摘要
智能故障诊断系统作为人工智能技术在故障诊断领域的应用,在实践中取得了较好的成效,但随着系统设备和功能的日益复杂化,各种故障现象成因越来越复杂,同时异常故障也时有发生,现有固定的诊断推理模型却难以满足复杂系统诊断面临的全部要求。针对故障诊断系统难以适应动态变化环境的缺点,基于免疫原理的多Agent故障诊断系统充分利用现有的或可获得的新诊断子系统和其它组元,可以快速自适应调整诊断过程、诊断功能和诊断能力,代表了故障诊断系统的发展方向。
论文围绕基于免疫机理的多Agent故障诊断系统这一关键问题,以国家自然科学基金项目“基于免疫应答理论重构故障诊断系统”和企业横向项目“励磁装置在线故障预测和故障诊断系统”为背景,就免疫机理在故障诊断系统研究中的诊断过程建模、诊断推理模型结构及其软硬件设计等重点问题进行了较为系统深入的研究,主要研究内容包括以下几个方面:
1在分析相关免疫学和故障诊断基本理论基础上,阐释了故障诊断系统的免疫学理论基础,结合Agent技术特点,提出了运用生物免疫机制解决故障诊断的方法。
2从系统工程的观点出发,将复杂设备系统按结构与功能进行层次分解,建立设备系统的解析模型用于描述系统的行为结构和故障状态空间,在生物免疫系统和故障诊断系统映射关系的基础上,借鉴生物免疫系统的免疫反应机理、体系结构和相关抽象算法,建立了故障诊断系统的免疫框架,分析了基于组件和智能体技术的软件实现方法,给出了相应智能体的结构模型,对该模型的功能实现进行了详细的分析。
3基于从已有的诊断经验事例中学习获取知识的思路,与经验事例的“相似性”是衡量知识好坏的根本因素,借鉴免疫理论的相关概念,采用信息熵作为经验事例相似性衡量的指标,建立了基于免疫机制的诊断知识获取模型,利用相关免疫算子,按照预定的优化目标函数,生成最优的诊断知识。
4针对项目“励磁装置在线故障预测和故障诊断系统”的特点,对基于免疫机理的多Agent故障诊断系统在电力系统故障诊断中的应用技术进行了研究,给出了诊断系统的硬件结构和软件结构,详细分析了诊断系统的工作流程和系统的主要模块设计。
5初步讨论了一种基于协同机制的多Agent故障预测系统模型,提出了故障预测策略和推理模型以及设计了系统流程。
As a kind of applied systems of AI in fault diagnosis, the intelligence fault diagnosis system gets very good effect in practice, with the gradual complication of the equipment and function of system, which leads to more complicated and exceptional faults, but, it can no longer meet all demand to diagnose fault of complex systems by existing static means. In the dynamic and changeable environment, it can no longer meet demand to diagnose fault used by the existing normal and fix reasoning model, fully utilizing existing or new diagnosis subsystems and other groups obtained, Multi-Agent fault diagnostic system based on Immune Mechanism (AIFDS) can fast self-adaptation adjust diagnosis courses, diagnosis functions and diagnosis abilities, which is a research trends on fault diagnosis with a dynamic adaptation ability of fault diagnosis.
Taking the national natural science program" Reconfigurable Fault Diagnosis System based on Immune Response Theory "and enterprises program" Fault Diagnostic and Prediction System On-line Research on a Excitation Device" as background, On the key problem of AIFDS, some important topics are lucubrated systematically and deeply in this dissertation including modeling methods for diagnosing system and its course, reasoning model and reconfigurable technology etc., the main contents of the dissertation is as follows:
1. On the basis of analyzing diagnoses basic theories and relevant immunology, the immunology theoretical foundation of AIFDS are explained, the method of using the immune mechanism of living beings is put forward to AIFDS. By combined with Agent technology, AIFDS’s goals based on immune mechanism are pointed out.
2. From the point of view of system engineering, a complex device system can be decomposed into layers according to its structure and function, and builds analytic model for device systems to describe systems’behavior structure and fault state space, On the base of the shine relation between biological immune systems and AIFDS, inspired by the immune response mechanism, system structure and relevant abstract algorithm, the immune frame of AIFDS is constructed, the software implementation methods based on the component and agent technology are analyzed, a diagnosis business course model is provided easy to reconstructed.
3. On the idea of learning from exiting diagnosis instances, it is a key factor of weighing the knowledge quality with similarity of experience examples, a new model of
引文
[1] 吴今培.智能故障诊断与专家系统[M].北京:科学出版社,1999: 5-11.
[2] 杨叔子.基于知识的故障诊断技术[M].北京:清华大学出版社,1993:23-27.
[3] 屈梁生,何正嘉.机械故障诊断学[M].上海:上海技术出版社,1996.
[4] 周东华,孙优贤.控制系统的故障检测与诊断技术[M].北京:清华大学出版社,1994.
[5] P.Vas, Parameter Estimation.Condition Monitoring and Diagnosis of Electrical Machines[M].Oxford:Clarendon Press,1993.
[6] F.M.Frank.Fault Diagnosis in Dynamic System Using Analytical and Knowledge-Based Redundancy-A Survey and Some New Results[J].Automatica,1990,26(3):459-474.
[7] B O. Wahlberg. Robust Frequency Domain Fault Detection/Diagnosis[A].Preprints 11th IFAC World Congress,Tallin,1990.
[8] Tylee J. L.On-Line Failure Detection in Nuclear Power Plant Instrumentation[J].IEEE Transactions On Automatic Control, 1983, AC-28(3):406-415.
[9] Chao K. R., Lang J. H, Umans,S. D. Detection of Broken Rotor Bars in Induction Motors Using State and Parameter Estimation[J].IEEE Transactions On Industry Applications,1992, 28(3):702-709.
[10] 虞和济,韩庆大,李沈等.设备故障诊断工程[M].北京:冶金工业出版社,2001.
[11] Frank P M, Ding X. Survey of Robust Residual Generation and Evaluation Methods in Observert-based Fault Detection Systems[J].Journal of Process Control,1997,7(60):403-424.
[12] 欧海涛.复杂系统中的多智能体协调机制的研究[D].上海交通大学博士学位论文,2000.
[13] 樊友平.可重构故障诊断系统的理论与方法及其应用研究[D].重庆大学博士学位论文,2003.
[14] 樊友平,xxx. 智能故障诊断技术的发展与思考[J].自然辩证法研究,2001,17(2):42-46.
[15] 王长琼,孙国正.故障诊断的智能模型研究述评[J].水利电力机械,1998,20(3):9-11.
[16] Fenton R J, Lopez-Toribio C J, Maguire L P.Fault diagnosis of electronic systems using intelligent techniques: a review[J]. IEEE Trans on systems,Man and cybemetics – Part CApplication and Reviews,2001,31(3):269-281.
[17] 杨良士.动态系统故障诊断的新方法——专家系统[J].信息与控制,1998,17(5):26-31.
[18] Pazzni M J.Failure Driven learing of fault diagnosis heuristics[J].IEEE Trans on systems Man and cybemetics,1998,17(3):380-384.
[19] Youn W C, Hammen J M .Aiding the operator during novel fault diagnosis[J].IEEE Trans on system, Man and cybernetics,1998,18(1):142-147.
[20] 张雪江.汽轮发电机组故障诊断专家系统知识处理技术的研究[J].振动工程学报,1996,9(3):230-236.
[21] 杨萍,吴捷.变权重模糊综合评判模型及其在故障诊断中的应用[J].控制理论与应用,2000,17(5):707-710.
[22] Tsukanmto Y, Terano T.Failure diagnosis by using fuzzy logic[A]. Proceedings IEEE decision and control. New Orlearns, England, 1987:1390-1395.
[23] 张建刚.模糊树模型及其在复杂系统辨识中应用[J].自动化学报,2000,26(3):378-381.
[24] Dasgupta D, Attoh – Okine N.Immutimy-based system:a surbey[A].In proc 1997 IEEE Int cof on System,Man and Cybernetic.Orlando,FL,USA,1997,1:369-374.
[25] Farmer, J D,Packard N H,Perelson A S.The immune system,adaptation, and Machine Learing,Physica 22D,1986:187-204.
[26] Jerne N K.Towards a network theory of the immune system[J].Annual Immunology.1974,125C:373-389.
[27] Tang Z, Yamaguchi T, Tashima K, Ishizuka O, Multiple T K.A Multiple valued immune network and its applications[J].IEICE Trans,Fundamentals,1999,E82-A(6):1102-1108.
[28] Smith D J, Forrest S ,Perelson A S.Immunological memory is associative[A].In:Workshop Notes,Workshop4:ImmunityBasedSystems,Intnl.conf.onMultiagnetSystems.Tokyo,Japan,1996:62-70.
[29] Forrest S, Perelson A S, Genetic algorithms and the immune system. In Hans-Paul Schwefel and R. M¨anner, editors, Parallel Problem Solving from Nature, Lecture Notes in Computer Science, Springer-Verlag, Berlin, Germany, 1991: 320–325.
[30] Chun J S, Kim M K, Jung H K. Shape Optimization of electromagnetic devices usingImmune Algorithm[J]. IEEE Transactions on Magnetics, 1997,33(2):86-99.
[31] Huang S J. An Immune-Based Optimization Method to Capacitor Placement in a Radial Distribution System[J]. IEEE Transactions on Powe Delivery, 2000,15(2):121-130.
[32] de Catro L N , Von Zuben F J. The Clonal Selection Algorithm with Engineering Applications. In Workshop Proceedings of GECC’00, Workshop on Artificial Immune Systems and Their Applications, Las Vegas, USA, 2000: 36-37.
[33] de Castro L N. Learning and Optimization Using the Clonal Selection Principle, IEEE Transaction on Evolutionary Computation, Special Issue on Artificial Immune Systems, 2001
[34] Wierzchoń S T. Function Optimization by the Immune Metaphor. TASK QUARTERLY,2002: 1-16.
[35] Costa A M, Vargas P A, et al. Makespan Minimization on Parallel Processors: An Immune-Based Approach. Evolutionary Computation, 2002:920 –925.
[36] Gaspar A, Collard P. Two Models of Immunation for Time Dependent Optimization. SMC 2000 IEEE International Conference on Systems, Man, and Cybernetics, Nashville, Tennessee, USA, 2000.
[37] Ishida Y. Fully Distributed Diagnosis by PDP Learning Algorithm: Towards Immune Network PDP Model[A]. Proceedings of ICNN90, San Diego,1990.
[38] Kayawa M, Sugita Y, Morooka. Sensor Diagnosis System Combing Immune Network and Learning Vector Quantization. Electrical Engineering in Japan, 1997:44-55.
[39] Dasgupta D, Forrest S. Artificial Immune Systems in Industrial Applications[A]. Proceedings of the Second International Conference on Intelligent Processing and Manufacturing of Materials,1999: 257-267.
[40] 龚非力.医学免疫学[M].北京:科学出版社, 2003.
[41] 莫宏伟.人工免疫系统原理与应用[M].黑龙江:哈尔滨工业大学出版社,2002.
[42] 漆安慎.免疫的非线性模型[M].上海:科技教育出版社,1993.
[43] Mattias Nyberg. Model Based Fault Diagnosis:Mthods,Theory,and Automotive Engine Applications[D].Sweden:Linkoping University,1999.
[44] Yu Yongxin,Fei Qi, Wang Hongwei. A Structural Analysis for Case-Based Models[J]. Journal of Huazhong University of Science and Technology, 1995, 23(8): 108-111.
[45] Sun R, Peterson T. Multi-agent reinforcement learning: weighting and partitioning[J]. Neural Network, 1990,12(6):727-753.
[46] 张跃,李德英.飞船系统故障诊断的知识模型[J].系统工程理论与实践,1996,16(9):51-57.
[47] 张跃,李德英.飞船系统故障诊断的层次分解模型[J].系统工程理论与实践,1996,16(10):38-43.
[48] 马笑潇,黄席樾.免疫 Agent 概念与模型[J].控制与决策,2002. 17(4):509-512.
[49] 陈善勇,何永勇,多Agent故障诊断原型系统研究[J].中国机械工程,2002,13(13):1084-1087.
[50] 刘弘,曾广阔.软件 Agent 的构筑[J].计算机科学,1998,25(2):24-28.
[51] Witting T. ARCHON: An architecture for multi-agent systems[M]. Chichester England: Ellis Horwood, 1992 .
[52] Forrest S ,Perelson A S ,Allen L. Self2nonself discrimination in a computer [A] . Proc IEEE symposium on research in security and privacy. Okaland ,CA :IEEE Press ,1994:202-212.
[53] 刘树林,张嘉钟等.基于免疫系统的旋转机械在线故障诊断[J].大庆石油学院学报,2001,25 (4) :69 - 72.
[54] 谷吉海,姜兴渭,刘树林等.免疫系统的反面选择算法在故障诊断中的应用[J].中国空间科学技术, 2002, (2):25-27.
[55] 卢嫦娟,董红斌.免疫识别器动态覆盖性的选择策略[J].计算机工程与应用,2003,(36):65-67.
[56] Sibylle D. Müller S D et al., Evolution Strategies for the Optimization of Microdevices, Proceedings of the Congress on Evolutionary Computation (CEC 2001), 2001:302-309.
[57] 杨荣华,董红斌,康立山等.人工免疫系统中免疫细胞的分离演化策略[J]. 计算机工程与应用, 2003, (36):58-60.
[58] Sang chan park ,Melinda T. Expanation-based Learning for Intelligent Process Planning[J] . IEEE Trans on SMC, 1993,23(6) :1597-1675.
[59] 石纯一.人工智能原理[M] .清华大学出版社,1993:176-267.
[60] 王磊,潘进,焦李成.免疫算法[J].电子学报,2000,28(7):74-78.
[61] Erich Gamma, Richard Helm, Ralph Johnson, John Vlissides. Design Patterns:Elements of Reusable Object-Oriented Software[M]. Addison Wesley, 1996.
[62] WANG Lei, PAN Jin, JIAO Li-cheng. The Immune Algorithms[J].ACTA ELECTRONICA SINICA, 2000, 28(7):74-78.
[63] 张著洪.人工免疫系统中智能优化及免疫网络算法理论与应用研究[D].重庆大学博士学位论文,2004.
[64] 张雪江,朱向阳,钟秉林等.基于模拟退火算法的知识获取方法的研究[J].控制与决策,1997,12(4):328-331.
[65] 彭志刚,张记会,徐心和. 基于遗传算法的知识获取及其在故障诊断中的应用研究[J].信息与控制,1999,28(5):391-395.
[66] 刘忠源.同步电机可控硅励磁系统[M].北京:水利电力出版社,1992.
[67] 孙雅明,王玉生,马群章等.发电机励磁系统在线功率晶闸管性能监测及其故障诊断专家系统[J] .中国电机工程学报,1991, 11(2): 37-46.
[68] SGE-III型励磁装置说明书[M].重庆西蜀电力控制设备公司,2002.
[69] 姜琳.同步发电机励磁系统状态监测与诊断研究[D].华中科技大学硕士学位论文,2003.
[70] 徐德鸿,马皓.电力电子装置故障自动诊断[M].北京:科学出版社,2001.
[71] 黄华.发电机励磁故障的原因和处理[J].机电工程技术,2002, 31(2):55-56.
[72] 肖志怀,张诚.励磁功率单元模型参考故障诊断方法[J].水电自动化与大坝监测,2002, 26(6): 10-13.
[73] 伍征团,杨菊元,廖海东等.发电机励磁控制系统故障的分析及改进[J].华东电力,2002, 30(2): 37-41.
[74] 孙宝赋,张科,廖志忠等.基于系统分层与故障树的故障诊断方法[J].火力与指挥控制,2002,27(4): 63-65.
[75] 黄文虎,江常伟,姜兴渭等.基于故障树模型的智能诊断的确定性推理[J].机械强度,1995,17(2): 38-42.
[76] 洪冶,蔡维由,乐振春.模糊故障树诊断及应用[J]. 武汉大学学报(工学版),2001,34(1): 93-94.
[77] 黄大荣.复杂系统的故障预测理论及其在励磁系统中的应用研究 [D].重庆大学博士学位论文,2006.