首先对声音进行快速傅里叶变换,根据三分之一倍频程对整个频率进行划分,并对生成的频率进行标准化处理;当空调内部的零部件发生损坏时,噪音峰值就会在某一段或几段频率中有升高,从而提取每个频程的峰值作为特征向量。通过“专家经验”得出一个逻辑关系R矩阵,使用matlab软件将特征向量与R矩阵进行点积运算得到故障诊断的结果,最后GUI界面呈现出结果。实验表明,本方法对空调噪音的识别诊断效果比较理想。最后给出本方法适用的一些条件。本方法的实现原理简单,在空调噪音故障诊断应用领域可以有一定的应用空间。     

基于随机矩阵最大特征值分布的频谱感知算法

现有频谱感知算法在低信噪比时检测性能较低且受虚警概率影响大,针对此问题,提出了一种基于wishart矩阵样本协方差矩阵最大特征值的分布特性的频谱感知算法。该算法利用最大特征值与几何平均特征值的比值,不需要主用户的先验知识,不敏感于噪声,对相关信号和独立同分布信号均具有较高的检测性能。仿真结果表明,所提算法受虚警概率的影响较小,检测性能高,并且在采样点数、协作用户数、信噪比及虚警概率较小的情况下,也能获得较好的检测性能。     

基于关联矩阵的短信自动分类

短信自动分类是短文本研究的热点问题。针对此问题,提出了关联强度和关联矩阵特征提取方法,并设计了基于关联矩阵的全监督学习算法。为了实现系统的自我学习,探讨了基于关联矩阵的半监督学习算法,其结合了人工矫正的主动学习算法。最后通过实例验证说明了算法的有效性。     

高校贫困学生辅助评价的研究

高校对贫困学生的认定存在认定成本高、可信度不强以及标准不一致等问题.本文通过分析高校一卡通的消费行为,刻画贫困学生的消费特征,应用马尔科夫模型对贫困学生消费行为进行建模,并提出了相似指标的概念和计算方法.通过对学生消费行为与贫困学生消费行为模型进行相似指标计算,对贫困学生进行认定.该方法具有计算效率高,速度快,计算成本低廉以及数据获取容易,在同一所学校中评价标准一致,对贫困学生的平均识别率达到90%以上等特点,可以作为高校在复评贫困学生的一个有力的辅助工具.     

社交网络用户影响力的模糊综合评价

社交网络平台信息传播迅速,为了有效地进行舆情预警,定量地评估用户在消息传播网络中重要性,将模糊综合评价方法引入用户影响力建模问题中. 通过对用户在社交平台上的行为分析,构造了包含用户活跃粉丝数以及平均转发数等五项指标在内的评价体系. 并针对传统的模糊综合评价算法在应用于计算评价指标权重方面的缺陷与不足,提出改进模糊合成算子的方法构建用户影响力评估模型. 利用新浪微博社交平台上的真实数据,结合对比实验和实际评估,改进的模糊合成算子能根据需求调整权重对评价结果的影响,同时该方法能较准确地反应社交网络中用户的实际影响力.     

基于矩阵分解的个性化推荐系统研究

随着社交网络的快速发展,用户在使用社交应用时会产生大量有价值的数据。通过对社交网络进行数据挖掘,发现隐藏在数据中关联用户与物品之间的偏好关系。然后对用户建模分析,选择合适的推荐引擎进行个性化物品推荐,这是一个非常有价值的研究方向。该文重点研究矩阵分解算法对处理大规模用户与物品评分矩阵的推荐效果,为了提高推荐的准确度展开了对用户社交关系和隐性反馈的研究,在组合预测模型中加入社交关系、人口统计学信息配置项、用户的消费记录等隐因子项,通过实验验证了扩展之后的混合预测模型在RMSE值上比SVD算法降低了0.259 475,在推荐性能有较大幅度的提高。     

基于异常数据驱动的WSN簇内数据融合方法

引入数据驱动的思想,提出了一种基于异常数据驱动的簇内数据融合方法.在节点数据采集过程中,仅当异常数据发生时才发送给簇头,减少了监测网络的数据传输量.在簇头数据融合过程中,建立了各传感器之间的相互支持度矩阵,支持度值较低的监测数据将被剔除,支持度值较高的监测数据进行最优加权融合,从而保证了融合结果的准确性和有效性.仿真实验结果表明,与算术平均值法及自适应加权融合法相比,本文方法能有效去除冗余信息,在融合精度、能量消耗方面具有明显的优势.