基于遗传—神经网络的电液伺服阀故障模式识别研究
|
摘要
电液伺服系统是一结构复杂而多耦合的机、电、液综合系统，是控制领域中的一个重要组成部分，具有功率大，响应快，精度高的特点，在工业生产领域得到了广泛的应用。而电液伺服阀是电液伺服系统的关键性元件，是系统中故障率最高的部位，其使用情况与性能的好坏直接影响到系统的工作性能。因此，研究一种切实可行的故障智能诊断方法，对电液伺服阀可能发生的故障进行预报、控制，对已发生的故障进行正确而迅速的智能诊断，提供可行的排除方法十分重要。
本文以电液伺服阀静、动态特性试验台为依托，在广泛研究国内外相关文献资料及工作的基础上，结合现代故障诊断理论、模式识别理论、人工智能理论和计算机科学，对动圈式电液伺服阀的故障智能诊断技术进行了较为深入的研究。
1．对电液伺服阀进行了抗污染研究。针对其先导级固定节流小孔易于堵塞的缺点，本文利用仿真软件AMESim建立了该阀的结构化的仿真模型，在改变节流孔大小、给节流孔串联和并联一个节流孔的情况下进行仿真，通过对特性曲线分析得出在阀体上开一个节流小孔，使之与固定节流小孔串联，并把两者的直径加大可以达到与原结构相同的性能，这样便可以适当提高该阀的抗污染能力。
2．对动圈式电液伺服阀的主要故障的机理进行了研究。故障机理被认为是故障诊断领域的“深知识”，研究故障机理是进行故障诊断的重要环节，本论文研究了动圈阀的主要故障模式—磨损、气蚀、液压卡紧、温升等的机理，探讨了减少动圈阀的各种故障模式的措施，分析了热楔效应对动圈阀性能的影响。
3．结合本实验室的实验条件对动圈式电液伺服阀进行了故障模拟实验：利用数字化电液伺服阀试验台，在系统压力为2MPa、2.5MPa、3MPa、3.5MPa、4MPa、4.5Mpa、5MPa时，分别测得电液伺服阀七种状态(正常、上下调零弹簧疲劳断裂、导阀一端固定节流孔堵塞、控制腔一端密封破损1／12，1／2与完全破损)下静态(压力特性曲线)特性曲线作为伺服阀的状态特征参数，并对得到的压力特性曲线进行了离散化处理。
4．建立了动圈式电液伺服阀故障智能诊断的网络模型。利用遗传算法和神经网络的特点和优点，将两者结合在一起而形成了GA-BP算法，并建立了用于电液伺服阀故障智能诊断的遗传—神经网络模型。该模型继承了传统遗传算法的优点，兼具神经网络强大的函数逼近功能，同时又克服了传统神经网络优化方法易陷入局部最优解的缺陷。实例的训练和模式识别结果表明：实数编码的遗传算法优化的神经网络模型模式识别精度较高，适合于电液伺服阀的故障模式识别。
5．在以上理论研究的基础上，利用面向对象、可视化的编程语言Visual Basic开发了基于遗传—神经网络的电液伺服阀智能故障诊断的系统。它包括知识库的建立和维护模块、推理和仿真模块等。
The electro-hydraulic servo system is a complex structure and much coupled system ,which synthesized the merit of the mechanical technology , electric technology and hydraulic technology and has been an important component in the control field. The system possesses some important features, such as great power, quick response, and high precision and so on, it has widely been used in the field of the industrial product And the electro-hydraulic servo valve is a key component for the system, which is the position with a highest fault rate in the system, its using condition and performance is bad or not directly influences the system performance. Consequently, the important problem now is how to research a really feasible method for intelligent fault diagnosis and use this method to predict , control the potential fault of the electro-hydraulic servo valve or rightly and quickly diagnose the happened fault and support the feasible method for elimination the happened fault.
This paper studies in depth the key technology of the electro-hydraulic servo valve, on the base of collecting broadly the relate literature and works, combined with the theories of modern fault diagnosis , recognizing the fault model , artificial intelligence and computer science, which relies on a new kind electric-hydraulic static-dynamistic character test station.
1. Antipollution research of the electro-hydraulic servo valve is proposed. The antipollution capability of electro-hydraulic servo valve is not well for the sake of its pilot valve's fixed throttle. In general, the fixed throttle's diameter is too small and is easily to be jammed. In the paper, the author establishes the electro-hydraulic servo valve's simulation model by using the software AMESim. We simulate by changing the two fixed throttle diameter and adding two throttle separately in series with the fixed two throttle and adding two throttle separately in parallel with the fixed two throttle. By analyzing the simulation curves , we can draw a conclusion as follows: the capability antipollution of the valve can be improved by adding a throttle in front of the fixed throttle and increasing their diameter.
2. The main fault mechanism of the electro-hydraulic servo valve is studied. The fault mechanism has been considered as the "deep knowledge" in the field of fault diagnosis. The study of the fault mechanism has been a key step for the fault diagnosis. The fault mechanism of the electro-hydraulic servo valve's main fault models, such as wearing , cavitation erosion, hydraulic lock, temperature rise and so on in this paper are studied , the measures to reduce all kinds of fault models of the electro-hydraulic servo valve are discussed and the thermal stress on how to affect the performances of the electro-hydraulic servo valve is analyzed.
3. The fault simulation testing for the electro-hydraulic servo valve combined with the
condition of our lab is done. By using digital electric-hydraulic servo valve's test station, on the pressures of system such as 2MPa, 2.5MPa, SMPa, 3 .SMPa, 4MPa, 4.5MPa, SMPa, we can measure static character carves (pressure gain) of servo valve as the static characteristic parameters, in seven conditions of electric-hydraulic servo valve (normal condition , up or down springs fatigue fracture, one side of fixed throttle hole of the pilot valve being blocked , one side of seal of bearer of control chamber being damaged 1/12, 1/2 and wholly. In order to meet the neural network model's input requirements, at first the author processes the pressure gain curves, and then makes the discrete data between -1 and 1.
4. The intelligent fault diagnosis' network framework for the electro-hydraulic servo valve is established. In this paper, the author combines an artificial neural network with the genetic algorithms and forms a new algorithm, GA-BP algorithm, and establishes a GA-ANN model for the intelligent fault diagnosis of the electro-hydraulic servo valve, which is used the features and virtues of ANN and GA. The model takes the advantages of traditi
本文以电液伺服阀静、动态特性试验台为依托，在广泛研究国内外相关文献资料及工作的基础上，结合现代故障诊断理论、模式识别理论、人工智能理论和计算机科学，对动圈式电液伺服阀的故障智能诊断技术进行了较为深入的研究。
1．对电液伺服阀进行了抗污染研究。针对其先导级固定节流小孔易于堵塞的缺点，本文利用仿真软件AMESim建立了该阀的结构化的仿真模型，在改变节流孔大小、给节流孔串联和并联一个节流孔的情况下进行仿真，通过对特性曲线分析得出在阀体上开一个节流小孔，使之与固定节流小孔串联，并把两者的直径加大可以达到与原结构相同的性能，这样便可以适当提高该阀的抗污染能力。
2．对动圈式电液伺服阀的主要故障的机理进行了研究。故障机理被认为是故障诊断领域的“深知识”，研究故障机理是进行故障诊断的重要环节，本论文研究了动圈阀的主要故障模式—磨损、气蚀、液压卡紧、温升等的机理，探讨了减少动圈阀的各种故障模式的措施，分析了热楔效应对动圈阀性能的影响。
3．结合本实验室的实验条件对动圈式电液伺服阀进行了故障模拟实验：利用数字化电液伺服阀试验台，在系统压力为2MPa、2.5MPa、3MPa、3.5MPa、4MPa、4.5Mpa、5MPa时，分别测得电液伺服阀七种状态(正常、上下调零弹簧疲劳断裂、导阀一端固定节流孔堵塞、控制腔一端密封破损1／12，1／2与完全破损)下静态(压力特性曲线)特性曲线作为伺服阀的状态特征参数，并对得到的压力特性曲线进行了离散化处理。
4．建立了动圈式电液伺服阀故障智能诊断的网络模型。利用遗传算法和神经网络的特点和优点，将两者结合在一起而形成了GA-BP算法，并建立了用于电液伺服阀故障智能诊断的遗传—神经网络模型。该模型继承了传统遗传算法的优点，兼具神经网络强大的函数逼近功能，同时又克服了传统神经网络优化方法易陷入局部最优解的缺陷。实例的训练和模式识别结果表明：实数编码的遗传算法优化的神经网络模型模式识别精度较高，适合于电液伺服阀的故障模式识别。
5．在以上理论研究的基础上，利用面向对象、可视化的编程语言Visual Basic开发了基于遗传—神经网络的电液伺服阀智能故障诊断的系统。它包括知识库的建立和维护模块、推理和仿真模块等。
The electro-hydraulic servo system is a complex structure and much coupled system ,which synthesized the merit of the mechanical technology , electric technology and hydraulic technology and has been an important component in the control field. The system possesses some important features, such as great power, quick response, and high precision and so on, it has widely been used in the field of the industrial product And the electro-hydraulic servo valve is a key component for the system, which is the position with a highest fault rate in the system, its using condition and performance is bad or not directly influences the system performance. Consequently, the important problem now is how to research a really feasible method for intelligent fault diagnosis and use this method to predict , control the potential fault of the electro-hydraulic servo valve or rightly and quickly diagnose the happened fault and support the feasible method for elimination the happened fault.
This paper studies in depth the key technology of the electro-hydraulic servo valve, on the base of collecting broadly the relate literature and works, combined with the theories of modern fault diagnosis , recognizing the fault model , artificial intelligence and computer science, which relies on a new kind electric-hydraulic static-dynamistic character test station.
1. Antipollution research of the electro-hydraulic servo valve is proposed. The antipollution capability of electro-hydraulic servo valve is not well for the sake of its pilot valve's fixed throttle. In general, the fixed throttle's diameter is too small and is easily to be jammed. In the paper, the author establishes the electro-hydraulic servo valve's simulation model by using the software AMESim. We simulate by changing the two fixed throttle diameter and adding two throttle separately in series with the fixed two throttle and adding two throttle separately in parallel with the fixed two throttle. By analyzing the simulation curves , we can draw a conclusion as follows: the capability antipollution of the valve can be improved by adding a throttle in front of the fixed throttle and increasing their diameter.
2. The main fault mechanism of the electro-hydraulic servo valve is studied. The fault mechanism has been considered as the "deep knowledge" in the field of fault diagnosis. The study of the fault mechanism has been a key step for the fault diagnosis. The fault mechanism of the electro-hydraulic servo valve's main fault models, such as wearing , cavitation erosion, hydraulic lock, temperature rise and so on in this paper are studied , the measures to reduce all kinds of fault models of the electro-hydraulic servo valve are discussed and the thermal stress on how to affect the performances of the electro-hydraulic servo valve is analyzed.
3. The fault simulation testing for the electro-hydraulic servo valve combined with the
condition of our lab is done. By using digital electric-hydraulic servo valve's test station, on the pressures of system such as 2MPa, 2.5MPa, SMPa, 3 .SMPa, 4MPa, 4.5MPa, SMPa, we can measure static character carves (pressure gain) of servo valve as the static characteristic parameters, in seven conditions of electric-hydraulic servo valve (normal condition , up or down springs fatigue fracture, one side of fixed throttle hole of the pilot valve being blocked , one side of seal of bearer of control chamber being damaged 1/12, 1/2 and wholly. In order to meet the neural network model's input requirements, at first the author processes the pressure gain curves, and then makes the discrete data between -1 and 1.
4. The intelligent fault diagnosis' network framework for the electro-hydraulic servo valve is established. In this paper, the author combines an artificial neural network with the genetic algorithms and forms a new algorithm, GA-BP algorithm, and establishes a GA-ANN model for the intelligent fault diagnosis of the electro-hydraulic servo valve, which is used the features and virtues of ANN and GA. The model takes the advantages of traditi
引文
[1] 陈新元．基于人工神经网络的电液伺服阀故障模式识别研究．武汉科技大学，1998，12
[2] 陈文伟．智能决策技术[M]．北京：电子工业出版社．1998
[3] Holland JH. Genetic Algorithms[J]. Scientific American, 1992(4):44～50
[4] 申敏．遗传算法与神经网络在模式分类中的应用．重庆邮电学院学报，1998，6(2)：14～16
[5] R. K. Below. Evolving Networks: using the genetic algorithm with connectionist learning[M]. Artificial Life Ⅱ, Addison Wesley Pub, 1992
[6] 王华伟，周经伦，沙基昌．基于遗传算法的一种新的神经网络结构模式研究．小型微型计算机系统，2002，4(4)：467～469
[7] 刘长年．液压伺服系统的分析与设计．北京：科学出版社．1985．4
[8] 焦素娟．基于模糊推理的电液动换向阀的故障诊断研究．1999，11
[9] 黄志坚．液压故障的特点与故障诊断的策略．液压与气动．2001(2)：29～31
[10] 神原正德．液压设备诊断技术现场适用．油压技术．1987
[11] 师占群．基于小波故障提取的电液伺服阀故障诊断．机械科学与技术，2000，19(增)：111～116
[12] 骆志明，冯庚斌．遗传BP网络在滚动轴承故障诊断中的应用．铁道学报，1999，10(5)：29～32
[13] 湛从昌．液压系统故障的模糊诊断方法．液压与气动，1994(6)：3-7
[14] 王垌．人工智能中的模糊启发式搜索技术．北京：机械工业出版社，1993，85～90
[15] 张乃尧，阎平凡．神经网络与模糊控制．北京：清华大学出版社．1997
[16] 王道平，张义忠．故障智能诊断系统的理论与方法．北京：冶金工业出版社，2001
[17] 陈章位．电液控制元件故障识别的新方法．液压与气动．1998(5)：71～72
[18] 曾昭军，何全成，史维祥．故障诊断神经网络的发展与前景．机械工程学报，1992，2(1)：1～6
[19] 王少萍，王占林．液压泵故障诊断的神经网络方法．北京航空航天大学学报．1997，23(6)：714～718
[20] 祝海林，胡爱萍．液压系统故障诊断的灰色理论方法探讨．石油化工高等学校学报，1998，9(3)：55～59
[21] 陈进才，郑守．人工智能发展的新趋势[J]．西安公路交通大学学报，1998，18(3)：94～97．
[22] Frank P metal. New Development Using AI in Fault Diagnosis. Engng. Applic. Artif. Intell, 1997, 10(1): 3～14
[23] Lu Yong xiang, Chen Zhang Wei. A Knowledge-Based electro hydraulic servo-system Fault diagnostic mode. Proc. Of the 4th Scandinavian Int. Conf. On fluid power Temper, 1995:686～694
[24] Branco, P. J Costa, Dente, J. A. On using fuzzy logic to integrate learning mechanisms in an electrohydraulic system-part Ⅰ.... IEEE Transactions on Systems, Man & Cybernetics: Part C-Applications & Reviews, Aug 2000, vol. 30 Issue 3, p317, 12p, 15 diagrams, 14 graphs, 2bw
[25] Goldberg DE. Genetic Algorithms in Search, Optimization and Machine learning[M]. Addison Wesley: Reading MA, 1989
[26] Window Betal. Neural Networks Application in industry, business and science. Communication of the Acm, 1994, 37:93～105
[27] Special Issue on Application of neural networks. Proc IEEE, 1996(84)
[28] Bagley JD. The behavior of adaptive systems which employ genetic and correlation algorithms. Dissertation abstracts international, 1967(28)
[29] Constantin Zopounidis. Special issue on artificial intelligence and decision support with multiple criteria[J]. Computers & Operations Research 2000(27):597～599
[30] R. J Patton, C. J Lopez-Toribio, F. J Uppal. Artificial intelligence approaches to fault diagnosis(C). IEE Colloquium on Condition Monitoring: Machinery, External Structures and Health (Ref. No. 1999/034). 1999:5/1～5/18
[31] 林建亚，何存兴．液压元件．北京：机械工业出版社，1988：8～30
[32] 孙文质．液压控制系统．国防工业出版社，1985：21～33
[33] 杨自厚．自动控制原理．北京：冶金工业出版社，1996：58～256
[34] 孙瑞利，周建桃．电液伺服阀的原理及其故障的排除．南京海运学院，1999(6)：49～51
[35] 王春行．液压伺服控制系统．北京：机械工业出版社，1981：96～180
[36] 虞和济，韩庆大，李沈．设备故障诊断工程．北京：冶金工业出版社，2001，6
[37] 中国机械工程学会材料学会主编．机械产品失效分析．北京：机械工业出版社，1986
[38] 湛从昌．液压可靠性与故障诊断．冶金工业出版社，1995，1
[39] 夏志新．液压系统污染控制．北京：机械工业出版社，1992
[40] 许耀铭．油膜理论与液压泵和马达的摩擦副设计．北京：机械工业出版社，1987
[41] 蔡倩，湛从昌．液压系统计算机辅助监测与故障诊断．武汉冶金科技大学学报．1997，20(3)：348～354
[42] 宋志军．基于人工神经网络的数据挖掘工具及应用．石油大学，2001，5
[43] 陈建安，郭大伟，徐乃平，孙云芝．遗传算法理论研究综述．西安电子科技大学学报，1998，6(3)：363～368
[44] 刘洁，魏连雨，杨春风．基于遗传．神经网络的交通量预测．长安大学学报，2003，3(1)：68～70
[45] 钟春生．人工神经网络在板形检测中的应用研究．西安建筑科技大学，2002，4
[46] 楼顺天，施阳．基于MATLAB的系统分析与设计．西安电子科技大学出版社，2000，11
[47] 邹岳华．人工神经网络BP算法及应用研究．上海理工大学，2000，1
[48] 赵军．基于遗传算法和神经网络模型的智能化拉深参数实时识别．燕山大学，2002，3
[49] 王宝树．遗传算法及神经网络在数据融合中的应．西安电子科技大学，2001，1
[50] 梁化楼，戴贵亮．人工神经网络与遗传算法的结合：进展及展望[J]．电子学报．1995，23(10)：194～200
[51] 倪远平．人工智能用于故障诊断的方法研究[J]．电器传动自动化．1998，20(4)：29～32
[52] 樊友平，黄席樾．智能诊断技术的发展和思考[J]．自然辨证法研究，2001，17(2)：42～46
[53] Irie B, Miyake S. Capabilities of three-layered perceptrons. Proceedings of the IEEE Inter Conf on Neural Network. 1998(1):641～648
[54] Health-NielsemR. The theory of back propagation neural network. In review, 1998
[55] 张立明．人工神经网络的模型及其应用．南京：复旦大学出版社．1994
[56] 郑志军，郑守淇．用基于实数编码的自适应遗传算法进化神经网络．计算机工程与应用．2000，9：36～37
[57] 张明辉，王尚锦．自适应搜索的改进遗传算法及其应用．西安交通大学学报，2002，3(3)：226～230
[58] DAI Xiao hui LI Min qiang KOU Ji song. Multi-objective optimization using genetic algorithm with local search. TransactionsofTianjinUniversity, 1998, 11(4):139～143
[59] Ralf Ostermark. . A hybrid genetic fuzzy neural network algorithm designed for classification problems involving several groups[J]. Fuzzy Sets and Systems, 2000, 114:311～324
[60] Kuo R. J, Xue K. C. Fuzzy neural networks with application to sales forecasting[J]. Fuzzy Sets and systems. 1999, 108:123～143
[61] 王正志，薄涛著．进化计算．长沙：国防科技大学出版社，2000
[62] Bob Reselman, Richard Peasley．实用Visual Basic 6教程．北京：清华大学出版社，2001
[63] Rob Krumm. ADO Programming For Dummies. BEIJING: Publishing House of Electronics Industry, 2001, 6
[2] 陈文伟．智能决策技术[M]．北京：电子工业出版社．1998
[3] Holland JH. Genetic Algorithms[J]. Scientific American, 1992(4):44～50
[4] 申敏．遗传算法与神经网络在模式分类中的应用．重庆邮电学院学报，1998，6(2)：14～16
[5] R. K. Below. Evolving Networks: using the genetic algorithm with connectionist learning[M]. Artificial Life Ⅱ, Addison Wesley Pub, 1992
[6] 王华伟，周经伦，沙基昌．基于遗传算法的一种新的神经网络结构模式研究．小型微型计算机系统，2002，4(4)：467～469
[7] 刘长年．液压伺服系统的分析与设计．北京：科学出版社．1985．4
[8] 焦素娟．基于模糊推理的电液动换向阀的故障诊断研究．1999，11
[9] 黄志坚．液压故障的特点与故障诊断的策略．液压与气动．2001(2)：29～31
[10] 神原正德．液压设备诊断技术现场适用．油压技术．1987
[11] 师占群．基于小波故障提取的电液伺服阀故障诊断．机械科学与技术，2000，19(增)：111～116
[12] 骆志明，冯庚斌．遗传BP网络在滚动轴承故障诊断中的应用．铁道学报，1999，10(5)：29～32
[13] 湛从昌．液压系统故障的模糊诊断方法．液压与气动，1994(6)：3-7
[14] 王垌．人工智能中的模糊启发式搜索技术．北京：机械工业出版社，1993，85～90
[15] 张乃尧，阎平凡．神经网络与模糊控制．北京：清华大学出版社．1997
[16] 王道平，张义忠．故障智能诊断系统的理论与方法．北京：冶金工业出版社，2001
[17] 陈章位．电液控制元件故障识别的新方法．液压与气动．1998(5)：71～72
[18] 曾昭军，何全成，史维祥．故障诊断神经网络的发展与前景．机械工程学报，1992，2(1)：1～6
[19] 王少萍，王占林．液压泵故障诊断的神经网络方法．北京航空航天大学学报．1997，23(6)：714～718
[20] 祝海林，胡爱萍．液压系统故障诊断的灰色理论方法探讨．石油化工高等学校学报，1998，9(3)：55～59
[21] 陈进才，郑守．人工智能发展的新趋势[J]．西安公路交通大学学报，1998，18(3)：94～97．
[22] Frank P metal. New Development Using AI in Fault Diagnosis. Engng. Applic. Artif. Intell, 1997, 10(1): 3～14
[23] Lu Yong xiang, Chen Zhang Wei. A Knowledge-Based electro hydraulic servo-system Fault diagnostic mode. Proc. Of the 4th Scandinavian Int. Conf. On fluid power Temper, 1995:686～694
[24] Branco, P. J Costa, Dente, J. A. On using fuzzy logic to integrate learning mechanisms in an electrohydraulic system-part Ⅰ.... IEEE Transactions on Systems, Man & Cybernetics: Part C-Applications & Reviews, Aug 2000, vol. 30 Issue 3, p317, 12p, 15 diagrams, 14 graphs, 2bw
[25] Goldberg DE. Genetic Algorithms in Search, Optimization and Machine learning[M]. Addison Wesley: Reading MA, 1989
[26] Window Betal. Neural Networks Application in industry, business and science. Communication of the Acm, 1994, 37:93～105
[27] Special Issue on Application of neural networks. Proc IEEE, 1996(84)
[28] Bagley JD. The behavior of adaptive systems which employ genetic and correlation algorithms. Dissertation abstracts international, 1967(28)
[29] Constantin Zopounidis. Special issue on artificial intelligence and decision support with multiple criteria[J]. Computers & Operations Research 2000(27):597～599
[30] R. J Patton, C. J Lopez-Toribio, F. J Uppal. Artificial intelligence approaches to fault diagnosis(C). IEE Colloquium on Condition Monitoring: Machinery, External Structures and Health (Ref. No. 1999/034). 1999:5/1～5/18
[31] 林建亚，何存兴．液压元件．北京：机械工业出版社，1988：8～30
[32] 孙文质．液压控制系统．国防工业出版社，1985：21～33
[33] 杨自厚．自动控制原理．北京：冶金工业出版社，1996：58～256
[34] 孙瑞利，周建桃．电液伺服阀的原理及其故障的排除．南京海运学院，1999(6)：49～51
[35] 王春行．液压伺服控制系统．北京：机械工业出版社，1981：96～180
[36] 虞和济，韩庆大，李沈．设备故障诊断工程．北京：冶金工业出版社，2001，6
[37] 中国机械工程学会材料学会主编．机械产品失效分析．北京：机械工业出版社，1986
[38] 湛从昌．液压可靠性与故障诊断．冶金工业出版社，1995，1
[39] 夏志新．液压系统污染控制．北京：机械工业出版社，1992
[40] 许耀铭．油膜理论与液压泵和马达的摩擦副设计．北京：机械工业出版社，1987
[41] 蔡倩，湛从昌．液压系统计算机辅助监测与故障诊断．武汉冶金科技大学学报．1997，20(3)：348～354
[42] 宋志军．基于人工神经网络的数据挖掘工具及应用．石油大学，2001，5
[43] 陈建安，郭大伟，徐乃平，孙云芝．遗传算法理论研究综述．西安电子科技大学学报，1998，6(3)：363～368
[44] 刘洁，魏连雨，杨春风．基于遗传．神经网络的交通量预测．长安大学学报，2003，3(1)：68～70
[45] 钟春生．人工神经网络在板形检测中的应用研究．西安建筑科技大学，2002，4
[46] 楼顺天，施阳．基于MATLAB的系统分析与设计．西安电子科技大学出版社，2000，11
[47] 邹岳华．人工神经网络BP算法及应用研究．上海理工大学，2000，1
[48] 赵军．基于遗传算法和神经网络模型的智能化拉深参数实时识别．燕山大学，2002，3
[49] 王宝树．遗传算法及神经网络在数据融合中的应．西安电子科技大学，2001，1
[50] 梁化楼，戴贵亮．人工神经网络与遗传算法的结合：进展及展望[J]．电子学报．1995，23(10)：194～200
[51] 倪远平．人工智能用于故障诊断的方法研究[J]．电器传动自动化．1998，20(4)：29～32
[52] 樊友平，黄席樾．智能诊断技术的发展和思考[J]．自然辨证法研究，2001，17(2)：42～46
[53] Irie B, Miyake S. Capabilities of three-layered perceptrons. Proceedings of the IEEE Inter Conf on Neural Network. 1998(1):641～648
[54] Health-NielsemR. The theory of back propagation neural network. In review, 1998
[55] 张立明．人工神经网络的模型及其应用．南京：复旦大学出版社．1994
[56] 郑志军，郑守淇．用基于实数编码的自适应遗传算法进化神经网络．计算机工程与应用．2000，9：36～37
[57] 张明辉，王尚锦．自适应搜索的改进遗传算法及其应用．西安交通大学学报，2002，3(3)：226～230
[58] DAI Xiao hui LI Min qiang KOU Ji song. Multi-objective optimization using genetic algorithm with local search. TransactionsofTianjinUniversity, 1998, 11(4):139～143
[59] Ralf Ostermark. . A hybrid genetic fuzzy neural network algorithm designed for classification problems involving several groups[J]. Fuzzy Sets and Systems, 2000, 114:311～324
[60] Kuo R. J, Xue K. C. Fuzzy neural networks with application to sales forecasting[J]. Fuzzy Sets and systems. 1999, 108:123～143
[61] 王正志，薄涛著．进化计算．长沙：国防科技大学出版社，2000
[62] Bob Reselman, Richard Peasley．实用Visual Basic 6教程．北京：清华大学出版社，2001
[63] Rob Krumm. ADO Programming For Dummies. BEIJING: Publishing House of Electronics Industry, 2001, 6