基于粗糙神经网络的WSN节点故障诊断
|
摘要
无线传感器网络(Wirelss Sensor Network,WSN)技术作为当前的新技术研究热点之一,在军用和民用等许多领域都具有极高的研究价值和广泛的应用前景。但是随着WSN系统的自动化程度不断提高,WSN系统的结构也变得越来越复杂,并且因为WSN主要工作于复杂条件和恶劣环境中,WSN节点要承受风吹、日晒、雨淋等诸多不利因素的影响,很容易发生故障,导致WSN节点不能按预定的设计完成原有的任务功能,更何况WSN的大多数节点的工作环境的条件差异不大,多数节点一同发生故障的概率较高,以致于造成WSN的整个网络瘫痪。因此,实时地对WSN节点的工作状态进行监控,准确及时的进行WSN节点的故障诊断,能有效提高WSN运行的可靠性和安全性,保证WSN能有效发挥设计潜能完成预定的监控任务。
本文首先对WSN节点的故障特点、故障类型及故障模型进行研究;然后在深入研究Rough Sets理论和神经网络算法各自特点的基础上,分析Rough Sets理论与神经网络算法集成的可能性,并深入研究Rough Sets理论与神经网络算法集成的主要方式。根据WSN节点故障诊断特点选择BP神经网络模型作为神经网络的典型代表与Rough Sets算法相集成,由于传统BP网络算法存在易陷入局部极小以及收敛速度慢的固有缺陷,本文在深入分析BP网络算法原理的基础上,提出一种附加动量项的自适应学习率的BP算法(Additional Momentum Self-Adaptive BP, AMSABP),并针对WSN实际应用中故障监测系统的输入条件属性值为连续数值的情况,创新性的提出了基于Rough Sets属性约简算法与AMSABP算法相集成的故障诊断算法,即RS-AMSABP故障诊断算法,RS-AMSABP算法首先利用Rough Sets的属性约简算法对WSN节点的故障诊断决策表进行约简,得到WSN节点的最简故障诊断决策表,并由该决策表建立决策规则,最后根据决策规则建立AMSABP网络模型;WSN节点故障诊断仿真实验表明,与其他故障诊断算法相比,本文提出的RS-AMSABP算法在大大提高了网络收敛速度的情况下,故障诊断的正确率也大幅提高,故障诊断正确率达到99.74%,并且该算法具有很好的容错性。
由于WSN实际工作于复杂及恶劣的环境中,要承受天气条件、环境干扰诸多不利因素的影响,WSN节点发生故障时,节点的故障监测系统的输入条件属性值极有可能为区间数,本文提出基于含有区间数的粗糙元神经网络(Interval-NumbersRough Neural Network, INRNN)故障诊断算法,该算法创新性引入含有区间数的粗糙决策分析方法,并提出采用区间数的上、下限端点值作为粗糙神经元的输入构造粗糙神经元,从而可以用含有区间数的粗糙元神经网络解决含有区间数的WSN节点故障诊断问题,基于INRNN网络的WSN节点故障诊断实验表明,与其他故障诊断算法相比,INRNN算法的网络收敛速度快,故障诊断正确率高达99.57%。
本文针对WSN节点的故障诊断问题,提出了一整套有效的解决方案,采用本文提出的RS-AMSABP算法和INRNN算法开发WSN节点的故障诊断软件系统,可以有效解决WSN实际应用中的节点故障问题,具有很好的实用价值。
Wireless Sensor Networks (WSN) is a current emerging and hot technique, its emergence changes the way of human beings interact with nature. WSN has high research value and broad application prospects in the military and civil ,and many other areas. However, as the increased of degree of automation in WSN, its structure has become more complex, and the WSN mainly work on the complex conditions and harsh environments, the nodes of WSN have to bear the wind, sun, rain and many other negative factors, it is prone to failure, so the original design features can not be complete. Moreover, the enviromental conditions in the monitored region in which the nodes of WSN were deployed are similar, most likely the majority of nodes simultaneously fail, and resulting in paralysis of the entire network of WSN. Therefore, it is extremly necessary to monitor the working status of the nodes in WSN in the real-time, the timely and accurate fault diagnosis of nodes in WSN can effectively improve the WSN operation reliability and safety, and ensure that WSN complete the scheduled tasks.
In this paper, firstly, studied the characteristics, types, levels of nodes’fault in WSN, and then researched the basis of individual characteristics of the Rough Sets theory and neural network algorithms in depth, study the possibility and the way of integrating the Rough Sets theory and Neural Network algorithms. Based on the characteristics of node in WSN, select the BP neural network to integrated with Rough Sets, because BP has the inherent defects as easy to fall into partial minimal and slow convergence, this paper proposed a new improved AMSABP algorithm, for the condition of the input attribute value of fault monitor system is contineous, this paper proposed the integrated RS-AMSABP fault diagnosis method. Firstly, this method get the most simple decision-making table of the fault diagnosis by the improved discriminate matrix, then established diagnosis rules by the table. Finally, constructed the AMSABP network model by the diagnosis rules, and trainning the network through the sample data. The expriment results of fault diagnosis of the node in WSN show that RS-AMSABP algrithom made the high diagnostic accurate to 99.74% and low calculate time compared with other diagnosis method.
Because WSN mainly work in the complex and bad enviroment, when the failure occurred in WSN, the input attribute value of fault monitor system is likely contineous, this paper proposed constructed the rough neuron by the two endpoints of the interval numbers of the input attribute of the fault monitor system, and applied rough decision-making analysis method constructed a decision information system of WSN fault diagnosis with the interval numbers, so the problem of the fault diagnosis of nodes in WSN with the interval numbers can be resolved by the the three-layers feed-forward rough neural network with the interval numbers. The simulation results show the diagnosis algrithom based on the Interval-Numbers Rough Neural Network improved the diagnostic accurate to 99.57% when the computing time was greatly reduced.
This paper proposed a whole solution scheme for the fault diagnosi of nodes in WSN, effectively meet the actual needs which the developing of WSN technology and application. It has high practical value.
本文首先对WSN节点的故障特点、故障类型及故障模型进行研究;然后在深入研究Rough Sets理论和神经网络算法各自特点的基础上,分析Rough Sets理论与神经网络算法集成的可能性,并深入研究Rough Sets理论与神经网络算法集成的主要方式。根据WSN节点故障诊断特点选择BP神经网络模型作为神经网络的典型代表与Rough Sets算法相集成,由于传统BP网络算法存在易陷入局部极小以及收敛速度慢的固有缺陷,本文在深入分析BP网络算法原理的基础上,提出一种附加动量项的自适应学习率的BP算法(Additional Momentum Self-Adaptive BP, AMSABP),并针对WSN实际应用中故障监测系统的输入条件属性值为连续数值的情况,创新性的提出了基于Rough Sets属性约简算法与AMSABP算法相集成的故障诊断算法,即RS-AMSABP故障诊断算法,RS-AMSABP算法首先利用Rough Sets的属性约简算法对WSN节点的故障诊断决策表进行约简,得到WSN节点的最简故障诊断决策表,并由该决策表建立决策规则,最后根据决策规则建立AMSABP网络模型;WSN节点故障诊断仿真实验表明,与其他故障诊断算法相比,本文提出的RS-AMSABP算法在大大提高了网络收敛速度的情况下,故障诊断的正确率也大幅提高,故障诊断正确率达到99.74%,并且该算法具有很好的容错性。
由于WSN实际工作于复杂及恶劣的环境中,要承受天气条件、环境干扰诸多不利因素的影响,WSN节点发生故障时,节点的故障监测系统的输入条件属性值极有可能为区间数,本文提出基于含有区间数的粗糙元神经网络(Interval-NumbersRough Neural Network, INRNN)故障诊断算法,该算法创新性引入含有区间数的粗糙决策分析方法,并提出采用区间数的上、下限端点值作为粗糙神经元的输入构造粗糙神经元,从而可以用含有区间数的粗糙元神经网络解决含有区间数的WSN节点故障诊断问题,基于INRNN网络的WSN节点故障诊断实验表明,与其他故障诊断算法相比,INRNN算法的网络收敛速度快,故障诊断正确率高达99.57%。
本文针对WSN节点的故障诊断问题,提出了一整套有效的解决方案,采用本文提出的RS-AMSABP算法和INRNN算法开发WSN节点的故障诊断软件系统,可以有效解决WSN实际应用中的节点故障问题,具有很好的实用价值。
Wireless Sensor Networks (WSN) is a current emerging and hot technique, its emergence changes the way of human beings interact with nature. WSN has high research value and broad application prospects in the military and civil ,and many other areas. However, as the increased of degree of automation in WSN, its structure has become more complex, and the WSN mainly work on the complex conditions and harsh environments, the nodes of WSN have to bear the wind, sun, rain and many other negative factors, it is prone to failure, so the original design features can not be complete. Moreover, the enviromental conditions in the monitored region in which the nodes of WSN were deployed are similar, most likely the majority of nodes simultaneously fail, and resulting in paralysis of the entire network of WSN. Therefore, it is extremly necessary to monitor the working status of the nodes in WSN in the real-time, the timely and accurate fault diagnosis of nodes in WSN can effectively improve the WSN operation reliability and safety, and ensure that WSN complete the scheduled tasks.
In this paper, firstly, studied the characteristics, types, levels of nodes’fault in WSN, and then researched the basis of individual characteristics of the Rough Sets theory and neural network algorithms in depth, study the possibility and the way of integrating the Rough Sets theory and Neural Network algorithms. Based on the characteristics of node in WSN, select the BP neural network to integrated with Rough Sets, because BP has the inherent defects as easy to fall into partial minimal and slow convergence, this paper proposed a new improved AMSABP algorithm, for the condition of the input attribute value of fault monitor system is contineous, this paper proposed the integrated RS-AMSABP fault diagnosis method. Firstly, this method get the most simple decision-making table of the fault diagnosis by the improved discriminate matrix, then established diagnosis rules by the table. Finally, constructed the AMSABP network model by the diagnosis rules, and trainning the network through the sample data. The expriment results of fault diagnosis of the node in WSN show that RS-AMSABP algrithom made the high diagnostic accurate to 99.74% and low calculate time compared with other diagnosis method.
Because WSN mainly work in the complex and bad enviroment, when the failure occurred in WSN, the input attribute value of fault monitor system is likely contineous, this paper proposed constructed the rough neuron by the two endpoints of the interval numbers of the input attribute of the fault monitor system, and applied rough decision-making analysis method constructed a decision information system of WSN fault diagnosis with the interval numbers, so the problem of the fault diagnosis of nodes in WSN with the interval numbers can be resolved by the the three-layers feed-forward rough neural network with the interval numbers. The simulation results show the diagnosis algrithom based on the Interval-Numbers Rough Neural Network improved the diagnostic accurate to 99.57% when the computing time was greatly reduced.
This paper proposed a whole solution scheme for the fault diagnosi of nodes in WSN, effectively meet the actual needs which the developing of WSN technology and application. It has high practical value.
引文
[1]任丰原,黄海宁,林闯.无线传感器网络.软件学报,2003,14(7):1282-1291
[2]宋文学等编著.无线传感器网络技术与应用.北京:电子工业出版社,2007:1-27
[3]孙利民,李建中.无线传感器网络.北京:清华大学出版社,2005:1-26
[4]孙雨耕,张静,孙永进,房朝晖.无线自组传感器网络.传感技术学报,2003.6:331-334.
[5]李晓维,徐勇军,任丰原.无线传感器网络技术.北京:北京理工大学出版社,2007:224-232
[6] Pottie GJ, Kaiser WJ. Wireless integrated network sensors. Communications of the ACM 2000,43(5):51-58
[7] Pister K, Smart Dust: Autonomous Sensing and Communication in a Cubic Milimeter, http:// robotics. eecs. berkeley.edu/～pister/ SmartDust
[8] Rockwell. Wireless Sensing Network (WSN), http :// wins.rsc.rockwell.com/
[9] Noury N, Herve T, Riale V, Virone G, Mercier E, Morey G, Moro A , Porcheron T. Monitoring behavior in home using a smart fail sensor. IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine and Biology, October 2000:607 - 610
[10]曾鹏,于海斌,梁英等.分布式无线传感器网络体系结构及应用支撑技术研究.信息与控制,2004,33(3):307-313
[11] B. Krishnamachari and S. Lyengar, Distributed Bayesian algorithms for fault-tolerant event region detection in wireless sensor networks, IEEE Transactions on Computers,2004, 53 (3):241-249
[12] T. Clouqueur, K.K. Saluja and P. Ramanathan. Fault tolerance in collaborative sensor networks for target detection, IEEE Transactions on Computers 53, 2004 (3): 320-333
[13] Ishida Y, Adachi N. Immune algorithm for multi-agent: application to adaptive noise neutralization[C]. In: IEEE International Conference on Intelligent Robots and System, 1996:1739-1746
[14]林瑞仲.面向目标跟踪的无线传感器网络研究[博士学位论文].浙江:浙江大学, 2005
[15]冯志鹏.计算智能在机械设备故障诊断中的应用研究.博士学位论文.大连:大连理工大学,2003,5.3
[16]姜万录,张淑清,王益群著.基于混沌和小波的故障信息诊断.北京:国防工业出版社,2005:210-224
[17] Timmis J, Neal M, Hunt J. Artificial immune system for data analysis. Bio systems, 2000, 55(1-3):143-150
[18] Werner-Allen, G.; Swieskowski, P.; Welsh, M.; MoteLab: a wireless sensor network testbed。Information Processing in Sensor Networks, 2005. IPSN 2005. Fourth International Symposium on 15 April 2005: 483– 488
[19] Tim nieberg.“Distributed Bayesisian Algorithms for Fault-Tolerant Event Region Detection in Wireless Sensor Networks”. Preprint submitted to Elsevier Science, 2003
[20] Kuo-Feng Ssu, Chih-Hsun Chou, Hewijin Christine Jiau and Wei-Te Hu. Detection and diagnosis of data inconsistency failures in wireless sensor networks. Computer Networks, In Press, Corrected Proof, Available online 8 August,2005
[21]张劼,景博,张宗麟,孙勇,陈明.无线传感器网络中基于比较的簇节点故障诊断算法.传感器技术学报,2007,20(8):1860-1864
[22]蒋鹏.一种改进的DFD无线传感器网络节点故障诊断算法研究.传感器技术学报,2008,21(8):1417-1421
[23] Stefano Chessa and Paolo Santi. Crash faults identification in wireless sensor networks. Computer Communications, Volume 25, Issue 14, 1 September 2002:1273-1282
[24] Chihfan Hsin and Mingyan Liu. Self-monitoring of wireless sensor networks. Computer Communications,2006,29(4):462-476
[25] Werner-Allen, G.; Swieskowski, P.; Welsh, M.; MoteLab: a wireless sensor network testbed。Information Processing in Sensor Networks, 2005. IPSN 2005. Fourth International Symposium on 15 April 2005: 483-488
[26]高建良,卢伟业,徐勇军,李晓维.无线传感器网络的容错研究.计算机科学,2007,Vol.34(9):311-313
[27] Farivar, R.; Fazeli, M.; Miremadi, S.G. Directed flooding: a fault-tolerant routing protocol for wireless sensor networks,Systems Communications, 2005. Proceedings 14-17 Aug. 2005 (s):395-399
[28] Petar Djukic and Shahrokh Valaee The Edward S. Rogers Sr.Minimum Energy Fault Tolerant Sensor Networks. Global Telecommunications Conference Workshops, 2004: 22-26
[29] Sinem Coleri and Pravin Varaiya. Fault Tolerant and Energy Efficient Routing for Sensor Networks.Global Telecommunications Conference Workshops, 2004: 10-15
[30] Pawalk Z. Rough sets [M]. Norwell , Netherlands: Kluwer Academic Publisher, 1991
[31]胡寿松,何亚群,等.粗糙决策理论与应用[M].北京:北京航空航天大学出版社,2006.
[32] PAWLAK Z, SKOWRON A. Rough sets and Boolean reasoning[J]. Information Sciences, 2007, 177(1): 41-73
[33]张文修,吴伟志,梁吉业等.粗糙集理论与方法[M].北京:科学出版社, 2005.
[34]阮晓钢.神经计算科学[M].北京:国防工业出版社,2006.
[35]雷霖,代传龙,王厚军.基于Rough set理论的无线传感器网络节点故障诊断.北京邮电大学学报,2008, 30(4): 69-73
[36]雷霖,代传龙,王厚军.粗糙集-神经网络集成的WSN节点故障诊断.电子科技大学学报,2008,37(4):565-568
[37]赵熙临,刘辉.粗糙集理论在故障诊断中的问题分析.计算机技术与发展,2008,18(1):132-135
[38]何明,冯博琴,马兆丰,傅向华.一种基于粗糙集的粗糙神经网络构造方法.西安交通大学学报,2004,38(12):1240-1243
[39]陈遵德. Rough Set神经网络智能系统及其应用.模式识别与人工智能,1999,12(1):1-5
[40] Ahn B.S. , Cho S. S. , Kim C. Y. . The integrated methodology of rough set theory and artificial neural network for business failure prediction. Expert system with Application, 2000,18:65-74
[41] Chandana, Sandeep(IEEE), Mayorqa, et al. Rough set theory based neural network architecture// International Joint Conference on Neural Networks 2006. IEEE International Conference on Neural Networks-Conference Proceedings,July16-July 21, 2006,Vancouver,BC, Canada
[42] Wang S. , Scott E. , Gamermann A. . Extract rules by using rough set and knowledge-based NN, Journal of Computer Science and Technology, 1998,13(3):279-284
[43] Hashei R. R. , LeBlance L. A. , etl. Hybired intelligent system for predicting bank holding structure. European Journal of Operational Research, 1999,109(2):390-402
[44] Lazar, Alina, Sethi, etl. Decision rule extraction from trained neural networks using rough sets. Intelligent Engineering Systems through Artificial Neural Networks, 1999,10:493-498
[45] Nguyon H. S. . Neural network design: Rough set approach to real-valued data. In:Komorowski, The First European Symposium on Principles of Data Mining and Knowledge Discovery(PKDD’97), Norway, Springer-Verlag, Berlin,1997:359-366
[46]王国胤.Rough集理论与知识获取.西安:西安交通大学出版社, 2001: 1-155
[47]苗夺谦,李道国.粗糙集理论、算法及应用.北京:清华大学出版社,2008:175-204
[48]乔斌.粗糙集理论分层递阶约简算法的研究:[博士学位论文].浙江:浙江大学, 2003
[49] Min Liu, Degang Chen, Cheng Wu, Hongxing Li. Reduction method based on a new fuzzy rough set in fuzzy information system and its applications to scheduling problems. Computers &Mathematics with Applications, 2005.5,51(9-10):1571-1583
[50]胡可云.基于概念格和粗糙集的数据挖掘方法研究[博士学位论文].北京:清华大学, 2001
[51]于兴网.粗糙集属性约简算法在数据挖掘中的应用:[硕士学位论文].重庆:重庆大学, 2003
[52]冷永刚.粗糙集理论约简算法的研究:[硕士学位论文].成都:电子科技大学, 2003
[53]蒋瑜.粗集决策表属性约简算法的研究:[硕士学位论文].兰州:兰州大学, 2005
[54]黄国顺.基于粗糙集的决策表知识约简研究:[博士学位论文].武汉:华中科技大学,2007
[55] Wen jin, Zhao Jia li, Luo si wei , Han Zhen The improvement of BP neural network learning algorithm Signal Processing Proceedings,2000 WCCC_ICSP 2000,5th International Conference on,Volume:3 2000,Page(s): 1647～1649
[56]张昌星.前馈神经网络的新学习算法研究及其应用.控制与决策,1997,Vol.12,No.3:213～216
[57] Chien-cheng ,Yun-ching Tang.To improve the training time of BP neural networks.Info-tech and Info-net,2001 International Conferences on,Volumn:3.2001,Page(s):473～479 Vol.3
[58]孙伯清,潘启树,冯英浚,张长胜.提高BP网络训练速度的研究.哈尔滨工业大学学报,2001,Vol,33,No.4:439～441
[59]陈玉芳,雷霖.提高BP网络收敛速度的又一种算法.计算机仿真,2004,21(11):74-79
[60]殷跃.基于BP神经网络的电力变压器故障诊断的研究:[硕士学位论文].长春:吉林大学,2007
[61] Kryszkiewcz. M. Rules in incomp lete info rmat ion system s[A]. In: Proceedings from the Third Joint Conference on Info rmat ion Sciences [C]. North Carolina,USA, 25,March, 1997
[62] Kryszkiewcz. M. Rough set approach to incomplete information systems[J]. Information Sciences, 1998,112: 39- 49
[63] Jouni J. Rough sets defined by to lerances[A]. In: J.Penjam (ed.) , Proceedings of the Fenno-Ugric Symposium FUSST’99[C]. 1999. 247- 257
[64] Skowron A , Stepaniuk J. Tolerance approximation spaces [J]. Fundamenta Informaticae, 1996, ( 27) :245-253
[65]徐泽水.求解不确定型多属性决策问题的一种新方法.系统工程学报,2002, 17(2): 177-181
[66] Yoon K. The propagation of errors in multiple-attribute decision analysis: a practical approach.Journal of the Operational Research Society, 1989, 40(7): 681-686
[67] Bryson N, Mobolurin A. An action learning evaluation procedure for multiple criteria decision making problems. European Journal of Operational Research, 1995, 96: 379-386
[68]何亚群,胡寿松,侯霞.具有区间数的粗糙神经网络故障认定方法.应用科学学报,2004,22(3):283-286
[69] Zhu Hai-yang,Lei Lin.Fault Diagnosis of Node in Wireless Sensor Network Based on the Interval-Numbers Rough Neural Network. The 2nd IEEE International Conference on Information management and engineering (IEEE ICEME 2010).
[2]宋文学等编著.无线传感器网络技术与应用.北京:电子工业出版社,2007:1-27
[3]孙利民,李建中.无线传感器网络.北京:清华大学出版社,2005:1-26
[4]孙雨耕,张静,孙永进,房朝晖.无线自组传感器网络.传感技术学报,2003.6:331-334.
[5]李晓维,徐勇军,任丰原.无线传感器网络技术.北京:北京理工大学出版社,2007:224-232
[6] Pottie GJ, Kaiser WJ. Wireless integrated network sensors. Communications of the ACM 2000,43(5):51-58
[7] Pister K, Smart Dust: Autonomous Sensing and Communication in a Cubic Milimeter, http:// robotics. eecs. berkeley.edu/～pister/ SmartDust
[8] Rockwell. Wireless Sensing Network (WSN), http :// wins.rsc.rockwell.com/
[9] Noury N, Herve T, Riale V, Virone G, Mercier E, Morey G, Moro A , Porcheron T. Monitoring behavior in home using a smart fail sensor. IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine and Biology, October 2000:607 - 610
[10]曾鹏,于海斌,梁英等.分布式无线传感器网络体系结构及应用支撑技术研究.信息与控制,2004,33(3):307-313
[11] B. Krishnamachari and S. Lyengar, Distributed Bayesian algorithms for fault-tolerant event region detection in wireless sensor networks, IEEE Transactions on Computers,2004, 53 (3):241-249
[12] T. Clouqueur, K.K. Saluja and P. Ramanathan. Fault tolerance in collaborative sensor networks for target detection, IEEE Transactions on Computers 53, 2004 (3): 320-333
[13] Ishida Y, Adachi N. Immune algorithm for multi-agent: application to adaptive noise neutralization[C]. In: IEEE International Conference on Intelligent Robots and System, 1996:1739-1746
[14]林瑞仲.面向目标跟踪的无线传感器网络研究[博士学位论文].浙江:浙江大学, 2005
[15]冯志鹏.计算智能在机械设备故障诊断中的应用研究.博士学位论文.大连:大连理工大学,2003,5.3
[16]姜万录,张淑清,王益群著.基于混沌和小波的故障信息诊断.北京:国防工业出版社,2005:210-224
[17] Timmis J, Neal M, Hunt J. Artificial immune system for data analysis. Bio systems, 2000, 55(1-3):143-150
[18] Werner-Allen, G.; Swieskowski, P.; Welsh, M.; MoteLab: a wireless sensor network testbed。Information Processing in Sensor Networks, 2005. IPSN 2005. Fourth International Symposium on 15 April 2005: 483– 488
[19] Tim nieberg.“Distributed Bayesisian Algorithms for Fault-Tolerant Event Region Detection in Wireless Sensor Networks”. Preprint submitted to Elsevier Science, 2003
[20] Kuo-Feng Ssu, Chih-Hsun Chou, Hewijin Christine Jiau and Wei-Te Hu. Detection and diagnosis of data inconsistency failures in wireless sensor networks. Computer Networks, In Press, Corrected Proof, Available online 8 August,2005
[21]张劼,景博,张宗麟,孙勇,陈明.无线传感器网络中基于比较的簇节点故障诊断算法.传感器技术学报,2007,20(8):1860-1864
[22]蒋鹏.一种改进的DFD无线传感器网络节点故障诊断算法研究.传感器技术学报,2008,21(8):1417-1421
[23] Stefano Chessa and Paolo Santi. Crash faults identification in wireless sensor networks. Computer Communications, Volume 25, Issue 14, 1 September 2002:1273-1282
[24] Chihfan Hsin and Mingyan Liu. Self-monitoring of wireless sensor networks. Computer Communications,2006,29(4):462-476
[25] Werner-Allen, G.; Swieskowski, P.; Welsh, M.; MoteLab: a wireless sensor network testbed。Information Processing in Sensor Networks, 2005. IPSN 2005. Fourth International Symposium on 15 April 2005: 483-488
[26]高建良,卢伟业,徐勇军,李晓维.无线传感器网络的容错研究.计算机科学,2007,Vol.34(9):311-313
[27] Farivar, R.; Fazeli, M.; Miremadi, S.G. Directed flooding: a fault-tolerant routing protocol for wireless sensor networks,Systems Communications, 2005. Proceedings 14-17 Aug. 2005 (s):395-399
[28] Petar Djukic and Shahrokh Valaee The Edward S. Rogers Sr.Minimum Energy Fault Tolerant Sensor Networks. Global Telecommunications Conference Workshops, 2004: 22-26
[29] Sinem Coleri and Pravin Varaiya. Fault Tolerant and Energy Efficient Routing for Sensor Networks.Global Telecommunications Conference Workshops, 2004: 10-15
[30] Pawalk Z. Rough sets [M]. Norwell , Netherlands: Kluwer Academic Publisher, 1991
[31]胡寿松,何亚群,等.粗糙决策理论与应用[M].北京:北京航空航天大学出版社,2006.
[32] PAWLAK Z, SKOWRON A. Rough sets and Boolean reasoning[J]. Information Sciences, 2007, 177(1): 41-73
[33]张文修,吴伟志,梁吉业等.粗糙集理论与方法[M].北京:科学出版社, 2005.
[34]阮晓钢.神经计算科学[M].北京:国防工业出版社,2006.
[35]雷霖,代传龙,王厚军.基于Rough set理论的无线传感器网络节点故障诊断.北京邮电大学学报,2008, 30(4): 69-73
[36]雷霖,代传龙,王厚军.粗糙集-神经网络集成的WSN节点故障诊断.电子科技大学学报,2008,37(4):565-568
[37]赵熙临,刘辉.粗糙集理论在故障诊断中的问题分析.计算机技术与发展,2008,18(1):132-135
[38]何明,冯博琴,马兆丰,傅向华.一种基于粗糙集的粗糙神经网络构造方法.西安交通大学学报,2004,38(12):1240-1243
[39]陈遵德. Rough Set神经网络智能系统及其应用.模式识别与人工智能,1999,12(1):1-5
[40] Ahn B.S. , Cho S. S. , Kim C. Y. . The integrated methodology of rough set theory and artificial neural network for business failure prediction. Expert system with Application, 2000,18:65-74
[41] Chandana, Sandeep(IEEE), Mayorqa, et al. Rough set theory based neural network architecture// International Joint Conference on Neural Networks 2006. IEEE International Conference on Neural Networks-Conference Proceedings,July16-July 21, 2006,Vancouver,BC, Canada
[42] Wang S. , Scott E. , Gamermann A. . Extract rules by using rough set and knowledge-based NN, Journal of Computer Science and Technology, 1998,13(3):279-284
[43] Hashei R. R. , LeBlance L. A. , etl. Hybired intelligent system for predicting bank holding structure. European Journal of Operational Research, 1999,109(2):390-402
[44] Lazar, Alina, Sethi, etl. Decision rule extraction from trained neural networks using rough sets. Intelligent Engineering Systems through Artificial Neural Networks, 1999,10:493-498
[45] Nguyon H. S. . Neural network design: Rough set approach to real-valued data. In:Komorowski, The First European Symposium on Principles of Data Mining and Knowledge Discovery(PKDD’97), Norway, Springer-Verlag, Berlin,1997:359-366
[46]王国胤.Rough集理论与知识获取.西安:西安交通大学出版社, 2001: 1-155
[47]苗夺谦,李道国.粗糙集理论、算法及应用.北京:清华大学出版社,2008:175-204
[48]乔斌.粗糙集理论分层递阶约简算法的研究:[博士学位论文].浙江:浙江大学, 2003
[49] Min Liu, Degang Chen, Cheng Wu, Hongxing Li. Reduction method based on a new fuzzy rough set in fuzzy information system and its applications to scheduling problems. Computers &Mathematics with Applications, 2005.5,51(9-10):1571-1583
[50]胡可云.基于概念格和粗糙集的数据挖掘方法研究[博士学位论文].北京:清华大学, 2001
[51]于兴网.粗糙集属性约简算法在数据挖掘中的应用:[硕士学位论文].重庆:重庆大学, 2003
[52]冷永刚.粗糙集理论约简算法的研究:[硕士学位论文].成都:电子科技大学, 2003
[53]蒋瑜.粗集决策表属性约简算法的研究:[硕士学位论文].兰州:兰州大学, 2005
[54]黄国顺.基于粗糙集的决策表知识约简研究:[博士学位论文].武汉:华中科技大学,2007
[55] Wen jin, Zhao Jia li, Luo si wei , Han Zhen The improvement of BP neural network learning algorithm Signal Processing Proceedings,2000 WCCC_ICSP 2000,5th International Conference on,Volume:3 2000,Page(s): 1647～1649
[56]张昌星.前馈神经网络的新学习算法研究及其应用.控制与决策,1997,Vol.12,No.3:213～216
[57] Chien-cheng ,Yun-ching Tang.To improve the training time of BP neural networks.Info-tech and Info-net,2001 International Conferences on,Volumn:3.2001,Page(s):473～479 Vol.3
[58]孙伯清,潘启树,冯英浚,张长胜.提高BP网络训练速度的研究.哈尔滨工业大学学报,2001,Vol,33,No.4:439～441
[59]陈玉芳,雷霖.提高BP网络收敛速度的又一种算法.计算机仿真,2004,21(11):74-79
[60]殷跃.基于BP神经网络的电力变压器故障诊断的研究:[硕士学位论文].长春:吉林大学,2007
[61] Kryszkiewcz. M. Rules in incomp lete info rmat ion system s[A]. In: Proceedings from the Third Joint Conference on Info rmat ion Sciences [C]. North Carolina,USA, 25,March, 1997
[62] Kryszkiewcz. M. Rough set approach to incomplete information systems[J]. Information Sciences, 1998,112: 39- 49
[63] Jouni J. Rough sets defined by to lerances[A]. In: J.Penjam (ed.) , Proceedings of the Fenno-Ugric Symposium FUSST’99[C]. 1999. 247- 257
[64] Skowron A , Stepaniuk J. Tolerance approximation spaces [J]. Fundamenta Informaticae, 1996, ( 27) :245-253
[65]徐泽水.求解不确定型多属性决策问题的一种新方法.系统工程学报,2002, 17(2): 177-181
[66] Yoon K. The propagation of errors in multiple-attribute decision analysis: a practical approach.Journal of the Operational Research Society, 1989, 40(7): 681-686
[67] Bryson N, Mobolurin A. An action learning evaluation procedure for multiple criteria decision making problems. European Journal of Operational Research, 1995, 96: 379-386
[68]何亚群,胡寿松,侯霞.具有区间数的粗糙神经网络故障认定方法.应用科学学报,2004,22(3):283-286
[69] Zhu Hai-yang,Lei Lin.Fault Diagnosis of Node in Wireless Sensor Network Based on the Interval-Numbers Rough Neural Network. The 2nd IEEE International Conference on Information management and engineering (IEEE ICEME 2010).