无线传感器网络故障诊断方法研究
发布时间:2018-05-24 23:27
本文选题:无线传感器网络 + 故障诊断 ; 参考:《上海电力学院》2015年硕士论文
【摘要】:随着在各种监测系统中越来越广泛的应用无线传感器网络(Wireless Sensor Networks,简称WSN),对无线传感器网络的研究也愈发重要。无线传感器网络节点部署之初起,便处于无人监控和检查的状态,传感器网络节点本身运行的状态我们无从得知,不可能对其进行实时监控或者经常检查,传感器网络一旦发生故障,就可能会对监测产生影响。因此,准确并且及时诊断出无线传感器网络的故障节点,尽早排除故障,能提高无线传感器网络运行的可靠性,保证应用无线传感器网络的监测系统完成预定的监测任务。本文对无线传感器网络故障诊断方法进行深入研究,研究内容有以下几个方面:(1)研究了粗糙集理论,将无线传感器网络故障节点的故障类型与对应故障特征属性做成相应的决策表,运用粗糙集理论对无线传感器网络故障诊断决策表进行约简,并对基于粗糙集理论的WSN故障诊断方法进行了仿真实验,结果证明了该方法的优越性,但同时也反应出基于粗糙集理论的WSN故障诊断方法的不足之处。(2)研究了基于BP算法的小波神经网络,针对其由于采用梯度算法导致的进化速度缓慢且目标函数容易陷入局部极小的问题,提出了在基于BP算法的小波神经网络中采用增加动量项和学习率自适应调整这种方法来对小波神经网络进行改进,通过训练实验证明了这种改进措施的可行性。最后,在WSN的故障诊断中应用这种改进的小波神经网络算法进行实验,通过实验不仅验证了改进的小波神经网络算法在WSN故障诊断中的可行性,更加体现出其良好的容错性能。(3)针对基于粗糙集理论的WSN故障诊断方法的容错能力不足和小波神经网络不能识别多余数据知识的缺点,本文将粗糙集理论与改进的小波神经网络集成来解决这个问题,并在WSN节点故障诊断仿真实验上对两者集成的RS-IWNN故障诊断算法进行仿真。与基于粗糙集理论的WSN故障诊断方法的实验结果比较,证明了RS-IWNN故障诊断算法的优越性。(4)针对DFD算法存在的能耗高及诊断为“正常”的条件苛刻这两个问题,对运行于CTP协议下的无线传感器网络所运用的DFD故障诊断算法进行改进,以减少故障诊断所消耗的能量,同时提高故障诊断正确率。通过仿真实验证明了该改进的故障诊断算法取得的良好效果。(5)针对目前的研究大多数集中于对WSN故障诊断算法的研究上,而忽略了对故障诊断系统设计的问题,提出一种WSN故障诊断系统的设计,并对设计内容进行了介绍。
[Abstract]:With the more and more extensive application of wireless sensor networks (WSNs) in various monitoring systems, the research on wireless sensor networks (WSNs) is becoming more and more important. Since the beginning of the deployment of wireless sensor network nodes, they have been in a state of unattended monitoring and inspection. We have no way to know the state of the nodes running in the sensor networks themselves, and it is impossible to monitor them in real time or to check them frequently. Sensor networks may have an impact on monitoring once they fail. Therefore, accurate and timely diagnosis of wireless sensor network fault nodes, early troubleshooting, can improve the reliability of the operation of wireless sensor networks, ensure the application of wireless sensor networks monitoring system to complete the scheduled monitoring tasks. In this paper, the methods of fault diagnosis in wireless sensor networks are deeply studied. The research contents are as follows: 1) the rough set theory is studied. The fault types of the fault nodes and the corresponding fault feature attributes of wireless sensor networks are made into the corresponding decision tables, and the rough set theory is used to reduce the fault diagnosis decision tables of wireless sensor networks. The simulation results of WSN fault diagnosis method based on rough set theory prove the superiority of the method. But it also reflects the deficiency of WSN fault diagnosis method based on rough set theory.) the wavelet neural network based on BP algorithm is studied. In view of the slow evolution speed caused by using gradient algorithm and the problem that the objective function is prone to fall into local minima, An improved wavelet neural network based on BP algorithm is proposed, which is based on the adaptive adjustment of momentum and learning rate. The feasibility of the improved method is proved by the training experiment. Finally, the improved wavelet neural network algorithm is applied to the fault diagnosis of WSN. The experiment not only verifies the feasibility of the improved wavelet neural network algorithm in WSN fault diagnosis. The fault tolerance ability of WSN fault diagnosis method based on rough set theory and the shortcoming of wavelet neural network can not recognize redundant data knowledge can be realized. In this paper, the rough set theory is integrated with the improved wavelet neural network to solve this problem, and the integrated RS-IWNN fault diagnosis algorithm is simulated in the WSN node fault diagnosis simulation experiment. Compared with the experimental results of WSN fault diagnosis method based on rough set theory, it is proved that the superiority of RS-IWNN fault diagnosis algorithm is to solve the two problems of high energy consumption and "normal" condition of DFD algorithm. In order to reduce the energy consumption of fault diagnosis and improve the accuracy of fault diagnosis, the DFD fault diagnosis algorithm used in wireless sensor networks running under CTP protocol is improved. Simulation results show that the improved fault diagnosis algorithm has a good effect. Aiming at the current research, most of the researches focus on the WSN fault diagnosis algorithm, but the design of the fault diagnosis system is ignored. The design of a WSN fault diagnosis system is presented, and the design content is introduced.
【学位授予单位】:上海电力学院
【学位级别】:硕士
【学位授予年份】:2015
【分类号】:TN929.5;TP212.9
【参考文献】
相关期刊论文 前7条
1 刘仁云,于繁华,叶欣;改进的小波神经网络及应用[J];长春师范学院学报;2004年10期
2 王建芳;李伟华;;基于扩展T-S模型的PSO神经网络在故障诊断中的应用[J];计算机科学;2009年09期
3 蔡用;胡绍华;胡立群;杨丽;杨保华;;关于粗糙集理论的一些探讨[J];科技广场;2008年12期
4 曹云峰;王耀才;王军威;;基于粗糙集的容错神经网络故障诊断系统[J];计算机工程与设计;2006年04期
5 胡耀垓,李伟,胡继明;一种改进激活函数的人工神经网络及其应用[J];武汉大学学报(信息科学版);2004年10期
6 陈健;杨志义;王敏;;无线传感器网络数据汇聚协议CTP的仿真与研究[J];现代电子技术;2011年06期
7 李宏;谢政;陈建二;李熙熙;向遥;;一种无线传感器网络分布式加权容错检测算法?[J];系统仿真学报;2008年14期
相关博士学位论文 前2条
1 李翠玲;粗糙集理论研究及其在虚拟装配系统中的应用[D];同济大学;2007年
2 金瑜;基于小波神经网络的模拟电路故障诊断方法研究[D];电子科技大学;2008年
相关硕士学位论文 前1条
1 徐洲;基于粗糙集的供应链企业之间知识流动影响要素识别研究[D];昆明理工大学;2010年
,本文编号:1931061
本文链接:https://www.wllwen.com/kejilunwen/wltx/1931061.html