Semi-physical simulation system and simulation method for multi-agent system in complex environment

Abstract

The application discloses a kind of complex environment under multi-agent system's semi-physical simulation system and simulation method, wherein simulation system includes: monitoring module, multi-agent control module, at least one agent motor;Multi-agent control module is directly connected with agent motor;Monitoring module is used to monitor the motion state of agent motor, configuration topological structure and time delay parameter;Multi-agent control module is used to issue instruction to agent motor;Multi-agent control module includes topological and time delay control module, and at least one agent control submodule, agent control submodule corresponds with agent motor one by one;Topological and time delay control module is used to generate topological and time delay data according to the configuration of topological structure and time delay parameter, and is forwarded to individual agent control submodule;Agent control submodule is used to adjust the motion of agent motor according to topological and time delay data.The system can demonstrate and simulate the influence of different time delays and topological switching on multi-agent system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of multi-agent simulation, and relates to a semi-physical simulation system and a simulation method for a multi-agent system in a complex environment. BACKGROUND

[0002] A multi-agent system is a complex network system composed of multiple agents, each of which has certain independence and autonomy. The agents can adjust their behaviors according to changes in external conditions to affect the environment, and can also transmit information data to adjacent agents to reach an agreement and complete a series of complex contents and tasks that cannot be completed by a single agent.

[0003] Topology switching and variable time delay have always been a difficult problem in the control of multi-agent systems. In actual multi-agent systems, due to the limitations of the information processing ability of the agents and the communication range, errors in the information transmission process, external uncertain disturbances and other factors, time delay problems exist when the multi-agents exchange information. Time delay can destroy the stability of the entire system, cause the multi-agents to produce unexpected behaviors, and hinder normal information exchange between them. If the time-varying time delay and the existence of topology in the system are not well described or considered in system modeling, design defects or incorrect conclusions will be obtained, which will affect the theoretical verification.

[0004] The data are processed by the Raspberry Pi, and finally the influence of the simulation topology and the time delay on the multi-agent system is simulated by the monitoring host, thereby providing a new solution for multi-agent theory and experimental teaching. SUMMARY

[0005] The application provides a semi-physical simulation system for a multi-agent system in a complex environment, which can demonstrate and simulate the influence of different time delays and topology switching on the multi-agent system.

[0006] The application discloses a semi-physical simulation system for a multi-agent system in a complex environment, comprising a monitoring module, a multi-agent control module and at least one agent motor.

[0007] The monitoring module and the agent motor are connected in a direct connection or relay mode, and the multi-agent control module and the agent motor are connected in a direct connection mode.

[0008] The monitoring module is used for monitoring the motion state of the agent motor, configuring a topology structure and a time delay parameter, and the multi-agent control module is used for issuing an instruction to the agent motor.

[0009] The multi-agent control module comprises a topology and time delay control module and at least one agent control submodule corresponding to the agent motor one by one.

[0010] The topology and time delay control module is configured to generate topology and time delay data according to the configuration of the topology structure and the time delay parameter, and forward the topology and time delay data to the agent control submodule.

[0011] The agent control submodule is configured to adjust the motion of the agent motor according to the topology and time delay data.

[0012] Further, a display device is further included, which is connected with the monitoring module and configured to display the topology structure and the motion state of the plurality of agent motors.

[0013] Further, the multi-agent control module is directly connected with the agent motor through RS485, EtherCAT or CAN.

[0014] Further, an agent motor motion state acquisition module is further included, which is configured to acquire the motion state of each agent motor and forward the motion state to the monitoring module.

[0015] Further, the monitoring module is implemented based on Raspberry Pi.

[0016] Further, the agent motor motion state acquisition module comprises a motor speed measurement module and a motor position measurement module.

[0017] The motor speed measurement module is configured to measure the rotating speed of the agent motor, and comprises a grating disc, a photoelectric door and a photoelectric sensor; when the agent motor drives the grating disc to rotate through a transmission part, the photoelectric door obtains a series of pulse signals, and the photoelectric sensor converts the pulse signals into electrical signals and inputs the electrical signals into a counter on the monitoring host; the counter counts the pulse signals, and calculates the real-time rotating speed of the agent motor according to the frequency of the pulse signals.

[0018] The motor position measurement module is configured to measure the real-time position of the agent motor, and acquires the current position through a potentiometer angle sensor and calculates the total displacement by using the monitoring host.

[0019] Further, the multi-agent control module further comprises a motor rotating speed control module configured to adjust the rotating speed of each agent motor; specifically,

[0020] In an offline state, the electromagnetic torque value corresponding to each torque angle output by the agent motor is calculated to form a torque value vector table.

[0021] During the rotation of the motor, real-time torque angle and torque value output by the agent motor are acquired; torque value is inquired in the torque value vector table according to the acquired real-time torque angle, and compared with the acquired real-time torque value, input voltage value of the agent motor is adjusted according to the comparison result, torque value is corrected, so that the rotation speed of the agent motor is adjusted.

[0022] In another aspect, the application discloses a multi-agent system semi-physical simulation method in a complex environment based on the semi-physical simulation system, comprising:

[0023] The monitoring module sets an adjacency matrix A(k) and a time delay matrix T(k) of the multi-agent topological structure, the adjacency matrix A(k)=[a ij (k)], wherein a ij (k)>0 indicates that there is communication from the agent j to the agent i at the kth sampling time, otherwise, the communication does not exist; the time delay matrix T(k)=[τ ij (k)], τ ij (k) indicates that the data of the agent j received by the agent i has a time delay of T0τ ij (k) at the kth sampling time, i, j∈{1, 2, …, Num}, Num is the number of agent motors; T0 is a basic sampling period;

[0024] The multi-agent control module realizes the communication between the agents according to the adjacency matrix and the time delay matrix at the kth sampling: if a ij (k)>0, then at the kth sampling, the agent j sends data to the agent i, and the sent data is the basic sampling data of the agent j at the f(k)-τ ij (k)th time, or the agent i obtains the basic sampling data of the agent j at the f(k)-τ ij (k)th time through the buffer; wherein f(k) is a sampling rate function, indicating the f(k)th basic sampling corresponding to the kth sampling.

[0025] Further, the basic sampling data of the agent is the data obtained by sampling the motion state of the agent motor with the basic sampling period T0 as the period.

[0026] Further, when sampling at a fixed rate, the sampling rate function f(k)=kn, wherein n is a positive integer constant, indicating the multiple of the actual sampling period and the basic sampling period, and the actual sampling period is nT0.

[0027] When sampling at a variable rate, the sampling rate function f(k) is assigned according to a preset kth sampling time T k , T k =f(k)T0, and f(k) is a positive integer.

[0028] Beneficial effects: the semi-physical simulation system of the multi-agent system in a complex environment disclosed by the application can demonstrate and simulate the influence of different time lags and topology switching on the multi-agent system, and intuitively and concisely present the simulation results. Based on the simulation system, the topology structure of the multi-agent system can be adjusted by modifying the adjacency matrix, and the time lag parameter between the agents can be adjusted by modifying the time lag matrix, and the adjustment of the topology structure and the time lag parameter can be realized by obtaining the current adjacency matrix and time lag matrix during simulation, which is simple to operate. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 The composition schematic diagram of the semi-physical simulation system of the multi-agent system in a complex environment disclosed by the application is shown in the figure.

[0030] Figure 2 The flowchart of the semi-physical simulation method of the multi-agent system in a complex environment disclosed by the application is shown in the figure.

[0031] Figure 3 The change curve diagram of the motion state of the four agents in the embodiment is shown in the figure. DETAILED DESCRIPTION

[0032] The application will be further illustrated in combination with the drawings and specific embodiments.

[0033] The application discloses a semi-physical simulation system of a multi-agent system in a complex environment, as shown in the figure, which comprises a monitoring module 100, a multi-agent control module 200, and at least one agent motor 300. Figure 1

[0034] The monitoring module 100 is connected with the agent motor 300 in a direct connection or relay mode, the multi-agent control module 200 is connected with the agent motor 300 in a direct connection mode, and the monitoring module is realized based on a Raspberry Pi.

[0035] The monitoring module 100 is used for monitoring the motion state of the agent motor 300, and configuring the topology structure and the time lag parameter; the multi-agent control module 200 is used for issuing instructions to the agent motor 300.

[0036] The multi-agent control module 200 comprises a topology and time lag control module 201 and at least one agent control submodule 202, and the agent control submodule 202 corresponds to the agent motor 300 one by one.

[0037] The topology and time lag control module is used for generating topology and time lag data according to the configuration of the topology structure and the time lag parameter, and forwarding the data to the agent control submodule.

[0038] The agent control submodule is used for adjusting the motion of the agent motor according to the topology and time lag data.​

[0039] The multi-agent control module and the agent motor can be directly connected through RS485, EtherCAT or CAN in this embodiment, and the instructions can be quickly sent to the agent motor.

[0040] In order to intuitively show the real-time topology and time delay results of the multi-agent system, the simulation system further comprises a display device 400 connected with the monitoring module, for displaying the topology structure and motion state of the plurality of agent motors.

[0041] The motion state of the agent motor is obtained by an agent motor motion state acquisition module 500, and the agent motor motion state acquisition module 500 further sends the obtained motion state data to the monitoring module. The agent motor motion state acquisition module comprises a motor speed measurement module 501 and a motor position measurement module 502.

[0042] The motor speed measurement module 501 is used for measuring the speed of the agent motor, and comprises a grating disc, a photoelectric gate and a photoelectric sensor. When the agent motor drives the grating disc to rotate through the transmission part, the photoelectric gate obtains a series of pulse signals, and the photoelectric sensor converts the pulse signals into electrical signals and inputs the electrical signals into a counter on the monitoring host. The counter counts the pulse signals, and the real-time speed of the agent motor is calculated according to the frequency of the pulse signals.

[0043] The motor position measurement module 502 is used for measuring the real-time position of the agent motor, and the current position is obtained by a potentiometer angle sensor, and the total displacement is calculated by the monitoring host.

[0044] The multi-agent control module 200 further comprises a motor speed control module 203 for adjusting the speed of each agent motor. Specifically,

[0045] In the offline state, the electromagnetic torque value corresponding to each torque angle output by the agent motor is calculated to form a torque value vector table. In the case of constant power voltage and stator magnetic field frequency, the real-time output torque value of the motor is proportional to the sine value of the torque angle, and thus the torque value vector table is calculated.

[0046] During the rotation of the motor, the real-time torque angle and torque value output by the agent motor are obtained. The torque value is queried in the torque value vector table according to the obtained real-time torque angle, and compared with the obtained real-time torque value. The input voltage value of the agent motor is adjusted according to the comparison result, and the torque value is corrected, so as to adjust the speed of the agent motor.

[0047] Based on the above simulation system, a semi-physical simulation method of a multi-agent system in a complex environment is provided, as shown in Figure 2 The method comprises the following steps:

[0048] Step 1, the monitoring module sets the adjacency matrix A(k) and the time delay matrix T(k) of the multi-agent topology structure, the adjacency matrix A(k)=[a ij (k)], where a ij (k)>0 represents that there is a communication from agent j to agent i at the kth sampling time, otherwise, there is no such communication; the time delay matrix T(k)=[τ ij (k)], τ ij (k) represents that the data received by agent i from agent j at the kth sampling time has a time delay of T0τ ij (k), i, j∈{1, 2, …, Num}, Num is the number of agent motors; T0 is the basic sampling period;

[0049] Step 2, the multi-agent control module realizes the communication between agents according to the adjacency matrix and the time delay matrix at the kth sampling time: if a ij (k)>0, then at the kth sampling time, agent j sends data to agent i, and the sent data is the f(k)-τ ij (k)th basic sampling data of agent j, or agent i obtains the f(k)-τ ij (k)th basic sampling data of agent j through the buffer.

[0050] The basic sampling data of the agent is the data obtained by sampling the motion state of the agent motor with a basic sampling period T0, such as the data obtained by sampling the position, speed, acceleration and other motion states of the motor. The basic sampling data of each agent is stored in the respective data buffer and updated with T0 as the period.

[0051] The communication between agent j and agent i is realized through the corresponding agent control submodule. The multi-agent control module synchronizes the clock of the agent control submodule, and realizes the time delay simulation by sending different basic sampling data.

[0052] f(k) is a sampling rate function, which represents the f(k)th basic sampling corresponding to the kth sampling. When sampling at a fixed rate, the sampling rate function f(k)=kn, where n is a positive integer constant, representing the multiple of the actual sampling period and the basic sampling period, and the actual sampling period is nT0;

[0053] When sampling at a variable rate, the sampling rate function f(k) is assigned according to the preset kth sampling time T k , T k =f(k)T0, and f(k) is a positive integer.

[0054] The sampling period subdivision technology is adopted, based on a basic sampling period, a more fine control time delay is realized, a variable rate sampling control can be simulated, and different sampling rates among multiple agents are allowed, that is, multi-rate sampling is realized.

[0055] The multi-agent system under different topologies and different time delays is simulated by adjusting the values of the adjacency matrix A(k) and the time delay matrix T(k).

[0056] In the embodiment, four electric machines are adopted to form four agents to realize the above scheme. The real-time average speed θ(t) of the four agents, the switching signal σ(t) and the maximum position error Y(t) are as shown in the following table. Figure 3 The switching signal σ(t) is a periodic pulse. When the switching signal σ(t) pulse appears, the state separation of the electric machine occurs, that is, the maximum position error Y(t) rapidly increases. Then, the state synchronization of the electric machine is realized again, and the maximum position error Y(t) gradually decreases. When Y(t) is substantially 0, that is, after the state synchronization is achieved, the speed of the electric machine is also not 0, thereby realizing the effect of dynamic synchronization of the electric machine.

Claims

1. A hardware-in-the-loop simulation method using a hardware-in-the-loop simulation system of a multi-agent system in a complex environment, characterized by, The semi-physical simulation system of the multi-agent system in the complex environment comprises a monitoring module, a multi-agent control module, and at least one agent motor; The monitoring module is connected with the agent motor in a direct connection or relay manner; the multi-agent control module is connected with the agent motor in a direct connection manner; The monitoring module is configured to monitor the motion state of the agent motor, configure a topological structure and a time delay parameter; the multi-agent control module is configured to issue an instruction to the agent motor; The multi-agent control module comprises a topological and time delay control module, and at least one agent control submodule, the agent control submodule corresponds to the agent motor one by one; The topological and time delay control module is configured to generate topological and time delay data according to the configuration of the topological structure and the time delay parameter, and forward the data to the agent control submodule; The agent control submodule is configured to adjust the motion of the agent motor according to the topological and time delay data; The multi-agent control module further comprises a motor speed control module for adjusting the speed of each agent motor; specifically: In an offline state, the electromagnetic torque value corresponding to each torque angle output by the agent motor is calculated to form a torque value vector table; During the rotation of the motor, the real-time torque angle and torque value output by the agent motor are obtained; the torque value is queried in the torque value vector table according to the obtained real-time torque angle, and compared with the obtained real-time torque value; the input voltage value of the agent motor is adjusted according to the comparison result, and the torque value is corrected, so as to adjust the speed of the agent motor; The semi-physical simulation method of the multi-agent system in the complex environment is based on a basic sampling period to realize more precise time delay control and allow variable rate sampling, and specifically comprises: The monitoring module sets the adjacency matrix of the multi-agent topology. and time delay matrix The adjacency matrix ,in Indicates the first The second sampling time exists from the intelligent agent arrive The communication must be established; otherwise, the communication will not exist. The time delay matrix... , Indicates the first The intelligent agent at the next sampling time Received intelligent agent The data has a size of The time lag, , The number of motors in the intelligent agent; The basic sampling period; The multi-agent control module in the first During the next sampling, communication between agents is achieved based on the adjacency matrix and the time delay matrix: if intelligent agent Send data; the data sent is for the intelligent agent. No. Secondary basic sampling data, or intelligent agent Retrieve the agent from the buffer No. Secondary basic sampling data; among which Let be the sampling rate function, representing the th The sampling corresponds to the first Secondary basic sampling. 2.The hardware-in-the-loop simulation method of using the hardware-in-the-loop simulation system of multi-agent system in complex environment according to claim 1, wherein, Further comprising a display device connected with the monitoring module, configured to display the topological structure and motion state of the plurality of agent motors. 3.The semi-physical simulation method of using the semi-physical simulation system of the multi-agent system in a complex environment according to claim 1, wherein, The multi-agent control module and the agent motor are directly connected through RS485, EtherCAT, or CAN.

4. The hardware-in-the-loop simulation method of claim 1, wherein, Further comprising an agent motor motion state acquisition module configured to acquire the motion state of each agent motor and forward the state to the monitoring module. 5.The semi-physical simulation method of using the semi-physical simulation system of the multi-agent system in a complex environment according to claim 1, wherein, The monitoring module is realized based on Raspberry Pi.

6. The hardware-in-the-loop simulation method of claim 4, wherein, The agent motor motion state acquisition module comprises a motor speed measurement module and a motor position measurement module; The motor speed measurement module is configured to measure the speed of the agent motor, comprising a grid disc, a photoelectric door, and a photoelectric sensor; when the agent motor drives the grid disc to rotate through a transmission part, the photoelectric door obtains a series of pulse signals, and the photoelectric sensor converts the signals into electrical signals and inputs the signals into a counter on the monitoring host; the counter counts the pulse signals, and the real-time speed of the agent motor is calculated according to the frequency of the pulse signals; The motor position measurement module is configured to measure the real-time position of the agent motor, acquire the current position through a potentiometer angle sensor, and calculate the total displacement by using the monitoring host. 7.The hardware-in-the-loop simulation method of using the hardware-in-the-loop simulation system of multi-agent system in complex environment according to claim 1, wherein, The basic sampling data of the intelligent agent is data obtained by sampling the motion state of the motor of the intelligent agent with a basic sampling period .

8. The semi-physical simulation method of claim 1, wherein When sampling at a fixed rate, the sampling rate function where is a positive integer constant representing the multiple of the base sampling period, the actual sampling period being ; When sampling at a variable rate, the sampling rate function According to a preset first Sampling time Assignment, , and is a positive integer.

Images