Double-layer collaborative security defense strategy and system for multi-robot system
By constructing network layer and control layer models of a multi-robot system, calculating consistency speed performance parameters, and establishing a two-layer defense strategy of network-control collaboration, the problem of topology changes in multi-robot systems under denial-of-service attacks is solved, thereby improving the robustness and consistency rate of the system.
Patent Information
- Application Number
- CN202511691978.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-18
- Publication Date
- 2026-02-03
AI Technical Summary
When facing denial-of-service attacks, the network layer topology of existing multi-robot systems is prone to change, which can lead to slower system consistency or failure to achieve consistency. The control layer is not robust enough, and existing methods cannot effectively improve system performance.
Construct network layer and control layer models for a multi-robot system, calculate consistent speed performance parameters, establish a two-layer defense strategy of network-control cooperation, model it as a Markov decision process, and use a proximal policy optimization algorithm to train and obtain the optimal security defense control strategy.
By maximizing the consistency speed performance of multi-robot systems, the impact of attacks on system performance is reduced, and the speed at which the system achieves consistency is improved.
Smart Images
Figure CN121462284A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of multi-robot system control technology, and in particular to a two-layer collaborative safety defense strategy and system for multi-robot systems. Background Technology
[0002] A multi-robot system consists of multiple individual robots, each with communication, information processing, and execution capabilities. However, the combined capabilities of each individual robot are limited. A multi-robot system, formed by the collaborative actions of these individual robots, can generate shared behaviors or accomplish tasks that a single robot cannot. To achieve collaborative control goals in a multi-robot system, it is necessary to establish rules based on the characteristics of these goals and design control protocols based on these rules as control inputs for each individual robot.
[0003] With the rapid development of information technology, network attacks have become one of the serious challenges facing society today. Due to the openness of the shared network in multi-robot systems, external attackers can easily attack the communication network of these systems to compromise system security. Currently, the mainstream network attack methods include Denial-of-Service (DoS) attacks, replay attacks, and error message injection attacks. Among these, DoS attacks, due to their simplicity and high threat level, are a key focus of research. DoS attacks inject a large number of malicious data packets into the channel, causing the channel's data traffic to exceed the receiver's processing capacity. Under a DoS attack, the communication channel between robots is broken, the topology of the multi-robot system changes, and the rate at which the multi-robot system achieves consistency slows down or even fails to achieve consistency. To address this problem, researchers have successively introduced adaptive control, fuzzy control, and robust control methods at the controller level to compensate for the adverse effects of attacks on system performance. However, most current research assumes that the network layer topology will inevitably be completely broken once attacked, making it impossible to further improve system performance by incorporating network layer components. Summary of the Invention
[0004] To overcome the shortcomings of the control layer in terms of robustness and the ease with which the topology can change, this invention provides a two-layer collaborative safety defense strategy and system for multi-robot systems.
[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: This invention provides a two-layer collaborative security defense strategy for multi-robot systems, comprising: Construct network layer and control layer models for a multi-robot system; Based on the network layer model and control layer model, calculate the consistent speed performance parameters of the multi-robot system. With the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established. The network-control cooperative two-layer defense strategy is modeled as a Markov decision process; The Markov decision process is trained using a proximal policy optimization algorithm to obtain the optimal speed performance parameters for consistency in the multi-robot system, thereby obtaining a safety defense control strategy.
[0006] Preferably, the network layer model and control layer model of the multi-robot system are constructed, including: Constructing multi-robot systems; Based on the aforementioned multi-robot system, a network layer model and a control layer model are constructed.
[0007] Preferably, a multi-robot system is constructed, including: The multi-robot system includes one leader model and N follower models, and the expression for the leader model is:
[0008] Where k is time, and A is the relationship between different states; The expression for the follower model is:
[0009] in, It is the system state vector of the i-th robot at time k. B is the control input vector of the i-th robot at time k, and B is the input matrix.
[0010] Preferably, based on the multi-robot system, a network layer model is constructed, including: Construct the traffic distribution matrix and its constraints, bandwidth distribution matrix and its constraints, and communication state constraints of the network layer model; Construct the traffic distribution matrix of the communication channel at time k. :
[0011] in, and They represent Distribution matrix of normal traffic and attack traffic for each communication channel in the time-domain topology network:
[0012]
[0013] when hour, ,when hour, ,in, Represents robots and The connectivity between them; The constraints of the flow distribution matrix are expressed as follows:
[0014] in, and All are constants greater than 0; Construct the bandwidth distribution matrix of the communication channel at time k. :
[0015] in, This represents the bandwidth of each communication channel, when hour, ,when hour, ; The constraints of the bandwidth distribution matrix are expressed as follows:
[0016] in, This represents the total system bandwidth, and satisfy ; When the bandwidth of the communication channel is greater than the traffic of the communication channel, the communication channel transmits information, and the communication state constraint is: .
[0017] Preferably, based on the multi-robot system, a control layer model is constructed, including: The control layer model includes a controller, an error system dynamics model, and a control objective; The controller is designed as follows:
[0018] in, Let be the feedback gain matrix to be solved. This is the gain coefficient. This indicates a connection with the leader; The expression for the dynamic model of the error system is as follows:
[0019] in, Indicates the Kronecker product. for 3D identity matrix It is a Laplace matrix; Construct control objectives: .
[0020] Preferably, based on the network layer model and the control layer model, the speed performance parameters of the multi-robot system consistency are calculated, including: Define the topology graph matrix:
[0021] in, It is a diagonal matrix; Based on Lyapunov functions, the control objective is achieved using the eigenvalues of the topological graph matrix and the controller, as follows:
[0022] Where P is a positive definite matrix and N is a positive definite matrix; Based on matrix theory, the following inequality is obtained:
[0023] in, and Represent The minimum and maximum eigenvalues; Transforming the inequality, we get:
[0024] Let the velocity performance parameter α representing the rate of change of the Lyapunov function in a multi-robot system be set as follows: Representing Lyapunov functions With speed magnitude as attenuation; According to the aforementioned error system dynamics model, let Let Γ(k) be the Jordan matrix corresponding to matrix Γ(k), where Let Γ(k) be the eigenvalue of matrix Γ(k); According to matrix theory, there exist nonsingular matrices. Make ,make We can obtain:
[0025] Further simplification of the above equation yields:
[0026]
[0027] Substituting the above equation into... By combining the properties of the Kronecker product, we can obtain:
[0028]
[0029] The error vector in the above equation Unfold, and you get: From the above equation, the relationship between the rate of change of the Lyapunov function, the system, and the controller is:
[0030] Multiply both sides of the above equation get:
[0031] Linearize the inequality; after transformation, let the matrix be... And according to Schul's complement lemma, we obtain the following expression:
[0032] Where α is the speed performance parameter for consistency of the multi-robot system.
[0033] Preferably, with the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established, including: The two-layer defense strategy is described as follows:
[0034]
[0035]
[0036]
[0037]
[0038] in, It is the set of edges in the system's topology when there is no DoS attack.
[0039] Preferably, the network-control cooperative two-layer defense strategy is modeled as a Markov decision process, including: The Markov decision process includes a state space, an action space, state transition probabilities, and a total reward function. The expression for the state space is as follows:
[0040] in, Network topology diagram of a multi-robot system; The expression for the action space is as follows:
[0041] The expression for the state transition probability is as follows: .
[0042] Preferably, the total reward function includes: link satisfaction rate, bandwidth utilization rate, connectivity reward, and control performance reward based on linear matrix inequalities and its constraints; The expression for the link satisfaction rate is as follows:
[0043] in, This represents the total number of edges in the network; The expression for the bandwidth utilization rate is as follows:
[0044] The connectivity reward is expressed as follows:
[0045] in, Indicates at time The number of robots that cannot be connected in the network topology of a multi-robot system; The control performance reward and its constraints based on linear matrix inequalities are expressed as follows:
[0046] The LMI constraint is as follows:
[0047] The expression for the total reward function is as follows:
[0048] in, It is an adjustable weighting coefficient that is greater than zero.
[0049] This invention also provides a two-layer collaborative security defense system for multi-robot systems, comprising: The model building module is used to build the network layer model and control layer model of a multi-robot system. The speed performance parameter calculation module is used to calculate the consistent speed performance parameters of the multi-robot system based on the network layer model and the control layer model. The dual-layer defense strategy establishment module is used to establish a network-control collaborative dual-layer defense strategy with the goal of maximizing the consistent speed performance parameters of the multi-robot system. The Markov decision process modeling module is used to model the network-control cooperative two-layer defense strategy as a Markov decision process; The safety defense control strategy acquisition module is used to train the Markov decision process using a near-end policy optimization algorithm to obtain the optimal speed performance parameters for consistency of the multi-robot system, and then obtain the safety defense control strategy.
[0050] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: This invention combines the network layer and the control layer to maximize the speed performance of consistency in multi-robot systems. By reducing the impact of attacks on system performance through bandwidth scheduling, it improves the speed at which multi-robot systems achieve consistency. Attached Figure Description
[0051] Figure 1 This is a flowchart illustrating a two-layer collaborative security defense strategy for a multi-robot system in Example 1. Figure 2 This is a schematic diagram of a two-layer collaborative security defense system for a multi-robot system in Example 3. Detailed Implementation
[0052] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent. To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions; It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.
[0053] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.
[0054] Example 1 This embodiment provides a two-layer collaborative security defense strategy for multi-robot systems, such as... Figure 1 As shown, it includes: Construct network layer and control layer models for a multi-robot system; Based on the network layer model and control layer model, calculate the consistent speed performance parameters of the multi-robot system. With the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established. The network-control cooperative two-layer defense strategy is modeled as a Markov decision process; The Markov decision process is trained using a proximal policy optimization algorithm to obtain the optimal speed performance parameters for consistency in the multi-robot system, thereby obtaining a safety defense control strategy.
[0055] In its implementation, this invention first constructs a network layer model and a control layer model for a multi-robot system. Then, based on these models, it calculates the consistent speed performance parameters of the multi-robot system. Next, with the goal of maximizing the consistent speed performance parameters, it establishes a network-control collaborative two-layer defense strategy. Finally, it models this strategy as a Markov decision process. Finally, it trains the Markov decision process using a near-end policy optimization algorithm to obtain the optimal consistent speed performance parameters for the multi-robot system, thereby acquiring a safety defense control strategy.
[0056] Example 2 This embodiment provides a two-layer collaborative security defense strategy for multi-robot systems, including: Construct network layer and control layer models for a multi-robot system; Based on the network layer model and control layer model, calculate the consistent speed performance parameters of the multi-robot system. With the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established. The network-control cooperative two-layer defense strategy is modeled as a Markov decision process; The Markov decision process is trained using a proximal policy optimization algorithm to obtain the optimal speed performance parameters for consistency in the multi-robot system, thereby obtaining a safety defense control strategy.
[0057] It should be noted that, in this embodiment, Constructing the network layer model and control layer model of the multi-robot system, including: Constructing multi-robot systems; Based on the aforementioned multi-robot system, a network layer model and a control layer model are constructed.
[0058] It should be noted that, in this embodiment, constructing a multi-robot system includes: The multi-robot system includes one leader model and N follower models, and the expression for the leader model is:
[0059] Where k is time, and A is the relationship between different states; The expression for the follower model is:
[0060] in, It is the system state vector of the i-th robot at time k. B is the control input vector of the i-th robot at time k, and B is the input matrix.
[0061] It should be noted that in this embodiment, the robot transmits information through a communication channel. However, these communication channels can also be exploited by attackers on the network to launch denial-of-service attacks, thereby disrupting normal communication processes.
[0062] Based on the aforementioned multi-robot system, a network layer model is constructed, including: Construct the traffic distribution matrix and its constraints, bandwidth distribution matrix and its constraints, and communication state constraints of the network layer model; Construct the traffic distribution matrix of the communication channel at time k. :
[0063] in, and They represent Distribution matrix of normal traffic and attack traffic for each communication channel in the time-domain topology network:
[0064]
[0065] when hour, ,when hour, ,in, Represents robots and The connectivity between them; In practical applications, attackers' energy is often limited. Therefore, it is assumed that the sum of traffic on each communication channel and the traffic on each communication channel are bounded. The constraints of the traffic distribution matrix are expressed as follows:
[0066] in, and All are constants greater than 0; Construct the bandwidth distribution matrix of the communication channel at time k. :
[0067] in, This represents the bandwidth of each communication channel, when hour, ,when hour, ; The constraints of the bandwidth distribution matrix are expressed as follows:
[0068] in, This represents the total system bandwidth, and satisfy ; When the bandwidth of the communication channel is greater than the traffic of the communication channel, the communication channel transmits information, and the communication state constraint is: .
[0069] Meanwhile, to ensure system stability, it is assumed that the network topology of the multi-robot system remains connected under a DoS attack, meaning that each robot can communicate directly or indirectly. At the network layer, the total system bandwidth is assumed to be... .
[0070] It should be noted that, in this embodiment, at the control level, a controller under linear matrix inequality constraints is designed. Under this controller, the system can achieve the corresponding performance parameters while satisfying the control objective.
[0071] Based on the aforementioned multi-robot system, a control layer model is constructed, including: The control layer model includes a controller, an error system dynamics model, and a control objective; The controller is designed as follows:
[0072] in, Let be the feedback gain matrix to be solved. This is the gain coefficient. This indicates a connection with the leader; The expression for the dynamic model of the error system is as follows:
[0073] in, Indicates the Kronecker product. for 3D identity matrix It is a Laplace matrix; Construct control objectives: .
[0074] It should be noted that, in this embodiment, the speed performance parameters of the multi-robot system consistency are calculated based on the network layer model and the control layer model, including: Define the topology graph matrix:
[0075] in, It is a diagonal matrix; Based on Lyapunov functions, the control objective is achieved using the eigenvalues of the topological graph matrix and the controller, as follows:
[0076] Where P is a positive definite matrix and N is a positive definite matrix; Based on matrix theory, the following inequality is obtained:
[0077] in, and Represent The minimum and maximum eigenvalues; Transforming the inequality, we get:
[0078] Let the velocity performance parameter α representing the rate of change of the Lyapunov function in a multi-robot system be set as follows: Representing Lyapunov functions With speed magnitude as attenuation; According to the aforementioned error system dynamics model, let Let Γ(k) be the Jordan matrix corresponding to matrix Γ(k), where Let Γ(k) be the eigenvalue of matrix Γ(k); According to matrix theory, there exist nonsingular matrices. Make ,make We can obtain:
[0079] Further simplification of the above equation yields:
[0080]
[0081] Substituting the above equation into... By combining the properties of the Kronecker product, we can obtain:
[0082]
[0083] The error vector in the above equation Unfold, and you get: From the above equation, the relationship between the rate of change of the Lyapunov function, the system, and the controller is:
[0084] Multiply both sides of the above equation get:
[0085] Linearize the inequality; after transformation, let the matrix be... And according to Schul's complement lemma, we obtain the following expression:
[0086] Where α is the speed performance parameter for consistency of the multi-robot system.
[0087] It should be noted that, in this embodiment, the traffic distribution matrix is based on the real-time communication channel. and bandwidth distribution matrix By determining the achievable system network topology under the traffic distribution matrix of the communication channel through bandwidth scheduling, the performance parameters can be optimized. Maximize, and then solve for the corresponding controller: With the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established, including: The two-layer defense strategy is described as follows:
[0088]
[0089]
[0090]
[0091]
[0092] in, It is the set of edges in the system's topology when there is no DoS attack.
[0093] It should be noted that, in this embodiment, the network-control cooperative two-layer defense strategy is modeled as a Markov decision process, including: The Markov decision process includes a state space, an action space, state transition probabilities, and a total reward function. The expression for the state space is as follows:
[0094] in, Network topology diagram of a multi-robot system; The expression for the action space is as follows:
[0095] The expression for the state transition probability is as follows: .
[0096] It should be noted that, in this embodiment, the total reward function includes: link satisfaction rate, bandwidth utilization rate, connectivity reward, and control performance reward based on linear matrix inequalities and its constraints; The expression for the link satisfaction rate is as follows:
[0097] in, This represents the total number of edges in the network; The expression for the bandwidth utilization rate is as follows:
[0098] The connectivity reward is expressed as follows:
[0099] in, Indicates at time The number of robots that cannot be connected in the network topology of a multi-robot system; The control performance reward and its constraints based on linear matrix inequalities are expressed as follows:
[0100] The LMI constraint is as follows:
[0101] The expression for the total reward function is as follows:
[0102] in, It is an adjustable weighting coefficient that is greater than zero.
[0103] Example 3 This embodiment provides a two-layer collaborative security defense system for multi-robot systems, used to implement the two-layer collaborative security defense strategy for multi-robot systems described in Embodiment 1 or 2, such as... Figure 2 As shown, it includes: The model building module is used to build the network layer model and control layer model of a multi-robot system. The speed performance parameter calculation module is used to calculate the consistent speed performance parameters of the multi-robot system based on the network layer model and the control layer model. The dual-layer defense strategy establishment module is used to establish a network-control collaborative dual-layer defense strategy with the goal of maximizing the consistent speed performance parameters of the multi-robot system. The Markov decision process modeling module is used to model the network-control cooperative two-layer defense strategy as a Markov decision process; The safety defense control strategy acquisition module is used to train the Markov decision process using a near-end policy optimization algorithm to obtain the optimal speed performance parameters for consistency of the multi-robot system, and then obtain the safety defense control strategy.
[0104] The same or similar labels correspond to the same or similar parts; The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent. Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A two-layer collaborative security defense strategy for multi-robot systems, characterized in that, include: Construct network layer and control layer models for a multi-robot system; Based on the network layer model and control layer model, calculate the consistent speed performance parameters of the multi-robot system. With the goal of maximizing the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy of network-control collaboration is established. The network-control cooperative two-layer defense strategy is modeled as a Markov decision process; The Markov decision process is trained using a proximal policy optimization algorithm to obtain the optimal speed performance parameters for consistency in the multi-robot system, thereby obtaining a safety defense control strategy.
2. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 1, characterized in that, Constructing the network layer model and control layer model of the multi-robot system, including: Constructing multi-robot systems; Based on the aforementioned multi-robot system, a network layer model and a control layer model are constructed.
3. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 2, characterized in that, Constructing a multi-robot system includes: The multi-robot system includes one leader model and N follower models, and the expression for the leader model is: Where k is time, and A is the relationship between different states; The expression for the follower model is: in, It is the system state vector of the i-th robot at time k. B is the control input vector of the i-th robot at time k, and B is the input matrix.
4. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 3, characterized in that, Based on the aforementioned multi-robot system, a network layer model is constructed, including: Construct the traffic distribution matrix and its constraints, bandwidth distribution matrix and its constraints, and communication state constraints of the network layer model; Construct the traffic distribution matrix of the communication channel at time k. : in, and They represent Distribution matrix of normal traffic and attack traffic for each communication channel in the time-domain topology network: when hour, ,when hour, ,in, Represents robots and The connectivity between them; The constraints of the flow distribution matrix are expressed as follows: in, and All are constants greater than 0; Construct the bandwidth distribution matrix of the communication channel at time k. : in, This represents the bandwidth of each communication channel, when hour, ,when hour, ; The constraints of the bandwidth distribution matrix are expressed as follows: in, This represents the total system bandwidth, and satisfy ; When the bandwidth of the communication channel is greater than the traffic of the communication channel, the communication channel transmits information, and the communication state constraint is: .
5. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 4, characterized in that, Based on the aforementioned multi-robot system, a control layer model is constructed, including: The control layer model includes a controller, an error system dynamics model, and a control objective; The controller is designed as follows: in, Let be the feedback gain matrix to be solved. This is the gain coefficient. This indicates a connection with the leader; The expression for the dynamic model of the error system is as follows: in, Indicates the Kronecker product. for 3D identity matrix It is a Laplace matrix; Construct control objectives: 。 6. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 5, characterized in that, Based on the network layer model and control layer model, calculate the consistent speed performance parameters of the multi-robot system, including: Define the topology graph matrix: in, It is a diagonal matrix; Based on Lyapunov functions, the control objective is achieved using the eigenvalues of the topological graph matrix and the controller, as follows: Where P is a positive definite matrix and N is a positive definite matrix; Based on matrix theory, the following inequality is obtained: in, and Represent The minimum and maximum eigenvalues; Transforming the inequality, we get: Let the velocity performance parameter α representing the rate of change of the Lyapunov function in a multi-robot system be set as follows: Representing Lyapunov functions With speed magnitude as attenuation; According to the aforementioned error system dynamics model, let Let Γ(k) be the Jordan matrix corresponding to matrix Γ(k), where Let Γ(k) be the eigenvalue of matrix Γ(k); According to matrix theory, there exist nonsingular matrices. Make ,make We can obtain: Further simplification of the above equation yields: Substituting the above equation into... By combining the properties of the Kronecker product, we can obtain: The error vector in the above equation Unfold, and you get: From the above equation, the relationship between the rate of change of the Lyapunov function, the system, and the controller is: Multiply both sides of the above equation get: Linearize the inequality; after transformation, let the matrix be... And according to Schul's complement lemma, we obtain the following expression: Where α is the speed performance parameter for consistency of the multi-robot system.
7. A two-layer collaborative security defense strategy for a multi-robot system according to claim 6, characterized in that, To maximize the consistent speed performance parameters of the multi-robot system, a two-layer defense strategy based on network-control collaboration is established, including: The two-layer defense strategy is described as follows: in, It is the set of edges in the system's topology when there is no DoS attack.
8. The dual-layer collaborative security defense strategy for a multi-robot system according to claim 7, characterized in that, The network-control cooperative two-layer defense strategy is modeled as a Markov decision process, including: The Markov decision process includes a state space, an action space, state transition probabilities, and a total reward function. The expression for the state space is as follows: in, Network topology diagram of a multi-robot system; The expression for the action space is as follows: The expression for the state transition probability is as follows: 。 9. A two-layer collaborative security defense strategy for a multi-robot system according to claim 8, characterized in that, The total reward function includes: link satisfaction rate, bandwidth utilization rate, connectivity reward, and control performance reward based on linear matrix inequalities and their constraints; The expression for the link satisfaction rate is as follows: in, This represents the total number of edges in the network; The expression for the bandwidth utilization rate is as follows: The connectivity reward is expressed as follows: in, Indicates at time The number of robots that cannot be connected in the network topology of a multi-robot system; The control performance reward and its constraints based on linear matrix inequalities are expressed as follows: The LMI constraint is as follows: The expression for the total reward function is as follows: in, It is an adjustable weighting coefficient that is greater than zero.
10. A two-layer collaborative security defense system for a multi-robot system, used to implement the two-layer collaborative security defense strategy for a multi-robot system as described in claims 1-9, characterized in that, include: The model building module is used to build the network layer model and control layer model of a multi-robot system. The speed performance parameter calculation module is used to calculate the consistent speed performance parameters of the multi-robot system based on the network layer model and the control layer model. The dual-layer defense strategy establishment module is used to establish a network-control collaborative dual-layer defense strategy with the goal of maximizing the consistent speed performance parameters of the multi-robot system. The Markov decision process modeling module is used to model the network-control cooperative two-layer defense strategy as a Markov decision process; The safety defense control strategy acquisition module is used to train the Markov decision process using a near-end policy optimization algorithm to obtain the optimal speed performance parameters for consistency of the multi-robot system, and then obtain the safety defense control strategy.