印刷机故障诊断专家系统的研究开发
|
摘要
随着计算机技术的发展及智能技术的应用，故障诊断技术已开始进入了一个新阶段，即智能化诊断阶段。这是一种基于专家知识和人工智能技术的诊断方法，对复杂大系统的诊断极为有效。
现代科技的综合运用使印刷机越来越复杂高效，加上印刷工艺要求高以及纸张油墨的可变因素大，所以印刷机产生故障的原因很多，故障诊断与排除越来越困难。当前印刷行业的维修人员主要靠感觉和个人经验对印刷机故障进行诊断，有时为了排除一个故障要花几天时间，充分说明应用在印刷领域的故障诊断技术已远远不能适应现代化印刷的要求。
利用人工智能技术的故障诊断专家系统是设备故障诊断的发展方向，但由于印刷工艺的复杂性，有许多不精确、不完全了解的因素致使专家系统的建立有较大困难，至今尚未见到印刷机故障诊断专家系统(Printing machine Fault Diagnosis Expert System，PMFDES)的问世。本文从印刷机实际出发，通过借鉴其它行业专家系统的成功经验，综合运用各种先进的故障诊断理论，初步开发出了PMFDES系统。具体所做工作如下：
1．在查阅大量资料及请教印刷机维修专家的基础上归纳总结了印刷机常见故障及故障处理方法等领域知识；
2．利用故障树分析法建立了几种常见印刷机故障的故障树，并在此基础上进行了分析；
3．对模糊诊断理论及四种模糊算子、两种诊断原则进行了讨论，在此基础上研究了如何运用模糊理论实现印刷机故障诊断；
4．介绍了神经网络的基本理论，讨论了前向多层神经网络特征和反向传播(BP)算法实现，研究了如何利用神经网络进行印刷机故障诊断；
5．对模糊理论和神经网络技术的特点及优缺点进行了归纳总结，在此基础上讨论了模糊神经网络理论，研究了取大与积型联想记忆网络学习算法及其如何实现印刷机故障诊断；
6．在上面讨论了基于数值计算的智能诊断方法后，本文从另一个角度，即基于符号推理的智能诊断方法对印刷机故障诊断进行了讨论。介绍了产生式表示法和模糊Petri网表示法，讨论了模糊Petri网如何表示产生式规则，研究了其推理算法及模糊Petri网如何进行印刷机故障诊断。
7．在以上研究分析的基础上，针对传统专家系统知识库维护困难的不足，本文采用Visual C++语言开发出了与数据库技术相结合的PMFDES系统，系统界面友好、操作方便、稳健高效、功能强大，具有良好的可扩充性和开放性。
With the development of computer technology and the application of intelligence technology, fault diagnosis technology has got into a new period, the period of intelligence diagnosis. This is a diagnosis method based on expert knowledge and artificial intelligence, which is especially efficient for the diagnosis of large complex system.
The comprehensive application of modern science and technology makes printing machine becoming more and more efficient and complex. This as well as the high requirement of typography and the large variable factors of paper and printing ink makes the printing machine having many fault causes, accordingly fault diagnosis becomes more and more difficult. At present, the maintenance personnel at the printing industry mainly depend on their perception and personal experiences to carry on printing machine fault diagnosis. Sometimes it will spend their several days to remove a fault. This fully indicates that the fault diagnosis technology used in the printing industry is far from meeting the requirement of modern printing.
The fault diagnosis expert system using the artificial intelligence technology is the progressing tiend of equipment fault diagnosis, however, as typography being so complex, there are still many uncertain and incomprehensive factors that hide the establishing of expert system. So far, there is still no report of Printing Machine Fault Diagnosis Expert System (PMFDES). Based on the actual conditions of printing machine, the author develops the preliminary PMFDES system through synthesizing many advanced fault diagnosis theories by drawing lessons from successful experiences of expert systems in other industries. The main works done in this paper are:
1.Based on consulting a lot of material and printing machine maintenance experts, this author induces and summarizes several common printing faults and their removing methods.
2.The author establishes the Fault Trees of several common printing machine faults by the Fault Tree Analysis (FTA), and carries on analysis accordingly.
3.The author inquires into the Fuzzy Diagnosis theory * four Fuzzy Operator and two Diagnosis Pnnciple, then studies how to use the Fuzzy theory diagnosing printing machine faults.
4.The author introduces the basic theory of Neural Network, inquires into the Forward Multi-layer Neural Network and the realization of Back Propagation (BP) algorithm, then studies how to use the Neural Network Theory diagnosing printing
machine faults.
5.The author induces and summarizes the characteristics and advantages as well as disadvantages af Fuzzy theory and Neural Network technology, then inquires into the Fuzzy Neural Network theory, studies the learning algorithm of Max-Multiply Fuzzy Associative Memories Network and how to use it diagnosing printing machine faults.
6.After discussing the artificial diagnosis methods based on numerical computing, the author then inquires into the artificial diagnosis method from another point of view to diagnosis printing machine fault. The author introduces the Production rule and Fuzzy Petri Net, inquires into how to representing the Production rule with the Fuzzy Petri Net, studies the reasoning algorithm and how to use the Fuzzy Petri Net diagnosing printing machine faults.
7.On the basis of research mentioned above, to counter the repository maintenance disadvantage of traditional expert system, the author uses the Visual C++ programming language to develop the PMFDES system combining with the Database. The system has a friendly interface. It is easy to handle and robust as well as efficient, at the same time, it is powerful and expansible as well as open.
现代科技的综合运用使印刷机越来越复杂高效，加上印刷工艺要求高以及纸张油墨的可变因素大，所以印刷机产生故障的原因很多，故障诊断与排除越来越困难。当前印刷行业的维修人员主要靠感觉和个人经验对印刷机故障进行诊断，有时为了排除一个故障要花几天时间，充分说明应用在印刷领域的故障诊断技术已远远不能适应现代化印刷的要求。
利用人工智能技术的故障诊断专家系统是设备故障诊断的发展方向，但由于印刷工艺的复杂性，有许多不精确、不完全了解的因素致使专家系统的建立有较大困难，至今尚未见到印刷机故障诊断专家系统(Printing machine Fault Diagnosis Expert System，PMFDES)的问世。本文从印刷机实际出发，通过借鉴其它行业专家系统的成功经验，综合运用各种先进的故障诊断理论，初步开发出了PMFDES系统。具体所做工作如下：
1．在查阅大量资料及请教印刷机维修专家的基础上归纳总结了印刷机常见故障及故障处理方法等领域知识；
2．利用故障树分析法建立了几种常见印刷机故障的故障树，并在此基础上进行了分析；
3．对模糊诊断理论及四种模糊算子、两种诊断原则进行了讨论，在此基础上研究了如何运用模糊理论实现印刷机故障诊断；
4．介绍了神经网络的基本理论，讨论了前向多层神经网络特征和反向传播(BP)算法实现，研究了如何利用神经网络进行印刷机故障诊断；
5．对模糊理论和神经网络技术的特点及优缺点进行了归纳总结，在此基础上讨论了模糊神经网络理论，研究了取大与积型联想记忆网络学习算法及其如何实现印刷机故障诊断；
6．在上面讨论了基于数值计算的智能诊断方法后，本文从另一个角度，即基于符号推理的智能诊断方法对印刷机故障诊断进行了讨论。介绍了产生式表示法和模糊Petri网表示法，讨论了模糊Petri网如何表示产生式规则，研究了其推理算法及模糊Petri网如何进行印刷机故障诊断。
7．在以上研究分析的基础上，针对传统专家系统知识库维护困难的不足，本文采用Visual C++语言开发出了与数据库技术相结合的PMFDES系统，系统界面友好、操作方便、稳健高效、功能强大，具有良好的可扩充性和开放性。
With the development of computer technology and the application of intelligence technology, fault diagnosis technology has got into a new period, the period of intelligence diagnosis. This is a diagnosis method based on expert knowledge and artificial intelligence, which is especially efficient for the diagnosis of large complex system.
The comprehensive application of modern science and technology makes printing machine becoming more and more efficient and complex. This as well as the high requirement of typography and the large variable factors of paper and printing ink makes the printing machine having many fault causes, accordingly fault diagnosis becomes more and more difficult. At present, the maintenance personnel at the printing industry mainly depend on their perception and personal experiences to carry on printing machine fault diagnosis. Sometimes it will spend their several days to remove a fault. This fully indicates that the fault diagnosis technology used in the printing industry is far from meeting the requirement of modern printing.
The fault diagnosis expert system using the artificial intelligence technology is the progressing tiend of equipment fault diagnosis, however, as typography being so complex, there are still many uncertain and incomprehensive factors that hide the establishing of expert system. So far, there is still no report of Printing Machine Fault Diagnosis Expert System (PMFDES). Based on the actual conditions of printing machine, the author develops the preliminary PMFDES system through synthesizing many advanced fault diagnosis theories by drawing lessons from successful experiences of expert systems in other industries. The main works done in this paper are:
1.Based on consulting a lot of material and printing machine maintenance experts, this author induces and summarizes several common printing faults and their removing methods.
2.The author establishes the Fault Trees of several common printing machine faults by the Fault Tree Analysis (FTA), and carries on analysis accordingly.
3.The author inquires into the Fuzzy Diagnosis theory * four Fuzzy Operator and two Diagnosis Pnnciple, then studies how to use the Fuzzy theory diagnosing printing machine faults.
4.The author introduces the basic theory of Neural Network, inquires into the Forward Multi-layer Neural Network and the realization of Back Propagation (BP) algorithm, then studies how to use the Neural Network Theory diagnosing printing
machine faults.
5.The author induces and summarizes the characteristics and advantages as well as disadvantages af Fuzzy theory and Neural Network technology, then inquires into the Fuzzy Neural Network theory, studies the learning algorithm of Max-Multiply Fuzzy Associative Memories Network and how to use it diagnosing printing machine faults.
6.After discussing the artificial diagnosis methods based on numerical computing, the author then inquires into the artificial diagnosis method from another point of view to diagnosis printing machine fault. The author introduces the Production rule and Fuzzy Petri Net, inquires into how to representing the Production rule with the Fuzzy Petri Net, studies the reasoning algorithm and how to use the Fuzzy Petri Net diagnosing printing machine faults.
7.On the basis of research mentioned above, to counter the repository maintenance disadvantage of traditional expert system, the author uses the Visual C++ programming language to develop the PMFDES system combining with the Database. The system has a friendly interface. It is easy to handle and robust as well as efficient, at the same time, it is powerful and expansible as well as open.
引文
[1] 成成、黄道等 基于小波网络非稳态故障诊断方法．华东理工大学学报．2000．10
[2] 孙佰清，潘启树等 医疗诊断系统专家知识的表达与获取方法．哈尔滨工业大学学报．2001．2
[3] Feret MP, Glasgow JI .Combining case-based and model-based reasoning for the diagnosis of complex devices. Journal of Applied Intelligence, 1997,7(1)
[4] 刘增良，刘有才．模糊逻辑与神经网络—理论研究与探索．北京航空航天大学出版社，1996
[5] Wedsker LR. Hybrid neural network and expert systems. Norwell: Academic Publishers 1994.
[6] Golding AR. Improving accuracy by combining rule-based and case-based reasoning.Artificial Intelligence,1996.87(1)
[7] Tiehua Cao,Arthur C,Sanbderson. Task Sequence Planning Using Fuzzy Petri Nets. IEEE Transaction on System. Man and Cybernetics, 1995,25(2)
[8] 王永庆．人工智能原理与方法．西安交通大学出版社．1998．
[9] 邢春晓，潘泉，张洪才．基于Fuzzy Petri网专家系统的研究 西北工业大学学报1998．3
[10] 吴今培，肖健华，智能故障诊断与专家系统．科学出版社1997
[11] 李博轩．Visual C++6.0数据库开发指南．清华大学出版社2000
[12] 博彦科技公司．Access开发教程．清华大学出版社．2001
[13] 吴泉源，刘江宁．人工智能与专家系统．国防科技大学出版社．1995
[14] 吴今培．模糊诊断理论及其应用．科学出版社．1995
[15] 杨杰，黄欣，郭英凯．用于故障诊断的集成型智能系统．上海交通大学学报1998．6
[16] 粱久祯，何新贵．模糊极大极小算子神经元网络的图灵等价性．北京航空航天大学出版社2001．8
[17] 庆胜，邬学礼．FMS故障诊断的模糊行为Petri网研究．电子技术应用．1997．5
[18] 廉小亲，王信义，朱小燕．故障诊断系统中模糊推理算法的研究．北京理工大学学报．1996．1
[19] 蔡之华．模糊Petri网及知识表示．计算机应用及软件1994．3
[20] 张海燕，武吉梅，印刷设备．西安理工大学(内部教材)．1998
[21] 冯焕玉，张子林．胶印疑难故障判断与排除．印刷工业出版社．1994
[22] zhang Bo, et al.Research and Implementation of SPMD-MOdeI-Task Development. Computer Research and Development, 1997,34(8)
[23] Lue N K, Dillon T. An Approach Towards the Verification of Expert Systems Using Numerical Petri Nets.lnt. J.of Intelligent System, 1991,35(6)
[24] 邝朴生等．现代机器故障诊断学．农业出版社．1991
[25] 张安华，同淑荣等．机电设备状态检测与故障诊断．西北工业大学出版社．1995
[26] 虞和济，陈长征，张省，周建男．基于神经网络的智能诊断．冶金工业出版社．2000
[27] 石博强等机械故障诊断的分形方法—理论与实践．冶金工业出版社．2001
[28] Hanna M M,et al. Fuzzy Petri Nets with Neural Networks to Model Products Quality from a CNC-millings Machining Center. IEEE Trans. On SMC-Part A, 1996,26(5)
[29] Kosko B. Fuzzy associative memories. In: Kondel(ed). Fuzzy Expert System Reading. MA:Addison Weley. 1987
[30] Blanco A,Delgado M,Requena Ⅰ. Identification of fuzzy relational equations by fuzzy neural networks. Fussy Sets and Systems, 1995,71
[31] Blanco A,Delgado M,Requena Ⅰ. Improved fuzzy neural networks for solving relational equations. Fussy Sets and Systems, 1995,72
[32] Simpson P K. Fuzzy min-max neural networks-part Ⅰ:classification. IEEE Trans. on Networks, 1992,3
[33] Simpson P K. Fuzzy min-max neural networks-part Ⅱ:clustering. IEEE Trans. Fuzzy Systems, 1993,1
[34] Chung F L,Lee T. Fuzzy competitive learning. Neural Networks, 1994,7
[35] Chung F L, Lee T. On FAM with multiple-rule storage capacity. IEEE Trans. Fuzzy systems, 1996,4
[36] Ishibuchi H, Tanake H.A learning algorithm of fuzzy neural networks with triangular fuzzy weights. Fuzzy Sets and Systems, 1995,71
[37] lshibuchi H, Tanake H. Fuzzy regression analysis using neural networks.Fuzzy Sets and Systems, 1992,50
[38] Ishibuchi H, Okada H,Tanaks H. Fuzzy neural networks with fuzzy weights and fuzzy biases. Proc ICNN'93(San Francisco).Vol Ⅲ, 1993
[39] 张冰，邱志强．模糊神经网络在雷达网数据融合中的研究．电子科技大学学报．2001．1
[40] 冯兰君，李健等．燃气轮机故障诊断专家系统的设计与实现．电子科技大学学报．2000．3
[41] 许颖原，俞金寿．DCS故障诊断专家系统的开发．华东理工大学学报．2001．10
[42] 朱新宇，沈颂华．飞机电源系统故障诊断专家系统．北京航空航天大学．2001．12
[43] 朱平，黄文虎等．多模型融合故障诊断技术的研究．哈尔滨工业大学学报．2000．8
[44] Kajior Watanage,et al. Diagnosis of Multiple Simultaneous Fault via Hierarchical Artificial Neural Networks. Journal of AlChE 1994;40(5)
[45] Kozo Osakada and goubin Yang. Application of Neural Networks to an Expert system for Cold Forging. Int J. Mach. Tools Manufacture. 1991.31(4)
[2] 孙佰清，潘启树等 医疗诊断系统专家知识的表达与获取方法．哈尔滨工业大学学报．2001．2
[3] Feret MP, Glasgow JI .Combining case-based and model-based reasoning for the diagnosis of complex devices. Journal of Applied Intelligence, 1997,7(1)
[4] 刘增良，刘有才．模糊逻辑与神经网络—理论研究与探索．北京航空航天大学出版社，1996
[5] Wedsker LR. Hybrid neural network and expert systems. Norwell: Academic Publishers 1994.
[6] Golding AR. Improving accuracy by combining rule-based and case-based reasoning.Artificial Intelligence,1996.87(1)
[7] Tiehua Cao,Arthur C,Sanbderson. Task Sequence Planning Using Fuzzy Petri Nets. IEEE Transaction on System. Man and Cybernetics, 1995,25(2)
[8] 王永庆．人工智能原理与方法．西安交通大学出版社．1998．
[9] 邢春晓，潘泉，张洪才．基于Fuzzy Petri网专家系统的研究 西北工业大学学报1998．3
[10] 吴今培，肖健华，智能故障诊断与专家系统．科学出版社1997
[11] 李博轩．Visual C++6.0数据库开发指南．清华大学出版社2000
[12] 博彦科技公司．Access开发教程．清华大学出版社．2001
[13] 吴泉源，刘江宁．人工智能与专家系统．国防科技大学出版社．1995
[14] 吴今培．模糊诊断理论及其应用．科学出版社．1995
[15] 杨杰，黄欣，郭英凯．用于故障诊断的集成型智能系统．上海交通大学学报1998．6
[16] 粱久祯，何新贵．模糊极大极小算子神经元网络的图灵等价性．北京航空航天大学出版社2001．8
[17] 庆胜，邬学礼．FMS故障诊断的模糊行为Petri网研究．电子技术应用．1997．5
[18] 廉小亲，王信义，朱小燕．故障诊断系统中模糊推理算法的研究．北京理工大学学报．1996．1
[19] 蔡之华．模糊Petri网及知识表示．计算机应用及软件1994．3
[20] 张海燕，武吉梅，印刷设备．西安理工大学(内部教材)．1998
[21] 冯焕玉，张子林．胶印疑难故障判断与排除．印刷工业出版社．1994
[22] zhang Bo, et al.Research and Implementation of SPMD-MOdeI-Task Development. Computer Research and Development, 1997,34(8)
[23] Lue N K, Dillon T. An Approach Towards the Verification of Expert Systems Using Numerical Petri Nets.lnt. J.of Intelligent System, 1991,35(6)
[24] 邝朴生等．现代机器故障诊断学．农业出版社．1991
[25] 张安华，同淑荣等．机电设备状态检测与故障诊断．西北工业大学出版社．1995
[26] 虞和济，陈长征，张省，周建男．基于神经网络的智能诊断．冶金工业出版社．2000
[27] 石博强等机械故障诊断的分形方法—理论与实践．冶金工业出版社．2001
[28] Hanna M M,et al. Fuzzy Petri Nets with Neural Networks to Model Products Quality from a CNC-millings Machining Center. IEEE Trans. On SMC-Part A, 1996,26(5)
[29] Kosko B. Fuzzy associative memories. In: Kondel(ed). Fuzzy Expert System Reading. MA:Addison Weley. 1987
[30] Blanco A,Delgado M,Requena Ⅰ. Identification of fuzzy relational equations by fuzzy neural networks. Fussy Sets and Systems, 1995,71
[31] Blanco A,Delgado M,Requena Ⅰ. Improved fuzzy neural networks for solving relational equations. Fussy Sets and Systems, 1995,72
[32] Simpson P K. Fuzzy min-max neural networks-part Ⅰ:classification. IEEE Trans. on Networks, 1992,3
[33] Simpson P K. Fuzzy min-max neural networks-part Ⅱ:clustering. IEEE Trans. Fuzzy Systems, 1993,1
[34] Chung F L,Lee T. Fuzzy competitive learning. Neural Networks, 1994,7
[35] Chung F L, Lee T. On FAM with multiple-rule storage capacity. IEEE Trans. Fuzzy systems, 1996,4
[36] Ishibuchi H, Tanake H.A learning algorithm of fuzzy neural networks with triangular fuzzy weights. Fuzzy Sets and Systems, 1995,71
[37] lshibuchi H, Tanake H. Fuzzy regression analysis using neural networks.Fuzzy Sets and Systems, 1992,50
[38] Ishibuchi H, Okada H,Tanaks H. Fuzzy neural networks with fuzzy weights and fuzzy biases. Proc ICNN'93(San Francisco).Vol Ⅲ, 1993
[39] 张冰，邱志强．模糊神经网络在雷达网数据融合中的研究．电子科技大学学报．2001．1
[40] 冯兰君，李健等．燃气轮机故障诊断专家系统的设计与实现．电子科技大学学报．2000．3
[41] 许颖原，俞金寿．DCS故障诊断专家系统的开发．华东理工大学学报．2001．10
[42] 朱新宇，沈颂华．飞机电源系统故障诊断专家系统．北京航空航天大学．2001．12
[43] 朱平，黄文虎等．多模型融合故障诊断技术的研究．哈尔滨工业大学学报．2000．8
[44] Kajior Watanage,et al. Diagnosis of Multiple Simultaneous Fault via Hierarchical Artificial Neural Networks. Journal of AlChE 1994;40(5)
[45] Kozo Osakada and goubin Yang. Application of Neural Networks to an Expert system for Cold Forging. Int J. Mach. Tools Manufacture. 1991.31(4)