WO2021114609A1 - 带有非对称信息的网络系统的最优分布式控制方法 - Google Patents

带有非对称信息的网络系统的最优分布式控制方法 Download PDF

Info

Publication number
WO2021114609A1
WO2021114609A1 PCT/CN2020/098449 CN2020098449W WO2021114609A1 WO 2021114609 A1 WO2021114609 A1 WO 2021114609A1 CN 2020098449 W CN2020098449 W CN 2020098449W WO 2021114609 A1 WO2021114609 A1 WO 2021114609A1
Authority
WO
WIPO (PCT)
Prior art keywords
controller
optimal
equations
equation
observation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/098449
Other languages
English (en)
French (fr)
Inventor
梁笑
张琦妍
卢晓
王海霞
张桂林
盛春阳
张治国
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong University of Science and Technology
Original Assignee
Shandong University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shandong University of Science and Technology filed Critical Shandong University of Science and Technology
Publication of WO2021114609A1 publication Critical patent/WO2021114609A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00Program-control systems
    • G05B19/02Program-control systems electric
    • G05B19/418Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
    • G05B19/41845Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by system universality, reconfigurability, modularity
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/30Nc systems
    • G05B2219/33Director till display
    • G05B2219/33273DCS distributed, decentralised controlsystem, multiprocessor
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Definitions

  • the invention relates to the field of control of network systems, in particular to an optimal distributed control method of a network system with asymmetric information.
  • the network control system is a closed-loop control system composed of a communication network.
  • control signals and observation signals can be interactively transmitted through system components.
  • the network control system has many advantages, such as low energy consumption, low maintenance cost and high flexibility.
  • the network control system can be divided into two kinds of network forms: centralized network and distributed network.
  • the centralized network has only one feedback closed-loop control loop, and all observation signals must be transmitted to a centralized controller, which will bring a great calculation burden to the centralized controller.
  • the centralized controller fails, The entire system will collapse.
  • another network form namely distributed networks, is studied.
  • the distributed network has multiple closed-loop control loops and multiple autonomous controllers, the computational burden of each controller is greatly reduced, and when one of the controllers fails, the other closed-loop circuits will not be affected.
  • Related research has proposed a new type of network framework that includes shared historical information. Under this network framework, some historical information (historical control information and historical observation information) can be shared between controllers. Based on this network framework, many research results have emerged. By assuming that the form of the system satisfies a particular form, the relevant research results give the optimal linear control strategy for the network. For this network framework, relevant researchers considered the optimal control problem with local controllers and remote controllers, and gave the optimal control strategy using dynamic programming.
  • the purpose of the present invention is to solve the above shortcomings and propose an optimal distributed control method for a network system with asymmetric information.
  • the method uses asymmetric observation information to provide optimal filters for two controllers. Using the principle of Poundriakin maximum, the solution based on the forward-backward stochastic difference equation is obtained. Based on this solution, the optimal controller with the smallest performance index is given. Finally, according to the special form of the optimal controller, The separation principle separates the optimal filter and the optimal controller and calculates offline.
  • the optimal distributed control method for a network system with asymmetric information includes:
  • Is the status signal Is the control signal of controller C1, Is the control signal of controller C2, with Is the observation signal of sensor 1 and sensor 2
  • c, D 1 , D 2 , F 1 , F 2 are deterministic matrices with appropriate dimensions, with Is the system noise and the observation noise, the mean value is zero, the covariance is Q ⁇ ,
  • the mean value of the initial state r 0 is ⁇ , and the variance is ⁇ , r 0 , ⁇ k , with They are all Gaussian and independent of each other.
  • the corresponding performance index is equation (4),
  • Q, T 1 , T 2 and P N+1 are positive semi-definite matrices, and E is for the random process ⁇ k ⁇ , And random variable r 0 to take the expected value;
  • the observation equation corresponding to the controller C1 is equation (5)
  • the optimal filter of the controller C2 is (20, (21),
  • the estimated error covariance (22) is calculated as:
  • Equation (22) is expressed as Equation (49)
  • the system state controlled by this control method has little fluctuation and has better performance than the system state controlled by a centralized controller.
  • Figure 1 is a comparison diagram of the influence of a centralized controller and a distributed controller on the system
  • Figure 2 is a comparison chart of estimated error covariance
  • the optimal distributed control method for a network system with asymmetric information includes:
  • Is the status signal Is the control signal of controller C1, Is the control signal of controller C2, with Is the observation signal of sensor 1 and sensor 2
  • C, D 1 , D 2 , F 1 , F 2 are deterministic matrices with appropriate dimensions, with Is the system noise and the observation noise, the mean value is zero, the covariance is Q ⁇ ,
  • the mean value of the initial state r 0 is ⁇ , and the variance is ⁇ , r 0 , ⁇ k , with They are all Gaussian and independent of each other.
  • the corresponding performance index is equation (4),
  • Q, T 1 , T 2 and P N+1 are positive semi-definite matrices, and E is for the random process ⁇ k ⁇ , And random variable r 0 to take the expected value;
  • the observation equation corresponding to the controller C1 is equation (5)
  • ⁇ k is the adjoint variable, P N+1 terminal matrix
  • equations (1) and (4) are rewritten into equations (14) and (15)
  • the estimated error covariance (22) is calculated as:
  • Equation (22) is expressed as Equation (49)

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Manufacturing & Machinery (AREA)
  • Quality & Reliability (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)

Abstract

一种带有非对称信息的网络系统的最优分布式控制方法,涉及网络系统的控制领域。首先利用非对称观测信息,给出关于两个控制器的最优滤波器,其次利用庞德里亚金极大值原理,得到了基于正倒向随机差分方程的解,基于该解,给出了使得性能指标最小的最优控制器,最后根据最优控制器的特殊形式,利用分离原理将最优滤波器和最优控制器进行分离并离线计算。

Description

带有非对称信息的网络系统的最优分布式控制方法 技术领域
本发明涉及网络系统的控制领域,具体涉及一种带有非对称信息的网络系统的最优分布式控制方法。
背景技术
网络控制系统是通过通信网络组成的闭环控制系统,在这样的网络系统中,控制信号和观测信号可以通过系统组件进行交互传输。与传统的点对点反馈控制系统相比,网络控制系统具有很多优势,比如低能耗、低维护成本以及高灵活性等。网络控制系统又可分为集中式网络和分布式网络两种网络形式。顾名思义,集中式网络只有一个反馈闭环控制回路,所有的观测信号都要传递给一个集中控制器,这样就会给集中控制器带来很大的计算负担,此外,当集中控制器发生故障时,整个系统就会崩溃。鉴于集中式网络的这些缺点,研究了另一种网络形式,即分布式网络。
由于分布式网络具有多个闭环控制回路以及多个自主控制器,这就使得每个控制器的计算负担大大减少,并且当其中一个控制器出现故障时,其他闭环回路也不会受影响。鉴于分布式网络的这些优点,近年来,越来越多的研究人员开始研究分布式网络的控制问题。相关研究提出了一种新型的包含共享历史信息的网络框架,在这种网络框架下,控制器之间可以共享部分历史信息(历史控制信息和历史观测信息)。基于这种网络框架,很多研究成果应运而生。通过假设系统形式满足某种特殊形式,相关研究成果给出了针对该网络的最优线性控制策略。针对该网络框架,相关研究人员考虑了带有当地控制器和远程控制器的最优控制问题,利用动态规划的方法给出了最优控制策略。
但是以往的研究工作都没有考虑噪声的影响,在现实生活中,噪声是无处不在的,如果忽略噪声的影响,会使得设计的控制器不能很好地控制现实系统,进而影响控制效果。
发明内容
本发明的目的是针对上述不足,提出了一种带有非对称信息的网络系统的最优分布式控制方法,该方法利用非对称观测信息,给出关于两个控制器的最优滤波器,利用庞德里亚金极大值原理,得到了基于正倒向随机差分方程的解,基于该解,给出了使得性能指标最小的最优控制器,最后根据最优控制器的特殊形式,利用分离原理将最优滤波器和最优控制器进行分离并离线计算。
本发明具体采用如下技术方案:
带有非对称信息的网络系统的最优分布式控制方法,包括:
定义带有非对称信息的网络系统的数学模型如式(1)-(3)所示,
Figure PCTCN2020098449-appb-000001
Figure PCTCN2020098449-appb-000002
Figure PCTCN2020098449-appb-000003
其中,
Figure PCTCN2020098449-appb-000004
为状态信号,
Figure PCTCN2020098449-appb-000005
为控制器C1的控制信号,
Figure PCTCN2020098449-appb-000006
为控制器C2的控制信号,
Figure PCTCN2020098449-appb-000007
Figure PCTCN2020098449-appb-000008
为传感器1和传感器2的观测信号,c,D 1,D 2,F 1,F 2为具有合适维数的确定矩阵,
Figure PCTCN2020098449-appb-000009
Figure PCTCN2020098449-appb-000010
为系统噪声和观测噪声,均值均为零,协方差为Q θ
Figure PCTCN2020098449-appb-000011
初始状态r 0的均值为μ,方差为∑,r 0,θ k
Figure PCTCN2020098449-appb-000012
Figure PCTCN2020098449-appb-000013
都是高斯的,且相互独立,对应的性能指标为式(4),
Figure PCTCN2020098449-appb-000014
其中,Q,T 1,T 2和P N+1是半正定矩阵,E是对随机过程{θ k},
Figure PCTCN2020098449-appb-000015
和随机变量r 0取期望值;
控制器C1获得传感器1和传感器2的观测信息,也就是
Figure PCTCN2020098449-appb-000016
Figure PCTCN2020098449-appb-000017
控制器C2获得传感器2的观测信息,即
Figure PCTCN2020098449-appb-000018
定义Z k={z 0,...,z k}作为控制器C1的观测信息,其中
Figure PCTCN2020098449-appb-000019
控制器C1对应的观测方程为式(5)
z k=Fr k+w k    (5)
其中F=[F 1′ F 2′]′和
Figure PCTCN2020098449-appb-000020
控制器C2对应的观测方程为式(3);
确定
Figure PCTCN2020098449-appb-000021
可测量的控制器
Figure PCTCN2020098449-appb-000022
Figure PCTCN2020098449-appb-000023
可测量的控制器
Figure PCTCN2020098449-appb-000024
使得式(4)所示的性能指标最小。
优选地,确定式(4)所示的性能指标最小的过程中,利用庞德里亚金极大值原理,将式(1)和(4)改写成式(14)和(15)
Figure PCTCN2020098449-appb-000025
Figure PCTCN2020098449-appb-000026
其中,D=[D 1 D 2],
Figure PCTCN2020098449-appb-000027
基于观测信号{z 0,...,z k},对于(1)和(5),控制器C1的最优滤波器为式(16)、(17)
Figure PCTCN2020098449-appb-000028
Figure PCTCN2020098449-appb-000029
其中,
Figure PCTCN2020098449-appb-000030
是估计误差协方差,并且满足式(18)、(19)
Figure PCTCN2020098449-appb-000031
Figure PCTCN2020098449-appb-000032
初始值为
Figure PCTCN2020098449-appb-000033
Figure PCTCN2020098449-appb-000034
基于观测信号
Figure PCTCN2020098449-appb-000035
对于(1)和(3),控制器C2的最优滤波器为式(20、(21),
Figure PCTCN2020098449-appb-000036
Figure PCTCN2020098449-appb-000037
其中
Figure PCTCN2020098449-appb-000038
是估计误差协方差,并且满足式(22)、(23)
Figure PCTCN2020098449-appb-000039
Figure PCTCN2020098449-appb-000040
初始值为
Figure PCTCN2020098449-appb-000041
Figure PCTCN2020098449-appb-000042
基于式(22),可知估计误差协方差
Figure PCTCN2020098449-appb-000043
与控制器
Figure PCTCN2020098449-appb-000044
是耦合的,通过计算,将
Figure PCTCN2020098449-appb-000045
Figure PCTCN2020098449-appb-000046
中解耦出来。
优选的,将
Figure PCTCN2020098449-appb-000047
Figure PCTCN2020098449-appb-000048
中解耦过程如下:
在给出最优控制器之前,首先将伴随方程(6)-(9)改写成(24)-(27)所示的方程
Figure PCTCN2020098449-appb-000049
Figure PCTCN2020098449-appb-000050
Figure PCTCN2020098449-appb-000051
Figure PCTCN2020098449-appb-000052
给出下面的耦合黎卡提方程:
Figure PCTCN2020098449-appb-000053
Figure PCTCN2020098449-appb-000054
其中,
K k=D′G k+1C,     (30)
γ k=D′G k+1D+T,     (31)
Figure PCTCN2020098449-appb-000055
Figure PCTCN2020098449-appb-000056
L k=D 1′Φ k+1C,     (34)
Λ k=D 1′Φ k+1D 1+T 1,    (35)
终端值为G N+1=Δ N+1=P N+1
然后给出最优分布式控制器的形式以及求解方法:
当k=N,...,0时,假设γ k和Λ k是可逆的,那么使得性能指标(4)最小的最优分布式控制器为(36)、(37):
Figure PCTCN2020098449-appb-000057
Figure PCTCN2020098449-appb-000058
相应的最优控制器
Figure PCTCN2020098449-appb-000059
最优控制器
Figure PCTCN2020098449-appb-000060
并且正倒向方程(14)和(24)的解满足式(38)
Figure PCTCN2020098449-appb-000061
估计误差协方差(22)计算为:
Figure PCTCN2020098449-appb-000062
其中
Figure PCTCN2020098449-appb-000063
Figure PCTCN2020098449-appb-000064
迭代计算;
式(22)表示为式(49)
Figure PCTCN2020098449-appb-000065
由此可见,
Figure PCTCN2020098449-appb-000066
Figure PCTCN2020098449-appb-000067
中解耦出来。
优选地,利用庞德里亚金极大值原理的具体计算过程为,给出如下所示的伴随方程:
Figure PCTCN2020098449-appb-000068
Figure PCTCN2020098449-appb-000069
Figure PCTCN2020098449-appb-000070
Figure PCTCN2020098449-appb-000071
其中λ k是伴随变量,P N+1为终端矩阵;
鉴于控制器
Figure PCTCN2020098449-appb-000072
的适应性,给出如下定义:
Figure PCTCN2020098449-appb-000073
Figure PCTCN2020098449-appb-000074
相应的
Figure PCTCN2020098449-appb-000075
Figure PCTCN2020098449-appb-000076
有如下性质:
Figure PCTCN2020098449-appb-000077
Figure PCTCN2020098449-appb-000078
本发明具有如下有益效果:
受该控制方法控制的系统状态波动很小,比受集中式控制器作用的系统状态拥有更好的性能。
附图说明
图1为集中式控制器与分布式控制器对系统影响对比图;
图2为估计误差协方差对比图
Figure PCTCN2020098449-appb-000079
and
Figure PCTCN2020098449-appb-000080
具体实施方式
带有非对称信息的网络系统的最优分布式控制方法,包括:
定义带有非对称信息的网络系统的数学模型如式(1)-(3)所示,
Figure PCTCN2020098449-appb-000081
Figure PCTCN2020098449-appb-000082
Figure PCTCN2020098449-appb-000083
其中,
Figure PCTCN2020098449-appb-000084
为状态信号,
Figure PCTCN2020098449-appb-000085
为控制器C1的控制信号,
Figure PCTCN2020098449-appb-000086
为控制器C2的控制信号,
Figure PCTCN2020098449-appb-000087
Figure PCTCN2020098449-appb-000088
为传感器1和传感器2的观测信号,C,D 1,D 2,F 1,F 2为具有合适维 数的确定矩阵,
Figure PCTCN2020098449-appb-000089
Figure PCTCN2020098449-appb-000090
为系统噪声和观测噪声,均值均为零,协方差为Q θ
Figure PCTCN2020098449-appb-000091
初始状态r 0的均值为μ,方差为∑,r 0,θ k
Figure PCTCN2020098449-appb-000092
Figure PCTCN2020098449-appb-000093
都是高斯的,且相互独立,对应的性能指标为式(4),
Figure PCTCN2020098449-appb-000094
其中,Q,T 1,T 2和P N+1是半正定矩阵,E是对随机过程{θ k},
Figure PCTCN2020098449-appb-000095
和随机变量r 0取期望值;
控制器C1获得传感器1和传感器2的观测信息,也就是
Figure PCTCN2020098449-appb-000096
Figure PCTCN2020098449-appb-000097
控制器C2获得传感器2的观测信息,即
Figure PCTCN2020098449-appb-000098
定义Z k={z 0,...,z k}作为控制器C1的观测信息,其中
Figure PCTCN2020098449-appb-000099
控制器C1对应的观测方程为式(5)
z k=Fr k+w k     (5)
其中F=[F 1′ F 2′]′和
Figure PCTCN2020098449-appb-000100
控制器C2对应的观测方程为式(3);
确定
Figure PCTCN2020098449-appb-000101
可测量的控制器
Figure PCTCN2020098449-appb-000102
Figure PCTCN2020098449-appb-000103
可测量的控制器
Figure PCTCN2020098449-appb-000104
使得式(4)所示的性能指标最小。
利用庞德里亚金极大值原理的具体计算过程为,给出如下所示的伴随方程:
Figure PCTCN2020098449-appb-000105
Figure PCTCN2020098449-appb-000106
Figure PCTCN2020098449-appb-000107
Figure PCTCN2020098449-appb-000108
其中λ k是伴随变量,P N+1终端矩阵;
鉴于控制器
Figure PCTCN2020098449-appb-000109
的适应性,给出如下定义:
Figure PCTCN2020098449-appb-000110
Figure PCTCN2020098449-appb-000111
相应的
Figure PCTCN2020098449-appb-000112
Figure PCTCN2020098449-appb-000113
有如下性质:
Figure PCTCN2020098449-appb-000114
Figure PCTCN2020098449-appb-000115
确定式(4)所示的性能指标最小的过程中,利用庞德里亚金极大值原理,将式(1)和(4)改写成式(14)和(15)
Figure PCTCN2020098449-appb-000116
Figure PCTCN2020098449-appb-000117
其中,D=[D 1 D 2],
Figure PCTCN2020098449-appb-000118
基于观测信号{z 0,...,z k},对于(1)和(5),控制器C1的最优滤波器为式(16)、(17)
Figure PCTCN2020098449-appb-000119
Figure PCTCN2020098449-appb-000120
其中,
Figure PCTCN2020098449-appb-000121
是估计误差协方差,并且满足式(18)、(19)
Figure PCTCN2020098449-appb-000122
Figure PCTCN2020098449-appb-000123
初始值为
Figure PCTCN2020098449-appb-000124
Figure PCTCN2020098449-appb-000125
基于观测信号
Figure PCTCN2020098449-appb-000126
对于(1)和(3),控制器C2的最优滤波器为式(20、(21),
Figure PCTCN2020098449-appb-000127
Figure PCTCN2020098449-appb-000128
其中
Figure PCTCN2020098449-appb-000129
是估计误差协方差,并且满足式(22)、(23)
Figure PCTCN2020098449-appb-000130
Figure PCTCN2020098449-appb-000131
初始值为
Figure PCTCN2020098449-appb-000132
Figure PCTCN2020098449-appb-000133
基于式(22),可知估计误差协方差
Figure PCTCN2020098449-appb-000134
与控制器
Figure PCTCN2020098449-appb-000135
是耦合的,通过计算,将
Figure PCTCN2020098449-appb-000136
Figure PCTCN2020098449-appb-000137
中解耦出来。
优选的,将
Figure PCTCN2020098449-appb-000138
Figure PCTCN2020098449-appb-000139
中解耦过程如下:
在给出最优控制器之前,首先对伴随方程(6)-(9)改写成(24)-(27)所示的方程
Figure PCTCN2020098449-appb-000140
Figure PCTCN2020098449-appb-000141
Figure PCTCN2020098449-appb-000142
Figure PCTCN2020098449-appb-000143
给出下面的耦合黎卡提方程:
Figure PCTCN2020098449-appb-000144
Figure PCTCN2020098449-appb-000145
其中,
K k=D′G k+1C,    (30)
γ k=D′G k+1D+T,    (31)
Figure PCTCN2020098449-appb-000146
Figure PCTCN2020098449-appb-000147
L k=D 1′Φ k+1C,     (34)
Λ k=D 1′Φ k+1D 1+T 1,      (35)
终端值为G N+1=Δ N+1=P N+1
然后给出最优分布式控制器的形式以及求解方法:
假设γ k和Λ k是可逆的,用数学归纳法证明最优控制器满足(36)和(37),伴随变量1β k-1满足(38)。从(27)和G N+1=Δ N+1=P N+1,可以得到当k=N+1时,(38)成立;
当k=N时,利用(14),(27)和(12),(25)变为
Figure PCTCN2020098449-appb-000148
基于(30)和(31),最优控制器g N计算为式(39)
Figure PCTCN2020098449-appb-000149
因此当k=N时,(36)成立,利用(14),(27)和(13),(26)变为
Figure PCTCN2020098449-appb-000150
利用(34)和(35),最优控制器
Figure PCTCN2020098449-appb-000151
计算为
Figure PCTCN2020098449-appb-000152
因此当k=N时,(37)成立;
利用(14),(39),(40)和(13),(24)可以写成
Figure PCTCN2020098449-appb-000153
利用(28)和(29),可以得到当k=N时,(38)成立;
利用数学归纳法,选取任意的l满足0≤l≤N,假设β k-1,g k
Figure PCTCN2020098449-appb-000154
满足(38),(36)和(37)对于所有的k≥l+1,现在将要证明(38),(36)和(37)在k=l时是成立的,首先,给出如下准备工作:
利用(20),(12)和(14)可以得到
Figure PCTCN2020098449-appb-000155
使用(16),(20),(12),(13),(14)和(41)可以得到
Figure PCTCN2020098449-appb-000156
由于(38)在k≥l+1时始终成立,当k=l+1时可以得到
Figure PCTCN2020098449-appb-000157
利用(41),(42),(43)和(12),(25)可以写成
Figure PCTCN2020098449-appb-000158
使用(30)和(31),最优控制器g l计算为式(44)
Figure PCTCN2020098449-appb-000159
因此(36)在k=l时成立,利用(41),(42),(43),(12)和(13),(26)计算为
Figure PCTCN2020098449-appb-000160
Figure PCTCN2020098449-appb-000161
利用(32)-(35),最优控制器
Figure PCTCN2020098449-appb-000162
可以计算为式(45)
Figure PCTCN2020098449-appb-000163
因此(37)在k=l时成立;
利用(41),(42),(43),(13),(44)和(45),(24)变为
Figure PCTCN2020098449-appb-000164
利用(28),(29),(30)和(32),可以得到(38)在k=l时成立,证明完毕;
很明显可以看到
Figure PCTCN2020098449-appb-000165
(22)耦合着
Figure PCTCN2020098449-appb-000166
注意到(37)的特殊形式以及(37)和(22)的关系,在以下计算中成功的将
Figure PCTCN2020098449-appb-000167
Figure PCTCN2020098449-appb-000168
中解耦出来。
估计误差协方差(22)计算为:
Figure PCTCN2020098449-appb-000169
其中
Figure PCTCN2020098449-appb-000170
Figure PCTCN2020098449-appb-000171
迭代计算;
从(22)中可以看到耦合项为
Figure PCTCN2020098449-appb-000172
Figure PCTCN2020098449-appb-000173
首先,可以计算得到
Figure PCTCN2020098449-appb-000174
同样可以计算得到
Figure PCTCN2020098449-appb-000175
式(22)表示为式(49)
Figure PCTCN2020098449-appb-000176
由此可见,
Figure PCTCN2020098449-appb-000177
Figure PCTCN2020098449-appb-000178
中解耦出来。
设定系统(3),(5),(14)和性能指标(15)的参数如下所示
C=2.7,D 1=1.2,D=[1.2 1.1],
Figure PCTCN2020098449-appb-000179
Figure PCTCN2020098449-appb-000180
相应的初始值和终端值为
Figure PCTCN2020098449-appb-000181
在图1中画出了分别受集中式控制器和分布式控制器作用的系统状态,从中可以看到受分布式控制器控制的系统状态波动很小,比受集中式控制器作用的系统状态拥有更好的性能。
从图2中可以看到,两个估计误差协方差都是渐进稳定的,估计器1的协方差比估计器2的协方差要小,说明估计器1的估计效果更好。
当然,上述说明并非是对本发明的限制,本发明也并不仅限于上述举例,本技术领域的技术人员在本发明的实质范围内所做出的变化、改型、添加或替换,也应属于本发明的保护范围。

Claims (4)

  1. 带有非对称信息的网络系统的最优分布式控制方法,其特征在于,包括:
    定义带有非对称信息的网络系统的数学模型如式(1)-(3)所示,
    Figure PCTCN2020098449-appb-100001
    Figure PCTCN2020098449-appb-100002
    Figure PCTCN2020098449-appb-100003
    其中,
    Figure PCTCN2020098449-appb-100004
    为状态信号,
    Figure PCTCN2020098449-appb-100005
    为控制器C1的控制信号,
    Figure PCTCN2020098449-appb-100006
    为控制器C2的控制信号,
    Figure PCTCN2020098449-appb-100007
    Figure PCTCN2020098449-appb-100008
    为传感器1和传感器2的观测信号,C,D 1,D 2,F 1,F 2为具有合适维数的确定矩阵,
    Figure PCTCN2020098449-appb-100009
    Figure PCTCN2020098449-appb-100010
    为系统噪声和观测噪声,均值均为零,协方差为Q θ
    Figure PCTCN2020098449-appb-100011
    初始状态r 0的均值为μ,方差为∑,r 0,θ k
    Figure PCTCN2020098449-appb-100012
    Figure PCTCN2020098449-appb-100013
    都是高斯的,且相互独立,对应的性能指标为式(4),
    Figure PCTCN2020098449-appb-100014
    其中,Q,T 1,T 2和P N+1是半正定矩阵,E是对随机过程{θ k},
    Figure PCTCN2020098449-appb-100015
    和随机变量r 0取期望值;
    控制器C1获得传感器1和传感器2的观测信息,也就是
    Figure PCTCN2020098449-appb-100016
    Figure PCTCN2020098449-appb-100017
    控制器C2获得传感器2的观测信息,即
    Figure PCTCN2020098449-appb-100018
    定义Z k={z 0,...,z k}作为控制器C1的观测信息,其中
    Figure PCTCN2020098449-appb-100019
    控制器C1对应的观测方程为式(5)
    z k=Fr k+w k  (5)
    其中F=[f 1′ F 2′]′和
    Figure PCTCN2020098449-appb-100020
    控制器C2对应的观测方程为式(3);
    确定
    Figure PCTCN2020098449-appb-100021
    可测量的控制器
    Figure PCTCN2020098449-appb-100022
    Figure PCTCN2020098449-appb-100023
    可测量的控制器
    Figure PCTCN2020098449-appb-100024
    使得式(4)所示的性能指标最小。
  2. 如权利要求1所示的带有非对称信息的网络系统的最优分布式控制方法,其特征在于,使得式(4)所示的性能指标最小的过程中,利用庞德里亚金极大值原理,将式(1)和(4)改写成式(14)和(15):
    Figure PCTCN2020098449-appb-100025
    Figure PCTCN2020098449-appb-100026
    其中,D=[D 1 D 2],
    Figure PCTCN2020098449-appb-100027
    基于观测信号{z 0,...,z k},对于(1)和(5),控制器C1的最优滤波器为式(16)、(17)
    Figure PCTCN2020098449-appb-100028
    Figure PCTCN2020098449-appb-100029
    其中,
    Figure PCTCN2020098449-appb-100030
    是估计误差协方差,并且满足式(18)、(19)
    Figure PCTCN2020098449-appb-100031
    Figure PCTCN2020098449-appb-100032
    初始值为
    Figure PCTCN2020098449-appb-100033
    Figure PCTCN2020098449-appb-100034
    基于观测信号
    Figure PCTCN2020098449-appb-100035
    对于(1)和(3),控制器C2的最优滤波器为式(20)、(21),
    Figure PCTCN2020098449-appb-100036
    Figure PCTCN2020098449-appb-100037
    其中
    Figure PCTCN2020098449-appb-100038
    是估计误差协方差,并且满足式(22)、(23):
    Figure PCTCN2020098449-appb-100039
    Figure PCTCN2020098449-appb-100040
    初始值为
    Figure PCTCN2020098449-appb-100041
    Figure PCTCN2020098449-appb-100042
    基于式(22),可知估计误差协方差
    Figure PCTCN2020098449-appb-100043
    与控制器
    Figure PCTCN2020098449-appb-100044
    是耦合的,通过计算,将
    Figure PCTCN2020098449-appb-100045
    Figure PCTCN2020098449-appb-100046
    中解耦出来。
  3. 如权利要求1所示的带有非对称信息的网络系统的最优分布式控制方法,其特征在于,将
    Figure PCTCN2020098449-appb-100047
    Figure PCTCN2020098449-appb-100048
    中解耦,过程如下:
    在给出最优控制器之前,首先将伴随方程(6)-(9)改写成(24)-(27)所示的方程
    Figure PCTCN2020098449-appb-100049
    Figure PCTCN2020098449-appb-100050
    Figure PCTCN2020098449-appb-100051
    Figure PCTCN2020098449-appb-100052
    给出下面的耦合黎卡提方程:
    Figure PCTCN2020098449-appb-100053
    Figure PCTCN2020098449-appb-100054
    其中,
    K k=D′G k+1C,   (30)
    γ k=D′G k+1D+T,   (31)
    Figure PCTCN2020098449-appb-100055
    Figure PCTCN2020098449-appb-100056
    L k=D 1′Φ k+1C,   (34)
    Λ k=D 1′Φ k+1D 1+T 1,   (35)
    终端值为G N+1=Δ N+1=P N+1
    然后给出最优分布式控制器的形式以及求解方法:
    当k=N,...,0时,假设γ k和Λ k是可逆的,那么使得性能指标(4)最小的最优分布式控制器为(36)、(37):
    Figure PCTCN2020098449-appb-100057
    Figure PCTCN2020098449-appb-100058
    相应的最优控制器
    Figure PCTCN2020098449-appb-100059
    最优控制器
    Figure PCTCN2020098449-appb-100060
    并且正倒向方程(14)和(24)的解满足式(38)
    Figure PCTCN2020098449-appb-100061
    估计误差协方差(22)计算为:
    Figure PCTCN2020098449-appb-100062
    由此可见,
    Figure PCTCN2020098449-appb-100063
    Figure PCTCN2020098449-appb-100064
    中解耦出来。
  4. 如权利要求2所示的带有非对称信息的网络系统的最优分布式控制方法,其特征在于,利用庞德里亚金极大值原理,给出如下所示的伴随方程:
    Figure PCTCN2020098449-appb-100065
    Figure PCTCN2020098449-appb-100066
    Figure PCTCN2020098449-appb-100067
    Figure PCTCN2020098449-appb-100068
    其中λ k是伴随变量,P N+1为终端矩阵;
    鉴于控制器
    Figure PCTCN2020098449-appb-100069
    的适应性,给出如下定义:
    Figure PCTCN2020098449-appb-100070
    Figure PCTCN2020098449-appb-100071
    相应的
    Figure PCTCN2020098449-appb-100072
    Figure PCTCN2020098449-appb-100073
    有如下性质:
    Figure PCTCN2020098449-appb-100074
    Figure PCTCN2020098449-appb-100075
PCT/CN2020/098449 2019-12-11 2020-06-28 带有非对称信息的网络系统的最优分布式控制方法 Ceased WO2021114609A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911262510.5 2019-12-11
CN201911262510.5A CN111123860B (zh) 2019-12-11 2019-12-11 带有非对称信息的网络系统的最优分布式控制方法

Publications (1)

Publication Number Publication Date
WO2021114609A1 true WO2021114609A1 (zh) 2021-06-17

Family

ID=70498411

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/098449 Ceased WO2021114609A1 (zh) 2019-12-11 2020-06-28 带有非对称信息的网络系统的最优分布式控制方法

Country Status (2)

Country Link
CN (1) CN111123860B (zh)
WO (1) WO2021114609A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111123860B (zh) * 2019-12-11 2020-11-10 山东科技大学 带有非对称信息的网络系统的最优分布式控制方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106325069A (zh) * 2016-08-28 2017-01-11 北京工业大学 一种无线网络控制系统最优线性控制策略设计方法
CN108153259A (zh) * 2017-12-21 2018-06-12 北京工业大学 一种基于卡尔曼滤波的多控制器最优状态估计控制策略设计方法
CN111123860A (zh) * 2019-12-11 2020-05-08 山东科技大学 带有非对称信息的网络系统的最优分布式控制方法

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102352743A (zh) * 2011-10-12 2012-02-15 中国石油大学(华东) 基于最优控制理论的聚合物驱方案优化设计方法
CN103547033B (zh) * 2013-10-25 2016-01-20 华南理工大学 基于无线传感器执行器网的大型智能灯光节能控制方法
CN104391445B (zh) * 2014-08-06 2017-10-20 华南理工大学 基于观测器的车队协同自主控制方法
US10131341B2 (en) * 2015-06-16 2018-11-20 Mississippi State University Bandwidth-based methodology for controlling and optimally designing a hybrid power system
CN106197428B (zh) * 2016-07-10 2019-03-22 北京工业大学 一种利用测量信息优化分布式ekf估计过程的slam方法
CN107994885B (zh) * 2017-11-21 2021-04-27 河南工业大学 一种同时估计未知输入和状态的分布式融合滤波方法
CN108959808B (zh) * 2018-07-23 2022-05-03 哈尔滨理工大学 一种基于传感器网络的优化分布式状态估计方法
CN110239362B (zh) * 2019-06-19 2020-10-02 北京理工大学 一种分布式电驱动车辆多性能优化力矩分配方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106325069A (zh) * 2016-08-28 2017-01-11 北京工业大学 一种无线网络控制系统最优线性控制策略设计方法
CN108153259A (zh) * 2017-12-21 2018-06-12 北京工业大学 一种基于卡尔曼滤波的多控制器最优状态估计控制策略设计方法
CN111123860A (zh) * 2019-12-11 2020-05-08 山东科技大学 带有非对称信息的网络系统的最优分布式控制方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
12 May 2018, article LIANG, XIAO: "Study on Optimal Estimation and Control for Stochastic Systems with Time Delay and Multiplicative Noises", pages: 1 - 141, XP055820177 *
DING, YAO: "An Asymmetric Coupling Wireless Sensor Network Synchronization Control Model", SIGNAL PROCESSING, vol. 27, no. 11, 1 November 2011 (2011-11-01), pages 1 - 4, XP055820168, ISSN: 1003-0530 *
LI HONGDAN; HAN CHUNYAN; ZHANG HUANSHUI; XIE LIHUA: "Optimal Control and Stabilization for Networked Systems With Input Delay and Markovian Packet Losses", IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS, IEEE, PISCATAWAY, NJ, USA, vol. 51, no. 7, 13 September 2019 (2019-09-13), Piscataway, NJ, USA, pages 4453 - 4465, XP011861184, ISSN: 2168-2216, DOI: 10.1109/TSMC.2019.2938792 *

Also Published As

Publication number Publication date
CN111123860A (zh) 2020-05-08
CN111123860B (zh) 2020-11-10

Similar Documents

Publication Publication Date Title
CN109507885B (zh) 基于自抗扰的无模型自适应auv控制方法
CN115981143A (zh) 一种多智能体系统领导跟随实用二分一致性控制方法
CN114296342B (zh) 一种分布式动态事件触发多智能体系统的一致性控制方法
CN109960149B (zh) 用于pH中和过程的自抗扰控制参数整定方法
CN106774379B (zh) 一种智能超螺旋强鲁棒姿态控制方法
CN110362103B (zh) 分布式自主水下航行器姿态协同优化控制方法
CN104570729A (zh) 一种改进型的Smith预估控制器
CN106054596B (zh) 一种基于改进性能指标的pid控制器参数优化整定方法
CN112666832B (zh) 一种非周期通信的水下滑翔机协同控制器系统及设计方法
CN109976150B (zh) 一类欠驱动多输入多输出系统的集中式自抗扰控制方法
CN116339315B (zh) 面向多智能体输入时滞系统的强化学习编队方法
CN108919639A (zh) 一种pid控制器参数最优比例模型建立方法
CN113110058A (zh) 一种通信受限的多智能体系统二分实用一致性控制方法
CN110609467B (zh) 基于pid的时滞多智能体系统的一致性控制方法
CN109946979A (zh) 一种伺服系统灵敏度函数的自适应调整方法
CN106354010A (zh) 自适应最优控制方法及自适应最优控制系统
CN110497418A (zh) 一种机器人自适应模糊控制方法及系统
CN119087871A (zh) 一种非线性多智能体系统输出反馈一致性控制方法
CN118192224A (zh) 一种规定时间的自适应动态规划控制方法
WO2021114609A1 (zh) 带有非对称信息的网络系统的最优分布式控制方法
CN114185273A (zh) 一种饱和受限下分布式前置时间一致性控制器的设计方法
CN118259673A (zh) 一种无人船预定义时间滑模轨迹跟踪控制方法
CN107102555B (zh) 一种镇定一阶惯性加纯滞后系统的线性自抗扰控制器设计方法
CN112445135B (zh) 非匹配扰动二阶多智能体系统有限时间一致性控制方法
CN104793496B (zh) 二输入二输出聚合反应器的解耦控制系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20900529

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20900529

Country of ref document: EP

Kind code of ref document: A1