Distributed model prediction control method and device based on state estimation

Abstract

The embodiment of the invention discloses a distributed model prediction control method and device based on state estimation. The method comprises the following steps: acquiring a discretization model of an identified agent; receiving reference state information in a neighborhood range of the intelligent agent; performing state prediction on the state of the neighbor agent according to the past time state to obtain a virtual estimation state; based on the reference state information and the virtual estimation state, constructing a cost optimal problem for solving; based on set constraint conditions, obtaining an optimal control input sequence and a virtual reference state; the method does not depend on centralized control, and can better deal with the problems of local information loss or communication failure and the like; each agent only performs local optimization calculation, so that the calculation burden is reduced, and the control efficiency is improved; the optimization result of the previous time is considered, and the robustness is good.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of multi-agent cluster control, and in particular, to a distributed model predictive control method and device based on state estimation. Background Art

[0002] Distributed model predictive control (DMPC) is a technology developed on the basis of traditional MPC, mainly for systems with a distributed structure, such as multi-robot systems, smart grids, groups of autonomous driving vehicles, UAV clusters, etc. In DMPC, each local controller makes decisions by optimizing its own objective function (such as local state tracking, energy consumption, stability, etc.), and these local objective functions work in coordination with other controllers by exchanging information. Each controller needs to coordinate with other controllers in addition to considering its own objectives to ensure the achievement of the global objective. However, the existing distributed model predictive control methods rely on centralized control, resulting in a large computational burden and low control efficiency. Summary of the Invention

[0003] The embodiments of the present application provide a distributed model predictive control method and device based on state estimation, which solve the problems that the existing distributed model predictive control methods rely on centralized control, have a large computational burden, and low control efficiency.

[0004] In a first aspect, the embodiments of the present application provide a distributed model predictive control method based on state estimation, and the method includes the following steps:

[0005] Obtain the discretized model of the identified agent;

[0006] Receive the reference state information within the neighborhood range of the agent;

[0007] Predict the state of the neighboring agent according to the past time state to obtain a virtual estimated state;

[0008] Based on the reference state information and the virtual estimated state, construct an optimal cost problem for solution;

[0009] Based on the set constraint conditions, obtain the optimal control input sequence and the virtual reference state; wherein, the constraint conditions include the constraint for ensuring that the predicted state at the initial moment of the agent is consistent with the current actual state, the constraint for ensuring that the state and the control satisfy their respective constraints, the constraint for ensuring that each state of the agent is within a predetermined range, and the terminal constraint.

[0010] Further, the predicting the state of the neighboring agent according to the past time state to obtain a virtual estimated state includes:

[0011] The calculation formula for the optimal state sequence is:

[0012]

[0013] Virtual estimation state Through to perform the calculation of state update:

[0014]

[0015] where A and B are state and input matrices, k|t represents the time for k-step optimal prediction at time t, and N p is the prediction horizon.

[0016]

[0017] Δt is the discrete time step, and τ is the inertia time constant.

[0018] Furthermore, the terminal constraint is:

[0019]

[0020] The constraint expressions for the optimal control input sequence and the virtual reference state are:

[0021] u min ≤u i ≤u max ;

[0022] ||x i -x j ||≥d ij ;

[0023] where u min and u max are the upper and lower limits of the input control; d ij is the minimum distance between the first agent and the second agent.

[0024] Furthermore, based on the set constraint conditions, obtaining the optimal control input sequence and the virtual reference state includes:

[0025] Each agent adopts model predictive control based on its own state and control input, and shares the states and prediction information of other agents through information exchange;

[0026] Within the given prediction horizon, minimize the tracking error, the change in control input, and the cost function for formation maintenance, with constraints including limitations on position, speed, and control input;

[0027] The first agent obtains the optimal control input through local prediction and optimization calculation based on its own model and current state;

[0028] In each control cycle, each agent calculates a control sequence over a period of time and selects the optimal control input at the current moment. The first element of

[0029] is executed as the actual control input.

[0030]

[0031] Among them, is the predicted state of the first agent, and are the virtual states of the first agent and its neighboring agents, and U i is the control input sequence of the first agent;

[0032] l i The function is defined as:

[0033]

[0034] Among them, Q i , R i , F i , G i are all symmetric non - negative definite weight matrices, is the target trajectory of agent i, i.e., the following trajectory, and Δ ij is the formation setting.

[0035] Furthermore, it also includes:

[0036] At each discrete moment, taking the predicted control input sequence U i as a variable, using the system dynamics model to recursively calculate the system states at the next N p moments of the system, performing optimization at time t, and taking the first element of the optimal control sequence of agent i as the actual control input to act on the first agent.

[0037] In a second aspect, an embodiment of the present application further provides a distributed model predictive control device based on state estimation, including:

[0038] A model acquisition module, configured to acquire the discretized model of the identified agent;

[0039] An information receiving module, configured to receive the reference state information within the neighborhood range of the agent;

[0040] A state acquisition module, configured to predict the states of neighboring agents based on the past time states to obtain virtual estimated states;

[0041] A problem-solving module for constructing and solving a cost-optimal problem based on reference state information and virtual estimated states;

[0042] A result-obtaining module for obtaining an optimal control input sequence and a virtual reference state based on set constraint conditions, where the constraint conditions include constraints for ensuring the consistency between the predicted state at the initial moment and the current actual state of the agent, constraints for ensuring that the states and controls satisfy their respective corresponding constraints, constraints for ensuring that each state of the agent is within a predetermined range, and terminal constraints.

[0043] In a third aspect, an embodiment of the present application further provides a computer device, including: a memory and one or more processors;

[0044] The memory is used for storing one or more programs;

[0045] When the one or more programs are executed by the one or more processors, the one or more processors implement a distributed model predictive control method based on state estimation as described above.

[0046] In a fourth aspect, an embodiment of the present application further provides a storage medium containing computer-executable instructions, and the computer-executable instructions are used for executing a distributed model predictive control method based on state estimation as described above when executed by a computer processor.

[0047] In a fifth aspect, an embodiment of the present application provides a computer program product, and the computer program product includes instructions that, when executed by a computer, cause the computer to implement the method described above.

[0048] The embodiment of the present application obtains the discretized model of the identified agent; receives the reference state information within the neighborhood range of the agent; predicts the states of neighbor agents according to the past time states to obtain virtual estimated states; constructs and solves a cost-optimal problem based on the reference state information and the virtual estimated states; obtains an optimal control input sequence and a virtual reference state based on the set constraint conditions; does not rely on centralized control and can better cope with problems such as local information loss or communication failures; each agent only performs local optimization calculations, reducing the computational burden and improving the control efficiency; considering the optimization results of the previous time period, it has good robustness. Description of the Drawings

[0049] Figure 1 is a flowchart of a distributed model predictive control method based on state estimation provided by an embodiment of the present application;

[0050] Figure 2 is a structural schematic diagram of a distributed model predictive control device based on state estimation provided by an embodiment of the present application;

[0051] Figure 3 It is a schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0052] In order to make the objectives, technical solutions, and advantages of the present application clearer, the following further describes the specific embodiments of the present application in detail with reference to the accompanying drawings. It can be understood that the specific embodiments described herein are only used to explain the present application, rather than limiting the present application. Additionally, it should be noted that for the sake of convenience of description, only parts related to the present application rather than all content are shown in the accompanying drawings. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operations are completed, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0053] The embodiment of the present application establishes a distributed model predictive control method based on state estimation to achieve a reasonable allocation of elevators to robots, and solve the problems that the existing distributed predictive control methods rely on centralized control, have a large computational burden, and low control efficiency.

[0054] The distributed model predictive control method based on state estimation provided in the embodiment can be executed by a distributed model predictive control device based on state estimation. The distributed model predictive control device based on state estimation can be implemented in a software and / or hardware manner and integrated in a distributed model predictive control device based on state estimation. Among them, the distributed model predictive control device based on state estimation can be a device such as a computer.

[0055] Figure 1 It is a flowchart of a distributed model predictive control method based on state estimation provided by an embodiment of the present application. Refer to Figure 1 , the method includes the following steps:

[0056] 100. Obtain the discretized model of the identified agent.

[0057] 200. Receive the reference state information within the neighborhood range of the agent.

[0058] Specifically, obtain the discretized model of the first intelligent agent i; at time t, when performing model predictive control on the first intelligent agent i, the states of neighboring intelligent agents within the prediction horizon cannot be accurately obtained; therefore, it is necessary to perform state prediction and estimation based on past time states and use virtual estimated states to replace x j .

[0059] 300. Predict the states of neighboring intelligent agents based on past time states to obtain virtual estimated states.

[0060] 400. Construct a cost-optimal problem for solution based on the reference state information and virtual estimated states.

[0061] 500. Based on the set constraint conditions, obtain the optimal control input sequence and virtual reference state; among them, the constraint conditions include constraints for ensuring that the predicted state at the initial moment of the intelligent agent is consistent with the current actual state, constraints for ensuring that the state and control satisfy their respective corresponding constraints, constraints for ensuring that each state of the intelligent agent is within a predetermined range, and terminal constraints.

[0062] In some embodiments, the predicting the states of neighboring intelligent agents based on past time states to obtain virtual estimated states includes:

[0063] The formula for the optimal state sequence is:

[0064]

[0065] Virtual estimated state is updated through for state calculation:

[0066]

[0067] where A and B are state and input matrices, k|t represents the time for k-step optimal prediction at time t, and N p is the prediction horizon.

[0068]

[0069] Δt is the discrete time step, and τ is the inertia time constant.

[0070] Furthermore, the terminal constraint is:

[0071]

[0072] The constraint expressions for the optimal control input sequence and virtual reference state are:

[0073] u min ≤u i≤u max ;

[0074] ||x i -x j ||≥d ij ;

[0075] where u min and u max are the upper and lower limits of the input control; d ij is the minimum distance between the first agent and the second agent.

[0076] In some embodiments, obtaining the optimal control input sequence and the virtual reference state based on the set constraint conditions includes:

[0077] Each agent adopts model predictive control based on its own state and control input, and shares the states and prediction information of other agents through information exchange;

[0078] Within the given prediction horizon, minimize the tracking error, the change of the control input, and the cost function for formation maintenance, and the constraints include the limitations of position, speed, and control input;

[0079] The first agent obtains the optimal control input through local prediction and optimization calculation based on its own model and current state;

[0080] Each agent calculates the control sequence for a period of time within each control cycle, and selects the first element of the optimal control input at the current moment as the actual control input for execution.

[0081] In some embodiments, the cost function is set as:

[0082]

[0083] where is the predicted state of the first agent, and are the virtual states of the first agent and its neighbor agents, U i is the control input sequence of the first agent;

[0084] The definition of the l i function is:

[0085]

[0086] where Q i , R i , F i , G i are all symmetric non-negative definite weight matrices, is the target trajectory of agent i, i.e., the following trajectory, Δ ij is the formation setting.

[0087] In some embodiments, it further includes:

[0088] At each discrete time instant, taking the predicted control input sequence U i as a variable, using the system dynamics model to recursively calculate the system states at the next N p time instants of the system to perform an optimization solution at time t, and taking the first element in the obtained optimal control sequence of agent i as the actual control input to act on the first agent.

[0089] As described above, the embodiments of the present application obtain the discretized model of the identified agent; receive the reference state information within the neighborhood range of the agent; predict the states of neighbor agents based on the past time states to obtain virtual estimated states; construct and solve a cost-optimal problem based on the reference state information and the virtual estimated states; obtain the optimal control input sequence and the virtual reference state based on the set constraint conditions; without relying on centralized control, it can better handle problems such as local information loss or communication failures; each agent only performs local optimization calculations, reducing the computational burden and improving the control efficiency; considering the optimization results of the previous time instant, it has good robustness.

[0090] Based on the above embodiments, please refer to Figure 2 , a distributed model predictive control device based on state estimation provided by the embodiments of the present application specifically includes: a model acquisition module 201, an information reception module 202, a state acquisition module 203, a problem solving module 204, and a result obtaining module 205.

[0091] The model acquisition module 201 is used to obtain the discretized model of the identified agent; the information reception module 202 is used to receive the reference state information within the neighborhood range of the agent; the state acquisition module 203 is used to predict the states of neighbor agents based on the past time states to obtain virtual estimated states; the problem solving module 204 is used to construct and solve a cost-optimal problem based on the reference state information and the virtual estimated states; the result obtaining module 205 is used to obtain the optimal control input sequence and the virtual reference state based on the set constraint conditions; where the constraint conditions include constraints for ensuring that the predicted state at the initial time of the agent is consistent with the current actual state, constraints for ensuring that the states and controls satisfy their respective constraints, constraints for ensuring that the states of the agent are within a predetermined range, and terminal constraints.

[0092] As described above, the embodiments of the present application obtain the discretized model of the identified agent; receive the reference state information within the neighborhood range of the agent; predict the state of the neighbor agent based on the past time state to obtain the virtual estimated state; construct and solve the cost-optimal problem based on the reference state information and the virtual estimated state; obtain the optimal control input sequence and the virtual reference state based on the set constraint conditions; without relying on centralized control, it can better handle problems such as local information loss or communication failures; each agent only performs local optimization calculations, reducing the computational burden and improving the control efficiency; considering the optimization results of the previous time period, it has better robustness.

[0093] The distributed model predictive control device based on state estimation provided by the embodiments of the present application can be used to execute the distributed model predictive control method based on state estimation provided by the above embodiments, and has the corresponding functions and beneficial effects.

[0094] The embodiments of the present application also provide a computer device, which can integrate the distributed model predictive control device based on state estimation provided by the embodiments of the present application. Figure 3 It is a schematic structural diagram of a computer device provided by the embodiments of the present application. Refer to Figure 3 , the computer device includes: an input device 33, an output device 34, a memory 32, and one or more processors 31; the memory 32 is used to store one or more programs; when the one or more programs are executed by the one or more processors 31, the one or more processors 31 implement the distributed model predictive control method based on state estimation provided by the above embodiments. Among them, the input device 33, the output device 34, the memory 32, and the processor 31 can be connected through a bus or other means, Figure 3 Taking the connection through the bus as an example.

[0095] The processor 31 executes various functional applications and data processing of the device by running software programs, instructions, and modules stored in the memory 32, that is, implements the above-mentioned distributed model predictive control method based on state estimation.

[0096] The above-provided computer device can be used to execute the distributed model predictive control method based on state estimation provided by the above embodiments, and has the corresponding functions and beneficial effects.

[0097] An embodiment of the present application further provides a storage medium containing computer-executable instructions, and the computer-executable instructions are used to execute a distributed model predictive control method based on state estimation when executed by a computer processor. The distributed model predictive control method based on state estimation includes: adding scheduling information through the background of the scheduling system, where the scheduling information includes point positions, paths, and action information; automatically generating a running path based on the starting point and target point of the robot and dispatching it to the robot; and according to the real-time task state and real-time working state of the elevator, combining the real-time action information of the robot, adjusting the running path in real time, and controlling the robot to run.

[0098] Storage medium - Any of various types of memory devices or storage devices. The term "storage medium" is intended to include: installation media such as CD-ROMs, floppy disks, or magnetic tape devices; computer device memories or random access memories such as DRAM, DDRRAM, SRAM, EDORAM, Rambus RAM, etc.; non-volatile memories such as flash memory, magnetic media (such as hard disks or optical storage); register or other similar types of memory elements, etc. The storage medium may also include other types of memories or combinations thereof. Additionally, the storage medium may be located in the first computer device in which the program is executed, or may be located in a different second computer device, and the second computer device is connected to the first computer device through a network (such as the Internet). The second computer device may provide program instructions to the first computer for execution. The term "storage medium" may include two or more storage media that may reside in different locations (such as in different computer devices connected through a network). The storage medium may store program instructions (such as specifically implemented as a computer program) executable by one or more processors.

[0099] Of course, for a storage medium containing computer-executable instructions provided by an embodiment of the present application, the computer-executable instructions are not limited to the distributed model predictive control method based on state estimation as described above, and can also execute related operations in the distributed model predictive control method based on state estimation provided by any embodiment of the present application.

[0100] The distributed model predictive control device, storage medium, and computer device provided in the above embodiments can execute the distributed model predictive control method provided by any embodiment of the present application. For technical details not described in detail in the above embodiments, reference can be made to the distributed model predictive control method provided by any embodiment of the present application.

[0101] The embodiments of the present application further provide a computer program product. The methods described in the embodiments of the present application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in the form of a computer program product in whole or in part. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are executed in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, a core network device, an OAM (Open Application Model), or other programmable devices.

[0102] The above are only the preferred embodiments of the present application and the technical principles applied. The present application is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions that can be made by those skilled in the art will not depart from the protection scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments. Without departing from the concept of the present application, it may also include more other equivalent embodiments, and the scope of the present application is determined by the scope of the claims.

Claims

1. A distributed model predictive control method based on state estimation, characterized in that: The method comprises the following steps: Obtaining a discretized model of the identified agent; Receive reference state information within the agent's neighborhood; Predict the state of neighboring agents based on the past state to obtain a virtual estimated state; Based on the reference state information and the virtual estimated state, a cost optimization problem is constructed and solved; Based on the set constraints, the optimal control input sequence and virtual reference state are obtained; wherein the constraints include constraints for ensuring that the initial predicted state of the intelligent agent is consistent with the current actual state, constraints for ensuring that the state and control satisfy their respective corresponding constraints, constraints for ensuring that each state of the intelligent agent is within a predetermined range, and terminal constraints.

2. The distributed model predictive control method based on state estimation according to claim 1, characterized in that: The step of predicting the state of the neighboring agent according to the state in the past time to obtain a virtual estimated state includes: The optimal state sequence calculation formula is: Virtual estimated state pass To calculate the state update: Among them, A and B are the state and input matrices, k|t represents the time for k-step optimal prediction at time t, and N p For the prediction time domain. Δt is the discrete time step and τ is the inertia time constant.

3. The distributed model predictive control method based on state estimation according to claim 1, characterized in that: The terminal constraints are: The constraint expressions of the optimal control input sequence and virtual reference state are: in min in i in max ; ||x i -x j ||≥d ij ; Among them, u min and u max is the upper and lower limits of input control; d ij is the minimum distance between the first agent and the second agent.

4. The distributed model predictive control method based on state estimation according to claim 1, characterized in that: The method of obtaining an optimal control input sequence and a virtual reference state based on the set constraint conditions includes: Each agent uses model predictive control based on its own state and control input, and shares the state and prediction information of other agents through information exchange; Minimize the cost function of tracking error, change of control input and formation maintenance within a given prediction horizon, with constraints including position, velocity and control input limits; The first agent obtains the optimal control input through local prediction and optimization calculation based on its own model and current state; Each agent calculates the control sequence within a period of time in each control cycle and selects the optimal control input at the current moment. The first element of is used as the actual control input for execution.

5. The distributed model predictive control method based on state estimation according to claim 4 is characterized in that: The cost function is set as: in, is the predicted state of the first agent, and is the virtual state of the first agent and its neighboring agents, U i is a control input sequence for the first agent; l i The function is defined as: Among them, Q i , R i 、F i , G i are all symmetric non-negative definite weight matrices, is the target trajectory of agent i, i.e., the following trajectory, Δ ij Set up for formation.

6. The distributed model predictive control method based on state estimation according to claim 1, characterized in that: Also includes: At each discrete moment, the predictive control input sequence U i As a variable, the system dynamics model is used to recursively infer the future N of the system p System status at a moment Perform the optimization solution at time t and solve the optimal control sequence of agent i The first element in acts as the actual control input to the first agent.

7. A distributed model predictive control device based on state estimation, characterized in that: include: A model acquisition module is used to obtain a discretized model of the identified intelligent agent; An information receiving module, used to receive reference state information within the neighborhood of the agent; The state acquisition module is used to predict the state of the neighboring agent based on the state of the past time and obtain the virtual estimated state; A problem solving module is used to construct and solve the cost optimization problem based on the reference state information and the virtual estimated state; The result obtaining module is used to obtain the optimal control input sequence and virtual reference state based on the set constraints; wherein the constraints include constraints for ensuring that the predicted state of the intelligent body at the initial moment is consistent with the current actual state, constraints for ensuring that the state and control satisfy their respective corresponding constraints, constraints for ensuring that each state of the intelligent body is within a predetermined range, and terminal constraints.

8. A computer device, characterized in that: include: memory and one or more processors; The memory is used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a distributed model predictive control method based on state estimation as described in any one of claims 1-6.

9. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute a distributed model predictive control method based on state estimation as described in any one of claims 1-6 when executed by a computer processor.

10. A computer program product, characterized in that The computer program product comprises instructions which, when executed by a computer, cause the computer to implement the method according to any one of claims 1-6.

Images