1、null nullcI|:1671- 637null (2005) 03-0047-03 * 0 !#pE“d肖元姣, null 苏广川, null 韩null 雷( v0“,null 100081)Knull null null1: null 对复杂的电子设备的故障进行快速准确的诊断和维护是一个急待解决的重要问题b在分析了单独应用专家系统和神经网络的优缺点的基础上,提出了一种集成式诊断专家系统模式,将神经网络与传统的专家系统进行有机结合b经实验证明,系统的推理效率明显提高,且具备良好的鲁棒性a启发性和透明性b由此可见,该集成方法对故障诊断十分有效b1nullonullM: null 神经网络
2、; null 故障诊断; null 专家系统; null 电子设备ms |: null V24 null null nullDS M : null ANeural network based fault diagnosis expert systemfor electronic equipmentXIAO Yuan-jiao, null SU Guang-chuan, null HAN Lei( Electronic Engineering Department, Beijing Institute of Technology , Beijing 100081, China)Abstract:
3、 null Itnull s very important to make fault diagnosis for electronic equipment and repaire it quickly and ac-curately. On the basis of analyzing the advantages and disadvantages when expert system or neural network isapplied separately to equipment fault diagnosis, the paper introduces an integrated
4、 fault diagnosis expert systemwhich combines the neural network with traditional expert system organically. It is proved that the inference ef-ficiency of the whole system is improved, and the system is robust, transparent and flexible. It is effective tofault diagnosis.Key Words: null neural networ
5、k; null fault diagnosis; null expert system; null electronic equipment0 null40 M , E“d(ES) | vZ, |) % J7.dESi5,1: M | 5a Mc a1 ? aw r qw bdLis) * (NN) ?Z, ?ES4 rZEb V NN“d ?aX:k ?as Ti) ? %ES MV Ua |iw 5b NN9l :2004-09-15 null null null :2004-11-09Te:o( 1980-) , o, , V 3,1Z_E“d, ?brz 1 2 g L a 2|B 2
6、 rz 2 rz 3.Y rz 4V3 null 3 =Y ?T#p“#pg1 g2 g3 g41 z 1 0. 9 0. 1 0. 1 0. 12 z 2 0. 1 0. 9 0. 1 0. 13 z 3 0. 1 0. 1 0. 9 0. 14 z 4 0. 1 0. 1 0. 1 0. 9null null:V0.9V#pi,0. 1V#pibBP * * “ , (4,7 c * “5? p V,YV k |11b0. 01, c0.8, c 0. 8 H 1 p,N H, 206Q, L=0. 00903b“, L= V4 UbV4null T“1 z1 z2 z3 z41 0. 9
7、 0. 1 0. 1 0. 1L=0. 9078 0. 0008 0. 0530 0. 07172 0. 1 0. 9 0. 1 0. 1L=0. 1662 0. 8828 0. 0672 0. 15023 0. 1 0. 1 0. 9 0. 1L=0. 1088 0. 0683 0. 8154 0. 04044 0. 1 0. 1 0. 1 0. 9L=0. 1709 0. 0796 0. 0519 0. 9261null null Vn,YV“Q, f L=M, Vr MY#p Tb #p“1, d:yg1 ( V0. 9) ,i Og2( V0. 1) ,i Og3( V0. 1) ,i
8、Og4( V0. 1) , #pz 1( V0. 9078)b3 null 1) gE“d * $ , | 1, 4NW J,4 “ TE“d+ ,iy B“ T0 !#pE“db2)YVv l“ T, BM s “,i|“ T“dFL=b HWw r q1 7 ib|.dE“d/ * /M ,/B“ T0 !#(/54:)493 null null null null null null null null null null nullo: null * 0 !#pE“dm4null mFig.4null Precision of trainingm5null ksmFig.5 null E
9、rrorof testingnull nullV m V A, = EP H, Xr , 9FM Ab ksm ,27 k“,“7 k, 20Bl0. 2,N_ * MY q: 74. 07%, MY (78. 4%b4 null YVsa_ V, * MY “S V, V T1 pb. “S+4 |, 4 b T. , VF “S , P “S ,null * / M #pE“d JnullS ,2001,34(5):24 nullZk, . * ?#pE“dJ. ZYv( j S) ,2001(2):2-3null5 null XU D, WU M, AN J W. Design of a
10、n expert system basedonneural network ensembles for missle fault diagnosisA.Proceedings of the 2003 IEEE International Conference onRobotics,Intelligent Systems and SignalProcessing C. 2003,Changsha, 903-908.6 null XU W. WANG D, ZHOU Z, CHEN H. Fault diagnosis ofpower transformers: application of fuzzy set theory, expertsystems andartificial neuralnetworksA . IEE Proceedings ,on science,measurement and technology C . Vol 144, Issue1, 39-44null54 null null null null null null; enull null null null null null null null null null null null null null null null null null null null null12
