CN115128961A - Self-triggering model prediction method and system based on LoRaWAN (Long Range network) - Google Patents

Self-triggering model prediction method and system based on LoRaWAN (Long Range network) Download PDF

Info

Publication number
CN115128961A
CN115128961A CN202211050547.3A CN202211050547A CN115128961A CN 115128961 A CN115128961 A CN 115128961A CN 202211050547 A CN202211050547 A CN 202211050547A CN 115128961 A CN115128961 A CN 115128961A
Authority
CN
China
Prior art keywords
control
triggering
time
self
lorawan
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.)
Granted
Application number
CN202211050547.3A
Other languages
Chinese (zh)
Other versions
CN115128961B (en
Inventor
吕亮
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.)
Ningbo Liangkong Information Technology Co ltd
Original Assignee
Ningbo Liangkong Information Technology Co ltd
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 Ningbo Liangkong Information Technology Co ltd filed Critical Ningbo Liangkong Information Technology Co ltd
Priority to CN202211050547.3A priority Critical patent/CN115128961B/en
Publication of CN115128961A publication Critical patent/CN115128961A/en
Application granted granted Critical
Publication of CN115128961B publication Critical patent/CN115128961B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
    • H04L67/025—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP] for remote control or remote monitoring of applications
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W84/00—Network topologies
    • H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Selective Calling Equipment (AREA)

Abstract

The invention relates to a self-triggering model prediction method and a self-triggering model prediction system based on a LoRaWAN (Long area network), which solve the problems of large calculation burden of a prediction control algorithm of a wireless network control system and high communication energy consumption of a wireless communication network, construct an extension triggering time based on a relaxation dynamic programming condition to the maximum extent based on a finite time domain value function by solving an optimal control input sequence obtained by solving a constrained finite time domain optimal control problem, and reduce the updating times of model prediction control as much as possible; meanwhile, the trigger interval is further adjusted within the self-triggering model predictive control convergence range by combining the communication protocol characteristics of the LoRaWAN network, so that the trigger interval is synchronous with the information receiving window of the controlled object, on one hand, the communication energy consumption is reduced, on the other hand, the effectiveness of the trigger interval of the self-triggering model predictive control is ensured, and thus, the calculated amount and the communication burden are reduced while the model predictive control accuracy is ensured.

Description

Self-triggering model prediction method and system based on LoRaWAN (Long Range network)
Technical Field
The invention relates to the technical field of model prediction control, in particular to a self-triggering model prediction method and system based on a LoRaWAN (LoRaWAN) network.
Background
Model Predictive Control (MPC) is a powerful constrained system optimal control technique, and the basic principle thereof is to solve an open-loop optimal control problem on line at each sampling time to obtain an optimal control sequence, and then apply a first control input element in the optimal control sequence to a controlled system.
LoRaWAN is a set of communication protocol and system architecture based on the supporting design of loRa long-distance communication technique. The existing MPC does not consider the limited distribution quantity of communication, when the controlled object is not in the information receiving window, the MPC information communication fails, and the communication failure causes the control input of the controlled object to exceed the control boundary of the MPC, so that the MPC process cannot be converged, and the accuracy of model prediction control is reduced.
Disclosure of Invention
The invention solves the problem of how to reduce the calculation amount and the communication burden of model predictive control on the basis of ensuring the accuracy of the model predictive control.
In order to solve the above problems, the present invention provides a self-triggering model prediction method based on a LoRaWAN network, including:
step 1, according to the time of the controlled object
Figure 100002_DEST_PATH_IMAGE002
The state input, the control input and the output of the control system construct a linear model of the control system;
step 2, constructing a constrained finite time domain optimal control problem based on a linear model of a control system, and solving the finite time domain optimal control problem to obtain the slave time
Figure 684490DEST_PATH_IMAGE002
Time of arrival
Figure 100002_DEST_PATH_IMAGE004
Optimal control input sequence of
Figure 100002_DEST_PATH_IMAGE006
，
Figure 100002_DEST_PATH_IMAGE008
Representing time
Figure 273735DEST_PATH_IMAGE002
Is input to the state of the mobile terminal,
Figure 100002_DEST_PATH_IMAGE010
representing a prediction step size;
step 3, constructing a finite time domain cost function for solving the finite time domain optimal control problem, and calculating a trigger interval based on a relaxation dynamic programming condition according to the finite time domain cost function
Figure 100002_DEST_PATH_IMAGE012
，
Figure 100002_DEST_PATH_IMAGE014
In order to trigger the moment of time,
Figure 100002_DEST_PATH_IMAGE016
for the state at the moment of triggering, to obtain an optimal control input sequence
Figure 974843DEST_PATH_IMAGE006
The number of control input elements meeting the linear model constraint of the control system;
step 4, calculating the next trigger time based on the calculated trigger interval
Figure 100002_DEST_PATH_IMAGE018
；
Step 5, judging the next trigger time according to the LoRaWAN protocol
Figure 838894DEST_PATH_IMAGE018
Whether the information is in the information receiving window of the controlled object or not, if so, entering a step 6; if not, entering step 7;
step 6, inputting the optimal control into a sequence
Figure 512321DEST_PATH_IMAGE006
Time of flight
Figure 334783DEST_PATH_IMAGE002
To the next trigger moment
Figure 70658DEST_PATH_IMAGE018
The control input element in between is sent to the controlled object and is triggered at the next moment
Figure 315564DEST_PATH_IMAGE018
Requesting the controlled object to collect and report new state input,
Figure 100002_DEST_PATH_IMAGE020
，
Figure 100002_DEST_PATH_IMAGE022
and returning to the step 2;
step 7, receiving window modification and correcting the next trigger time according to the information of the controlled object
Figure 100002_DEST_PATH_IMAGE024
And inputting the optimal control into the sequence
Figure 312339DEST_PATH_IMAGE006
Time of flight
Figure 100002_DEST_PATH_IMAGE026
To the next trigger moment
Figure 100002_DEST_PATH_IMAGE028
The control input element in between is sent to the controlled object and is triggered at the next moment
Figure 649910DEST_PATH_IMAGE028
Requesting the controlled object to collect and report new state input,
Figure 100002_DEST_PATH_IMAGE030
，
Figure 935398DEST_PATH_IMAGE022
and returns to step 2.
The self-triggering model prediction method based on the LoRaWAN network has the beneficial effects that: an optimal control input sequence is obtained by solving a constrained finite time domain optimal control problem, and a relaxation-based dynamic programming condition is constructed based on a finite time domain cost function to use the optimal control input sequence as much as possible
Figure 672410DEST_PATH_IMAGE006
The control input element in (1) so as to prolong the triggering time to the maximum extent and reduce the updating times of model predictive control as much as possible; meanwhile, the trigger interval is further adjusted in the self-triggering model prediction control convergence range by combining the protocol characteristics of the LoRaWAN network, so that the next trigger time of the self-triggering model prediction control is matched and synchronized with the information receiving window of the controlled object, on one hand, the communication energy consumption is reduced, on the other hand, the effectiveness of the trigger interval of the self-triggering model prediction control is ensured, and therefore, the calculated amount and the communication burden are reduced while the model prediction control accuracy is ensured.
Preferably, the expression of the linear model of the control system constructed in step 1 is as follows:
Figure 100002_DEST_PATH_IMAGE032
Figure 100002_DEST_PATH_IMAGE034
in the formula (I), the compound is shown in the specification,
Figure 913904DEST_PATH_IMAGE008
representing time
Figure 343748DEST_PATH_IMAGE002
The status of (2) is input;
Figure 100002_DEST_PATH_IMAGE036
representing time
Figure 54215DEST_PATH_IMAGE002
A control input of (2);
Figure 100002_DEST_PATH_IMAGE038
representing time
Figure 257836DEST_PATH_IMAGE002
An output of (d);
Figure 100002_DEST_PATH_IMAGE040
a parameter matrix representing a linear model of the control system.
Preferably, the mathematical expression for constructing the constrained finite time domain optimal control problem in step 2 is as follows:
Figure 100002_DEST_PATH_IMAGE042
s.t.
Figure 100002_DEST_PATH_IMAGE044
Figure 100002_DEST_PATH_IMAGE046
Figure 100002_DEST_PATH_IMAGE048
Figure 100002_DEST_PATH_IMAGE050
Figure 100002_DEST_PATH_IMAGE052
；
in the formula (I), the compound is shown in the specification,
Figure 42253DEST_PATH_IMAGE010
represents a prediction step size;
Figure 100002_DEST_PATH_IMAGE054
，
Figure 100002_DEST_PATH_IMAGE056
，
Figure 100002_DEST_PATH_IMAGE058
a set of constraints representing the state input,
Figure 100002_DEST_PATH_IMAGE060
a set of constraints representing control inputs;
Figure 100002_DEST_PATH_IMAGE062
is based on time
Figure 548058DEST_PATH_IMAGE002
State input of
Figure 824450DEST_PATH_IMAGE008
And control input set
Figure 100002_DEST_PATH_IMAGE064
The constructed prediction step length is a value function of N steps;
thereby obtaining the time from
Figure 372106DEST_PATH_IMAGE002
Time of arrival
Figure 729138DEST_PATH_IMAGE004
Optimal control input sequence of
Figure 969626DEST_PATH_IMAGE006
The expression is as follows:
Figure 100002_DEST_PATH_IMAGE066
。
preferably, the finite time domain cost function expression constructed in step 3 is:
Figure 100002_DEST_PATH_IMAGE068
；
the finite time domain cost function is used for constructing a mathematical expression based on a relaxation dynamic programming condition, wherein the mathematical expression is as follows:
Figure 100002_DEST_PATH_IMAGE070
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE072
to trigger the moment, satisfy
Figure 100002_DEST_PATH_IMAGE074
；
Figure 100002_DEST_PATH_IMAGE076
(ii) a In the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE078
for a given scalar:
Figure 100002_DEST_PATH_IMAGE080
，
Figure 100002_DEST_PATH_IMAGE082
sequence of
Figure 100002_DEST_PATH_IMAGE084
For the error term:
Figure 100002_DEST_PATH_IMAGE086
；
due to the next moment of triggering
Figure 100002_DEST_PATH_IMAGE088
Has a value of
Figure 12275DEST_PATH_IMAGE014
Is unknown, for which reason it is based on an optimal control input sequence
Figure 894781DEST_PATH_IMAGE006
Construction of Shift control sequences
Figure 100002_DEST_PATH_IMAGE090
：
Figure 100002_DEST_PATH_IMAGE092
；
Thereby obtaining
Figure 100002_DEST_PATH_IMAGE094
Is estimated value of
Figure 100002_DEST_PATH_IMAGE096
：
Figure 100002_DEST_PATH_IMAGE098
；
Figure 100002_DEST_PATH_IMAGE100
For shifting control sequences
Figure 100002_DEST_PATH_IMAGE102
Closed loop output after being applied to a controlled object;
therefore, the trigger interval is calculated based on the relaxed dynamic programming condition
Figure 388210DEST_PATH_IMAGE012
The method specifically comprises the following steps:
Figure 100002_DEST_PATH_IMAGE104
(ii) a Satisfies the following conditions:
Figure 100002_DEST_PATH_IMAGE106
for the
Figure 100002_DEST_PATH_IMAGE108
Is provided with
Figure 100002_DEST_PATH_IMAGE110
。
Preferably, the next trigger time in step 4
Figure 100002_DEST_PATH_IMAGE112
The calculation formula of (2) is as follows:
Figure 100002_DEST_PATH_IMAGE114
。
preferably, in step 7, the next trigger time is modified and corrected according to the information acceptance window of the controlled object
Figure 783288DEST_PATH_IMAGE028
Comprises the following steps:
Figure 100002_DEST_PATH_IMAGE116
in the formula (I), the reaction is carried out,
Figure DEST_PATH_IMAGE118
the function is a remainder function and is a function,
Figure DEST_PATH_IMAGE120
the window period duration is the message acceptance window.
Preferably, in step 7, the next trigger time is modified and corrected according to the information acceptance window of the controlled object
Figure 860703DEST_PATH_IMAGE028
The method specifically comprises the following steps:
judgment of
Figure DEST_PATH_IMAGE122
Whether or not it is greater than or equal to
Figure DEST_PATH_IMAGE124
If yes, then:
Figure DEST_PATH_IMAGE126
；
if not, then the mobile terminal can be switched to the normal mode,
Figure DEST_PATH_IMAGE128
；
in the formula, D2 represents the communication time of the information receiving window 2 of the Class B type terminal, and D1 represents the communication time of the information receiving window 1 of the Class B type terminal.
Self-triggering model prediction system based on LoRaWAN network includes:
the remote control terminal is used for executing the self-triggering model prediction method based on the LoRaWAN network, and comprises a control system module used for constructing a linear model of the control system, a computing unit used for executing the self-triggering model prediction control based on the linear model of the controlled system and solving to obtain a triggering interval, the next triggering moment and a control input sequence used for ensuring the convergence of the model prediction control, an edge control unit used for computing the control input sequence, and a LoRaWAN network communication unit used for communication, wherein the edge control unit adjusts the control input sequence based on an information receiving window and the next triggering moment of the controlled object;
the wireless network communication module is used for storing a LoRaWAN protocol;
at least 1 controlled object, each controlled object is communicated with the remote control terminal through the wireless network communication module respectively, the controlled object comprises a LoRaWAN network/bus for establishing communication, a data processing unit, an actuator, a controlled device and a sensor unit, the data processing unit receives a control input sequence from the remote control terminal over the LoRaWAN network/bus, and sequentially sending control input elements in a control input sequence to the actuator, wherein the actuator controls the controlled equipment to work, the sensor unit processes the acquired data parameters through the data processing unit, reporting to the remote control terminal through the wireless network communication module as a new status input, and the remote control terminal executes the self-triggering model prediction method based on the LoRaWAN network again based on the received state input.
The self-triggering model prediction system based on the LoRaWAN network has the advantages that: solving the finite time domain optimal control problem through a remote control terminal to obtain an optimal control input sequence, and a control input sequence which can be used for controlling the object is calculated based on the relaxation dynamic programming condition is constructed through a finite time domain cost function, while ensuring that the model predictive control is in an effective control range, the control input elements in the optimal control input sequence are used as much as possible, the calculation burden of a remote control terminal is reduced, meanwhile, the remote control terminal adjusts the control input sequence applied to the controlled object through LoRaWAN protocol, so that the next trigger time of the remote control terminal is matched and synchronized with the information receiving window of the controlled object, thereby reducing communication energy consumption on one hand, ensuring the effectiveness of the trigger interval of self-triggering model predictive control on the other hand, therefore, the accuracy of model prediction control is guaranteed, and meanwhile, the calculation amount and the communication load are reduced.
Drawings
FIG. 1 is a diagram illustrating a Class A type terminal correcting a next trigger time according to an embodiment of the present invention
Figure DEST_PATH_IMAGE130
A schematic view;
FIG. 2 is a diagram illustrating modification of the next trigger time by a Class B type terminal according to another embodiment of the present invention
Figure 156687DEST_PATH_IMAGE130
A schematic view;
FIG. 3 is a diagram illustrating that a Class B type terminal preferably corrects the next trigger time in another embodiment of the present invention
Figure 98098DEST_PATH_IMAGE130
A schematic diagram;
fig. 4 is a system framework diagram of embodiment 2 of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
According to another embodiment provided by the present invention, the present embodiment provides a self-triggering model prediction method based on a LoRaWAN network, which is applied to a large-scale data center room cooling scenario, where the scenario includes a data center control end and multiple large data center rooms, each large data center room includes at least 600 server cabinets and a cooling system composed of at least 10 precision air conditioners, the data center control end communicates with the cooling system through the LoRaWAN network, the data center control end performs self-triggering model prediction by acquiring a state of the cooling system, and then transmits an output of the self-triggering model prediction to the cooling system through the LoRaWAN network for control; meanwhile, LoRaWAN is a remote wide area network protocol and is used for communicating a plurality of controlled objects with a data center control end far away from the controlled objects at a low speed, and the self-triggering model prediction method specifically comprises the following steps:
step 1, according to the time of the controlled object
Figure 211547DEST_PATH_IMAGE002
The state input, the control input and the output of the control system construct a linear model of the control system; in this embodiment, the data center control end determines the time of the controlled object
Figure 995833DEST_PATH_IMAGE002
State input of
Figure 220140DEST_PATH_IMAGE008
Control input
Figure 750479DEST_PATH_IMAGE036
And output
Figure 80835DEST_PATH_IMAGE038
Constructing a linear model of a controlled system, wherein a controlled object is a cooling system, and the established linear model expression of the control system is as follows:
Figure DEST_PATH_IMAGE032A
Figure DEST_PATH_IMAGE034A
in the formula (I), the compound is shown in the specification,
Figure 821258DEST_PATH_IMAGE008
representing time
Figure 396727DEST_PATH_IMAGE002
Is input to the state of the mobile terminal,
Figure 171785DEST_PATH_IMAGE036
representing time
Figure 361458DEST_PATH_IMAGE002
The control input of (a) is performed,
Figure 995701DEST_PATH_IMAGE038
representing time
Figure 561812DEST_PATH_IMAGE002
An output of (d);
Figure 706223DEST_PATH_IMAGE040
a parameter matrix representing a linear model of the control system, in this particular embodiment,
Figure DEST_PATH_IMAGE132
the specific matrix parameters of the parameter matrix are obtained by calculating a heat exchange calculation model in the prior art; the state input comprises the temperature of each server cabinet and the energy consumption of each cabinet; the control input comprises a temperature set value and/or a set air supply quantity of the cooling system; the output includes cooling system time
Figure 863535DEST_PATH_IMAGE002
A cost function of (a);
step 2, the data center control end constructs a constrained finite time domain optimal control problem based on a control system linear model, in the specific embodiment, sampling time is adopted
Figure 250654DEST_PATH_IMAGE002
Solving the optimal control problem of finite time domain to obtain the input sequence of the cooling system
Figure 354876DEST_PATH_IMAGE006
，
Figure 589680DEST_PATH_IMAGE008
Representing time
Figure 386734DEST_PATH_IMAGE002
Is input to the state of the mobile terminal,
Figure 792308DEST_PATH_IMAGE010
representing a prediction step size; optimal control input sequence
Figure 965800DEST_PATH_IMAGE006
The interval duration between each control input element is 30min, specifically:
the mathematical expression for constructing the constrained finite time domain optimal control problem is as follows:
Figure DEST_PATH_IMAGE042A
Figure DEST_PATH_IMAGE134
Figure 22487DEST_PATH_IMAGE046
Figure DEST_PATH_IMAGE048A
Figure 131388DEST_PATH_IMAGE050
Figure 961941DEST_PATH_IMAGE052
；
in the formula (I), the compound is shown in the specification,
Figure 673545DEST_PATH_IMAGE010
representing a prediction step size;
Figure DEST_PATH_IMAGE136
，
Figure 178214DEST_PATH_IMAGE058
a set of constraints representing the state input,
Figure 317071DEST_PATH_IMAGE060
a set of constraints representing control inputs;
Figure 634920DEST_PATH_IMAGE062
is based on time
Figure 478111DEST_PATH_IMAGE002
State input of
Figure 463384DEST_PATH_IMAGE008
And control input set
Figure 320613DEST_PATH_IMAGE064
The constructed prediction step length is a value function of N steps;
obtaining the cooling system slave time based on the above solving finite time domain optimal control problem
Figure 391337DEST_PATH_IMAGE002
Time of arrival
Figure 913585DEST_PATH_IMAGE004
Optimal control input sequence of
Figure 753365DEST_PATH_IMAGE006
The expression is as follows:
Figure 93080DEST_PATH_IMAGE066
；
step 3, the data center control end constructs a finite time domain cost function for solving the finite time domain optimal control problem, and calculates trigger intervals based on the relaxation dynamic programming conditions according to the finite time domain cost function
Figure 385521DEST_PATH_IMAGE012
，
Figure 711460DEST_PATH_IMAGE014
In order to trigger the moment of time,
Figure 405747DEST_PATH_IMAGE016
for the state at the moment of triggering, to obtain an optimal control input sequence
Figure 103313DEST_PATH_IMAGE006
The number of control input elements that satisfy the linear model constraint of the control system, thereby using the optimal control input sequence as much as possible
Figure 7684DEST_PATH_IMAGE006
A control input element of (1); specifically, the method comprises the following steps:
the constructed finite time domain cost function expression is as follows:
；
the finite time domain cost function is used for constructing a mathematical expression based on a relaxation dynamic programming condition, wherein the mathematical expression is as follows:
Figure 402893DEST_PATH_IMAGE070
in the formula (I), the compound is shown in the specification,
Figure 420528DEST_PATH_IMAGE014
the time for solving the finite time domain optimal control problem for the triggering moment, namely the time of the data center control end meets
Figure DEST_PATH_IMAGE138
；
Figure 56040DEST_PATH_IMAGE076
(ii) a In the formula (I), the compound is shown in the specification,
Figure 323073DEST_PATH_IMAGE078
for a given scalar:
Figure 115448DEST_PATH_IMAGE080
，
Figure 518748DEST_PATH_IMAGE082
sequence of
Figure DEST_PATH_IMAGE140
For the error term:
Figure 512112DEST_PATH_IMAGE086
；
due to the next moment of triggering
Figure 109184DEST_PATH_IMAGE088
Has a value of
Figure 642934DEST_PATH_IMAGE014
Is unknown, for which reason it is based on an optimal control input sequence
Figure 166319DEST_PATH_IMAGE006
Construction of Shift control sequences
Figure 330584DEST_PATH_IMAGE090
：
Figure 306630DEST_PATH_IMAGE092
；
Thereby obtaining
Figure 191541DEST_PATH_IMAGE088
Is estimated value of
Figure DEST_PATH_IMAGE142
：
Figure DEST_PATH_IMAGE144
；
Figure DEST_PATH_IMAGE146
For shifting control sequences
Figure 100591DEST_PATH_IMAGE090
Closed loop output after application to a cooling system;
therefore, the trigger interval is calculated based on the relaxed dynamic programming condition
Figure 544079DEST_PATH_IMAGE012
The method specifically comprises the following steps:
Figure DEST_PATH_IMAGE148
(ii) a Satisfies the following conditions:
Figure DEST_PATH_IMAGE150
；
for the
Figure 69739DEST_PATH_IMAGE108
Is provided with
Figure 617395DEST_PATH_IMAGE110
；
In this embodiment, the trigger interval is calculated to be 7 units based on the dynamic relaxation planning condition, and the current trigger time is defined
Figure DEST_PATH_IMAGE152
That is, solving the optimal control problem of the finite time domain at the control end of the current data center to obtain a control input sequence which can be directly applied to the cooling system is as follows:
Figure DEST_PATH_IMAGE154
the body isEmbodiments to enable optimal control input sequences to be applied as much as possible
Figure 990738DEST_PATH_IMAGE006
The triggering interval is calculated by constructing a relaxed dynamic programming condition through a finite time domain cost function
Figure 667445DEST_PATH_IMAGE012
Thereby using as many optimal control input sequences as possible
Figure 883663DEST_PATH_IMAGE006
The control input element in (1) so as to reduce the triggering times of model predictive control of the control end of the data center and reduce the calculation burden of the control end of the data center;
step 4, calculating the time for solving the finite time domain optimal control problem at the data center control end next time based on the calculated trigger interval, namely the next trigger moment
Figure 313638DEST_PATH_IMAGE018
：
Figure 666122DEST_PATH_IMAGE114
；
The present embodiment
Figure DEST_PATH_IMAGE156
；
Step 5, in order to ensure the communication reliability, the data center control end judges the next trigger moment of the communication between the cooling system and the data center control end according to the LoRaWAN protocol
Figure DEST_PATH_IMAGE158
Whether the information is in the information receiving window of the controlled object or not, if so, entering a step 6; if not, entering step 7;
step 6, the data center control end inputs the optimal control sequence
Figure 936566DEST_PATH_IMAGE006
Time of flight
Figure 374501DEST_PATH_IMAGE002
To the next trigger moment
Figure 263960DEST_PATH_IMAGE158
The control input element in between is sent to the cooling system, which is dependent on time
Figure 736529DEST_PATH_IMAGE002
To the next trigger moment
Figure 630405DEST_PATH_IMAGE158
The control input elements between the control input elements sequentially control the precise air conditioner to supply cold to the data center machine room according to the temperature set value and/or the air supply quantity, and the precise air conditioner supplies cold to the data center machine room at the next trigger moment
Figure 149111DEST_PATH_IMAGE158
Requesting the controlled object to collect and report new state input parameters,
Figure DEST_PATH_IMAGE160
，
Figure DEST_PATH_IMAGE162
and returning to the step 2;
step 7, receiving window modification and correcting the next trigger moment according to the information of the controlled object
Figure 107840DEST_PATH_IMAGE028
In this embodiment, the controlled object communicates with the data center control end by using the LoRaWAN protocol, and the controlled object is used as the LoRaWAN terminal device, and may be a Class a terminal or a Class B terminal, where:
the cooling system in the context of an embodiment, a Class a type terminal, remotely communicating with the data center control terminal via the LoRaWAN protocol, opens an information acceptance window at a time interval D1 after the send window, allowing data information to be received from the data center control terminal, as shown in fig. 1;
Figure DEST_PATH_IMAGE164
in the formula (I), the compound is shown in the specification,
Figure 575861DEST_PATH_IMAGE118
the function is a remainder function and is a function,
Figure 594633DEST_PATH_IMAGE120
the communication period is the communication period of a Class A type terminal;
the Class A type terminal is adopted, so that the service energy consumption can be reduced, and the Class A type terminal is adopted in the embodiment, so that the calculation time delay of the next trigger moment is smaller;
the cooling system in the scenario of another embodiment is a Class B type terminal, as shown in fig. 2, opening information accepting window 1 at an interval D1 after the send window, and opening information accepting window 2 at an interval D2 after the send window, allowing data information to be received from the data center control at both information accepting windows;
modifying and correcting the next trigger moment
Figure 7160DEST_PATH_IMAGE028
As shown in fig. 2:
Figure 612322DEST_PATH_IMAGE116
in the formula (I), wherein,
Figure 793905DEST_PATH_IMAGE120
receiving a window period duration for the information;
more preferably, when
Figure 45895DEST_PATH_IMAGE122
Is greater than or equal to
Figure 414559DEST_PATH_IMAGE124
Preferably, the next trigger time is correctedThe start time of port 2, as shown in fig. 3:
Figure 980669DEST_PATH_IMAGE126
；
wherein D2 is the communication time of the information receiving window 2, D1 is the communication time of the information receiving window 1;
in the embodiment, a Class B type terminal is adopted, the number of information receiving windows is increased, a control input sequence can be applied to a controlled object to the maximum extent on the premise of self-triggering model prediction control convergence, and the number of times of self-touch model prediction is reduced;
cooling system based on slave time
Figure 626546DEST_PATH_IMAGE026
To the next trigger moment
Figure DEST_PATH_IMAGE166
The control input elements between the control input elements sequentially control the precise air conditioner to supply cold to the data center machine room according to the temperature set value and/or the air supply quantity, and the precise air conditioner supplies cold to the data center machine room at the next trigger moment
Figure 721541DEST_PATH_IMAGE166
Requesting the controlled object to collect and report new state input,
Figure DEST_PATH_IMAGE168
，
Figure 905397DEST_PATH_IMAGE162
and returns to step 2.
Meanwhile, the working contents of the sending window in fig. 1 and fig. 2 in this embodiment specifically include: the data center control end sends a request communication to the cooling system, and meanwhile, the cooling system sends the acquired state input to the data center control end for calculation; the information acceptance window in fig. 1 and 2 is used for the data center control to transmit the calculated control input elements to the cooling system.
In addition, in step 5, the cooling system is judged to be in communication with the control end of the data centerNext trigger moment of message
Figure 141379DEST_PATH_IMAGE158
The method for judging whether the object is in the information receiving window of the controlled object is
Figure 297554DEST_PATH_IMAGE122
If equal to 0, if yes, the next trigger moment
Figure 563450DEST_PATH_IMAGE158
Is in the information receiving window, otherwise, the next trigger moment
Figure 172286DEST_PATH_IMAGE158
Not within the information acceptance window;
in addition, in order to ensure communication and facilitate calculation, the correction trigger time in this embodiment is the start time of the calculation information acceptance window.
In the embodiment, the communication between the data center control end and the cooling system is further adjusted and controlled by combining the LoRaWAN protocol, so that the data center control end executes the self-triggering model predictive control and always converges, and meanwhile, the next triggering time is corrected according to a Class A type terminal or a Class B type terminal in the self-triggering model predictive control convergence range by combining the protocol characteristics of the LoRaWAN network, so that the next triggering time of the self-triggering model predictive control is matched and synchronized with the information receiving window of the controlled object, and the communication energy consumption is greatly reduced.
According to another embodiment of the present invention, as shown in fig. 4, in a specific embodiment 2, a self-triggered model prediction system based on a LoRaWAN network is provided, and in this specific embodiment, the self-triggered model prediction system based on a LoRaWAN network is applied to a large-scale data center room cooling scenario, where the large-scale data center room cooling scenario includes a data center control end and multiple large data center rooms, each large data center room includes at least 600 server cabinets and a cooling system composed of at least 10 precision air conditioners; wherein:
the remote control terminal, that is, the data center control terminal in this embodiment, is configured to execute a self-triggering model prediction method based on a LoRaWAN network, where the remote control terminal includes a control system module, a computing unit, an edge control unit, and a LoRaWAN network communication unit;
the control system module is used for constructing a control system module of a linear model of the control system; in the embodiment, the control system module constructs a linear system model according to a cooling system in a data center machine room;
the computing unit executes self-triggering model prediction control based on a controlled system linear model and solves to obtain a triggering interval and the next triggering moment; in this embodiment, the computing unit first solves the finite time domain optimal control problem to obtain an optimal control input sequence
Figure 142516DEST_PATH_IMAGE006
(ii) a Next, a finite time domain cost function is constructed
Figure DEST_PATH_IMAGE170
And based on finite time-domain cost functions
Figure 497405DEST_PATH_IMAGE170
Constructing a dynamic planning condition based on relaxation:
Figure DEST_PATH_IMAGE172
(ii) a The trigger interval is then calculated based on the relaxation dynamic programming conditions:
Figure DEST_PATH_IMAGE174
(ii) a And solving by using the trigger interval to obtain the time for the data center control end to execute the model predictive control next time, namely the next trigger moment:
Figure 793257DEST_PATH_IMAGE114
(ii) a Further obtaining a control input sequence which can make the model prediction control converged;
the edge control unit is used for further adjusting a control input sequence capable of enabling model predictive control to be converged according to an information receiving window by using a LoRaWAN protocol so as to enable the control input sequence to be synchronous with the information receiving window of the controlled object; on one hand, communication energy consumption is reduced, and on the other hand, effectiveness of a trigger interval of self-triggering model predictive control is guaranteed, so that calculation amount and communication burden are reduced while model predictive control accuracy is guaranteed;
the data center control end sends the adjusted control input sequence to the cooling system through the LoRaWAN network communication unit through the wireless network communication module;
the wireless network communication module is used for storing a LoRaWAN protocol;
at least 1 controlled object, namely, a cooling system in a data center machine room in this embodiment, each cooling system communicates with a data center control end through a wireless network communication module, and the cooling system includes a LoRaWAN network/bus for establishing communication, a data processing unit, an actuator, a controlled device, and a sensor unit; wherein:
the data processing unit receives a control input sequence from a remote control terminal through a LoRaWAN network/bus and sequentially sends control input elements in the control input sequence to an actuator; the control input elements of the embodiment comprise a temperature set value and/or a set air supply quantity of the precise air conditioner, and the actuator controls the precise air conditioner to work and supply cold according to the control input elements until the last control input element is sent;
the actuator controls the controlled equipment to work, and the controlled equipment in the specific embodiment is a precise air conditioner forming a cooling system;
the sensor unit comprises a temperature sensor for sensing the temperature of the server cabinet and a current transformer for calculating the energy consumption of the server cabinet, the sensor unit takes the collected temperature data and the collected current energy consumption as state input, the state input is processed by the data processing unit and reported to a control system module of the remote control terminal as new state input through the wireless network communication module, and the remote control terminal executes the self-triggering model prediction method based on the LoRaWAN network again based on the received new state input.
At each trigger moment, the sensor unit collects new state input and reports the new state input to the remote control terminal through the LoRaWAN network communication unit, the remote control terminal solves the finite time domain optimal control problem to obtain an optimal control input sequence, the control input sequence which can be used for a controlled object is calculated based on a relaxation dynamic programming condition through a finite time domain cost function, control input elements in the optimal control input sequence are used as much as possible while model prediction control is ensured in an effective control range, the calculation burden of the remote control terminal is reduced, meanwhile, the remote control terminal adjusts the control input sequence through the type terminal of the controlled object and sends the adjusted control input sequence to the controlled object, the next trigger moment of the remote control terminal is matched and synchronized with an information receiving window of the controlled object, and on one hand, the communication energy consumption is reduced, on the other hand, the effectiveness of the trigger interval of the self-triggering model predictive control is ensured, so that the accuracy of the model predictive control is ensured, and meanwhile, the calculated amount and the communication load are reduced.
Although the present disclosure has been described above, the scope of the present disclosure is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present disclosure, and such changes and modifications will fall within the scope of the present invention.

Claims (8)

1. A self-triggering model prediction method based on a LoRaWAN (Long area network) is characterized by comprising the following steps:
step 1, according to the time of the controlled object
Figure DEST_PATH_IMAGE002
The state input, the control input and the output of the control system construct a linear model of the control system;
step 2, constructing a constrained finite time domain optimal control problem based on a linear model of a control system, and solving the finite time domain optimal control problem to obtain the slave time
Figure 276993DEST_PATH_IMAGE002
Time of arrival
Figure DEST_PATH_IMAGE004
Optimal control input sequence of
Figure DEST_PATH_IMAGE006
，
Figure DEST_PATH_IMAGE008
Representing time
Figure 395122DEST_PATH_IMAGE002
Is input into the state of (a) or (b),
Figure DEST_PATH_IMAGE010
representing a prediction step size;
step 3, constructing a finite time domain cost function for solving the finite time domain optimal control problem, and calculating a trigger interval based on a relaxation dynamic programming condition according to the finite time domain cost function
Figure DEST_PATH_IMAGE012
，
Figure DEST_PATH_IMAGE014
In order to trigger the moment of time,
Figure DEST_PATH_IMAGE016
for the state at the moment of triggering, to obtain an optimum control input sequence
Figure 571894DEST_PATH_IMAGE006
The number of control input elements meeting the linear model constraint of the control system;
step 4, trigger interval based on calculation
Figure 179593DEST_PATH_IMAGE012
Calculating to obtain the next trigger time
Figure DEST_PATH_IMAGE018
；
Step 5, judging the next trigger moment according to the LoRaWAN protocol
Figure 136922DEST_PATH_IMAGE018
Whether the information is in the information receiving window of the controlled object or not, if so, entering a step 6; if not, entering step 7;
step 6, inputting the optimal control into a sequence
Figure 730846DEST_PATH_IMAGE006
Time of flight
Figure 919382DEST_PATH_IMAGE002
To the next trigger moment
Figure 557037DEST_PATH_IMAGE018
The control input element in between is sent to the controlled object and is triggered at the next moment
Figure 470504DEST_PATH_IMAGE018
Requesting the controlled object to collect and report new state input,
Figure DEST_PATH_IMAGE020
，
Figure DEST_PATH_IMAGE022
and returning to the step 2;
step 7, receiving window modification and correcting the next trigger moment according to the information of the controlled object
Figure DEST_PATH_IMAGE024
And inputting the optimal control into the sequence
Figure 868118DEST_PATH_IMAGE006
Time of flight
Figure DEST_PATH_IMAGE026
To the next triggerTime of day
Figure 504636DEST_PATH_IMAGE024
The control input element in between is sent to the controlled object and is triggered at the next moment
Figure 890355DEST_PATH_IMAGE024
Requesting the controlled object to collect and report new state input,
Figure DEST_PATH_IMAGE028
，
Figure 651638DEST_PATH_IMAGE022
and returns to step 2.
2. The LoRaWAN network-based self-triggering model prediction method of claim 1, wherein the linear model expression of the control system constructed in the step 1 is as follows:
Figure DEST_PATH_IMAGE030
Figure DEST_PATH_IMAGE032
in the formula (I), the compound is shown in the specification,
Figure 554741DEST_PATH_IMAGE008
representing time
Figure 452290DEST_PATH_IMAGE002
The status of (2) is input;
Figure DEST_PATH_IMAGE034
representing time
Figure 775955DEST_PATH_IMAGE002
A control input of (2);
Figure DEST_PATH_IMAGE036
representing time
Figure 867276DEST_PATH_IMAGE002
An output of (d);
Figure DEST_PATH_IMAGE038
a parameter matrix representing a linear model of the control system.
3. The LoRaWAN network-based self-triggering model prediction method of claim 2, wherein the mathematical expression for constructing the constrained finite time domain optimal control problem in the step 2 is as follows:
Figure DEST_PATH_IMAGE040
s.t.
Figure DEST_PATH_IMAGE042
Figure DEST_PATH_IMAGE044
Figure DEST_PATH_IMAGE046
Figure DEST_PATH_IMAGE048
Figure DEST_PATH_IMAGE050
；
in the formula (I), the compound is shown in the specification,
Figure 121540DEST_PATH_IMAGE010
representing a prediction step size;
Figure DEST_PATH_IMAGE052
，
Figure DEST_PATH_IMAGE054
a set of constraints representing the state input,
Figure DEST_PATH_IMAGE056
a set of constraints representing control inputs;
Figure DEST_PATH_IMAGE058
is based on time
Figure 372131DEST_PATH_IMAGE002
State input of
Figure 929014DEST_PATH_IMAGE008
And control input set
Figure DEST_PATH_IMAGE060
The constructed prediction step length is a value function of N steps;
thereby obtaining the time from
Figure 868151DEST_PATH_IMAGE002
Time of arrival
Figure 4735DEST_PATH_IMAGE004
Optimal control input sequence of
Figure 673613DEST_PATH_IMAGE006
The expression is as follows:
Figure DEST_PATH_IMAGE062
。
4. the LoRaWAN network-based self-triggering model prediction method of claim 3, wherein the finite time domain cost function expression constructed in the step 3 is as follows:
Figure DEST_PATH_IMAGE064
；
the finite time domain cost function is used for constructing a mathematical expression based on a relaxation dynamic programming condition, wherein the mathematical expression is as follows:
Figure DEST_PATH_IMAGE066
in the formula (I), the compound is shown in the specification,
Figure 650665DEST_PATH_IMAGE014
to trigger the moment, satisfy
Figure DEST_PATH_IMAGE068
；
Figure DEST_PATH_IMAGE070
(ii) a In the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE072
for a given scalar:
Figure DEST_PATH_IMAGE074
，
Figure DEST_PATH_IMAGE076
sequence of
Figure DEST_PATH_IMAGE078
For the error term:
Figure DEST_PATH_IMAGE080
；
due to the next moment of triggering
Figure DEST_PATH_IMAGE082
Has a value of
Figure 185421DEST_PATH_IMAGE014
Is unknown, for which reason it is based on an optimal control input sequence
Figure 30755DEST_PATH_IMAGE006
Construction of Shift control sequences
Figure DEST_PATH_IMAGE084
：
Figure DEST_PATH_IMAGE086
；
Thereby obtaining
Figure 350878DEST_PATH_IMAGE082
Is estimated value of
Figure DEST_PATH_IMAGE088
：
Figure DEST_PATH_IMAGE090
；
Figure DEST_PATH_IMAGE092
For shifting control sequences
Figure 124930DEST_PATH_IMAGE084
Closed loop output after being applied to a controlled object;
therefore, the trigger interval is calculated based on the relaxed dynamic programming condition
Figure 779656DEST_PATH_IMAGE012
The method specifically comprises the following steps:
Figure DEST_PATH_IMAGE094
(ii) a Satisfies the following conditions:
Figure DEST_PATH_IMAGE096
for the
Figure DEST_PATH_IMAGE098
Is provided with
Figure DEST_PATH_IMAGE100
。
5. A LoRaWAN network-based self-triggering model prediction method according to any one of claims 1 to 4, characterized in that the next triggering time in step 4 is
Figure 726884DEST_PATH_IMAGE018
The calculation formula of (2) is as follows:
Figure DEST_PATH_IMAGE102
。
6. the LoRaWAN network-based self-triggering model prediction method of claim 5, wherein in step 7, the next triggering time is modified and corrected according to the information acceptance window of the controlled object
Figure 275415DEST_PATH_IMAGE024
Comprises the following steps:
Figure DEST_PATH_IMAGE104
in the formula (I), wherein,
Figure DEST_PATH_IMAGE106
the function is a remainder function and is a function,
Figure DEST_PATH_IMAGE108
the window period duration is the message acceptance window.
7. The LoRaWAN network-based self-triggering model prediction method of claim 6, wherein in step 7, the next triggering time is modified and corrected according to the information acceptance window of the controlled object
Figure 954789DEST_PATH_IMAGE024
The method specifically comprises the following steps:
judgment of
Figure DEST_PATH_IMAGE110
Whether or not it is greater than or equal to
Figure DEST_PATH_IMAGE112
If so, then
Figure DEST_PATH_IMAGE114
；
If not, then,
Figure DEST_PATH_IMAGE116
；
in the formula, D2 represents the communication time of the information receiving window 2 of the Class B type terminal, and D1 represents the communication time of the information receiving window 1 of the Class B type terminal.
8. A self-triggering model prediction system based on a LoRaWAN (Long Range WAN) network is characterized by comprising:
a remote control terminal for executing the self-triggering model prediction method based on the LoRaWAN network according to any one of claims 1 to 7, wherein the remote control terminal comprises a control system module for constructing a linear model of a control system, a calculation unit for executing self-triggering model prediction control based on the linear model of the controlled system and solving to obtain a triggering interval, a next triggering moment and a control input sequence for ensuring convergence of the model prediction control, an edge control unit for calculating the control input sequence, the edge control unit for adjusting the control input sequence based on an information acceptance window and the next triggering moment of the controlled object, and a LoRaWAN network communication unit for communication;
the wireless network communication module is used for storing a LoRaWAN protocol;
at least 1 controlled object, each controlled object is communicated with the remote control terminal through the wireless network communication module respectively, the controlled object comprises a LoRaWAN network/bus for establishing communication, a data processing unit, an actuator, a controlled device and a sensor unit, the data processing unit receives a control input sequence from the remote control terminal over the LoRaWAN network/bus, and sequentially sending control input elements in a control input sequence to the actuator, wherein the actuator controls the controlled equipment to work, the sensor unit processes the acquired data parameters through the data processing unit, reporting to the remote control terminal through the wireless network communication module as a new status input, and the remote control terminal executes the self-triggering model prediction method based on the LoRaWAN network again based on the received state input.
CN202211050547.3A 2022-08-31 2022-08-31 Self-triggering model prediction method and system based on LoRaWAN (Long Range network) Active CN115128961B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211050547.3A CN115128961B (en) 2022-08-31 2022-08-31 Self-triggering model prediction method and system based on LoRaWAN (Long Range network)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211050547.3A CN115128961B (en) 2022-08-31 2022-08-31 Self-triggering model prediction method and system based on LoRaWAN (Long Range network)

Publications (2)

Publication Number Publication Date
CN115128961A true CN115128961A (en) 2022-09-30
CN115128961B CN115128961B (en) 2023-01-10

Family

ID=83387226

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211050547.3A Active CN115128961B (en) 2022-08-31 2022-08-31 Self-triggering model prediction method and system based on LoRaWAN (Long Range network)

Country Status (1)

Country Link
CN (1) CN115128961B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160170429A1 (en) * 2014-12-15 2016-06-16 King Fahd University Of Petroleum And Minerals System and method for non-linear model predictive control of multi-machine power systems
CN111708277A (en) * 2020-05-29 2020-09-25 中国科学技术大学 Adaptive time domain event trigger model prediction control method
CN114423020A (en) * 2022-01-21 2022-04-29 温州大学乐清工业研究院 LoRaWAN network downlink route control method and system
CN114600127A (en) * 2019-09-10 2022-06-07 辉达公司 Architecture searching method based on machine learning for neural network
CN114662850A (en) * 2022-02-22 2022-06-24 大连海事大学 Electric energy prediction distribution system based on LoRaWAN network cloud monitoring
CN114746872A (en) * 2020-04-28 2022-07-12 辉达公司 Model predictive control techniques for autonomous systems

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160170429A1 (en) * 2014-12-15 2016-06-16 King Fahd University Of Petroleum And Minerals System and method for non-linear model predictive control of multi-machine power systems
CN114600127A (en) * 2019-09-10 2022-06-07 辉达公司 Architecture searching method based on machine learning for neural network
CN114746872A (en) * 2020-04-28 2022-07-12 辉达公司 Model predictive control techniques for autonomous systems
CN111708277A (en) * 2020-05-29 2020-09-25 中国科学技术大学 Adaptive time domain event trigger model prediction control method
CN114423020A (en) * 2022-01-21 2022-04-29 温州大学乐清工业研究院 LoRaWAN network downlink route control method and system
CN114662850A (en) * 2022-02-22 2022-06-24 大连海事大学 Electric energy prediction distribution system based on LoRaWAN network cloud monitoring

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
赵志浩等: "基于LoRa无线通信的变电站火灾报警系统设计", 《电子与封装》 *

Also Published As

Publication number Publication date
CN115128961B (en) 2023-01-10

Similar Documents

Publication Publication Date Title
JP6356978B2 (en) Use of predictive means in process control systems with wireless or intermittent process measurements
JP6356977B2 (en) Improved Kalman filter in process control systems
CN108594886B (en) Self-optimization control method and device for refrigeration system based on simulation and operation big data
US11199824B2 (en) Reducing controller updates in a control loop
CN101655272A (en) Energy-saving control management system of network central air conditioner and method thereof
CN112147955B (en) Distributed control system
Hunek et al. New results in generalized minimum variance control of computer networks
CN115128961B (en) Self-triggering model prediction method and system based on LoRaWAN (Long Range network)
CN113064480B (en) Multi-data center collaborative energy saving method based on multi-agent reinforcement learning
Dhar et al. On an integrated approach to networked climate control of a smart home
CN109976144B (en) Reducing controller updates in a control loop
CN112147956B (en) Distributed control system
CN115327901B (en) Non-affine multi-agent dynamic event trigger tracking control method under asynchronous frame
CN104601424B (en) The passive transacter of master and method of probabilistic model are utilized in equipment control net
KR102634553B1 (en) Apparatus and method for managing energy of building using reinforcement learning
Mao et al. Nobody but you: Sensor selection for voltage regulation in smart grid
JP7014091B2 (en) Communication systems, communication devices and communication methods
CN103812634B (en) Fieldbus networks control system and its Networked-induced delay computational methods
CN112702280A (en) Network data delay control method
CN111880419A (en) Control method and system of household appliance, household appliance and computing device
CN112262404A (en) Global control system for consumption of energy resources based on IOT technology
Liu et al. Modeling and analysis of networked supervisory control systems with multiple control channels
Ignaciuk et al. Mitigating the impact of delay uncertainty on the performance of LQ optimal remote control systems
Liu et al. Modeling and analysis of networked discrete event systems with multiple control channels
CN207716621U (en) A kind of energy-saving heating and ventilating equipment integrated control system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant