铅酸蓄电池容量光纤在线智能传感器研究
|
摘要
铅酸蓄电池是一种结构简单、使用方便、性能可靠的化学电源。铅蓄电池用途极为广泛,它是国民经济各部门必不可少的一种化学电源产品。随着生产力和科学技术的发展,蓄电池作为一种性能可靠的化学电源,其应用价值与日俱增,使用范围极为广泛,其作用日益显著,市场不断拓宽。蓄电池同时也是电动汽车的能量来源,为确保电池组性能良好并延长电池使用寿命,需对电池进行必要的管理和控制。因此,管理好蓄电池对于电动车来说是至关重要的,其中蓄电池容量的检测成为关键。
较准确和可靠地获得电池荷电状态(SOC)即铅酸蓄电池容量是电池管理系统中最基本和最首要的任务。因为SOC值的大小直接反映了电池所处的状态,由此可限定电池的最大放电电流和预测电动车的续驶里程。根据各节电池的SOC值,可以识别电池组中各电池间的性能差异,并依次进行均衡充电,以保持电池性能的均匀性,最终达到延长电池寿命的目的。
本文的主要研究内容如下:
⑴提出了一种基于光纤回归反射能量衰减技术的铅酸蓄电池容量检测方法。从Lambert-Beer定律可知,光的吸收系数与介质的浓度、通过吸收介质的光程和入射光强有关,本文研究正是从这一物理现象和理论出发,将接收到的光反射能量的大小与具有实用意义的电解液的浓度测量联系起来,而电解液的浓度又与铅酸蓄电池容量有着一一对应的函数关系,通过测量铅酸蓄电池电解液浓度来获得蓄电池的容量状态。
⑵在建立的测量方法的基础上,设计了铅酸蓄电池剩余容量在线测量传感器,并对传感器的组成原理、光路和实现技术进行了研究,开展了相关的实验研究。理论和实验研究表明:采用这种方法来测量蓄电池容量是可行的,单光头结构的测量相对误差≤8.42%,在测量过程中电解液温度的变化对铅酸蓄电池容量的测量有较大的影响。
⑶针对电解液温度对测量的影响,提出并设计了具有温度补偿的传感器。对原单光头传感器进行了结构改进,引入双光头探头结构。通过改进增加了传感器的测量功能:铅酸蓄电池容量测量、溶液温度测量。改进后的传感器一路信号用于测量待测电解液浓度,另一路通过测量相同环境下参考液温度变化,为消除温度变化对容量测量的影响奠定了基础。
⑷为消除温度变化对容量测量的影响,研究中采用神经网络及支持向量机等,建立了铅酸蓄电池剩余容量的估计模型。估计模型仿真研究表明:SVM和BP神经网络都具有任意精度逼近函数的学习性能,在样本个数较少的情况下,SVM的精度明显高于神经网络(BP)的逼近精度,且SVM训练所用的时间远远小于BP神经网络所用的时间。通过改进后的具有温度补偿的双光头结构传感器测量系统,采用估计模型仿真测量的相对误差可减小到≤3.25%,因此,改进后的结构要优于单光头结构。
针对铅酸蓄电池剩余容量测试的研究工作,只涉及到了铅酸蓄电池充放电过程的BP人工神经网络模型以及支持向量回归机估计模型的建立,目前,展开的铅酸蓄电池剩余容量在线测量传感器的研究,只涉及到电解液浓度(容量)和温度信号参数,更多的参数测量的融合技术还有待进一步研究。
Lead-acid battery is one kind of chemic power supplies which has simple structure, convenient use and credible capability. Lead-acid battery is used broadly, which can be used on industry, agriculture, traffic and transportation, and national defense etc. With the development of productivity and technology, the lead-acid battery as the credible chemic power, at the same time, it is also the energy source of electric vehicle, so we should do some necessary administering and controlling to make sure the pile capacity keep good and long life. The battery capacity measurement becomes the key in the electric vehicle administering and controlling.
The paper analyzed the relevant factors for battery capacity measurement, it turns out the discharge current, electrolyte temperature and electrolyte concentration can affect the measuring result. The paper also does some qualitative analysises on the charge-discharge voltage characteristic curve, the lead-acid battery capacity online measurement sensor based on fiber and intelligent information fuse technology is proposed.
The main research content as follows:
(1). Puts forward a testing method for lead-acid battery capacity based on fiber reflecting energy attenuation technology. Seen from the Lambert-Beer law, the optical absorption coefficient is related to the medium consistence, optical path and incidence energy, the paper research just based on the physics phenomena and theory, related the receiving optical energy to the electrolyte consistency, at the same time, the electrolyte consistency is related to the battery capacity, so we can get the battery capacity state by testing the lead-acid battery electrolyte consistency.
(2) Based on the built measuring method, the author designed the lead-acid battery capacity on-line measuring sensor, researched on the sensor constitute theory, optical path and realization technology, carried out related experiment research. The result turned out the new method used on testing the battery capacity is feasible, the single probe structure sensor’s relative error is less than 8.42%, in the testing process, and the electrolyte temperature change influenced the lead-acid battery capacity measuring result badly.
(3) Aiming at the electrolyte temperature influencing the measuring result, put forward and designed the sensor with temperature compensation, improved the former single probe structure and introduced the double probe structure. This way make the sensor added measuring functions, which is battery capacity measure and liquor temperature measure. The improved sensor has two channels, one is used to measure electrolyte consistency and the other is used to measure reference liquid temperature change under the same measuring condition, which is established the base for eliminating temperature factor for measuring result.
(4) In order to eliminate the temperature factor for measuring result, The author adopted some fitting means such as neural network (NN) and support vector machine (SVM) to build lead-acid battery capacity estimating module, the simulation result turned out that NN and SVM all have learning performance of fitting function with any precision. Under the conditions of few sample number, the SVM has better fitting precision than the NN, at the same time the SVM use fewer training time than BP NN. The improved double-probe structure sensor measuring system adopts estimating module to do simulation measure and the testing relative error can reduced to less than 3.25%. So the improved structure sensor is better than the former single probe structure sensor.
Aiming at research work of lead-acid battery residual capacity, the paper just involves the BP NN fitting and SVM fitting modeling of lead-acid battery charge-discharge course, and the study parameter just involves measuring signal voltage and reference liquid output signal voltage, the more parameters need deeply research.
较准确和可靠地获得电池荷电状态(SOC)即铅酸蓄电池容量是电池管理系统中最基本和最首要的任务。因为SOC值的大小直接反映了电池所处的状态,由此可限定电池的最大放电电流和预测电动车的续驶里程。根据各节电池的SOC值,可以识别电池组中各电池间的性能差异,并依次进行均衡充电,以保持电池性能的均匀性,最终达到延长电池寿命的目的。
本文的主要研究内容如下:
⑴提出了一种基于光纤回归反射能量衰减技术的铅酸蓄电池容量检测方法。从Lambert-Beer定律可知,光的吸收系数与介质的浓度、通过吸收介质的光程和入射光强有关,本文研究正是从这一物理现象和理论出发,将接收到的光反射能量的大小与具有实用意义的电解液的浓度测量联系起来,而电解液的浓度又与铅酸蓄电池容量有着一一对应的函数关系,通过测量铅酸蓄电池电解液浓度来获得蓄电池的容量状态。
⑵在建立的测量方法的基础上,设计了铅酸蓄电池剩余容量在线测量传感器,并对传感器的组成原理、光路和实现技术进行了研究,开展了相关的实验研究。理论和实验研究表明:采用这种方法来测量蓄电池容量是可行的,单光头结构的测量相对误差≤8.42%,在测量过程中电解液温度的变化对铅酸蓄电池容量的测量有较大的影响。
⑶针对电解液温度对测量的影响,提出并设计了具有温度补偿的传感器。对原单光头传感器进行了结构改进,引入双光头探头结构。通过改进增加了传感器的测量功能:铅酸蓄电池容量测量、溶液温度测量。改进后的传感器一路信号用于测量待测电解液浓度,另一路通过测量相同环境下参考液温度变化,为消除温度变化对容量测量的影响奠定了基础。
⑷为消除温度变化对容量测量的影响,研究中采用神经网络及支持向量机等,建立了铅酸蓄电池剩余容量的估计模型。估计模型仿真研究表明:SVM和BP神经网络都具有任意精度逼近函数的学习性能,在样本个数较少的情况下,SVM的精度明显高于神经网络(BP)的逼近精度,且SVM训练所用的时间远远小于BP神经网络所用的时间。通过改进后的具有温度补偿的双光头结构传感器测量系统,采用估计模型仿真测量的相对误差可减小到≤3.25%,因此,改进后的结构要优于单光头结构。
针对铅酸蓄电池剩余容量测试的研究工作,只涉及到了铅酸蓄电池充放电过程的BP人工神经网络模型以及支持向量回归机估计模型的建立,目前,展开的铅酸蓄电池剩余容量在线测量传感器的研究,只涉及到电解液浓度(容量)和温度信号参数,更多的参数测量的融合技术还有待进一步研究。
Lead-acid battery is one kind of chemic power supplies which has simple structure, convenient use and credible capability. Lead-acid battery is used broadly, which can be used on industry, agriculture, traffic and transportation, and national defense etc. With the development of productivity and technology, the lead-acid battery as the credible chemic power, at the same time, it is also the energy source of electric vehicle, so we should do some necessary administering and controlling to make sure the pile capacity keep good and long life. The battery capacity measurement becomes the key in the electric vehicle administering and controlling.
The paper analyzed the relevant factors for battery capacity measurement, it turns out the discharge current, electrolyte temperature and electrolyte concentration can affect the measuring result. The paper also does some qualitative analysises on the charge-discharge voltage characteristic curve, the lead-acid battery capacity online measurement sensor based on fiber and intelligent information fuse technology is proposed.
The main research content as follows:
(1). Puts forward a testing method for lead-acid battery capacity based on fiber reflecting energy attenuation technology. Seen from the Lambert-Beer law, the optical absorption coefficient is related to the medium consistence, optical path and incidence energy, the paper research just based on the physics phenomena and theory, related the receiving optical energy to the electrolyte consistency, at the same time, the electrolyte consistency is related to the battery capacity, so we can get the battery capacity state by testing the lead-acid battery electrolyte consistency.
(2) Based on the built measuring method, the author designed the lead-acid battery capacity on-line measuring sensor, researched on the sensor constitute theory, optical path and realization technology, carried out related experiment research. The result turned out the new method used on testing the battery capacity is feasible, the single probe structure sensor’s relative error is less than 8.42%, in the testing process, and the electrolyte temperature change influenced the lead-acid battery capacity measuring result badly.
(3) Aiming at the electrolyte temperature influencing the measuring result, put forward and designed the sensor with temperature compensation, improved the former single probe structure and introduced the double probe structure. This way make the sensor added measuring functions, which is battery capacity measure and liquor temperature measure. The improved sensor has two channels, one is used to measure electrolyte consistency and the other is used to measure reference liquid temperature change under the same measuring condition, which is established the base for eliminating temperature factor for measuring result.
(4) In order to eliminate the temperature factor for measuring result, The author adopted some fitting means such as neural network (NN) and support vector machine (SVM) to build lead-acid battery capacity estimating module, the simulation result turned out that NN and SVM all have learning performance of fitting function with any precision. Under the conditions of few sample number, the SVM has better fitting precision than the NN, at the same time the SVM use fewer training time than BP NN. The improved double-probe structure sensor measuring system adopts estimating module to do simulation measure and the testing relative error can reduced to less than 3.25%. So the improved structure sensor is better than the former single probe structure sensor.
Aiming at research work of lead-acid battery residual capacity, the paper just involves the BP NN fitting and SVM fitting modeling of lead-acid battery charge-discharge course, and the study parameter just involves measuring signal voltage and reference liquid output signal voltage, the more parameters need deeply research.
引文
[1] 肖永清,肖军. 汽车蓄电池的使用与维修[M]. 中国电力出版社, 2005:5-13
[2] 肖永清,杨忠敏. 汽车的发展与未来[M]. 化学工业出版社, 2004:11-24
[3] 徐曼珍. 新型蓄电池原理与应用[M]. 人民邮电出版社, 2005:27-59
[4] 张荣甫,高树发,孙霞. 通信蓄电池容量检测仪设计[J]. 军事通信技术,2004.25(2): 56-58
[5] 李兴虎.电动汽车概念[M].北京:北京理工大学, 2005:36-77
[6] 陈红雨,吴玲,任超.从欧洲蓄电池会议看国外铅酸蓄电池的研究动态[J].蓄电池.2000, (4):35-37
[7] 门智刚,胡信国,戴常松,王金玉.电动车用阀控铅酸蓄电池的进展[J].蓄电池.1993,(3):36-41
[8] Jaegul L,Soonjae L,Namgoong E. Dynamic state battery model with Self-adaptive aging factor for EV and HEV applications [A].Procee-Dings of the 15th International Electric Vehicle Symposium [C].1998.
[9] Morio K, Kazuhiro H, Anil P. Battery SOC and distance to empty of The Honda EV Plus [A]. Proceedings of the 14th International Electric Vehicle Symposium [C]. 1997.
[10] 齐国光,李建民,郏航等. 电动汽车电量计量技术的研究[J].清华大学学报(自然科学版).1997, 37(3):46-49
[11] 桂长清.密封铅蓄电池荷电态预测技术[J]. 电池 2003, 33 (5): 319-321
[12] Kozlowski J D. Model2based predictive biagnostics for primary and secondary batteries. The Battery Man, 2001, 4:14-29.
[13] 闫智刚等.电动自行车用铅酸动力电池的发展状况[J]. 蓄电池 2002, (3): 97-102
[14] 孙力. 电动车蓄电池电量监测技术研究[D].硕士论文. 北方工业大学, 2002
[15] James H A, Alfred T, Barry W J. A battery state of charge indicator for electric wheelchairs [A]. IEEE Transactions on Industrial Electronics [C].1992.10
[16] Huet F. A review of impedance measurements for determination of the state-of-charge or state-of-health of secondary batteries. J Power Sources, 1998, 70(1):59-69
[17] 周系维等.现代化学电源[M]. 河北大学出版社.1991:201-222
[18] CHEN Quan shi(陈全世), LIN Yong jun(林拥军),ZHANG Dong min(张东民).电动汽车用铅酸电池放电特性的研究[J].Automobile Technology(汽车技术),1996, (8): 7-11
[19] 贺乃宝.对铅酸蓄电池容量检测方法的探讨[J].淮海工学院学报.2003, 12 (3):28-30
[20] 赵明富, 陈艳, 罗渝微等. 光纤铅酸蓄电池容量传感器研究[J].压电与声光.2007, 29(1):15-18
[21] 赵明富, 廖强, 陈艳等. 光纤生物量浓度在线检测传感器[J].光学精密工程.2007, 15(4):478-483
[22] 赵明富. 生物膜滴滤塔净化低浓度有机废气过程的测量与控制问题研究[D]. 博士论文. 重庆大学, 2007: 61-66
[23] 李槟, 陈全世.混合动力电动车中电池特性的研究[J]. 汽车技术.1999(10): 21-31
[24] Donald W. Corson. High power battery systems for hybrid vehicles. Journal of Power Source 105(2002) 110-113
[25] 朱松然.蓄电池手册[M].天津: 天津大学出版社, 1998:20-29
[26] 袁方伟, 陈思忠.电动汽车电池管理系统的研究[J].试验与研究, 2003,(3):41-44
[27] 沈维祥.基于开路电压和环境温度的铅酸蓄电池充电状态的数学模型[J].太阳能学报, 1997,18(3):327-332
[28] James H A, Alfred T, Barry WJ. A battery state of charge indicator for electric wheelchairs [A]. IEEE Transactions on Industrial Electronics [C].1992.10
[29] 林成涛, 王军平, 陈全世.电动汽车 SOC 估计方法原理与应用[J]. 电池.2004, 34 (5):376-378
[30] Shinya Sato, Atsuo Kawamura. A New Estimation Method of State of Charge using Terminal Voltage and Internal Resistance for Lead Acid Battery. Department of Electrical and Computer Engineering Yokohama National University
[31] Ehret C, Piller S, Schroer W et al. State-of-charge determination for lead-acid batteries in PV-applications [A]. Proceedings of the 16th European Photovoltaic Solar Energy Conference [C]. Gasgow: 2000
[32] 裴锋, 黄向东, 罗玉涛等. 电动汽车动力电池变流放电特性与荷电状态实时估计[J], 中国电机工程学报, 2005, 25(9):164-168
[33] 宫 学 庚 , 齐 铂 金 等 . 电 动 汽 车 动 力 电 池 模 型 和 SOC 估 算 策 略 [J]. 电 源 技 术 , 2004,128(10):637-640
[34] 郁道银, 谈恒英. 工程光学[M]. 北京: 机械工业出版社, 1998.11:67-89
[35] 王姨平, 张金利, 黄群武.化工测试技术[M]. 天津: 天津大学出版社 2005.3:36-100
[36] 籍延坤, 郭 红.溶液折射率公式的一种验证方法[J]. 物理与工程, 2001:37-38.
[37] 王池.流量测量不确定度分析[M]. 北京:中国计量出版社, 2002:80-91
[38] 马科斯.玻恩, 埃米尔.沃耳夫. 光学原理(第七版上册)[M].北京: 电子工业出版社, 2005:25-70
[39] Zhao Mingfu, Wang Qingbing, Luo Song. Research of the photoelectric on-line sensor for detecting the residual capacity in a lead acid storage battery[C],The International Conference on Mechatronics and Information Technology'2005, Proc.of SPIE Vol.6041, 60411V-1-6.
[40] 国家技术监督局. GB2900.00-1988 蓄电池名词术语.1998
[41] 屈莉莉, 杨兆华, 杨振坤等. 一种新型蓄电池充放电专用节能装置[J].电源技术, 2003, 27(5): 468~469, 483
[42] 李国勇. 智能控制及其 MATLAB 实现[M]. 电子工业出版社, 2005: 21-65
[43] Martin T. Hagan, Howard B. Demuth, Mark H. Beale. Neural Network Design [J]. China Machine Press, 2002.
[44] K. M. Hornik, M. Stinchcombe, H. White. Multilayer feedforward networks are universal approximators. Neural Networks. 1989, 2 (5):359-366
[45] Pham D T, Liu X. Training of Elman networks and dynamic system modeling [J]. International Journal of SystemsScience: 1996.27(2): 221~226
[46] 张星昌. 前馈神经网络的新学习算法研究及应用[J].控制与决策:1997.12(3): 213~216
[47] MT Hagan, etai(美), 戴葵等译. 神经网络设计[M]. 北京:机械工业出版社, 2002, 9:11-24
[48] Vapnik V. The nature of Statistical Learning Theory[M]. Springer - Verlag, New York, 1999.
[49] 高异. 基于支持向量机的非线性系统建模与控制: [D], 硕士论文.西安理工大学, 2006.3:12-21
[50] 奉国和. SVM 与神经网络在时间序列预测中的比较[M]. 现代管理学. 2006.9:40-41
[2] 肖永清,杨忠敏. 汽车的发展与未来[M]. 化学工业出版社, 2004:11-24
[3] 徐曼珍. 新型蓄电池原理与应用[M]. 人民邮电出版社, 2005:27-59
[4] 张荣甫,高树发,孙霞. 通信蓄电池容量检测仪设计[J]. 军事通信技术,2004.25(2): 56-58
[5] 李兴虎.电动汽车概念[M].北京:北京理工大学, 2005:36-77
[6] 陈红雨,吴玲,任超.从欧洲蓄电池会议看国外铅酸蓄电池的研究动态[J].蓄电池.2000, (4):35-37
[7] 门智刚,胡信国,戴常松,王金玉.电动车用阀控铅酸蓄电池的进展[J].蓄电池.1993,(3):36-41
[8] Jaegul L,Soonjae L,Namgoong E. Dynamic state battery model with Self-adaptive aging factor for EV and HEV applications [A].Procee-Dings of the 15th International Electric Vehicle Symposium [C].1998.
[9] Morio K, Kazuhiro H, Anil P. Battery SOC and distance to empty of The Honda EV Plus [A]. Proceedings of the 14th International Electric Vehicle Symposium [C]. 1997.
[10] 齐国光,李建民,郏航等. 电动汽车电量计量技术的研究[J].清华大学学报(自然科学版).1997, 37(3):46-49
[11] 桂长清.密封铅蓄电池荷电态预测技术[J]. 电池 2003, 33 (5): 319-321
[12] Kozlowski J D. Model2based predictive biagnostics for primary and secondary batteries. The Battery Man, 2001, 4:14-29.
[13] 闫智刚等.电动自行车用铅酸动力电池的发展状况[J]. 蓄电池 2002, (3): 97-102
[14] 孙力. 电动车蓄电池电量监测技术研究[D].硕士论文. 北方工业大学, 2002
[15] James H A, Alfred T, Barry W J. A battery state of charge indicator for electric wheelchairs [A]. IEEE Transactions on Industrial Electronics [C].1992.10
[16] Huet F. A review of impedance measurements for determination of the state-of-charge or state-of-health of secondary batteries. J Power Sources, 1998, 70(1):59-69
[17] 周系维等.现代化学电源[M]. 河北大学出版社.1991:201-222
[18] CHEN Quan shi(陈全世), LIN Yong jun(林拥军),ZHANG Dong min(张东民).电动汽车用铅酸电池放电特性的研究[J].Automobile Technology(汽车技术),1996, (8): 7-11
[19] 贺乃宝.对铅酸蓄电池容量检测方法的探讨[J].淮海工学院学报.2003, 12 (3):28-30
[20] 赵明富, 陈艳, 罗渝微等. 光纤铅酸蓄电池容量传感器研究[J].压电与声光.2007, 29(1):15-18
[21] 赵明富, 廖强, 陈艳等. 光纤生物量浓度在线检测传感器[J].光学精密工程.2007, 15(4):478-483
[22] 赵明富. 生物膜滴滤塔净化低浓度有机废气过程的测量与控制问题研究[D]. 博士论文. 重庆大学, 2007: 61-66
[23] 李槟, 陈全世.混合动力电动车中电池特性的研究[J]. 汽车技术.1999(10): 21-31
[24] Donald W. Corson. High power battery systems for hybrid vehicles. Journal of Power Source 105(2002) 110-113
[25] 朱松然.蓄电池手册[M].天津: 天津大学出版社, 1998:20-29
[26] 袁方伟, 陈思忠.电动汽车电池管理系统的研究[J].试验与研究, 2003,(3):41-44
[27] 沈维祥.基于开路电压和环境温度的铅酸蓄电池充电状态的数学模型[J].太阳能学报, 1997,18(3):327-332
[28] James H A, Alfred T, Barry WJ. A battery state of charge indicator for electric wheelchairs [A]. IEEE Transactions on Industrial Electronics [C].1992.10
[29] 林成涛, 王军平, 陈全世.电动汽车 SOC 估计方法原理与应用[J]. 电池.2004, 34 (5):376-378
[30] Shinya Sato, Atsuo Kawamura. A New Estimation Method of State of Charge using Terminal Voltage and Internal Resistance for Lead Acid Battery. Department of Electrical and Computer Engineering Yokohama National University
[31] Ehret C, Piller S, Schroer W et al. State-of-charge determination for lead-acid batteries in PV-applications [A]. Proceedings of the 16th European Photovoltaic Solar Energy Conference [C]. Gasgow: 2000
[32] 裴锋, 黄向东, 罗玉涛等. 电动汽车动力电池变流放电特性与荷电状态实时估计[J], 中国电机工程学报, 2005, 25(9):164-168
[33] 宫 学 庚 , 齐 铂 金 等 . 电 动 汽 车 动 力 电 池 模 型 和 SOC 估 算 策 略 [J]. 电 源 技 术 , 2004,128(10):637-640
[34] 郁道银, 谈恒英. 工程光学[M]. 北京: 机械工业出版社, 1998.11:67-89
[35] 王姨平, 张金利, 黄群武.化工测试技术[M]. 天津: 天津大学出版社 2005.3:36-100
[36] 籍延坤, 郭 红.溶液折射率公式的一种验证方法[J]. 物理与工程, 2001:37-38.
[37] 王池.流量测量不确定度分析[M]. 北京:中国计量出版社, 2002:80-91
[38] 马科斯.玻恩, 埃米尔.沃耳夫. 光学原理(第七版上册)[M].北京: 电子工业出版社, 2005:25-70
[39] Zhao Mingfu, Wang Qingbing, Luo Song. Research of the photoelectric on-line sensor for detecting the residual capacity in a lead acid storage battery[C],The International Conference on Mechatronics and Information Technology'2005, Proc.of SPIE Vol.6041, 60411V-1-6.
[40] 国家技术监督局. GB2900.00-1988 蓄电池名词术语.1998
[41] 屈莉莉, 杨兆华, 杨振坤等. 一种新型蓄电池充放电专用节能装置[J].电源技术, 2003, 27(5): 468~469, 483
[42] 李国勇. 智能控制及其 MATLAB 实现[M]. 电子工业出版社, 2005: 21-65
[43] Martin T. Hagan, Howard B. Demuth, Mark H. Beale. Neural Network Design [J]. China Machine Press, 2002.
[44] K. M. Hornik, M. Stinchcombe, H. White. Multilayer feedforward networks are universal approximators. Neural Networks. 1989, 2 (5):359-366
[45] Pham D T, Liu X. Training of Elman networks and dynamic system modeling [J]. International Journal of SystemsScience: 1996.27(2): 221~226
[46] 张星昌. 前馈神经网络的新学习算法研究及应用[J].控制与决策:1997.12(3): 213~216
[47] MT Hagan, etai(美), 戴葵等译. 神经网络设计[M]. 北京:机械工业出版社, 2002, 9:11-24
[48] Vapnik V. The nature of Statistical Learning Theory[M]. Springer - Verlag, New York, 1999.
[49] 高异. 基于支持向量机的非线性系统建模与控制: [D], 硕士论文.西安理工大学, 2006.3:12-21
[50] 奉国和. SVM 与神经网络在时间序列预测中的比较[M]. 现代管理学. 2006.9:40-41