DOS Attacks and Unknown Time-Varying Gain-Based Swarm Containment Control Methods
By using the adaptive parameter update law and the distributed cooperative control of the Nussbaum function, the instability problem of unmanned surface vessel (USV) swarms under DOS attacks and actuator failures was solved, and safe asymptotic convergence and fully distributed adaptive disturbance rejection control of USV swarms were achieved.
Patent Information
- Application Number
- CN202611618089.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing unmanned surface vessel (USV) swarm containment control methods are prone to instability when faced with DoS attacks, actuator polarity reversal, and complex ocean disturbances. Furthermore, they rely on global topology information and lack fully distributed autonomous capabilities, leading to control system divergence or collision accidents.
A distributed cooperative control architecture based on adaptive parameter update law and Nussbaum function is adopted. By constructing a communication link state switching model and calculating local errors, the safe course dynamic guidance of the unmanned surface vessel swarm is realized, reducing communication overhead and improving robustness to actuator failures.
Under DOS attacks and actuator failures, it ensures the asymptotic convergence of the unmanned surface vessel swarm, avoids runaway collisions, achieves fully distributed adaptive disturbance rejection control, and possesses both flexibility and security for engineering deployment.
Smart Images

Figure CN123113841A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of multi-agent cooperative control and marine unmanned equipment technology, and more specifically, to a method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gains. Background Technology
[0002] Unmanned surface vessel (USV) swarm collaborative control technology has demonstrated significant engineering application potential in fields such as marine environmental monitoring, resource exploration, and operations in hazardous waters. Among numerous collaborative control tasks, inclusive control aims to construct a safe, dynamic zone using multiple lead USVs with advanced detection capabilities, driving all follower USVs undertaking specific tasks to enter and remain within this safe zone. Compared to traditional single-leader follower control, inclusive control provides the swarm with a wider and safer operating space, making it particularly suitable for complex and hazardous marine environments filled with unknowns.
[0003] Existing unmanned surface vessel (USV) swarm containment control methods typically rely on ideal, continuous, high-frequency communication interactions to acquire state information from neighboring nodes. However, in the harsh and complex real-world marine operating environment, open wireless communication networks not only face bandwidth limitations but are also highly vulnerable to malicious cyber-physical attacks, especially Denial of Service (DoS) attacks. Traditional strategies for dealing with DoS attacks largely rely on the macroscopic assumption of a complete network shutdown, assuming that communication across all nodes will be synchronously interrupted in the event of an attack. In actual combat or complex sea conditions, DoS attacks often only result in the disruption of local communication links. In cases of localized damage, if the attacked node uses only traditional zero-order hold mechanisms with severely lagging historical interaction data, it will generate huge, distorted tracking errors. If the underlying adaptive law continues to use these distorted errors for integral updates, this will lead to error integral saturation, ultimately causing the control system to diverge. Furthermore, most existing distributed control protocols still heavily rely on the eigenvalues of the Laplace matrix of the global communication network topology when calculating feedback coupling gains, lacking fully distributed autonomous capabilities.
[0004] Furthermore, during prolonged and high-load actual maritime operations, unmanned surface vessels (USVs) frequently experience unknown physical wear, aging, and even partial failure of their underlying power actuators due to the harsh environment. This inevitably leads to serious malfunctions such as unknown actuator control gain or even polarity reversal. When the control direction of the USV system experiences unpredictable random fluctuations (i.e., unknown polarity of the variable control coefficients), traditional adaptive or robust control algorithms that strictly rely on prior knowledge of the control direction will completely fail. If conventional control strategies are still used when the control gain is unknown, it can easily lead to system divergence, or even serious accidents such as uncontrolled collisions of swarmed USVs.
[0005] Therefore, how to design an adaptive cooperative control strategy that does not require global topology information, has the ability to cope with DOS attacks, and can strictly guarantee the asymptotic convergence of inclusive control under the combined extreme conditions of unknown marine environmental interference, frequent DOS attacks, and especially the polarity reversal of unmanned surface vessel actuators, is a key technical problem that urgently needs to be solved in the field of unmanned system control. Summary of the Invention
[0006] To overcome the problems of DoS attacks, strong dependence on global network topology, and susceptibility to instability in existing marine unmanned surface vessel (USV) swarm cooperative control systems when faced with actuator polarity reversal and complex marine disturbances, this invention provides a USV swarm containment control method under DoS attacks and unknown time-varying gain. This method constructs a distributed cooperative control architecture including an adaptive parameter update law and a Nussbaum function to achieve dynamic and safe heading guidance for multiple USV swarms. This significantly reduces system communication overhead and substantially improves robustness to underlying actuator failures while ensuring zero steady-state containment error for the swarm.
[0007] To achieve the above objectives, the present invention adopts the following technical solution: Step 1: Obtain the communication topology and initial state of the unmanned surface vessel (USV) swarm, and collect the heading angle and heading angular velocity of the following USVs; Step 2: Establish a communication link state switching model under a DOS attack. At each communication sampling time, determine whether the neighbor state data packet arrives within the preset reception time limit. If it arrives on time, update the neighbor state data. If it does not arrive on time, determine that the corresponding communication link is interrupted and trigger the zero-order hold to continue using the most recently successfully received neighbor state data. Step 3: Based on the real-time neighbor state data or zero-order hold-up state data obtained in Step 2, calculate the local tracking error of the heading angle in a distributed manner according to the topological weights of the directed communication graph. Step 4: Construct a first-order virtual control law using the backstepping method based on the local tracking error of the heading angle and the preset gain constant; Step 5: Calculate the local tracking error of the heading angular velocity according to the topology weights, and extract the comprehensive tracking error variable by subtracting it from the first-order virtual control law; Step 6: Update the dual adaptive parameters and Nussbaum state adjustment variables of the node communication status control determined in Step 2; update according to the corresponding update law when communication is normal, pause the update when communication is interrupted, and continue the update after communication is restored; Step 7: Generate an auxiliary control signal from the integrated tracking error variable and adaptive gain, and generate a continuous heading rudder angle control command through Nussbaum function direction probing; Step 8: Send the heading rudder angle control command to the bottom-level actuator to make the following unmanned surface vessel gradually converge into the safety convex hull formed by the lead unmanned surface vessel.
[0008] Further, step 1 specifically includes: S1.1: establishing a directed graph model of communication including the lead unmanned surface vessel (USV) and the follower USV, used to describe the information interaction network of unidirectional topological flow between nodes within the USV swarm system; S1.2: using the USV's onboard sensing equipment to collect real-time state data during the USV's daily navigation process; S1.3: the real-time state data includes the heading angle state and the heading angular velocity state, used to characterize the heading dynamic characteristics of the USV under a single degree of freedom.
[0009] Further, step 2 specifically includes: S2.1: Establishing a communication topology switching model under a DoS attack, equating the DoS attack to the switching of the available state of the communication link; at each communication sampling time, determining whether the following unmanned surface vessel (USV) has received a valid status data packet sent by a neighboring node within a preset reception time limit; S2.2: If a valid status data packet is received within the preset reception time limit, the corresponding communication link is marked as normal, and the neighboring state is updated using the currently received data; if no valid status data packet is received within the preset reception time limit, the corresponding communication link is marked as interrupted, triggering a zero-order hold to latch and reuse the neighboring state data that was last successfully received by the link; S2.3: Determining the node communication state based on the on / off state of each receiving link of the following USV, and using the node communication state as a trigger signal for normal parameter update or paused update.
[0010] Further, step 3 specifically includes: S3.1: obtaining the effective neighbor state data determined in step 2; the effective neighbor state data includes the real-time state data received when communication is normal, and the state data of the most recently successfully received state held by the zero-order hold when communication is interrupted; S3.2: calculating the local tracking error of the bow angle of the following unmanned surface vessel using the effective neighbor state data according to the adjacency matrix weight of the communication directed graph. This error integrates the relative bow angle deviation between the following unmanned surface vessel and the adjacent following unmanned surface vessel and the lead unmanned surface vessel.
[0011] Further, step 4 specifically includes: S4.1: Introducing a strictly positive feedback gain constant; S4.2: Combining the local tracking error of the heading angle, constructing a first-order virtual control law of the backstepping method, wherein the first-order virtual control law is used to provide an ideal reference convergence trajectory for the heading angular velocity of the following unmanned surface vessel.
[0012] Further, step 5 specifically includes: S5.1: Calculate the local tracking error of the bow angular velocity of the following unmanned surface vessel based on the communication network topology weight and its own real-time bow angular velocity; S5.2: Extract the comprehensive tracking error variable by subtracting the local tracking error of the bow angular velocity from the first-order virtual control law obtained in step 4.
[0013] Further, step 6 specifically includes: S6.1: To compensate for unknown bounded external disturbances and nonlinear hydrodynamic damping in the marine environment, a dual adaptive update law is designed for the comprehensive tracking error variable; S6.2: The first adaptive update law uses the square of the comprehensive tracking error variable as the driving term and the adaptive rate as the update gain for integral update, used to estimate the gain of the upper bound of the system's lumped uncertainty online; S6.3: The second adaptive update law uses the product of the comprehensive tracking error variable and the hyperbolic tangent function as the driving term for integral update, where the independent variable of the hyperbolic tangent function is the quotient of the comprehensive tracking error variable and the exponential decay function, used to smooth and compensate for the error; S6.4: The update of the dual adaptive parameters and Nussbaum state adjustment variables is controlled according to the node communication status determined in step S2; when node communication is normal, updates are performed according to the corresponding update law; when node communication is interrupted, the update rate of the corresponding variable is set to zero and the value before the communication interruption is maintained; when node communication is restored, parameter updates continue, thereby preventing the continuous accumulation of parameters caused by long-term use of historical data.
[0014] Further, step 7 specifically includes: S7.1: Constructing an auxiliary control signal based on the adaptive gain output by the dual adaptive update law in step 6; S7.2: Calculating the Nussbaum adjustment variable using the product of the auxiliary control signal and the comprehensive tracking error variable as the integral input, and selecting a Nussbaum function whose amplitude increases with the increase of the independent variable and whose sign periodically reverses; S7.3: Inputting the auxiliary control signal into the Nussbaum function to generate the final continuous heading rudder angle control command, so as to dynamically offset the adverse effects caused by the unknown control gain or polarity reversal of the unmanned surface vessel actuator.
[0015] Further, step 8 specifically includes: S8.1: converting the continuous heading angle control command generated in step 7 into a level drive signal that the underlying actuator can recognize; S8.2: driving the heading angle actuator of the following unmanned surface vessel so that when encountering unknown marine environmental disturbances and actuator polarity reversal failures, the heading angle can still asymptotically converge to the safe heading sector formed by the lead unmanned surface vessel, achieving zero steady-state error cluster cooperative inclusive control.
[0016] This invention also provides a distributed resilient containment control system for unmanned surface vessel (USV) swarms. The airborne controller deployed on each following USV includes: a data acquisition module for acquiring swarm communication topology information and collecting the heading angle and heading angular velocity of the following USVs; a communication status determination and latching module for determining the communication link status based on whether neighbor status data packets arrive within a preset reception time limit; updating neighbor status data when communication is normal, and triggering a zero-order hold to retain the most recently successfully received neighbor status data when communication is interrupted; and an error calculation module for distributed calculation of the heading angle according to the topology weights of the communication directed graph. The system includes: local tracking error of the angular velocity and local tracking error of the bow angular velocity; a virtual control module for constructing a first-order virtual control law using the backstepping method and extracting the comprehensive tracking error variable; an adaptive estimation module for updating the dual adaptive parameters and Nussbaum state adjustment variables based on the node communication status, pausing updates when communication is interrupted and resuming updates after communication is restored; and a rudder angle command generation module for fusing the comprehensive tracking error variable and adaptive gain to generate auxiliary control signals, generating continuous bow rudder angle control commands through Nussbaum function direction probing and sending them to the underlying actuators.
[0017] The present invention also provides a computer-readable storage medium storing a distributed resilient containment control program for an unmanned surface vessel (USV) swarm, wherein the USV swarm distributed resilient containment control program, when executed by a processor, implements the aforementioned USV swarm distributed resilient containment control method.
[0018] Compared with the prior art, the present invention has the following beneficial technical effects: First, this invention innovatively proposes a link state determination and data latching mechanism to address DoS attacks, solving the problem of network loss of control caused by damage in complex sea conditions. A DoS attack is transformed into an interruption of inter-node communication links. When a receiving node does not receive a valid status data packet from a neighboring node within a preset reception time limit, the corresponding communication link is determined to be interrupted. Only the receiving node corresponding to the interrupted link triggers a zero-order hold, using the most recently successfully received neighbor status data. Other uninterrupted communication links continue normal data transmission, maximizing the preservation of the cluster's local autonomy and collaborative capabilities.
[0019] Secondly, this invention effectively overcomes the instability risk caused by unmanned surface vessel (USV) actuator failure. Addressing the issue of unknown control gain or even polarity reversal in USVs due to wear and aging, this invention introduces the Nussbaum gain function into an inclusive control architecture. The system does not require prior knowledge of the control direction; it can automatically offset the damage caused by unknown time-varying coefficients through dynamic sign probing and gain adjustment, ensuring the absolute safety of the control system under extreme hardware failures.
[0020] Third, this invention achieves fully distributed adaptive disturbance rejection control that is free from global topology dependence. Unlike traditional control protocols that require prior knowledge of the global network topology (such as the eigenvalues of the Laplace matrix), the adaptive update law designed in this invention relies only on local neighbor node interaction information. The system does not require any prior knowledge of the upper bound of external ocean wind and wave disturbances or unmodeled nonlinear damping, and can dynamically estimate and smoothly compensate for lumped disturbances online. This enables unmanned surface vessel swarms to have truly fully distributed, plug-and-play engineering deployment capabilities. Attached Figure Description
[0021] Figure 1 This is a flowchart illustrating the overall implementation of the present invention; Figure 2 This is a digital simulation of the unmanned surface vessel node communication topology diagram of this invention; Figure 3 This is a diagram of the heading angle containment control trajectory obtained through digital simulation of the present invention. Figure 4 This is a diagram of the bow angular velocity containment control trajectory obtained through digital simulation of this invention. Figure 5 The diagrams show the errors in heading angle and heading angular velocity in the digital simulation of this invention, where (a) is the error diagram of heading angle in the digital simulation and (b) is the error diagram of heading angular velocity in the digital simulation. Figure 6 This is a plot of the Nussbaum function trajectory from the digital simulation of this invention; Figure 7 The diagram shows the adaptive parameter trajectory of the digital simulation of the present invention, wherein (a) is the first adaptive parameter trajectory and (b) is the second adaptive parameter trajectory. Figure 8 This is a comparison diagram of the heading angle containment control trajectory between this method and the traditional PID method. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention.
[0023] To facilitate understanding, the main terms involved in this invention will first be explained: (1) DOS attack: Denial of service attack refers to a malicious network physical attack in which the attacker occupies or blocks communication resources, making the target node unable to send and receive network data normally for a period of time; (2) Zero-order hold: A signal holding mechanism that holds the sampled value at the last moment before the network outage as a constant during communication interruption; (3) Enclosed control: a distributed cooperative control task that drives multiple following agents to converge and remain within a dynamic convex hull region enclosed by multiple leading agents; (4) Safety convex hull: The smallest convex area defined by the heading angles of multiple pilot unmanned surface vessels, i.e., the safe heading sector; (5) Backstepping method: A nonlinear control design method that constructs virtual control laws step by step and backsteps to obtain the actual control input for strictly feedback nonlinear systems; (6) Nussbaum function: a type of gain trial function with sign-changing characteristics. Its function value alternates between positive and negative as the independent variable increases and its amplitude increases without bounds. It is used to detect the true control direction online when the control direction is unknown. (7) Time-varying control gain: The gain coefficient of the input channel of the unmanned surface vessel actuator changes with time. Its unknown polarity means that the control direction is unknown. When the polarity is negative, it corresponds to the actuator polarity reversal fault. (8) Laplace matrix: A matrix that describes the global topology of a communication network, whose eigenvalues depend on the network connectivity. (9) Lumped uncertainty: The uncertainty term is composed of marine environmental disturbances and unmodeled nonlinear hydrodynamic damping in the unmanned surface vessel model.
[0024] First, establish the first [unmanned surface vessel] system. A single-degree-of-freedom nonlinear bow motion mathematical model for a following unmanned surface vessel (USV). This invention employs a nonlinear Norrbin model, which accurately reflects the nonlinear hydrodynamic characteristics of the rudder angle, to describe the bow dynamics of the USV in complex marine environments. Considering unknown failures such as actuator aging, wear, or polarity reversal, the first... The kinematic and dynamic equations of a following unmanned surface vessel can be uniformly expressed in the following state-space form:
[0025] in, Indicates the first The actual heading angle of the following unmanned surface vessel; Indicates the first The actual heading angular velocity state of the following unmanned surface vessel; This represents the inherent unmodeled nonlinear hydrodynamic damping function of the unmanned surface vessel; This represents the continuous heading angle control command calculated by the controller, i.e., the actual control input of the system; The unknown time-varying control gain coefficient represents the bow-side actuator at the bottom of the unmanned surface vessel; This refers to the unknown, bounded, lumped external environmental disturbances encountered by unmanned surface vessels during navigation, caused by sea winds, waves, and ocean currents.
[0026] This invention provides a mathematical model of DoS attacks along a timeline. Assume the... The communication network following the unmanned surface vessel throughout the entire operation time The system suffered multiple localized DoS attacks. (Definition of the first...) The time interval of this DOS attack is: ,in Indicates the first The moment the attack began Indicates the first The moment when the attack ends and communication is restored. Therefore, the node The set of times when an attack occurs can be represented as Accordingly, nodes The set of times in normal communication state is .
[0027] Accordingly, nodes Neighbor nodes used for controlling computation The valid state data is represented as :
[0028] To counter DoS attacks, a communication link between any two nodes is arbitrarily selected and severed during each attack period. It's important to note that the DoS attack time interval is used to describe the mathematical model of communication link disruption. The controller determines whether the corresponding receiving link is interrupted online based on whether neighbor status data packets arrive within a preset reception time limit. The channel disruption caused by a DoS attack manifests as a timeout in the reception of neighbor status data packets.
[0029] Based on the above model, a specific implementation of the distributed resilient containment control method of the present invention in each step is as follows.
[0030] Step 1 specifically includes: S1.1: Establish a directed graph model of communication including the lead unmanned surface vessel (USV) and the follower USV. The group of unmanned surface vessels is represented as , It is a directed graph model for communication, used to abstractly represent the communication network inside the entire unmanned surface vessel swarm system, consisting of nodes (unmanned surface vessels) and edges (communication links); Let be the set of vertices of the graph, representing all unmanned surface vessel nodes in the cluster (including navigators and followers). The set of edges in the graph represents the directed communication links between nodes of the unmanned surface vessel. This is the adjacency matrix of a directed graph, used to quantify the weights of communication links. The set of pilot unmanned surface vessels is represented as... , used to describe the information interaction network between nodes in an unmanned surface vessel swarm system with unidirectional topological flow; S1.2: Utilize the underlying sensing devices mounted on each following unmanned surface vessel (USV) to collect real-time and continuous single-degree-of-freedom bow motion physical state data of each USV during the collaborative maritime operations of the USV swarm. S1.3: The real state data includes the current time. Actual bow angle state (Corresponding to the actual bow angle of the unmanned surface vessel in the physical sense, in radians (rad)) and the actual bow angular velocity state. (Corresponding to the actual heading rate of the unmanned surface vessel in the physical sense, in rad / s), this is used to characterize the heading dynamics of the unmanned surface vessel under a single degree of freedom, and input it into the onboard microprocessor control motherboard in real time.
[0031] Step 2 specifically includes: S2.1: Let the communication sampling time be... ,in The communication sampling period is defined as follows: neighboring nodes send status data packets containing time stamps at each communication sampling time, and receiving nodes set preset reception time limits. S2.2: Node Determine whether the node was received within the preset reception time limit. Send valid status data packets and define communication link state variables. ;
[0032] when At that time, determine the communication link Normal; when When this happens, the communication link is determined to be interrupted.
[0033] S2.3: Define Nodes The actual neighbor status used is . For nodes In the previous moment Neighbor nodes used Valid state data; For nodes At any moment Real-time status data; S2.4: Based on nodes Communication state variables of each receiving link determining the node's communication state. When node When all necessary receiving links are functioning correctly, the node's communication state variable is set to 1; when any necessary receiving link is interrupted, the node's communication state variable is set to 0, and this is used as a trigger signal for updating or pausing the dual adaptive parameters and the Nussbaum state adjustment variable. This can be written as... , For a moment link Link communication state variables.
[0034] Step 3 specifically includes: S3.1: Obtain the valid neighbor state data obtained in step 2. When the communication link is normal, the valid neighbor state data is the current real-time state of the neighbor node; when the communication link is interrupted, the valid neighbor state data is the neighbor state of the most recently successfully received communication stored in the zero-order hold. S3.2: Communication weights based on the directed topological adjacency matrix The tracking error of the local bow angle of the outer ring of each following unmanned surface vessel is calculated, and its specific calculation expression is as follows: This error incorporates the relative angular deviation between the current following node and its neighboring unmanned surface vessels (USVs) and multiple lead USVs, and is used to accurately measure the relative orientation of the current following node within the safe heading range of the entire cluster.
[0035] Step 4 specifically includes: S4.1: Introduce a strictly positive feedback gain adjustment constant into the algorithm. This parameter, as an adjustable controller design index, directly determines the asymptotic convergence speed of the outer-loop position tracking error of the multi-unmanned surface vessel system. S4.2: Combining the local heading angle tracking error of the outer ring of the position. Construct a first-order virtual control law using the backstepping method, the mathematical expression of which is as follows: .
[0036] Step 5 specifically includes: S5.1: Combining the topological connection weights of the directed adjacency matrix in the communication network, calculate the dynamic local bow angular velocity tracking error of the following unmanned surface vessel. Its specific expression is as follows: ;x i,2 For the first The actual heading angular velocity of the unmanned surface vessel following the unmanned surface vessel. For nodes The actual neighbor nodes used The effective state of the heading angular velocity; S5.2: The calculated local tracking error of the heading angular velocity. Compared with the first-order ideal reference virtual control law generated in step 4 Perform a difference operation to extract the combined tracking error variable of the coupled position deviation and velocity deviation. It serves as the core driving signal of the algorithm, directly guiding the generation of adaptive disturbance rejection estimation and actual heading angle control quantity.
[0037] Step 6 specifically includes: S6.1: Addressing unknown bounded lumped external disturbances caused by wind, waves, and currents in extreme marine operating environments. In addition to the nonlinear hydrodynamic damping dynamics not modeled in the unmanned surface vessel model, a system based on comprehensive tracking error is constructed. The driver's real-time parameter estimation and update mechanism; S6.2: Design the first adaptive estimation update law. The differential equation form of the first adaptive parameter is defined as follows: , The first adaptive learning rate is used to estimate the boundary gain of the system's lumped uncertainty online, so that the controller can complete the compensation without any prior knowledge of the disturbance boundary. S6.3: Design the second adaptive estimation update law. The differential equation form of the second adaptive parameter is defined as follows: ,in, The second adaptive learning rate, Indicates the adaptive learning rate. This represents a predefined exponentially decaying, continuously smooth envelope function. Let be the initial amplitude constant of the envelope function; this update law, based on Barbalat's lemma in control theory, introduces a hyperbolic tangent function to replace the traditional sign function to eliminate the high-frequency chattering phenomenon of the underlying servo motor that easily occurs in robust control, but it will introduce nonlinear approximation error. The control is set to an exponentially decaying term to achieve continuous approximation and asymptotic convergence of the system; S6.4: Based on the node communication status determined in step 2 Control the updating of the dual adaptive parameters and the Nussbaum state adjustment variables. When At this time, node communication is normal, and the dual adaptive parameters and Nussbaum state adjustment variables are updated normally according to the corresponding update law; when When node communication is interrupted, the update rate of the above variables is set to zero to maintain their values before the communication interruption, while the effective neighbor state data output by the zero-order hold is used to calculate control commands; when node communication is restored .... Restore the update of the corresponding variables.
[0038] Step 7 specifically includes: S7.1: Integrated tracking error in deep fusion step 5 The dual adaptive time-varying estimation gain calculated online in step 6 is used to construct a continuous auxiliary control signal for stabilizing state deviation. Its specific calculation expression is as follows: ; S7.2: Calculate the Nussbaum state adjustment variable using the product of the auxiliary control signal and the integrated tracking error as input. derivative relationship = And select a continuously differentiable high-gain Nussbaum function. Its specific form is ; S7.3: Multiply the auxiliary control signal by the Nussbaum gain term to calculate the final output continuous heading angle control command. Addressing the extreme defects of unknown control gain and polarity reversal in unmanned surface vessel actuators caused by long-term seawater erosion and aging mechanical failures, this mechanism dynamically offsets unknown time-varying coefficients through the sign-changing and active sign-probing capabilities of the Nussbaum function. The system achieves online physical adaptive correction of the control direction to prevent instability and damage caused by the control system.
[0039] Step 8 specifically includes: S8.1: Calculate the continuous heading angle control commands obtained by the central processing unit in step 7. The digital-to-analog converter on the control motherboard converts the signals into drive voltage, current, or pulse width modulation commands that the forward actuator can directly recognize. S8.2: Drive each following unmanned surface vessel to perform underlying physical heading angle maneuvering adjustments, so that even under extreme sea conditions such as strong external wave impacts, limited local communication bandwidth, and actuator direction polarity reversal, the heading angle and heading angular velocity trajectories of all following unmanned surface vessels can still asymptotically converge with zero steady-state error and remain within the absolutely safe heading area (i.e., inside the geometric convex hull region) dynamically defined by multiple lead unmanned surface vessels, ultimately completing the fault-resistant distributed collaborative containment control task of the swarm.
[0040] To further verify the practical engineering feasibility and specific beneficial technical effects of the control method described in this invention, this embodiment conducts a detailed digital simulation analysis based on the physical parameters of the unmanned surface vessel formation at the scale of an actual physical model: The digital simulation verification environment configuration for an unmanned surface vessel (USV) swarm containment control method under DOS attacks and unknown time-varying gain provided by this invention is as follows: The system is configured with five follower USVs (one to five) and two lead USVs serving as virtual command boundaries. To comprehensively test the online hard correction capability of this invention against actuator polarity reversal faults, the time-varying unknown control gain coefficient of the system is... Perform the following extreme injection settings: The aforementioned unmanned surface vessels (USVs) one, two, and five maintain normal positive control polarity; however, for USVs three and four, a reverse fault control polarity is intentionally injected (i.e., simulating a reverse servo motor fault). The dynamic reference boundary lines generated by the two pilot unmanned surface vessels are set as a combination of sine and cosine trajectories, respectively. radian and Arc. The initial yaw pose of each vessel is randomly constructed in... Within the scope of... The interference design in this paper is as follows: The parameters of the adaptive control law here are: , , , And the feedback parameter in the backstepping method is At the same time at the initial moment and The system then adjusts according to the magnitude of the error. To ensure the objectivity and repeatability of the comparative experiment, the parameters of the traditional distributed PID controller are set as follows: proportional coefficient. Integral coefficient Differential coefficients The reference parameter has been properly tuned with the system in its normal polarity.
[0041] Figure 2 The simulation shows the communication topology of unmanned surface vessel nodes under normal conditions. During a DoS attack, a communication link is randomly selected and severed. The simulation evolution results are as follows: Figures 3 to 6 As shown. Figure 3 and Figure 4The heading angle and heading angular velocity tracking trajectories of the cluster are displayed. It can be clearly observed that in the initial stage, due to the completely unknown control polarity, the heading trajectories of the two following unmanned surface vessels (USVs) experiencing reverse control polarity faults exhibit momentary fluctuations. This fully demonstrates that the Nussbaum function introduced in this invention is efficiently probing the sign change direction under the drive of the underlying microprocessor. The red lines simulate the DOS attack interval; overall, although the curve fluctuations are affected during the DOS attack, it does not affect the final achieved result. Once the online physical adaptive hedging correction of the control direction is completed through variable integration, the severe oscillations immediately and rapidly subside with a very steep slope. Ultimately, the heading angle trajectories of all five following USVs asymptotically approach and remain stably within the dynamic safe heading interval (i.e., the safe heading sector) defined by Navigator 1 and Navigator 2 with zero steady-state error. Figure 5 The local error plots representing the bow angle and bow angular velocity of the unmanned surface vessel show that the overall error tends towards zero. While the impact of the DOS attack causes some variation in the error, it still generally approaches zero. From a quantitative perspective, after experiencing the initial stage... to During the intense probing phase of the Nussbaum probe, a transient overshoot occurred in the bow angular velocity due to polarity reversal, with a maximum peak value of approximately [missing information]. The system control direction was successfully corrected. Subsequently, the cluster... It can quickly achieve heading convergence. After entering the steady-state operation phase, even in the range where local DOS attacks occur frequently, the absolute tracking error of each following UAV from the safe heading boundary formed by the lead UAV still tends to zero.
[0042] Figure 6 The Nussbaum function was demonstrated. The internal evolution curves show that the gains of each boat rapidly and smoothly converge to a bounded constant state after the transient search ends, confirming the overall bounded stability of the control system. Figure 7 The adaptive parameter estimation curves show that and The estimation is monotonically increased only when dealing with external wave disturbances and unknown polarity. Once the uncertainty is completely dominated by the parameters, the estimation action automatically stops, proving that the dual adaptive mechanism is completely decoupled from the knowledge of the upper bound of prior environmental disturbances.
[0043] Figure 8This paper demonstrates a comparison of the containment performance of our proposed method with that of traditional distributed PID control strategies. Because traditional PID control lacks a mechanism for detecting unknown control directions, its inherent negative feedback regulation deteriorates into positive feedback when encountering a polarity reversal fault, causing the system to diverge rapidly (as shown in Figure 6, following the unmanned surface vessel). Simultaneously, this local divergence, through the coupling of the communication topology, forcibly drags healthy nodes (as shown in Figure 7, following the unmanned surface vessel) out of the safe containment region. In contrast, our proposed method introduces a Nussbaum function for dynamic probing and gain adjustment, ensuring that all following unmanned surface vessels can still converge smoothly and with zero error within the safe containment region even under extreme conditions of polarity reversal.
[0044] In one embodiment, the present invention also provides a distributed resilient containment control system for unmanned surface vessel (USV) swarms, including a data acquisition module, a communication latching module, an error calculation module, a virtual control module, an adaptive estimation module, and a rudder angle command generation module deployed in the onboard controllers of each following USV. The data acquisition module acquires the communication topology and initial state information of the USV swarm, and collects the actual heading angle and actual heading angular velocity of the following USVs in real time. The communication state determination and latching module determines the communication link state based on whether neighbor status data packets arrive within a preset reception time limit. When communication is normal, it updates the neighbor status data; when communication is interrupted, it triggers a zero-order hold to continue using the most recently successfully received neighbor status data. The adaptive estimation module updates the dual adaptive parameters and Nussbaum state adjustment variables based on the node communication status. Updates are paused when communication is interrupted and resumed after communication is restored. The error calculation module combines the topological weights of the directed graph of communication to distribute and calculate the local tracking error of the heading angle and the local tracking error of the heading angular velocity. The virtual control module constructs a first-order virtual control law using the backstepping method based on the local tracking error of the heading angle and extracts the comprehensive tracking error variable. The adaptive estimation module drives the update of adaptive parameters and Nussbaum parameters by the comprehensive tracking error variable to estimate unknown environmental disturbances online, and triggers a parameter pause mechanism to suspend parameter updates when a node suffers a DoS attack. The rudder angle command generation module integrates the comprehensive tracking error variable and adaptive gain to generate auxiliary control signals, performs direction probing and gain adjustment using the Nussbaum function to generate continuous heading rudder angle control commands, and sends them to the underlying actuators. All modules collaboratively execute steps 1 to 8 of the above method to achieve the same beneficial effects as the method implementation.
[0045] In one embodiment, the present invention also provides a computer-readable storage medium storing a distributed resilient containment control program for an unmanned surface vessel (USV) swarm. When executed by a processor, this program implements steps 1 to 8 of the method described above. The processor can be the main control chip in the USV's onboard microprocessor control motherboard.
[0046] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain, characterized by: Step 1: Obtain the communication topology and initial state of the unmanned surface vessel (USV) swarm, and collect the heading angle and heading angular velocity of the following USVs; Step 2: At each communication sampling time, determine whether the following unmanned surface vessel has received the status data packet of the neighbor node within the preset reception time limit; if it is received on time, update the neighbor status data; If the data is not received on time, the corresponding receiving link is determined to be interrupted, and the zero-order hold is triggered to continue using the neighbor state data that was most recently successfully received. Step 3: Based on the real-time neighbor state data or zero-order hold-up state data obtained in Step 2, calculate the local tracking error of the heading angle in a distributed manner according to the nominal topological weights of the directed graph of the communication. Step 4: Construct a first-order virtual control law using the backstepping method based on the local tracking error of the heading angle and the preset gain constant; Step 5: Calculate the local tracking error of the heading angular velocity according to the topology weights, and extract the comprehensive tracking error variable by subtracting it from the first-order virtual control law; Step 6: Normal communication nodes update the adaptive and Nussbaum parameters based on the comprehensive tracking error variables; When communication is interrupted, the node suspends the updating of the corresponding variables and resumes updating after communication is restored. Step 7: Generate an auxiliary control signal from the integrated tracking error variable and adaptive gain, and generate a continuous heading rudder angle control command through Nussbaum function direction probing; Step 8: Send the heading rudder angle control command to the bottom-level actuator to make the following unmanned surface vessel gradually converge into the safety convex hull formed by the lead unmanned surface vessel.
2. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 1, characterized in that, Step 1 specifically includes: S1.1: Establish a directed graph model of communication including the lead unmanned surface vessel (USV) and the follower USV to describe the information interaction network between the nodes in the USV swarm system with unidirectional topological flow. S1.2: Use the onboard sensing equipment of the unmanned surface vessel to collect real-time status data during the daily navigation process of the unmanned surface vessel; S1.3: The real state data includes the heading angle state and the heading angular velocity state, which are used to characterize the heading dynamics of the unmanned surface vessel under a single degree of freedom.
3. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 2, characterized in that, Step 2 specifically includes: S2.1: Establish a communication topology switching model under a DOS attack. At each communication sampling time, determine whether the following unmanned surface vessel has received a valid status data packet sent by the adjacent node within the preset reception time limit. S2.2: If a valid status data packet is received within the preset reception time limit, the corresponding communication link is marked as normal, and the neighbor status is updated using the currently received data; if no valid status data packet is received within the preset reception time limit, the corresponding communication link is marked as interrupted, triggering the zero-order hold, latching and using the neighbor status data that was last successfully received by the link. S2.3: Determine the node communication status based on the on / off status of each receiving link of the following unmanned surface vessel, and use the node communication status as a trigger signal for parameter update or pause.
4. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 3, characterized in that, Step 3 specifically includes: S3.1: Obtain the valid neighbor status data determined in step 2; the valid neighbor status data includes the real-time status data received when communication is normal, and the status data of the most recently successfully received data held by the zero-order hold when communication is interrupted; S3.2: Calculate the local heading angle tracking error of the following unmanned surface vessel (USV) based on the adjacency matrix weights of the directed communication graph. This error incorporates the relative heading angle deviation between the following USV node and its neighboring USVs and the lead USV.
5. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 4, characterized in that, Step 4 specifically includes: S4.1: Introduce a strictly positive feedback gain constant; S4.2: Combining the local tracking error of the heading angle, a first-order virtual control law of the backstepping method is constructed. The first-order virtual control law is used to provide a reference convergence trajectory for the heading angular velocity of the following unmanned surface vessel.
6. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 5, characterized in that, Step 5 specifically includes: S5.1: Calculate the local tracking error of the bow angular velocity of the following unmanned surface vessel based on the communication network topology weight and its own real-time bow angular velocity; S5.2: Subtract the local tracking error of the heading angular velocity from the first-order virtual control law obtained in step 4 to extract the comprehensive tracking error variable.
7. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 6, characterized in that, Step 6 specifically includes: S6.1: To compensate for unknown bounded external disturbances and nonlinear hydrodynamic damping in the marine environment, a dual adaptive update law is designed for the comprehensive tracking error variable; S6.2: The first adaptive update law uses the square of the comprehensive tracking error variable as the driving term and the adaptive rate as the update gain for integral update, and is used to estimate the gain of the upper bound of the system's lumped uncertainty online. S6.3: The second adaptive update law uses the product of the comprehensive tracking error variable and the hyperbolic tangent function as the driving term for integral update. The independent variable of the hyperbolic tangent function is the quotient of the comprehensive tracking error variable and the exponential decay function, which is used to smooth and compensate for the error. S6.4: Update the node communication status control parameters determined in step 2; when the node communication status is interrupted, set the update rate of the dual adaptive parameters and Nussbaum status adjustment variables to zero; when the node receives valid neighbor status data again, resume the update of the corresponding variables.
8. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 7, characterized in that, Step 7 specifically includes: S7.1: Construct an auxiliary control signal based on the adaptive gain output by the dual adaptive update law in step 6; S7.2: Using the product of the auxiliary control signal and the comprehensive tracking error variable as the integral input, calculate the Nussbaum adjustment variable, and select the Nussbaum function whose amplitude increases with the increase of the independent variable and whose sign reverses periodically; S7.3: Input the auxiliary control signal into the Nussbaum function to generate the final continuous heading rudder angle control command, dynamically offsetting the effects caused by unknown control gain or polarity reversal of the unmanned surface vessel actuator.
9. The method for containment control of unmanned surface vessel swarms under DOS attacks and unknown time-varying gain as described in claim 8, characterized in that, Step 8 specifically includes: S8.1: Convert the continuous heading rudder angle control command generated in step 7 into a level drive signal that can be recognized by the underlying actuator; S8.2: Drives the bow angle actuator of the following unmanned surface vessel, so that when it encounters unknown marine environmental disturbances and actuator polarity reversal failures, the bow angle can still asymptotically converge to the safe heading sector formed by the lead unmanned surface vessel, achieving zero steady-state error swarm cooperative inclusive control.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a distributed resilient containment control program for an unmanned surface vessel (USV) swarm. When executed by a processor, the USV swarm containment control program implements the USV swarm containment control method under DOS attacks and unknown time-varying gain as described in any one of claims 1 to 9.
