Undisturbed attack detection and safe dynamic positioning method for switching unmanned ship system

Abstract

The invention discloses an undisturbed attack detection and safe dynamic positioning method for switching an unmanned ship system, and the method comprises the steps: obtaining a state and mode switching signal based on a quality switching model of a hysteresis quantizer, and designing an event triggering protocol with an interpolation interval for sampling transmission; constructing a state observer to generate a state residual error, constructing an undisturbed rate residual error based on exponential interpolation to eliminate derivative interference caused by switching, further designing an attack detection function, and generating a safety controller according to a detection result; a controller-actuator event triggering protocol is established, and a control signal is protected in combination with a watermark encryption mechanism; constructing a closed-loop system by integrating two types of trigger protocols, constructing and satisfying implementation conditions of safe dynamic positioning, and completing safe positioning control of the quality switching unmanned ship; according to the method, detection false alarm and missing alarm caused by rate residual buffeting caused by switching behaviors are overcome, and it is ensured that the unmanned ship effectively resists cheating attacks under mass switching.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cooperative dynamic positioning of unmanned ships, and particularly relates to a method for detecting non-interference attacks and safe dynamic positioning of an unmanned ship system. BACKGROUND

[0002] The ocean, as a vast area covering more than 70% of the earth's surface, contains natural resources and strategic value that are increasingly prominent. An important factor in accelerating the exploration of the ocean is to develop efficient and applicable ocean exploration and transportation tools. As a new type of autonomous water vehicle, unmanned ships have shown great application prospects in the fields of ocean exploration, marine rescue, water quality monitoring, etc. Compared with traditional manned ships, unmanned ships have the advantages of autonomy, flexibility and cost-effectiveness, and have become the focus of attention and research of many researchers and practitioners. However, the model parameters of the motion characteristics of unmanned ships are easily affected by changes in the ocean environment and working conditions, and present a multi-dynamic and multi-modal situation, so the unmanned ship system is difficult to accurately describe by a single model. When a marine vehicle performs tasks such as salvaging sunken ships, scheduling and recovering autonomous underwater probes, loading and unloading, rescuing drowning people, or deploying mines, its mass may change greatly, bringing challenges to system modeling and control strategy design. It should be mentioned that in tasks such as salvage ships that require accurate positioning, it is necessary to control the unmanned ship to maintain and adjust its position and heading in order to safely and effectively salvage sunken ships. Obviously, when the mass of the unmanned ship changes, dynamic positioning is required to complete these tasks. Therefore, for unmanned ship systems with continuously changing mass, more and more researchers have studied the corresponding switching control and dynamic positioning problems.

[0003] The mode of using the network in combination with the unmanned ship may be subject to network attacks, among which the common one is the deception attack, which can greatly threaten the stability of the system by tampering with the data packets of network communication. In particular, when the unmanned ship is modeled as a switching system through mass changes, both the mode and the state information may be tampered with by deception attacks during transmission to the remote controller through the network, and when the switching signal changes, the difference between the estimated state information and the original state information of the system will also have a sudden change due to the change in the mode, resulting in false positives and false negatives of attack detection; at the present stage, for deception attack detection of the unmanned ship switching system, the common ones are state estimation-based detection, cumulative and detection, and game theory detection, which have the following problems in actual application: 1. Unable to effectively deal with the influence of mass disturbance caused by wind, waves and currents, which may lead to frequent mutations of the mass switching signal; 2. Unable to solve the situation where the mass of a large ship changes due to the influence of wind, waves, currents and other ocean environments and the task itself during the execution of a specific task; 3、Traditional methods for detecting spoofing attacks focus on designing observers or filters to detect whether the state signal is attacked, without considering that the change of the switching signal can also cause the residual to change suddenly, leading to false positives in attack detection; Therefore, the existing spoofing attack detection of the switching system of the unmanned ship is limited by the environment and signal transformation, and cannot achieve fast and accurate attack detection, so the safety of the unmanned ship system is limited. SUMMARY

[0004] The present application provides a disturbance-free attack detection and safe dynamic positioning method for switching an unmanned ship system to overcome the above technical problems.

[0005] To achieve the above purpose, the technical scheme of the present application is: A disturbance-free attack detection and safe dynamic positioning method for switching an unmanned ship system, for signal transmission between the sampler-controller and the controller-actuator of the unmanned ship and the shore-based control station, comprising: S1: introducing a hysteresis quantizer, establishing an unmanned ship mass switching model based on the hysteresis quantizer, and obtaining the state signal and mass modal switching signal of the unmanned ship; S2: constructing an event-triggered protocol with interpolation intervals to obtain the triggered state signal and triggered mass modal switching signal transmitted by the sampler-controller of the unmanned ship to the shore-based control station; S3: constructing a state observer to obtain a state residual signal based on the state observer and the triggered state signal; constructing a disturbance-free rate residual based on exponential interpolation, which is used to eliminate the influence of the triggered mass modal switching signal on the derivative of the state residual signal; S4: constructing a spoofing attack detection function based on the disturbance-free rate residual to detect the triggered state signal and the triggered mass modal switching signal, obtaining a spoofing attack detection result, and constructing a safe controller based on the disturbance-free spoofing attack detection based on the spoofing attack detection result; the spoofing attack detection result includes state detection results and modal detection results; S5: establishing a controller-actuator event-triggered protocol to obtain the control signal returned by the safe controller from the shore-based control station to the unmanned ship, and then designing a watermark mechanism for the controller-actuator channel based on the control signal to resist spoofing attacks on the controller-actuator channel; the watermark mechanism can encrypt and decrypt the control signal to obtain the decrypted control output; S6: constructing an error function based on the event-triggered protocol with interpolation intervals and the controller-actuator event-triggered protocol, and constructing a closed-loop system combining the error function, the state signal of the unmanned ship, the decrypted control output, the unmanned ship mass switching model and the state observer; S7: Based on the closed-loop system, construct the conditions for achieving safe dynamic positioning of the unmanned vessel with quality switching, and under the constructed conditions, achieve safe dynamic positioning of the unmanned vessel with quality switching through the closed-loop system.

[0006] Beneficial effects: This invention provides a non-intrusive attack detection and safe dynamic positioning method for switching unmanned surface vessel systems, which has the following advantages: 1. Modeling mass switching of unmanned vessels using a hysteresis quantizer: The hysteresis quantizer uses its hysteresis characteristics to introduce a certain dwell interval at the switching boundary, thereby effectively suppressing the frequent switching phenomenon caused by continuous disturbances such as wind, waves, and currents. 2. A non-disruptive deception attack detection strategy is proposed, which uses exponential interpolation to construct non-disruptive rate state residuals. The non-disruptive deception attack detection strategy is used to deal with the false alarms and false alarms that may be caused by the mode switching behavior of unmanned vessels. The non-disruptive rate state residuals are used to detect deception attacks on the state and mode, which overcomes the false alarms and false alarms caused by rate residual jitter caused by the switching behavior. The detection strategy only alarms when a deception attack occurs. 3. Establish an event triggering protocol with rate residual interpolation: This triggering protocol avoids errors caused by unmanned surface vessel mass mode switching and ensures that the unmanned surface vessel mass switching model effectively resists deception attacks; 4. Estimate the quality modes and build a security controller based on watermarking technology: By estimating the quality modes, the problem of continuous asynchronous switching caused by deception attacks between subsystems and controllers is reduced. The security controller adopts watermarking technology, which can effectively reduce the interference of deception attacks on the security controller. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This invention provides a flowchart of a method for non-intrusive attack detection and secure dynamic positioning of an unmanned surface vessel system. Figure 2 A block diagram of a mass-switching unmanned surface vessel system provided by the present invention; Figure 3 A comparison diagram of switching signals generated based on a uniform quantizer and a hysteresis quantizer; Figure 4 A schematic diagram of the event triggering interval for an event triggering protocol with interpolation intervals; Figure 5The switching signal of the unmanned ship and the estimated switching signal diagram; Figure 6 The interpolation interval analysis diagram; Figure 7 The , and residual and disturbance-free rate residual comparison diagram; Figure 8 The , and residual and disturbance-free rate residual comparison diagram; Figure 9 The state and modal attack detection result comparison diagram using , and ; Figure 10 The attack detection evaluation function evolution diagram based on residual, rate residual and disturbance-free rate residual; Figure 11 The event triggering interval diagram of the controller-actuator event triggering protocol; Figure 12 The comparison diagram of the watermarked and non-watermarked control input ; Figure 13 The state response diagram of the position , , the heading angle , the speed , and the angular velocity . DETAILED DESCRIPTION

[0009] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0010] The embodiment provides a disturbance-free attack detection and safe dynamic positioning method for switching an unmanned ship system, which is used for signal transmission between a sampler-controller and a controller-actuator between an unmanned ship and a shore-based control station, and comprises the following steps: S1: introducing a hysteresis quantizer, establishing an unmanned ship mass switching model based on the hysteresis quantizer, and obtaining state signals and mass modal switching signals of the unmanned ship; S2: constructing an event-triggered protocol with interpolation interval to obtain the triggered state signal and the triggered mass modal switching signal transmitted by the unmanned ship to the shore-based control station through the sampler-controller; S3: constructing a state observer to obtain a state residual signal according to the state observer and the triggered state signal; constructing an undisturbed rate residual based on exponential interpolation, which is used to eliminate the influence of the triggered mass modal switching signal on the derivative of the state residual signal; S4: constructing a spoofing attack detection function based on the undisturbed rate residual to detect the triggered state signal and the triggered mass modal switching signal, obtaining a spoofing attack detection result, and constructing a secure controller based on the undisturbed spoofing attack detection based on the spoofing attack detection result; the spoofing attack detection result includes state detection result and modal detection result; S5: establishing a controller-actuator event-triggered protocol to obtain the control signal returned by the secure controller from the shore-based control station to the unmanned ship, and then designing a watermark mechanism for the controller-actuator channel based on the control signal to resist spoofing attacks on the controller-actuator channel; the watermark mechanism is to encrypt and decrypt the control signal to obtain the decrypted control output; S6: constructing an error function based on the event-triggered protocol with interpolation interval and the controller-actuator event-triggered protocol, and constructing a closed-loop system combining the error function, the state signal of the unmanned ship, the decrypted control output, the mass switching model of the unmanned ship and the state observer; S7: constructing conditions to achieve safe dynamic positioning of the mass switching unmanned ship based on the closed-loop system, and achieving safe dynamic positioning of the mass switching unmanned ship through the closed-loop system under the constructed conditions.

[0011] Specifically, as shown in Figure 2 , the unmanned ship and its shore-based controller station communicate using a two-end network. A hysteresis quantizer is introduced, and an unmanned ship mass switching model is established based on the hysteresis quantizer. When the and signals of the mass switching unmanned ship model need to be transmitted to the shore-based control station through the network, in order to ensure that interpolation can be smoothly executed, an event-triggered protocol with interpolation interval is constructed; the network has a transmission time delay, where the transmission time delay of the sampler-controller and the controller-actuator channel is and , respectively; Secondly, the network also suffers from non-periodic spoofing attacks, which occur in the sensor-controller channel and the shore-based control station-actuator channel. During the attack process, the and transmitted by the sensor-controller channel, and the control output All of these could be tampered with by deception attacks. Therefore, in order to detect the signals received by the shore-based control station... and To determine whether the state signal has been tampered with by a deception attack, a corresponding state observer was designed. The state residual signal was obtained based on the state observer and the triggered state signal; a perturbation-free rate residual based on exponential interpolation was constructed to eliminate the influence of the quality mode switching signal on the derivative of the state residual signal. Furthermore, a deception attack detection function based on unperturbed rate residuals is constructed to detect the triggered state signal and the triggered quality mode switching signal, obtain the deception attack detection result, and construct a security controller based on unperturbed deception attack detection based on the deception attack detection result; the deception attack detection result includes state detection result and mode detection result; a controller-actuator event triggering protocol is established to obtain the control signal returned by the security controller from the shore-based control station to the unmanned vessel, and then a watermarking mechanism for the controller-actuator channel is designed based on the control signal to resist deception attacks on the controller-actuator channel; the watermarking mechanism is to encrypt and decrypt the control signal to obtain the decrypted control output; Finally, an error function is constructed based on the event triggering protocol with interpolation interval and the controller-actuator event triggering protocol. A closed-loop system is constructed by combining the error function, the unmanned vessel's state signal, the decrypted control output, the unmanned vessel's mass switching model, and the state observer. Based on the closed-loop system, the conditions for realizing the safe dynamic positioning of the mass-switching unmanned vessel are constructed, and the safe dynamic positioning of the mass-switching unmanned vessel is realized through the closed-loop system under the constructed conditions.

[0012] In a specific embodiment, a hysteresis quantizer is introduced, and an unmanned surface vessel (USV) mass switching model is established based on the hysteresis quantizer to obtain the USV's state signals and mass mode switching signals. The scheme is as follows: S11. The formula for the change in mass of the unmanned vessel is shown in formula (35). (35) in, Indicates the change in mass of the unmanned vessel. For the unloaded hull mass, For load variation quality, and These represent the maximum mass reduction during the unmanned vessel's unloading process and the maximum mass increase during the unmanned vessel's loading process, respectively. By analyzing the actual dynamic positioning tasks of unmanned surface vessels, in order to accurately quantify mass and load variation mass, this study aims to... As shown in formula (36), (36) in, This is the continuously changing part, describing the mass changes caused by disturbances such as wind, waves, and currents. , , and All are constants; The part indicating the change in the unit of mass is used to describe loading and unloading. The weight changes of goods during the loading process. , It is the smallest unit of mass change caused by load variation; S12, Regarding the changing mass of the unmanned surface vessel Introduce the following hysteresis quantizer, (37) in, For hysteresis quantizer, , This represents the width of the quantization interval; For the quantification function of quality, The mass of the first item loaded or unloaded; For loading and unloading The quality of each item; The unmanned vessel's changing mass is converted into a mass mode switching signal based on the formula hysteresis quantizer, as shown in formula (38). (38) To ensure that the dynamic positioning of the unmanned vessel can be achieved during changes in quality, The following switching conditions must be met (39) in, , Given the minimum residence time of the subsystem, For the index of the quality mode switching signal, , For the set of event triggering states, , All of these are matrices to be determined. , For in set Remove elements A set; This is the k-th switching time. S13. Construct the mass switching model for the unmanned vessel, as shown in formula (40). (40) in, Indicates control output. The state vector is a constant matrix. , a position vector of the unmanned ship, a velocity vector of the unmanned ship, a dynamic positioning position of the unmanned ship is , a dynamic positioning position of the unmanned ship in the x-axis, a dynamic positioning position of the unmanned ship in the y-axis, a mass modal switching signal of the unmanned ship switching model is , a number of mass modal switching signals; a corresponding switching sequence thereof is set as , is the number of unmanned ship switching, is the kth switching time; is a state signal, is a control input, is a disturbance input; , , , , and are a state matrix, an input matrix and an input disturbance matrix respectively, is an inverse matrix of an inertia matrix, is a damping matrix, is a thruster configuration matrix, is a disturbance matrix.

[0013] As shown in Figure 3 , when a uniform quantizer is used to convert mass changes into modal signals, mass disturbances caused by wind, waves and currents result in frequent modal switching phenomena, therefore, the present scheme uses a hysteresis quantizer to model the mass switching of the unmanned ship. The hysteresis quantizer uses its hysteresis characteristics to introduce a certain residence interval at the switching boundary, thereby effectively suppressing the frequent switching phenomenon caused by continuous disturbances such as wind, waves and currents.

[0014] In specific embodiments, the scheme for constructing an event-triggered protocol with interpolation intervals to obtain the triggered state signals and the triggered mass modal switching signals transmitted by the sampler-controller of the unmanned ship to the shore-based control station is: When the and signals of the mass switching unmanned ship system need to be transmitted to the shore-based control station through the network, in order to ensure that interpolation can be smoothly executed, an event-triggered protocol with interpolation intervals is constructed, as shown in equation (41), (41) wherein the error condition is (42) the modal matching condition is (43) in, It is the first Switching intervals The Next trigger time Indicates the sampling interval. , Indicates in The next Each sampling time, ; , is the error triggering condition. ;constant Given a threshold, and Let be the weight matrix to be determined; For state error, , for The next The system state at each sampling time; For the first Switching intervals The The system state at the moment of the next trigger. This refers to the modal matching condition, i.e., the trigger time for subsystem switching.

[0015] The unmanned surface vessel (USV) communicates with its shore-based controller station via a two-way network. This network suffers from transmission delays and is vulnerable to aperiodic spoofing attacks. The data transmission delays triggered in the sampler-controller and controller-actuator channels are denoted as... and ,in , , , , , ,thereby The definition detects the first... The activation and dormancy intervals for spoofing attacks on the secondary sensor-controller channel are respectively... and During the attack, the sensor-controller channel transmission as well as It can be attacked and tampered with as well as Because deception attacks have an energy limit, the energy level of a deceptive attack that alters the data is limited. The upper realm , Given a matrix.

[0016] like Figure 4 As shown, Figure 4For the triggering time and the triggering interval of the event-triggered mechanism in this embodiment, in this scheme, the event-triggered protocol with interpolation interval is constructed to ensure the smooth execution of interpolation.

[0017] In specific embodiments, a state observer is constructed, and a state residual signal is obtained according to the state observer and the state signal after triggering; the scheme for constructing a disturbance-free rate residual based on exponential interpolation is to eliminate the influence of the mass modal switching signal on the derivative of the state residual signal, which is: In order to detect whether the and received by the shore-based control station are tampered with by a spoofing attack, a state observer is constructed, as shown in formula (44), (44) wherein, is the state observation interval, , is the state transmission delay, is the observation value of , is the control signal to be solved, , is the controller modal signal based on spoofing attack detection, which is 0 or 1 according to the modal attack detection result to judge whether to use or the estimated , and the switching sequence is denoted as , represents the total number of switches on , by comparing the state observer estimated and with the and received by the shore-based control station which may be tampered with, the detection of the attack is realized. is the kth switching time of the estimated mode; is the observation state matrix, is the observation input matrix, is the observation error matrix, is the spoofing attack function; The state residual signal is obtained, as shown in formula (45), (45) wherein, is the state signal after triggering received by the shore-based control station; The residual signal is generally used for spoofing attack detection, however, in the unmanned ship mass switching model, when is tampered with, the modal adopted by the observer will change indirectly through the state observer​ The state observer will be discontinuous at the triggering point, which will directly affect the state residual and cause a sharp fluctuation. Therefore, compared with the state residual , the residual rate is more appropriate for detecting the mass switching model spoofing attack. On the other hand, directly using to detect spoofing attacks will cause to be tampered with by the switching and spoofing attacks, and the similar effects on will make it difficult for the attack detection to distinguish, resulting in false positives and false negatives of the attack. Therefore, the influence of the mass modal switching signal after triggering on needs to be removed before attack detection; Exponential interpolation is used to achieve smooth transition before and after the switching point , and construct the undisturbed rate residual, as shown in equation (46), (46) where is the time when the shore-based control station receives the mass modal change of the unmanned ship mass switching model, is the left limit of ; is the undisturbed function, as shown in equation (47), (47) In equations (47)-(48), is the residual rate, , , , and are the disturbance interval and the undisturbed interval, respectively, is the interpolation duration, in order to complete the interpolation of in equation (47) before the next triggering transmission to the shore-based control station , according to the modal matching condition, the waiting error condition is re-judged, so is satisfied; in order to make the interpolation interval exist, i.e. to satisfy , according to the network time delay , we can get , is the derivative of the state residual, and represent the minimum value of the time delay and the maximum value of the time delay, respectively.

[0018] The triggering time of the subsystem switching generated by the modal matching condition The time it takes for the shore-based control station to receive the mass switch from the unmanned vessel can be approximated as... Therefore, (47) and (48) are in Start interpolation, and you can find out from (48) exist Before and after time, , Substituting it into (47), we get as well as ,therefore exist The time frame is smooth and there is no chattering due to changes in the mass mode.

[0019] On the other hand, it can be seen from (48) that Substituting it into (47) yields as well as ,therefore At the end of the interpolation The front and back are smooth.

[0020] In the interpolation interval Inside, through the perturbation-free function in (48) And combine the integral term with the unperturbed rate residual at the current moment. Estimate Moment Thus achieving exist A smooth transition within. Therefore, with In comparison, using unperturbed rate residuals The impact of system switching can be ruled out, and it can be used. Perform non-intrusive deception attack detection.

[0021] like Figure 5 As shown, from Figure 5 As can be seen, by using unperturbed rate residuals to construct unperturbed modal attack detection, the asynchronous time caused by deceptive attacks tampering with the modality is shortened, thereby effectively avoiding the occurrence of continuous asynchronous switching; like Figure 6 As shown, the image illustrates perturbation-free rate residuals based on exponential interpolation. These residuals can be used to eliminate interference from handover behavior on spoofing attack detection, thus achieving perturbation-free spoofing attack detection. In the figure, event triggering mechanism 1 is an event triggering protocol with interpolation intervals. Figure 6 It can be seen that, in order to achieve unperturbed rate residuals Interpolation is performed before the next trigger reaches the shore-based control station, with the interpolation interval... The value of is related to the minimum residence time of the subsystem. Establish constraint relationships as Moreover, the minimum residence time condition satisfied by the given subsystem. This also provides a premise for the range of values ​​for the interpolation interval. Chattering caused by mass modal changes is... At that moment, the shore-based control station received the modal change at... At time, therefore the rate residual exist Bumps will occur at any moment. To achieve a smooth transition, the unperturbed rate residual must be... Interpolation is performed on the data to eliminate the impact of switching from the real system.

[0022] Through the Figure 7 and Figure 8 The comparison reveals that, compared to the residuals Rate residual The handover process exhibits less jitter, and exponential interpolation effectively eliminates the handover-induced unobstructed rate residual. The resulting chattering.

[0023] In a specific embodiment, a deception attack detection function based on unperturbed rate residuals is constructed to detect the triggered state signal and the triggered quality mode switching signal, obtain the deception attack detection result, and construct a security controller based on unperturbed deception attack detection based on the deception attack detection result; the scheme in which the deception attack detection result includes state detection result and mode detection result is as follows: S41. For the detection of deception attacks on system state signals, the evaluation function is set using the unperturbed rate residual, as shown in formula (48). (48) in, For the evaluation function, It is the time constant of the attack detection window size; Will With a given threshold The following state detection results were obtained through comparison: (49) in, A value of 0 indicates that the status signal has not been tampered with by a deception attack. A value of 1 indicates that the status signal has been tampered with by a deception attack; Using state detection results and The formula for calculating the safety status received by the controller is as follows: (50) S42. For deception attack detection of system state, modal estimation values ​​are obtained based on the security state, and then modal detection results are obtained, as shown in formula (51). (51) wherein, is the modal detection result, , is the system modal received by the state observer, is the triggered mass modal switching signal received by the shore-based control station, the modal estimation value is (52) is the left limit of , is the interval at the initial time, , and respectively represent the error condition and the error boundary condition, as shown in formulas (53) and (54), (53) (54) wherein the positive definite matrix , is the matrix to be solved; ; is the boundary of ; Using the modal detection result and the modal estimation value, the safe modal received by the controller is calculated, and the formula is (55) S43, combined with the safe state received by the controller and the safe modal received by the controller, a safe controller based on undisturbed spoofing attack detection is designed, as shown in formula (56), (56) wherein, is the controller gain matrix to be solved, and the controller switching sequence is denoted as , represents the total number of switching on

[0024] Figure 9 The state attack detection results of and are compared , and the modal attack detection results are compared . The interpolation interval of the undisturbed rate residual is 0.3, and the comparison shows that the detection result based on the undisturbed rate residual is more accurate, and the false positive rate and the false negative rate of attack detection are reduced.

[0025] Figure 10 ​​For the attack detection evaluation function evolution based on residual, rate residual and non-disturbance rate residual, it can be seen from the figure that the evaluation function based on can effectively reduce the rate residual mutation caused by switching compared with the evaluation function based on .

[0026] In specific embodiments, a controller-actuator event-triggered protocol is established to obtain the control signal returned by the safety controller from the shore-based control station to the unmanned ship, and then a watermark mechanism of the controller-actuator channel is designed according to the control signal to resist the spoofing attack of the controller-actuator channel; the watermark mechanism is to encrypt and decrypt the control signal to obtain the decrypted control output, including, The control output signal is transmitted through the network channel of the controller-actuator, in order to reduce the influence of the spoofing attack on , an event-triggered protocol of the controller-actuator is established, as shown in equation (57), (57) Wherein, is the triggering time, represents the after the sampling time, the error condition is , , represents the control input error, is a known constant threshold, and are weight matrices; is the control output sent by the controller to the actuator; is the control output sent by the controller to the actuator at the after the sampling time; the lower bound of the interval between adjacent triggers in this event-triggered protocol is the sampling interval , it is impossible to occur infinitely many times in a finite time, so the Zeno phenomenon is excluded.

[0027] As shown in Figure 11 , the Figure 11 is the triggering time and triggering interval of the controller-actuator event-triggered protocol in this embodiment; In order to resist the spoofing attack in the controller-actuator network channel, the watermark technology is introduced for the transmission of the control output : The control output sent by the controller to the actuator is encrypted to obtain the encrypted control output, as shown in equation (58), (58) wherein, is a reversible matrix for multiplication encryption, ; is a sampling interval of the control output; is a modal transmission delay; the control output of the controller-actuator channel will be tampered by an attack , since the energy upper limit of the tampering of the attack by deception, the upper bound of the tampering of the attack by deception , is a given matrix of; and decrypts the control signal transmitted to the actuator to obtain the decrypted control output, as shown in equation (59), (59) wherein, is an inverse matrix of the multiplication encryption matrix, is a control signal to the actuator.

[0028] As shown in Figure 12 , the fluctuation of the control input caused by the attack by deception during the operation of the unmanned ship is effectively reduced by the proposed watermarking technology.

[0029] In specific embodiments, the scheme for constructing an error function based on an event-triggered protocol with interpolation intervals and a controller-actuator event-triggered protocol, combining the error function, the state signal of the unmanned ship, the decrypted control output, the mass switching model of the unmanned ship, and the state observer to construct a closed-loop system is: S61, construct an error function, as shown in equation (60), (60) signal holding interval and can be divided into sub-intervals (61) (62) wherein, and are intermediate variables, specifically as follows: , ; Construct the control input delay and state delay based on the input time delay method, as shown in equation (63), (63) wherein, and represent the control input delay and state delay; satisfy , and , ; Rewrite the state signal and the decrypted control output as shown in equation (64), (64) S62, combine the unmanned ship mass switching model and the state observer, and substitute (64) into the combined equation to obtain a closed-loop system as shown in equation (65), (65) wherein, represents a composite switching signal, and the initial conditions are , represents the initial time, represents the initial value lower bound; represents an error state, represents an error state matrix, represents an attack error state matrix, represents an input time delay matrix, represents a state time delay matrix, represents an attack state time delay matrix, represents an error disturbance matrix, represents a state error function matrix, represents an attack error function matrix, represents an input error function matrix, represents an output matrix, represents a state matrix, represents an input matrix.

[0030] In specific embodiments, conditions for achieving safe dynamic positioning of the mass switching unmanned ship are constructed based on the closed-loop system, and a scheme for achieving safe dynamic positioning of the mass switching unmanned ship through the closed-loop system under the constructed conditions is: Based on the above discussion, the dynamic positioning problem of the mass switching unmanned ship can be converted into two control objectives of the closed-loop system: (i) When , the closed-loop system is asymptotically stable.

[0031] (ii) When , under the initial condition , , the closed-loop system has a bounded performance, i.e., satisfies , is a positive constant, It is a non-negative function. .

[0032] Handover behavior analysis: For any adjacent subsystem handover interval and Assuming the subsystem quality mode switching signal exist from Switch to And sub-controller mode Different modal signals will be used based on the results of the attack detection mechanism. .

[0033] Depending on whether the deception attack covers the trigger time And different attack durations, switching intervals superior and The changes can be divided into six categories. and These represent the times when the sub-controller switchover was triggered by the detection of a spoofing attack and when it ended, respectively. , , The following section analyzes the switching behavior between the subsystem and the subcontroller in detail: Case 1: This deception attack did not cover the trigger time. , , .exist The timing controller receives the mode that has not been tampered with, and the sub-controller mode. Switch to .exist When a deception attack is detected, due to the construction of modal attack detection in the non-intrusive deception attack detection, the sub-controller modality... Switch to modal estimates When the end of the attack is detected Sub-controller mode Switching to subsystem mode Therefore, modality As follows: in the interval superior, In the interval , and superior, On the interval , .

[0034] Case 2: The secondary deception attack covers part of the trigger time. The deception attack did not cover the trigger moment. ,exist The timing controller receives the mode that has not been tampered with, and the sub-controller mode. Switch to .exist When a deception attack is detected, the sub-controller modality is adjusted based on the modal attack detection mechanism. Switch to modal estimates When the end of the attack is detected as well as And the trigger time Overridden by deception attack, sub-controller mode Switch to the tampered modality Therefore, modality As follows: in the interval superior, On the interval , On the interval , , .

[0035] Case 3: The secondary deception attack covers part of the trigger time. The deception attack did not cover the trigger moment. ,exist The timing controller receives the mode that has not been tampered with, and the sub-controller mode. Switch to . When a deception attack is detected, the sub-controller modality is adjusted based on the modal attack detection mechanism. Switch to modal estimates When the end of the attack is detected Sub-controller mode Switch to the tampered modality Triggering time There was no cover from the deception attack, and the attack ended at the moment of detection. Sub-controller mode Switch to the tampered modality Therefore, modality As follows: in the interval superior, On the interval , , On the interval , .

[0036] Case 4: The secondary deception attack covers part of the trigger time. Deception attack trigger time , When a deception attack is detected, the sub-controller modality is adjusted based on the modal attack detection mechanism. Switch to modal estimates When the end of the attack is detected Sub-controller mode Switch to Therefore, modality As follows: in the interval superior, In the interval and superior, On the interval , .

[0037] Case 5: The secondary deception attack covers part of the trigger time. The deception attack did not cover the trigger moment. ,exist When a deception attack is detected, due to the construction of modal attack detection in the non-intrusive deception attack detection, the sub-controller modality... Switch to modal estimates When the end of the attack is detected And the trigger time Overridden by deception attack, sub-controller mode Switch to the tampered modality At the moment the attack ended was detected. And the trigger time Not covered by deception attack, sub-controller modality Switching to subsystem mode Therefore, modality As follows: in the interval superior, In the interval superior, In the interval superior, On the interval , .

[0038] Case 6: This deception attack covers all trigger moments. and switching range Sub-controller modes across the entire interval Switch to modal estimates Therefore, modality As follows: in the interval On the interval , .

[0039] For the convenience of subsequent analysis, the following lemmas are given: Definition Lemma 1: For a symmetric matrix , a symmetric matrix and a constant , there exists ; Definition Lemma 2: There exist a symmetric matrix , and a vector function such that the following inequality holds ; where denotes the delay function, denotes the upper bound of the delay constant; Definition Lemma 3: For a given symmetric matrix , , , , the following three inequalities are equivalent ; , ; , ; Under the constraints of formula (40) and lemmas 1-3, the conditions for realizing the safe dynamic positioning of unmanned ships with quality switching are constructed, and the specific conditions are as follows: Condition 1: Under the constraints of formula (40) and lemma 2, the closed-loop system is asymptotically stable and has a bounded performance and a dwell time , for a given constant , , , , , , , , , , there exists a positive definite matrix , , , a symmetric matrix , , , , , , a controller gain matrix and an indirect variable matrix satisfy (66) (67) (68) (69) where the parameters , , , , denote the estimated modalities at the k-th spoofing attack, , , , , , the stability matrix is composed of the following block matrices:

[0040] , , , , , , , , , , ,

[0041] , , , , , , , , , ,

[0042] , , , , , , , , , , , , , , , , , , , ; , ; Since the matrix inequality (65) is nonlinear, this embodiment gives a sufficient condition with solvability; Condition 2: Under the constraints of Lemma 1 and Lemma 3 of formula (5), the closed-loop system is asymptotically stable and has bounded performance and dwell time , for given constant , , , , , , , , , , , there exists a positive definite matrix , , , a symmetric matrix , , , , , , a matrix and an indirect variable matrix satisfy (70) where the parameters , , , , , , stability matrix is composed of the following block matrices:

[0043] , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,

[0044] , , , , , , , , , , , .

[0045] As shown in the switching rule and the safety controller, Figure 13 converges to , , Converges to zero gradually, which shows that the dynamic positioning problem is solved, and the control target is met.

[0046] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for non-disruptive attack detection and secure dynamic positioning of a switching unmanned surface vessel (USV) system, used for signal transmission between the USV and a shore-based control station via sampler-controller and controller-actuator, characterized in that... include: S1: Introduce a hysteresis quantizer, establish an unmanned vessel mass switching model based on the hysteresis quantizer, and obtain the unmanned vessel's state signal and mass mode switching signal; S2: Construct an event triggering protocol with interpolation intervals to obtain the triggered status signals and triggered mass mode switching signals transmitted by the unmanned vessel to the shore-based control station through the sampler-controller; S3: Construct a state observer and obtain the state residual signal based on the state observer and the state signal after triggering; construct an unperturbed rate residual based on exponential interpolation, which is used to eliminate the influence of the mass mode switching signal after triggering on the derivative of the state residual signal. S4: Construct a deception attack detection function based on unperturbed rate residuals to detect the state signal and the quality mode switching signal after triggering, obtain the deception attack detection result, and construct a security controller based on unperturbed deception attack detection based on the deception attack detection result; the deception attack detection result includes state detection result and mode detection result; S5: Establish a controller-actuator event triggering protocol to obtain the control signal returned by the safety controller from the shore-based control station to the unmanned vessel, and then design a watermarking mechanism for the controller-actuator channel based on the control signal to resist deception attacks on the controller-actuator channel; the watermarking mechanism can encrypt and decrypt the control signal to obtain the decrypted control output; S6: Construct an error function based on the event triggering protocol with interpolation interval and the controller-actuator event triggering protocol. Combine the error function, the unmanned vessel's state signal, the decrypted control output, the unmanned vessel's mass switching model, and the state observer to construct a closed-loop system. S7: Based on the closed-loop system, construct the conditions for achieving safe dynamic positioning of the unmanned vessel with quality switching, and under the constructed conditions, achieve safe dynamic positioning of the unmanned vessel with quality switching through the closed-loop system.

2. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 1, characterized in that, A hysteresis quantizer is introduced, and an unmanned surface vessel (USV) mass switching model is established based on the hysteresis quantizer to obtain the USV's state signals and mass mode switching signals, including: S11. The formula for the changing mass of the unmanned vessel is shown in formula (1). （1） in, Indicates the change in mass of the unmanned vessel. For the unloaded hull mass, For load variation quality, and These represent the maximum mass reduction during the unmanned vessel's unloading process and the maximum mass increase during the unmanned vessel's loading process, respectively. As shown in formula (2), （2） in, This is the continuously changing part, describing the mass changes caused by disturbances such as wind, waves, and currents. , , and All are constants; The part indicating the change in the unit of mass is used to describe loading and unloading. The weight changes of goods during the loading process. , It is the smallest unit of mass change caused by load variation; S12, Regarding the changing mass of the unmanned surface vessel Introduce the following hysteresis quantizer, （3） in, For hysteresis quantizer, , This represents the width of the quantization interval; For the quantification function of quality, The mass of the first item loaded or unloaded; For loading and unloading The quality of each item; The unmanned vessel's changing mass is converted into a mass mode switching signal based on the formula hysteresis quantizer, as shown in formula (4). （4） The following switching conditions must be met （5） in, , Given the minimum residence time of the subsystem, For the index of the quality mode switching signal, , For the set of event triggering states, , All of these are matrices to be determined. , For in set Remove elements A set; This is the k-th switching time. S13. Construct the mass switching model for the unmanned vessel, as shown in formula (6). （6） in, Indicates control output. The state vector is a constant matrix. , Let be the position vector of the unmanned vessel. The velocity vector of the unmanned vessel is given, and its dynamic positioning position is given. , This represents the dynamic positioning position of the unmanned vessel on the x-axis. The dynamic positioning position of the unmanned surface vessel (USV) on the y-axis is given by the mass mode switching signal of the USV's switching model. , The number of quality mode switching signals; its corresponding switching sequence is set to , It refers to the number of times the unmanned vessel switches. This is the k-th switching time. This is a status signal. To control the input, For disturbance input; , , , , and These are the state matrix, input matrix, and input perturbation matrix, respectively. It is the inverse of the inertia matrix. Here is the damping matrix. Configure the matrix for the thrusters. Let be the perturbation matrix.

3. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 2, characterized in that, Construct an event triggering protocol with interpolation intervals, including: Construct an event triggering protocol with interpolation intervals, as shown in formula (7). （7） The error condition is as follows: （8） Modal matching conditions are （9） in, It is the first Switching intervals The Next trigger moment Indicates the sampling interval. , Indicates in The next Each sampling time, ; This is the error triggering condition. ;constant Given a threshold, and Let be the weight matrix to be determined; For state error, , for The next The system state at each sampling time; For the first Switching intervals The The system state at the moment of the next trigger. This refers to the modal matching condition, i.e., the trigger time for subsystem switching.

4. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 3, characterized in that, Construct a state observer, and obtain the state residual signal based on the state observer and the state signal after triggering, including: Construct a state observer as shown in equation (10). （10） in, For the state observation interval, , For state transmission delay, yes The observed values, The control signal to be determined, , It is a controller modal signal based on deception attack detection, and its switching sequence is denoted as... , express superior Total number of switching times To estimate the k-th mode switching time; The observed state matrix, For the observation input matrix, The observation error matrix, This is a deception attack function; Obtain the state residual signal, as shown in formula (11). （11） in, This refers to the status signal received by the shore-based control station after triggering.

5. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 4, characterized in that, Constructing perturbation-free rate residuals based on exponential interpolation includes: Unperturbed rate residuals are constructed using exponential interpolation, as shown in formula (12). （12） in, This is the moment when the shore-based control station receives a change in the mass mode of the unmanned vessel's mass switching model. for The left limit; The function is a perturbation-free function, as shown in formula (13). （13） In formula (13)-(14), For residual rate, , , , and These are the disturbed interval and the undisturbed interval, respectively. It is the interpolation duration interval, satisfying ; According to network time delay , can be obtained , The derivative of the state residual. and These represent the minimum and maximum delay values, respectively.

6. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 5, characterized in that, A deception attack detection function based on unperturbed rate residuals is constructed to detect the triggered state signal and the triggered quality mode switching signal, obtain the deception attack detection result, and construct a security controller based on unperturbed deception attack detection based on the deception attack detection result, including: S41. For the detection of deception attacks on system status signals, the evaluation function is set using the unperturbed rate residual, as shown in formula (14). （14） in, For the evaluation function, It is the time constant of the attack detection window size; Will With a given threshold The following state detection results were obtained through comparison. （15） in, A value of 0 indicates that the status signal has not been tampered with by a deception attack. A value of 1 indicates that the status signal has been tampered with by a deception attack; Using state detection results and The formula for calculating the safety status received by the controller is as follows: （16） S42. For deception attack detection of system modes, modal estimation values ​​are obtained based on the security state, and then modal detection results are obtained, as shown in formula (17). （17） in, For modal detection results, , The system modes received by the state observer, The mode estimate is the quality mode switching signal received by the shore-based control station after triggering. for （18） for The left limit, the initial condition for the modal estimate is , The interval at the initial time. , and Let (19) and (20) represent the error condition and the error boundary condition, respectively. （19） （20） Where the positive definite matrix , Let be the matrix to be determined; ; Using the modal detection results and modal estimation values, the safety modes received by the controller are calculated, and the formula is as follows: （21） S43. Design a security controller based on non-intrusive deception attack detection by combining the security status and security modes received by the controller, as shown in formula (22). （22） in, Let be the controller gain matrix to be determined, and let be the controller switching sequence. , express superior Total number of switching times.

7. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 6, characterized in that, Establish a controller-actuator event triggering protocol, including: Establish a controller-actuator event triggering protocol, as shown in formula (23). （23） in, It is the first Next trigger moment Indicates in The next At each sampling time, the error condition is: , , This indicates the control of input error. Given a constant threshold, and This is the weight matrix; The control output sent from the controller to the actuator; In order to be in The next The controller sends the control output to the actuator at each sampling time.

8. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 7, characterized in that, Design a watermarking mechanism for the controller-actuator channel based on the control signal to resist spoofing attacks on the controller-actuator channel, including: Control output sent from the controller to the actuator Encryption is performed to obtain the encrypted control output, as shown in formula (24). （24） in, It is an invertible matrix used for multiplication encryption. ; To control the sampling range of the output; For modal transmission delay; The control signal input to the actuator is decrypted to obtain the decrypted control output, as shown in formula (25). （25） in, It is the inverse of the multiplication encryption matrix. The control signal that reaches the actuator.

9. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 8, characterized in that, An error function is constructed based on an event-triggered protocol with interpolation intervals and a controller-actuator event-triggered protocol. A closed-loop system is then built by combining the error function, the unmanned surface vessel's (USV) state signals, the decrypted control output, the USV mass switching model, and a state observer, including: S61. Construct the error function, as shown in formula (26). （26） The control input delay and state delay are constructed based on the input delay method as shown in formula (27). （27） in, and This indicates the control input delay and the status delay; The status signal and the decrypted control output are rewritten as shown in formula (28). （28） S62. Combine the unmanned vessel mass switching model and state observer, and substitute (28) into the combined equations to obtain the closed-loop system, as shown in formula (29). （29） in, Indicates a composite switching signal. and The initial conditions are , Indicates the initial time. Indicates the lower bound of the initial value; Indicates the error status. Represents the error state matrix. Represents the attack error state matrix. Represents the input delay matrix. Represents the state delay matrix. Represents the attack state delay matrix. Represents the error perturbation matrix. Represents the state error function matrix. Represents the attack error function matrix. Represents the input error function matrix. Indicates the output matrix. Represents the state matrix, This represents the input matrix.

10. The method for non-intrusive attack detection and safe dynamic positioning of a switching unmanned surface vessel system according to claim 9, characterized in that, The conditions for achieving safe dynamic positioning of unmanned surface vessels with quality switching based on the aforementioned closed-loop system include: Lemma 1: For a symmetric matrix symmetric matrix and constant ,have ； Lemma 2: There exists a symmetric matrix , sum vector function This makes the following inequalities true. ； in, Represents the time delay function. Indicates the upper bound of the time delay constant; Lemma 3: For a given symmetric matrix , , , The following three inequalities are equivalent. ； ， ； ， ； Under the constraints of formula (5) and lemma 1-3, the conditions for achieving safe dynamic positioning of the unmanned vessel with mass switching are constructed as follows: Condition 1: Under the constraints of formula (5) and lemma 2, for a given constant , , , , , , , , , There exists a positive definite matrix. , , symmetric matrix , , , , , ,matrix and indirect variable matrix satisfy （30） （31） （32） （33） Among them, parameters , , , , Indicates the first Estimated mode during a deception attack , , , Stable matrix It consists of the following block matrices: ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ； ， ； Condition 2: Under the constraints of Lemmas 1 and 3 of formula (5), for a given constant , , , , , , , , , , There exists a positive definite matrix. , , symmetric matrix , , , , , Controller gain matrix and indirect variable matrix satisfy （34） Where parameters , , , , , Stable matrix It consists of the following block matrices: ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， 。

Images