基于FCS的船舶机舱智能监控系统的研究
|
摘要
本文针对当前多微处理器多层结构的船舶机舱自动监控系统存在的系统结构
复杂、设备互操作性差、系统维护和扩展不便、缺乏对系统状态信息进行智能预
测分析等有待进一步研究的问题，对船舶机舱自动监控系统的硬件和软件进行了
分析，提出了一种基于现场总线技术的船舶机舱智能监控系统的设计方案。
利用现场总线技术，建立由管理计算机和智能节点组成的两层网络结构，取
消原系统中间层的通信转发站，数据采集和通信由安装在现场的智能节点完成，
简化了系统结构，增强了系统的可靠性。并对基于LonWorks现场总线的监控系
统各主要功能模块进行了设计。
采用模块化、通用化的软件设计思想，提出利用数据库技术管理系统监测参
数的方案，设计了多种用于参数显示的控件，使系统监控软件的组态和维护容易，
增强了系统的通用性、灵活性和可扩展性。
提出一种结构简单、学习速度快捷、适合于动态过程在线多步预测分析的
DRNN预测模型及其在线学习的TD-DBP复合学习算法。解决了非线性动态
过程基于前馈网络或反馈网络的预测模型需要事先确定系统模型阶数的问题。
把基于人工神经网络的预测分析模块集成到船舶机舱监控系统软件中，为船
舶机舱监控系统增加对监测参数在线动态趋势分析的功能，提高监控系统的可靠
性和安全性。
对本文提出的设计方案，在上海海运学院自动化机舱实验中心，通过安装调
试，取得了令人满意的效果。本设计方案，能简化系统结构，增强系统的信息处
理能力，提高系统的可靠性，增强系统的通用性、灵活性和可扩展性，能方便地
把船舶机舱自动化系统与船舶其它自动化系统集成为一个完整的船舶综合信息网
络，实现信息共享。
Practically, the reliability, maintainability, real-time control ability and
interoperability are required by the monitor and control system for marine engine room.
However the monitor system based on the multi-microprocessors and the hierarchical
structure is not so satisfied in such requirements, owing to its complex structure and
devoted communication sub-stations. In order to solve this problem, a design method
of intelligent monitor system based on field-bus control system (FCS) for marine
engine room is presented in this dissertation.
To simplify the structure of the monitor and control system and to enhance its
reliability, a monitoring network composed of a management computer and the
multiple intelligent nodes is constructed using field-bus technology. In this system,
each intelligent node can be installed at any control or measurement field to realize the
functions of both data-acquisition and communication. Because the FCS features much
higher performances as compared with DCS, it can be used as the updated substitute of
DCS-based monitor system for marine engine room. In this dissertation, the design of
the main functional units as well as the configuration of the monitor system is
described.
To supply much easier maintenance and configuration to the monitor system, a
design method of the system software based on database technology is presented in
this dissertation. All process parameters in the monitor and control system are
managed by the databases, with one being used specially to manage the process points
including the number and name of each sensor, the range of each measurement, the
alarm limit and the measuring channel etc. Besides, some universal display modules
are designed for the monitor software.
A new adaptive optimal predictor based on diagonal recurrent neural network
(DRNN) is present in this dissertation. And a new TD-DBP algorithm combined
temporal difference algorithm with dynamic back-propagation algorithm is proposed
for DRNN on-line learning. Usually, some important parameters must be determined
firstly in a predictor. For example, the order of the system must be determined at
advance for an adaptive predictor based on the forward or feedback neural network.
But it might be difficult or impossible in some cases. Now the new predictor does not
require deciding on any system parameter. For this reason, it is able to realize system
modeling, parameter modifying and trends forecasting on line via learning ftom the
time series of a real dynamic process. This predictor has the advantage in simple
3
network topology and short learning time. A program of predictive analysis based on
DRNN is integrated in the monitoring software.
All the theoretical methods proposed in this dissertation have been applied to build
up an experimental monitor system at the Lab Center of Automatic Marine Engine
Room in Shanghai Maritime University. The satisfactory results have been obtained by
the experiments. Such design of the monitor and control system can simplify its
structure, and enhance its ability of information processing, and promote its reliability
and flexibility. Besides, this type of monitor system for marine engine room can be
easily combined with other marine automatic systems to realize the information
network of the total ship.
复杂、设备互操作性差、系统维护和扩展不便、缺乏对系统状态信息进行智能预
测分析等有待进一步研究的问题，对船舶机舱自动监控系统的硬件和软件进行了
分析，提出了一种基于现场总线技术的船舶机舱智能监控系统的设计方案。
利用现场总线技术，建立由管理计算机和智能节点组成的两层网络结构，取
消原系统中间层的通信转发站，数据采集和通信由安装在现场的智能节点完成，
简化了系统结构，增强了系统的可靠性。并对基于LonWorks现场总线的监控系
统各主要功能模块进行了设计。
采用模块化、通用化的软件设计思想，提出利用数据库技术管理系统监测参
数的方案，设计了多种用于参数显示的控件，使系统监控软件的组态和维护容易，
增强了系统的通用性、灵活性和可扩展性。
提出一种结构简单、学习速度快捷、适合于动态过程在线多步预测分析的
DRNN预测模型及其在线学习的TD-DBP复合学习算法。解决了非线性动态
过程基于前馈网络或反馈网络的预测模型需要事先确定系统模型阶数的问题。
把基于人工神经网络的预测分析模块集成到船舶机舱监控系统软件中，为船
舶机舱监控系统增加对监测参数在线动态趋势分析的功能，提高监控系统的可靠
性和安全性。
对本文提出的设计方案，在上海海运学院自动化机舱实验中心，通过安装调
试，取得了令人满意的效果。本设计方案，能简化系统结构，增强系统的信息处
理能力，提高系统的可靠性，增强系统的通用性、灵活性和可扩展性，能方便地
把船舶机舱自动化系统与船舶其它自动化系统集成为一个完整的船舶综合信息网
络，实现信息共享。
Practically, the reliability, maintainability, real-time control ability and
interoperability are required by the monitor and control system for marine engine room.
However the monitor system based on the multi-microprocessors and the hierarchical
structure is not so satisfied in such requirements, owing to its complex structure and
devoted communication sub-stations. In order to solve this problem, a design method
of intelligent monitor system based on field-bus control system (FCS) for marine
engine room is presented in this dissertation.
To simplify the structure of the monitor and control system and to enhance its
reliability, a monitoring network composed of a management computer and the
multiple intelligent nodes is constructed using field-bus technology. In this system,
each intelligent node can be installed at any control or measurement field to realize the
functions of both data-acquisition and communication. Because the FCS features much
higher performances as compared with DCS, it can be used as the updated substitute of
DCS-based monitor system for marine engine room. In this dissertation, the design of
the main functional units as well as the configuration of the monitor system is
described.
To supply much easier maintenance and configuration to the monitor system, a
design method of the system software based on database technology is presented in
this dissertation. All process parameters in the monitor and control system are
managed by the databases, with one being used specially to manage the process points
including the number and name of each sensor, the range of each measurement, the
alarm limit and the measuring channel etc. Besides, some universal display modules
are designed for the monitor software.
A new adaptive optimal predictor based on diagonal recurrent neural network
(DRNN) is present in this dissertation. And a new TD-DBP algorithm combined
temporal difference algorithm with dynamic back-propagation algorithm is proposed
for DRNN on-line learning. Usually, some important parameters must be determined
firstly in a predictor. For example, the order of the system must be determined at
advance for an adaptive predictor based on the forward or feedback neural network.
But it might be difficult or impossible in some cases. Now the new predictor does not
require deciding on any system parameter. For this reason, it is able to realize system
modeling, parameter modifying and trends forecasting on line via learning ftom the
time series of a real dynamic process. This predictor has the advantage in simple
3
network topology and short learning time. A program of predictive analysis based on
DRNN is integrated in the monitoring software.
All the theoretical methods proposed in this dissertation have been applied to build
up an experimental monitor system at the Lab Center of Automatic Marine Engine
Room in Shanghai Maritime University. The satisfactory results have been obtained by
the experiments. Such design of the monitor and control system can simplify its
structure, and enhance its ability of information processing, and promote its reliability
and flexibility. Besides, this type of monitor system for marine engine room can be
easily combined with other marine automatic systems to realize the information
network of the total ship.
引文
1 汤天浩，王启宇．船舶综合自动化系统的展望与探索．见：张湘伟主编．跨世纪人才论中国交通．重庆：重庆大学出版社，1995．755～759
2 汤天浩，李杰仁．智能控制—21世纪船舶自动化系统的核心．见：邵维文等主编．第五届全国海事技术研讨会论文集．北京：海洋出版社，1999．125～130
3 周又丁．浅谈现场总线控制系统的出现及特点．自动化仪器仪表，1998(6)
4 王锦标．现场总线控制系统．微计算机信息，1996，12(6)
5 苏倩．基于现场总线技术的全开放分散控制系统．测控技术，1999，18(5)
6 阳宪惠．现场总线技术及其应用．北京：清华大学出版社，1999
7 杨育红．LON 网络控制技术及应用．西安：西安电子科技大学出版社，1999
8 孙自强．生产过程自动化及仪表．上海：华东理工大学出版社，1999
9 周振环，凌志浩等．一种新颖的分布式控制网络．自动化仪表，1997，18(7)
10 孙来业．基于 LonWorks 的分散智能控制系统的研究．工业控制计算机，1999，12(6)：36～38
11 凌志浩等．基于 Neuron 芯片的远程数据采集装置的设计．计算机自动测量与控制，1999,7(4)：26～29
12 吴誉，张卫东．现场总线技术展望及我国的发展策略．测控技术，1999，18(11)
13 刘秉瀚．学习控制系统的设计与实现．自动化仪表，1997，18(10)
14 吴今培．实用时序分析．长沙：湖南科学技术出版社，1989
15 T. Soderstrom, M. Mossberg Performance evaluation of methods for identifying continuous-time. Autoregressive processes. Automatica, 2000, 36(1): 53～59
16 杨位钦、顾岚．时间序列分析与动态数据建模．北京：北京工业学院出版社，1986．
17 Doglas M．Bates 等著．非线性回归分析及其应用(韦博成 万方焕等译)．北京：中国统计出版社 1997．
18 J. Mari et al. A covariance extension approach to identification of time series. Automatica, 2000, 36(2): 379～398
19 武宏伟，戴琼海等．基于神经网络的非线性系统的观测器设计．清华大学学报，2000，40(3)：44～47
20 焦李成．神经网络系统理论．西安：西安电子科技大学出版社，1990
21 何耀华，夏志忠．BP网络的快速自适应学习算法．系统工程理论与实践，2000(1)：93～98
22 张立明．人工神经网络的模型及其应用．上海：复旦大学出版社，1993
23 G. J. Kulawski, M. A. Brdys. Stable adaptive control with recurrent networks Automatica, 2000, 36(1): 5～22
24 沈清，胡德文等．神经网络应用技术．长沙：国防科技大学出版社，1993
25 C. Kambhampati et al. A stable one-step-ahead predictive control of non-linear system. Automatica, 2000, 36(2): 485～495
26 王玉涛，周清波等．基于神经网络的时间序列预测算法及其应用．信息与控制，1998，27(6)：413～417
27 王清印，崔援民．不确定性系统预测理论的基本问题．系统工程理论与实践 2000(3)：93～98
28 Tianhao Tang, et al. A RNN-based adaptive predictor for fault prediction and incipient diagnosis. Proceedings of UKACC International Conference on Control 2000，University Cambridge, UK, 4-7, Sept. 2000
29 王科俊，王克成．神经网络建模、预报与控制．哈尔滨：哈尔滨工程大学出版社，1996
30 沈轶等．时间离散的回归神经网络的定性分析．系统工程理论与实践，2000(1)：57～61
31 Box, G. E. P. and G. M. Jenkins. Time series analysis of forecasting and control Holden-day, San Francisco, 1970
32 汤天浩，刘以建等．一种基于人工神经网络在线学习的自适应预测方法．计算机辅助工程 1997，6(3)：13～18
33 Ku C C and Lee K Y. Diagonal recureent neural networks and robust time series prediction.. IEEE trans, on NN, 1994, 5(2)
34 Kang Li and Stephen Thompson. A Cascaded Neural Network and Its Application to Modelling Power Plant Pollutant Emission. Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun 28-July 2,2000 Hefei China IEEE Conference Publication No.992-997.
35 Tianhao Tang, M. Xiong ANN-based Nonlinear Timer Series Models in Fault Detection and Prediction. IFAC confrence CAMS'98 Control Application in Matin System, 1998
36 Pandit, S. M. and S. M. Wu. Time series and system analysis with applications, the United States: John Wiley and sons. 1983
37 Wittenmark, B. A self-tuning predictor. IEEE Trans. Autom, Cont, 1974, 19 (6): 848-851
38 Keyser, R.M.C. and van A.R. Cauwenberghe, A self-tuning multistep predictor application. Automatica, 1984, 17(1): 167-174
39 姜启源，孙义军．ARMAX模型两种自适应 K 步超前预报形式的比较．自动化学报，1993，19(3)：544-551
40 Hush, D., et al. Recursive neural networks for signal processing and control. in Proc. of 1st IEEE-SP Workshop on Neural Networks for Signal Processing, Princeton, NJ, 1991: 523-532
41 Lou Falong and Bao Zheng. Neural network for a class of nonlinear adaptive filters. Advances on Modeling and Simulation, 1992, 31 (1): 109-120
42 冯恩波等．一种基于时序预报神经网络的故障预报方法及其应用．自动化学报，1995，12(3)：348-352
43 Angeline, P. J., et al. An evolutionary algorithm that constructs recurrent neural networks. IEEE Trans. on Neural Networks. 1994, 5 (1): 54-65
44 McDonnell, J. R. and D. Waagen. Evolving recurrent perceptrons for time-series modeling. IEEE Trans. on Neural Network. 1994, 5, (1): 24-37
45 Parlos, A. G. et al. Application of recurrent multilayer perceptron in modeling complex process dynamics. IEEE Trans. on Neural Networks. 1994, 5, (2): 255-266
46 Connor, J. T. et al. Recurrent neural networks and robust time series prediction. IEEE Trans. on Neural Networks, 1994, 5, (2): 240-254
47 Tang Tianhao, et al. An integrated intelligent method for marine monitoring system Preprints of 14th World Congress of IFAC, vol. Q, pp, 55～60, Beijing, China, 1999
48 Williams, R. J. et al. An efficient gradient-based algorithm for on-line training of recurrent network trajectories. Neural Computation, 1990, 2 (4): 490-501
49 Li Jieren, and J. Mindykowski, Development of Engine Room Monitoring and Alarm System, Proceedings of the 1997 International Marine Electrotechnology Conference and Exhibition IMECE'97, China, Shanghai 1997: 131～135
50 刘辉，赵永生．船舶机舱监测报警系统的设计．大连海事大学学报．1999，25(4)：65～69
51 王新全．船舶柴油机微机监测系统及操作管理．北京：人民交通出版社，1995
52 曹善文．机舱自动化与微型计算机及其管理．北京：人民交通出版社，1993
53 凌峰，陈祥献等．工业测控软件中报警系统的设计与实现．测控技术，2000，19(2)：17～20
54 胡武，毕建权等．计算机监控软件中实时数据库系统的研制．测控技术，2000，19(2)：21～23
55 钱为民．测控软件发展的一些思考．测控技术，2000，19(2)：6～8
2 汤天浩，李杰仁．智能控制—21世纪船舶自动化系统的核心．见：邵维文等主编．第五届全国海事技术研讨会论文集．北京：海洋出版社，1999．125～130
3 周又丁．浅谈现场总线控制系统的出现及特点．自动化仪器仪表，1998(6)
4 王锦标．现场总线控制系统．微计算机信息，1996，12(6)
5 苏倩．基于现场总线技术的全开放分散控制系统．测控技术，1999，18(5)
6 阳宪惠．现场总线技术及其应用．北京：清华大学出版社，1999
7 杨育红．LON 网络控制技术及应用．西安：西安电子科技大学出版社，1999
8 孙自强．生产过程自动化及仪表．上海：华东理工大学出版社，1999
9 周振环，凌志浩等．一种新颖的分布式控制网络．自动化仪表，1997，18(7)
10 孙来业．基于 LonWorks 的分散智能控制系统的研究．工业控制计算机，1999，12(6)：36～38
11 凌志浩等．基于 Neuron 芯片的远程数据采集装置的设计．计算机自动测量与控制，1999,7(4)：26～29
12 吴誉，张卫东．现场总线技术展望及我国的发展策略．测控技术，1999，18(11)
13 刘秉瀚．学习控制系统的设计与实现．自动化仪表，1997，18(10)
14 吴今培．实用时序分析．长沙：湖南科学技术出版社，1989
15 T. Soderstrom, M. Mossberg Performance evaluation of methods for identifying continuous-time. Autoregressive processes. Automatica, 2000, 36(1): 53～59
16 杨位钦、顾岚．时间序列分析与动态数据建模．北京：北京工业学院出版社，1986．
17 Doglas M．Bates 等著．非线性回归分析及其应用(韦博成 万方焕等译)．北京：中国统计出版社 1997．
18 J. Mari et al. A covariance extension approach to identification of time series. Automatica, 2000, 36(2): 379～398
19 武宏伟，戴琼海等．基于神经网络的非线性系统的观测器设计．清华大学学报，2000，40(3)：44～47
20 焦李成．神经网络系统理论．西安：西安电子科技大学出版社，1990
21 何耀华，夏志忠．BP网络的快速自适应学习算法．系统工程理论与实践，2000(1)：93～98
22 张立明．人工神经网络的模型及其应用．上海：复旦大学出版社，1993
23 G. J. Kulawski, M. A. Brdys. Stable adaptive control with recurrent networks Automatica, 2000, 36(1): 5～22
24 沈清，胡德文等．神经网络应用技术．长沙：国防科技大学出版社，1993
25 C. Kambhampati et al. A stable one-step-ahead predictive control of non-linear system. Automatica, 2000, 36(2): 485～495
26 王玉涛，周清波等．基于神经网络的时间序列预测算法及其应用．信息与控制，1998，27(6)：413～417
27 王清印，崔援民．不确定性系统预测理论的基本问题．系统工程理论与实践 2000(3)：93～98
28 Tianhao Tang, et al. A RNN-based adaptive predictor for fault prediction and incipient diagnosis. Proceedings of UKACC International Conference on Control 2000，University Cambridge, UK, 4-7, Sept. 2000
29 王科俊，王克成．神经网络建模、预报与控制．哈尔滨：哈尔滨工程大学出版社，1996
30 沈轶等．时间离散的回归神经网络的定性分析．系统工程理论与实践，2000(1)：57～61
31 Box, G. E. P. and G. M. Jenkins. Time series analysis of forecasting and control Holden-day, San Francisco, 1970
32 汤天浩，刘以建等．一种基于人工神经网络在线学习的自适应预测方法．计算机辅助工程 1997，6(3)：13～18
33 Ku C C and Lee K Y. Diagonal recureent neural networks and robust time series prediction.. IEEE trans, on NN, 1994, 5(2)
34 Kang Li and Stephen Thompson. A Cascaded Neural Network and Its Application to Modelling Power Plant Pollutant Emission. Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun 28-July 2,2000 Hefei China IEEE Conference Publication No.992-997.
35 Tianhao Tang, M. Xiong ANN-based Nonlinear Timer Series Models in Fault Detection and Prediction. IFAC confrence CAMS'98 Control Application in Matin System, 1998
36 Pandit, S. M. and S. M. Wu. Time series and system analysis with applications, the United States: John Wiley and sons. 1983
37 Wittenmark, B. A self-tuning predictor. IEEE Trans. Autom, Cont, 1974, 19 (6): 848-851
38 Keyser, R.M.C. and van A.R. Cauwenberghe, A self-tuning multistep predictor application. Automatica, 1984, 17(1): 167-174
39 姜启源，孙义军．ARMAX模型两种自适应 K 步超前预报形式的比较．自动化学报，1993，19(3)：544-551
40 Hush, D., et al. Recursive neural networks for signal processing and control. in Proc. of 1st IEEE-SP Workshop on Neural Networks for Signal Processing, Princeton, NJ, 1991: 523-532
41 Lou Falong and Bao Zheng. Neural network for a class of nonlinear adaptive filters. Advances on Modeling and Simulation, 1992, 31 (1): 109-120
42 冯恩波等．一种基于时序预报神经网络的故障预报方法及其应用．自动化学报，1995，12(3)：348-352
43 Angeline, P. J., et al. An evolutionary algorithm that constructs recurrent neural networks. IEEE Trans. on Neural Networks. 1994, 5 (1): 54-65
44 McDonnell, J. R. and D. Waagen. Evolving recurrent perceptrons for time-series modeling. IEEE Trans. on Neural Network. 1994, 5, (1): 24-37
45 Parlos, A. G. et al. Application of recurrent multilayer perceptron in modeling complex process dynamics. IEEE Trans. on Neural Networks. 1994, 5, (2): 255-266
46 Connor, J. T. et al. Recurrent neural networks and robust time series prediction. IEEE Trans. on Neural Networks, 1994, 5, (2): 240-254
47 Tang Tianhao, et al. An integrated intelligent method for marine monitoring system Preprints of 14th World Congress of IFAC, vol. Q, pp, 55～60, Beijing, China, 1999
48 Williams, R. J. et al. An efficient gradient-based algorithm for on-line training of recurrent network trajectories. Neural Computation, 1990, 2 (4): 490-501
49 Li Jieren, and J. Mindykowski, Development of Engine Room Monitoring and Alarm System, Proceedings of the 1997 International Marine Electrotechnology Conference and Exhibition IMECE'97, China, Shanghai 1997: 131～135
50 刘辉，赵永生．船舶机舱监测报警系统的设计．大连海事大学学报．1999，25(4)：65～69
51 王新全．船舶柴油机微机监测系统及操作管理．北京：人民交通出版社，1995
52 曹善文．机舱自动化与微型计算机及其管理．北京：人民交通出版社，1993
53 凌峰，陈祥献等．工业测控软件中报警系统的设计与实现．测控技术，2000，19(2)：17～20
54 胡武，毕建权等．计算机监控软件中实时数据库系统的研制．测控技术，2000，19(2)：21～23
55 钱为民．测控软件发展的一些思考．测控技术，2000，19(2)：6～8