Unmanned ship safe dynamic positioning method based on switching fuzzy wavelet network

Abstract

The invention discloses an unmanned ship safe dynamic positioning method based on a switching fuzzy wavelet network, and the method comprises the steps: building and constructing a switching LPV networked unmanned ship system according to the number of shipborne equipment, designing an event triggering mechanism, and constructing an attack detection observer and a detector in a shore-based system for possible spoofing attacks in communication; designing a safety controller in combination with a switching fuzzy wavelet network and an attack detector, introducing a shore-ship event triggering mechanism, further designing a special attack detector and a control input compensator, and generating a safe event triggering control input; a closed-loop system is constructed by integrating an attack detection observer, an unmanned ship system and control input, and a stable condition of the system is given through theoretical analysis, so that safe dynamic positioning of the unmanned ship under potential network attacks is realized; according to the method, the estimation accuracy can be improved, the safety is improved, and safe dynamic positioning of the networked unmanned ship is realized.

Description

Technical Field

[0001] This invention relates to the field of unmanned vessel dynamic positioning technology, and in particular to an unmanned vessel dynamic positioning method based on a switched fuzzy wavelet network. Background Technology

[0002] With the development of autonomous driving technology, the development of unmanned surface vessels (USVs) has also attracted great attention from numerous scholars and technology companies both domestically and internationally. Networked USV systems based on remote control stations are developing rapidly with the advancement of the digital age. During navigation, the total mass of the USV changes in various ways due to cargo loading and unloading, onboard equipment performing tasks, salvage operations, and resource consumption. Therefore, to accurately characterize the USV's position and navigation status, it is necessary to use a switching system approach to meticulously describe the USV model based on these mass changes. While networking brings rapid advancements to USV development, researchers must focus on security control schemes for USV switching models under deception attacks to ensure good control performance. In today's highly interconnected world, the open nature of networks provides great convenience for information exchange and resource sharing. However, this openness also brings a series of security challenges. Especially for switching systems that rely on stable communication networks, the threat of cyberattacks is increasingly prominent. Therefore, many researchers have begun to focus on the security control issues of switching systems under cyber deception attacks. While networking brings advantages to USVs, it also means that networked USVs are vulnerable to deception attacks when communicating with remote control stations. This means that transmitted data may be tampered with or replaced, causing the controller to fail to receive accurate information, thus significantly impacting the stability and performance of networked unmanned vessels. Compared to achieving control objectives by reducing control performance under the influence of deception attacks, proactive defense against deception attacks can enable the system to achieve better control performance. Some researchers have considered detecting network deception attacks, allowing the system to detect the occurrence time of malicious attacks and thus take defensive measures. Intuitively, if attacks can be detected in a timely manner, the damage to the system can be minimized. Therefore, reliable attack detection schemes are crucial for establishing secure contactless communication systems.

[0003] Existing research on modeling networked switching unmanned surface vessels (USVs) based on mass changes has not received much attention. Furthermore, in the limited literature available, the mass changes of USVs are observed to vary between a few fixed mass data points, leading to significant conservatism in the modeling and failing to reflect the continuous mass changes encountered during actual operation. Therefore, how to model networked USVs as switching models to effectively reduce conservatism and increase flexibility is a topic that should be addressed and researched. When the output signal of the hull-shore channel is detected to have been tampered with by a deception attack, the security output of the shore-based system will use the observer's observed output to replace the tampered output signal. However, this replacement method has poor security in the initial stage of system operation. This is because, on the one hand, in the initial stage of system operation, the state error between the observer and the networked USV is relatively large, resulting in a large error between the observer's output and the networked USV's output, leading to low accuracy of the security output; on the other hand, during asynchronous behavior, the networked USV mode and the observer mode are inconsistent, inevitably resulting in a deviation between their states. Summary of the Invention

[0004] This invention provides a method for safe dynamic positioning of unmanned vessels based on switching fuzzy wavelet networks, in order to overcome the technical problems that when the output signal is attacked, the tampered output signal has a large error with the actual output of the networked unmanned vessel, the accuracy of the safe output is low, the mode after the attack is inconsistent with the mode of the networked unmanned vessel, and there will inevitably be a deviation between the two states, resulting in poor security in the initial stage of system operation.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows: A method for safe dynamic localization of unmanned surface vessels based on switched fuzzy wavelet networks includes: S1: Obtain the number of shipborne equipment, use the switching LPV modeling method to construct a switching LPV networked unmanned ship system based on the number of shipborne equipment, obtain the unmanned ship mode and the unmanned ship's regulation output, and transmit the unmanned ship mode and the unmanned ship's regulation output to the sensor for sampling; S2: Based on the unmanned vessel's regulation output and unmanned vessel mode, an event triggering mechanism is constructed. Based on the event triggering mechanism, the regulation output and unmanned vessel mode after being deceived are constructed and transmitted to the shore-based system for observation of the deception attack to establish a deception attack detection observer. A deception attack detector is constructed based on the time window chi-square detection technology. S3: Construct a switching fuzzy wavelet network and build a security controller based on the switching fuzzy wavelet network and the detector for spoofing attacks; S4: Design a shore-based-ship channel event triggering mechanism, combine the security controller and the shore-based-ship channel event triggering mechanism to construct the security controller output and security controller mode after a deception attack, and then design a deception attack detector for the security controller; S5: The deception attack detector based on the security controller is based on the construction of a control input compensator, and on the control input compensator and the security controller, an event-triggered control input is constructed. S6: Construct an error system by combining a deception attack detection observer, a switching LPV networked unmanned surface vessel system, and an event-triggered control input; and construct a closed-loop system based on the error system and the switching LPV networked unmanned surface vessel system. S7: Based on the closed-loop system, construct the conditions for realizing the dynamic positioning of the unmanned vessel, and realize the dynamic positioning of the unmanned vessel through the closed-loop system under the constructed conditions.

[0006] Beneficial effects: This invention provides a method for safe dynamic positioning of unmanned vessels based on switched fuzzy wavelet networks, which has the following advantages: 1. A deception attack defense mechanism based on a switching fuzzy wavelet network is proposed. The deception attack function is estimated using a fuzzy wavelet network with a switching adaptive law. On the one hand, the switching adaptive law ensures that the input data of the fuzzy wavelet network is the data of the system under the current mode after the system mode changes, thereby improving the accuracy of the estimation. On the other hand, when the transmitted data is detected to be tampered with, the constructed defense mechanism compensates the transmitted data with the estimated deception attack approximation value, thereby increasing its security. 2. Construct a safe dynamic positioning control scheme based on the deception attack defense mechanism of switching fuzzy wavelet network. Based on the deception attack detection results, the safety controller under the estimation of switching fuzzy wavelet network, and the event triggering mechanism, provide sufficient conditions for networked unmanned vessels to achieve safe dynamic positioning. 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 A flowchart of a method for safe dynamic positioning of unmanned vessels based on switching fuzzy wavelet networks provided by the present invention; Figure 2 Block diagram for switching to an LPV networked unmanned surface vessel system; Figure 3 For the quality of networked unmanned ships Modal Number parameter vector A schematic diagram; Figure 4 For the number of unmanned shipboard devices and State curve diagram at time; Figure 5 For the number of unmanned shipboard devices and State curve diagram at time; Figure 6 A diagram illustrating the event triggering interval for the event triggering mechanism; Figure 7 Controller modal estimates and its mean square error Schematic diagram; Figure 8 Deception attack detection function and test results Schematic diagram; Figure 9 For safety controller mode A schematic diagram of its asynchronous duration; Figure 10 For attack function Estimates under switching adaptive law and non-switching adaptive law Schematic diagram; Figure 11 Controller output with / without deception attack A schematic diagram; Figure 12 State curves for switching to networked unmanned vessels using LPV. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] This embodiment provides a flowchart of a method for safe dynamic positioning of unmanned vessels based on a switched fuzzy wavelet network, as shown below. Figure 1 As shown, it includes: S1: Obtain the number of shipborne equipment, use the switching LPV modeling method to construct a switching LPV networked unmanned ship system based on the number of shipborne equipment, obtain the unmanned ship mode and the unmanned ship's regulation output, and transmit the unmanned ship mode and the unmanned ship's regulation output to the sensor for sampling; S2: Based on the unmanned vessel's regulation output and unmanned vessel mode, an event triggering mechanism is constructed. Based on the event triggering mechanism, the regulation output and unmanned vessel mode after being deceived are constructed and transmitted to the shore-based system for observation of the deception attack to establish a deception attack detection observer. A deception attack detector is constructed based on the time window chi-square detection technology. S3: Construct a switching fuzzy wavelet network and build a security controller based on the switching fuzzy wavelet network and the detector for spoofing attacks; S4: Design a shore-based-ship channel event triggering mechanism, combine the security controller and the shore-based-ship channel event triggering mechanism to construct the security controller output and security controller mode after a deception attack, and then design a deception attack detector for the security controller; S5: The deception attack detector based on the security controller is based on the construction of a control input compensator, and on the control input compensator and the security controller, an event-triggered control input is constructed. S6: Construct an error system by combining a deception attack detection observer, a switching LPV networked unmanned surface vessel system, and an event-triggered control input; and construct a closed-loop system based on the error system and the switching LPV networked unmanned surface vessel system. S7: Based on the closed-loop system, construct the conditions for realizing the dynamic positioning of the unmanned vessel, and realize the dynamic positioning of the unmanned vessel through the closed-loop system under the constructed conditions.

[0011] Specifically, such as Figure 2 As shown, Figure 2 The structure of a networked unmanned surface vessel (USV) is demonstrated. A switching LPV model is established based on the variation in the number of onboard devices of the USV, obtaining the USV modes and the adjusted output of the USV. The USV modes and the adjusted output of the USV are then transmitted to sensors for sampling. A deception attack defense mechanism based on a switching fuzzy wavelet network is proposed. The deception attack function is estimated using a fuzzy wavelet network with a switching adaptive law. On the one hand, the switching adaptive law ensures that the input data of the fuzzy wavelet network is the data of the system under the current mode after the system mode changes, thereby improving the accuracy of the estimation. On the other hand, when the transmitted data is detected to be tampered with, the constructed defense mechanism compensates the transmitted data with the estimated deception attack approximation value, thereby increasing its security.

[0012] In a specific embodiment, the following scheme is employed: First, the number of onboard devices is obtained. Then, a switching LPV modeling method is used to construct a switching LPV networked unmanned surface vessel (USV) system based on the number of onboard devices. This yields the USV's modes and its regulation output. Finally, the USV's modes and regulation output are transmitted to sensors for sampling. S11. Construct an unmanned vessel model, as shown in formula (38). (38) Among them, surge velocity sway speed yaw rate Forming a vector Position coordinates , and yaw angle Forming a vector Control input vector , , , These represent the sway force, the roll force, and the pitching moment, respectively. , The inertia matrix; in It is hydrodynamics that increases mass. This represents the longitudinal component of the difference vector from the origin to the ship's center of gravity in the ship's fixed reference frame. It is the inertial matrix, the total mass of the unmanned vessel. ,satisfy ,in and It is the total mass of the unmanned vessel when it is empty and fully loaded. This represents the payload of the unmanned vessel as it changes over time. Indicates the surge mass as it changes over time; , Let be the damping matrix, where It is a linear damping coefficient; , For transformation matrix; in the dynamic positioning process of unmanned vessels, such as stopping and anchoring when performing tasks like scientific characterization and exploration, the yaw angle is typically used. It is small enough. Therefore, without loss of generality, we can assume... ,thereby , ,at this time ; Disturbance signal It is square-integrable, the disturbance signal. It is generated by the external system below, as shown in formula (39). (39) in, The disturbance matrix is ​​denoted by ; this is due to wind, waves, and ocean currents present in the marine environment. To describe the continuous impact of wind, waves, and currents on the unmanned vessel, we assume... All eigenvalues ​​have a real part of zero, meaning the external system (39) is neutral and stable; S12. Obtain the total number of shipborne equipment. In order to eliminate the frequent switching problem caused by using the partition modeling method, the change in the total number of shipborne equipment on the unmanned vessel is used to reflect the change in the unmanned vessel mode. The unmanned vessel mode is shown in formula (40). (40) in, This indicates the total number of onboard devices carried by the unmanned surface vessel. , This indicates the maximum number of devices on a fully loaded unmanned vessel. , Represents the smallest unit mass of shipboard equipment; the... The quantity and unit mass of the shipborne equipment are respectively and , ; When the first When shipborne equipment is released or recovered, It may decrease or increase accordingly; The total mass of the unmanned vessel is redefined using the unmanned vessel modes and the total number of onboard equipment, as shown in formula (41). (41) in ; Reconstruct the matrix using unmanned vessel modes and the number of onboard devices. The matrix is ​​as follows:

[0013] Among them, Reconstructed as

[0014] in, It is the total unloaded mass of the unmanned vessel. The corresponding hydrodynamic increase in mass, It is the total unloaded mass of the unmanned vessel. The corresponding inertia matrix, It is the total unloaded mass of the unmanned vessel. The corresponding linear damping coefficient; It is the first Unit mass of shipborne equipment The corresponding hydrodynamic increase in mass, It is the first Unit mass of shipborne equipment The corresponding inertia matrix, It is the first Unit mass of shipborne equipment The corresponding linear damping coefficient; Based on formulas (40) and (41), formulas (38) and (39) are rewritten as switching LPV networked unmanned vessel systems, as shown in formula (42). (42) in, ; Represents the state matrix; ; ; Represents the input matrix, Represents the perturbation matrix; Indicates the system status. Indicates adjusting the output. and These are the output matrix and the output perturbation matrix, respectively. It is the measurement output of the dynamic positioning of the unmanned vessel under disturbance.

[0015] Number of unmanned surface vessel onboard equipment and At that time, unmanned ship mode At this point, the output adjustment of the LPV networked unmanned vessel system (42) may be unsolvable, meaning the state of the unmanned vessel (38) may diverge due to factors such as thruster failure. When the number of onboard devices of the unmanned vessel and from Change to At that time, unmanned ship mode Depend on Become At this point, we can say that the quantity and... The moment of change is the modality switching point. In terms of quantity and... At the moment of change, the state of the unmanned vessel (38) may exhibit a divergent / convergent trend, i.e., the transition point between unstable and stable modes. Unmanned vessel modes Corresponding switching sequence exist Number of switches on Equal to stable subsequence Number of switching and unstable subsequences Number of switching The sum of .

[0016] In a specific embodiment, the scheme involves constructing an event-triggered mechanism based on the unmanned surface vessel's (USV) regulation output and USV modes, building the USV regulation output and USV modes after a deception attack based on the event-triggered mechanism, transmitting them to a shore-based system for deception attack observation to establish a deception attack detection observer, and defining a deception attack detector based on time window chi-square detection technology. S21. Use sensors to sample the switching LPV networked unmanned surface vessel system to obtain the regulated output. and Construct an event triggering mechanism, as shown in formula (43). (43) in, This is the trigger time for the (k+1)th event. For the (k+1)th time, the error condition is satisfied. The triggering time, For the (k+1)th time to satisfy the modal condition Triggering time: ; , It is the triggering time. The next One sampling point, ; Output error , This is the weight matrix to be designed. It is a given threshold parameter; for The modality of time, for Each sampling time, for Adjusting output at any time for Adjusting the output at any time; S22. Based on the event triggering mechanism, establish the regulation output and unmanned vessel mode after being deceived and tampered with by the deception attack, as shown in formulas (44) and (45). (44) (45) in, The adjusted output after being deceived and tampered with. To tamper with the output of a deceptive attack. The unmanned ship mode was altered after being deceived and attacked. For the first The unmanned ship mode at any time, Deceptive attacks aimed at altering the mode of unmanned vessels; S23. Use the adjusted output received by the shore-based system to observe the status of the unmanned vessel system and detect spoofing attacks in the hull-shore channel, and construct a spoofing attack detection observer as shown in formula (46). (46) in, These are state observations. It outputs the observed values. It is the observer gain matrix that is obtained by substitution. It is the modal estimate of the safety controller. , To adjust the output residual signal for detecting spoofing attacks in the ship-to-shore channel, Output matrix for observation; S24. Use time window chi-square detection technology to determine whether the deception attack has tampered with the regulation output and the unmanned vessel mode. The detector is shown in formula (47). (47) in, , For time windows The set detection function, The detection threshold; When the detection function At that time, the test results If both the regulation output and the unmanned surface vessel (USV) mode are detected to have been tampered with, then it is confirmed that they have not been tampered with. The attack detection for the regulation output and the USV mode shares a single detection function. Formula (47) detected the first The activation interval of a deception attack and the dormancy interval are ,in It is the first The duration of each detected spoofing attack, plus... , , This represents the number of times a spoofing attack was detected.

[0017] In a specific embodiment, the scheme of constructing a switched fuzzy wavelet network and building a security controller based on the switched fuzzy wavelet network and the detector for spoofing attacks is as follows: S31. Establish fuzzy rules: rules :if yes , , yes The output of the fuzzy rule is shown in formula (48). (48) in, , It is the total number of fuzzy rules. It is the first The input of the fuzzy rule, , It is the first The output of the fuzzy rules, It is the weight matrix of the output layer. It is the first The first fuzzy rule Wavelet functions, For fuzzy sets, The Gaussian fuzzy membership function is The design parameters of the Gaussian membership function; and The expansion and translation parameters are the same as those of the wavelet function, respectively. S32. Construct the output vector of the fuzzy wavelet network, as shown in formula (49). (49) in, These represent the state, scale, translation parameters, and weights of the fuzzy wavelet network, respectively. , , ; For fuzzy wavelet basis function vectors, ; No. The fuzzy wavelet basis functions are: ,in , ; Because fuzzy wavelet networks, as a universal approximator, can achieve arbitrary precision... Approximation definition in compact set Any continuous function on Therefore, (49) can be written as Among them, the approximation error The expected weight, inflation, and translation parameter vectors , , satisfy , , ; S33. Obtain the deception attack estimate based on the output vector of the fuzzy wavelet network, as shown in formula (50). (50) in, Describes the basis functions for estimation; , , They represent respectively to , , The estimated value; , , The following switching adaptive law is satisfied: (51) in, For weight parameters, , , and For intermediate parameters; ; ; , ; , , , , , These are, respectively, state parameters, control and detection parameters, error parameters, estimation parameters, Lyapunov parameters, and time delay matrix parameters. , Let represent the basis functions for estimating the scaling and translation parameters, respectively; Matrix parameters , , , , , With unmanned ship mode Estimated controller modes Safety controller mode Relatedly, on the one hand, tampering through deception attacks is unavoidable, and shore-based systems cannot obtain absolutely accurate unmanned vessel modes. Switching to a fuzzy wavelet network can only obtain the security controller mode. On the other hand, safety controller mode This is the closest to the unmanned ship mode under the defense mechanism of this invention. The result. Therefore, we can let the switching adaptive law (51) be... Thus, the matrix parameters in (51) are only related to the safety controller mode. Related. When After the switch occurs, the matrix parameters of the switching adaptive law (15) also change accordingly, and the weight, dilation, and translation parameter vectors used in (50) are also changed. , , It has switching characteristics. Therefore, a fuzzy wavelet network with this switching adaptive law (51) can be called a switched fuzzy wavelet network.

[0018] Note 1: The spoofing attack estimate (50) of the switched fuzzy wavelet network output can be closer to the true value in the switched system. In the study of fuzzy wavelet networks in non-switched systems, the parameters used in the adaptive law of the fuzzy wavelet network are fixed. However, in the switched LPV system considered in this invention, the system state used in the adaptive law is different. In networked unmanned vessel mode After a switch occurs, a sharp change will happen. If matrix parameters that do not match the current system mode are continued to be used, it will lead to changes in the weight, inflation, and translation parameter vectors. , , The deviation is so great that it deceives the output of the attack estimate. The accuracy is reduced. The weight, dilation, and translation parameter vectors used in the deception attack estimate are... , , The switching adaptive law (51) in the safety controller mode After the switch occurs, the corresponding matrix parameters , , , , , Switching to the appropriate mode can reduce the weight, dilation, and translation parameter vectors. , , The deviation.

[0019] S34. The detection results of the deception attack detector. The deception attack estimate and the modified regulation output obtained by the deception attack are combined to obtain the security controller output and the security controller mode, as shown in formulas (52) and (53). (52) (53) When the regulation output received by the shore-based system and the unmanned vessel mode are not subjected to spoofing attacks, the safety output uses the trigger output. The safety controller mode uses a sensor-to-controller network. Otherwise, the security output uses a spoofing attack estimate to compensate for the signal transmitted through the sensor-to-controller network. The safety controller modality uses the controller modality estimate. ; Note 2: The security output (52) based on the switching fuzzy wavelet network defense mechanism is more reliable. In previous conclusions, when a spoofing attack was detected, the decision mechanism would select the observer's observation output. Replace the transmitted data that has been attacked and tampered with As a reliable output signal. However, on the one hand, in the initial stage of system operation, the state error between the observer and the unmanned vessel... The error is relatively large at first, and it will only approach zero as the system continues to operate. At this point, the observer output... Output of unmanned ships Large error, safe output The accuracy is low; on the other hand, during asynchronous behavior, the unmanned vessel mode... With observer mode If they are inconsistent, then there must be a deviation between their states. Therefore, the accuracy of the observed output is relatively low in the early stages of system operation. On the one hand, the safety output (52) does not use an alternative method, but directly transmits the output through the network. Subtract the deception attack estimate from the deception attack function. This eliminates the problem of large errors in the early stages of system operation. On the other hand, the weight, dilation, and translation parameter vectors used in the deception attack estimation... , , It is based on the current safety controller mode. The chosen switching adaptive law can largely eliminate deviations caused by asynchronous behavior.

[0020] S35. Construct a safety controller based on the safety output and safety controller modes to ensure the stability of the unmanned vessel, as shown in formula (54). (54) in, The controller gain matrix to be determined is the safety controller mode. Corresponding switching sequence exist Number of switches on Equal to stable subsequence Number of switching and unstable subsequences Number of switching The sum of .

[0021] In a specific embodiment, a shore-based-ship channel event triggering mechanism is designed. This mechanism, combined with the security controller, is used to construct the security controller output and modality after a deception attack. The scheme for designing a deception attack detection standard for the security controller is as follows: S41. The control signal of the safety controller is transmitted to the actuator as the control input for switching the LPV networked unmanned vessel. In order to save network resources, an event triggering mechanism is used in the shore-to-ship channel, that is, the shore-to-ship channel event triggering mechanism is designed as shown in formula (55). (55) in, Indicates the first A trigger moment, It is a sufficiently small positive scalar. To trigger the counter, the counter is incremented by 1 after the event triggering mechanism is triggered; Under the shore-based-ship passage event triggering mechanism, for have Due to network latency Safety controller mode and safety controller output At any moment Upon reaching the side of the hull, among which, , , , ; S42. Based on the event triggering mechanism combined with the security controller and the shore-based-ship channel event triggering mechanism, the security controller output and security controller mode that have been tampered with by the deception attack are constructed, as shown in formulas (56) and (57). (56) (57) in, For the first The safety controller output at each trigger moment For the first Safety controller modality at each trigger moment; For deception attack; S43. Compare the safety controller modes and the number of shipboard devices. To determine whether the security controller output has been tampered with by a deception attack, a detector is constructed as shown in formula (58). (58) Among them, the test results This means that a spoofing attack was detected in the output of the security controller. If the input is tampered with, it needs to be compensated; otherwise, it will be detected that it has not been tampered with.

[0022] In this scheme, the constructed switching adaptive law ensures that after the system mode changes, the input data of the fuzzy wavelet network is the data of the system under the current mode, thereby improving the accuracy of estimation. On the other hand, when the transmitted data is detected to be tampered with, the constructed defense mechanism compensates the transmitted data with the estimated deception attack approximation value, thereby increasing its security.

[0023] In a specific embodiment, the deception attack detector based on the security controller is based on a scheme that constructs a control input compensator and constructs event-triggered control inputs based on the control input compensator and the security controller: Based on the detection results and the output of the security controller that has been tampered with by the deception attack, a control input compensator is constructed to obtain the compensated control input, as shown in formula (59). (59) in, For the compensated control input, The maximum event trigger interval to be determined is... Previously, when a spoofing attack was detected, the control input compensator used... Compensate the safety controller output; otherwise, continue using the safety controller output. Combining formulas (44), (52), (54), and (59), the event-triggered control input for switching LPV networked unmanned surface vessel systems under deception attacks is constructed, as shown in formula (60). (60) in, Let be the controller gain matrix to be determined; in this scheme, substitute formula (44) into formula (52) and replace the value in formula (52). Then substitute the replaced formula (52) into formula (54) and replace the formula in formula (54). Finally, substitute the replaced formula (54) into formula (59) to obtain formula (60); interval Divide into sub-intervals, as follows:

[0024] in ; Using the input delay method on the above-described sub-intervals, the lower bound of the delay is derived. , ; Based on the divided sub-intervals, redefine the event trigger control input as shown in formula (61). (61).

[0025] In a specific embodiment, the scheme of constructing an error system by combining a deception attack detection observer, a switching LPV networked unmanned surface vessel system, and an event-triggered control input, and constructing a closed-loop system based on the error system and the switching LPV networked unmanned surface vessel system is as follows: S61. To prove the effectiveness of the deception attack detection observer, the LPV networked unmanned vessel system and the deception attack detection observer are switched together. Substituting the redefined event-triggered control input, an error system is constructed as shown in formula (62). (62) in, Here is the observer gain matrix, and the observation error is... Let the estimation error of the deception attack be... Thus, we obtain formula (63). (63) in, , , For weighted error, It is the error basis function; To approximate the error; To derive an adaptive learning theorem using Lyapunov's direct method, it is necessary to... Therefore, Taylor series expansion linearization techniques are used to extend the linearization. Expanding the formula, we get formula (64). (64) In the formula , , yes about The partial derivative, yes about The partial derivative, It is a higher-order term in a Taylor expansion. and To separate scale error and translation error; Substituting equations (63) and (64) into equation (62), the error system is rewritten as shown in equation (65). (65) in, intermediate parameters ; S62. Based on the estimation error of the error system and the construction of the LPV networked unmanned vessel system, a closed-loop system is constructed, as shown in formula (66). (66) in, Represents the joint state of the closed-loop system, initially , ; , , , , , , These are, respectively, state parameters, control and detection parameters, error parameters, estimation parameters, Lyapunov parameters, and time delay matrix parameters. , , and For intermediate parameters; ; ; ; ; , , ; ; Introducing a coordinate transformation matrix within the output adjustment framework and To eliminate the closed-loop system that is about to be established Item, matrix , and To meet the requirements of formula (67), (67).

[0026] The construction of the safety controller mode and network latency cause asynchronous behavior between the safety controller of the shore-based system and the switching LPV networked unmanned surface vessel. This is due to the detection of a spoofing attack activation interval. Duration and unmanned vessel mode arbitrary activation interval The uncertainty of the relationship between durations, intervals The duration may be greater or less than the interval. Duration. During the interval Duration less than interval Time, interval In the interval Different internal positions will lead to different safety controller modes Different values ​​of result in complex asynchronous behavior. Therefore, it is necessary to discuss the different scenarios of asynchronous behavior in detail. Without loss of generality, assume that Above, unmanned ship mode ;exist Above, unmanned ship mode , The error system (62) is in the interval Estimated controller modes , .according to and The changes can be categorized into three cases: Scenario A: In It was detected on the top This is a deception attack, and none of the deception attacks cover the controller mode switching time. That is Due to the construction of the security controller modality, within the activation range where a spoofing attack is detected, Equal to the estimated controller mode In the dormant interval where no deception attack was detected, Equal to unmanned ship mode Therefore, modality as follows: exist superior, , ; exist superior, ; exist superior, .

[0027] Scenario B: In It was detected on the top The second deception attack, and the third The secondary deception attack covers the controller's mode switching time. That is Similar to case A, the modality As follows: superior, , ;exist superior, ;exist superior, .

[0028] Case C: Detected spoofing attack activation range cover ,Right now At this moment, the entire On the interval Both are equal to the estimated controller modes Therefore, mode As follows: In superior, .

[0029] The dynamic positioning of the constructed switching LPV networked unmanned surface vessel system is transformed into the output regulation problem of a closed-loop system, namely: 1) when When the closed-loop system is exponentially stable; 2) when When, for any initial conditions The regulation output of the closed-loop system satisfies .

[0030] In a specific embodiment, the scheme for realizing the dynamic positioning of the unmanned vessel based on the closed-loop system, and realizing the dynamic positioning of the unmanned vessel through the closed-loop system under the constructed conditions, is as follows: To establish the conditions for achieving dynamic positioning of unmanned vessels, under the constraints of formula (67), for a given positive parameter... , , , , , , Dynamic positioning of LPV-networked unmanned surface vessels (USVs) based on switching the number of onboard devices is feasible under the following conditions: Condition 1: There exists a matrix , , , , , , , Make (68) (69) and It is a constant; Condition 2: For known parameters , , , , There exists a matrix , , To meet the requirements of formulas (70) and (71), (70) (71) in, These are the delay matrices for no delay, first-order delay, and second-order delay, respectively. Condition 3: Modes of the unmanned vessel The required number of stable / unstable unmanned surface vessel switching points needs to be met. Number of stable / unstable controller switching points Quantity restrictions between: (72) And the average length of stay criteria: (73) in dimensional block matrix It consists of the following block matrices, , , , , , , , , , , , , , , , The block matrices at the remaining positions are zero matrices of appropriate dimension. , , , , , , .

[0031] Prove and verify the validity of each condition: Proof: Due to the introduction of the switching fuzzy wavelet network, it is necessary to consider candidate Lyapunov functions in two cases. When and When choosing, select the Lyapunov function: (74) in, , , , .

[0032] Differentiating (74) along the closed-loop system, we get: (75) (76) (77) (78) Applying the lemma, from condition (69), we have in (77) (79) (80) because , , They are all constants, so , , Therefore, in (78) there is (81) By combining (75) and (81), and based on the switching adaptive law and condition (68), we can obtain the relationship between the candidate Lyapunov function and its derivative: (82) in ,

[0033] when or When choosing, select the Lyapunov function: (83) in , , .

[0034] Combining (75)-(77) and (79)-(80), using the same derivation method as (83), we can obtain (84) in .

[0035] Analyze the Lyapunov function in arbitrary intervals under three cases. Changes on.

[0036] Scenario A: For , At this point, the unmanned surface vessel system and the safety controller exhibit asynchronous behavior; therefore, the following selection is made. Integrating both sides of (82) gives: (85) for , Select , Using the same integration method as (82), we get: (86) for , , Select , ,have: (87) for , Select , ,have: (88) for , Select , ,have: (89) for , , Select , ,have: (90) for , Select , ,have: (91) exist The controller mode switching caused by network latency and spoofing attacks necessitates a discussion of the timing of mode switching. , , The changes in the Lyapunov function before and after. For Due to different coordinate transformations, Combining (79) and ,have .

[0037] Depend on get According to (70), .

[0038] Using the same method, there exists , , .therefore (92) Using the same derivation method, for , ,have: (93) for , ,have: (94) for , ,have: (95) for , ,have: (96) By combining (85) and (96), we can obtain any interval. Above: (97) Case B: Similar to Case A, for , Select ,have:

[0039] for , Select , ,have: ; for , , Select , ,have: ; for , , Select , ,have: ; for , Select , ,have: .

[0040] Due to the A deception attack activation zone Controller mode switching time caused by overlay network latency Therefore, in this case, we only need to discuss the mode switching time. , The changes in the Lyapunov function before and after. For , ,have:

[0041] for , ,have:

[0042] for , ,have:

[0043] for , ,have:

[0044] By simultaneously solving all formulas in case B, we can obtain the result for any interval. Above: (98) Case C: Due to the first A deception attack activation zone Covering the entire interval Therefore, for , Select ,have: .

[0045] Then, the networked unmanned surface vessel modes were analyzed. Switching time The relationship between the Lyapunov functions before and after. From condition (71), we get:

[0046] Based on formulas (97), (98) and the two formulas in case C, the lemma can be used to derive the following: superior

[0047] in , , Representing an interval Duration of internal attacks Representing an interval The synchronization duration within. Because of conditions (72) and (73). , It can be deduced that...

[0048] in ,

[0049] Second, due to the lemma, we have and Make This means Due to the closed-loop system , Therefore, the conditions prove that the dynamic positioning of the unmanned vessel can be satisfied.

[0050] In summary, it has been proven that the output regulation of the closed-loop system under the external system (39) is solvable, that is, the dynamic positioning of the constructed networked unmanned vessel switching LPV model is solvable.

[0051] To verify the effectiveness of this invention, an example of an unmanned surface vessel with varying mass is used to verify the effectiveness of the proposed model conversion method, detection method, and safety control scheme.

[0052] Considering the unloaded mass of the unmanned vessel Generally, the payload capacity of an unmanned surface vessel (USV) is designed based on its own weight, typically ranging from 30% to 100% of its weight. Therefore, the full-load mass of the USV must be considered. The first type of shipborne equipment quality ; Type 2 shipboard equipment quality their respective quantities .

[0053] in addition, Figure 4 Surge quality in Its corresponding quantity parameter The specific quantity, mass, and modal correspondence of the unmanned surface vessel's onboard equipment during navigation are shown in Table 1, along with their changes. Figure 4 The unmanned vessel mode was showcased. The total mass of the unmanned vessel. for , , .

[0054] Table 1. Quantity, mass, and modal correspondence of onboard equipment for networked unmanned surface vessels.

[0055] Table 2 Parameters of the unmanned surface vessel under three conditions

[0056] Since the surge mass is negligible compared to the mass of the networked unmanned surface vessel (USV), its impact on the coefficient matrix can be ignored. Therefore, using the USV parameters in Table 2, the matrix... and for , , , , , .

[0057] At this point, the unmanned surface vessel (USV) has been transformed into a switching LPV networked USV system comprising two subsystems:

[0058] in coefficient matrix , , for: , , , , , , matrix It is a 6-dimensional identity matrix. and Selected as: .

[0059] Without loss of generality, choose the initial state. Other parameters , , , , , , , , , , , Based on the duration of the detected spoofing attack, there are , Based on the given parameters and according to condition (37), the maximum and minimum average dwell times can be calculated. , Select the range that meets this average dwell time. The switching rules, such as Figure 2 As shown, the number of shipborne devices on networked unmanned vessels is... It should remain constant throughout the time period. From Assumption 2, the gain matrix of the detector / observer is:

[0060] Applying Theorem 1, the gain matrix of the safety controller can be obtained as follows: , .

[0061] This embodiment's simulation considers the dynamic positioning of a single unmanned vessel, and the data curves are as follows: Figures 3-12 As shown. Figure 3 As shown, the number of shipborne devices on networked unmanned vessels changes in The range is acceptable. From Figures 4-5 As can be seen from this, switching to networked LPV unmanned vessels alone affects the number of devices and , Position error during operation and Yaw angle Surge speed sway speed yaw rate The curves cannot all approach zero, meaning that switching to a networked LPV-based unmanned vessel cannot achieve dynamic positioning. Figure 6 The event triggering mechanism and its triggering interval in the hull-shore channel are shown.

[0062] exist Figure 7 In the middle, the controller modal estimate The values ​​are given, along with estimates and unmanned surface vessel modes. The mean square error between the estimated values ​​can be seen from the mean square error of the estimated values. Only in unmanned vessel mode While there may be errors in the switching points, their accuracy is reliable. Figure 8 A deception attack detection function is given. and test results This means that the detection standard can detect the activation range of deception attacks. Figure 9 Demonstrates safety controller modality The asynchronous duration between the networked unmanned surface vessel and its controller. A real deception attack function. And a spoofing attack function using switched fuzzy wavelet network estimation. exist Figure 10 The estimated value can be seen in the display. Aside from some fluctuations at the beginning and end of the attack, the accuracy is reliable at other times. Furthermore, the estimation of the deception attack function under both switching and non-switching adaptive laws is also demonstrated. Compared to real deception attacks The error shows that the estimation result under the switching adaptive law is closer to the real deception attack function than that under the non-switching adaptive law. Through Figure 11The controller output signals of a networked unmanned surface vessel (USV) are compared under three scenarios: no deception attack, deception attack with a deception attack detection observer, and deception attack with a switched fuzzy wavelet network. Control signals including yaw angle yaw rate control signal and control signal for swing speed It can be seen that in the early stages of system operation, under the condition of being subjected to a deception attack and being detected using a deception attack detection observer, the security output... Significant fluctuations occur during deception attacks, while the secure output of a security controller built during a deception attack is protected. Only slight fluctuations occurred in the initial stage of the deception attack activation, which verifies that the security output under the constructed switching fuzzy wavelet network defense mechanism is more secure and reliable. Figure 12 The position error shown and Yaw angle Surge speed sway speed yaw rate The curve approaching zero indicates that dynamic positioning of the networked unmanned vessel can be achieved.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for safe dynamic positioning of unmanned surface vessels based on switched fuzzy wavelet networks, characterized in that, include: S1: Obtain the number of shipborne equipment, use the switching LPV modeling method to construct a switching LPV networked unmanned ship system based on the number of shipborne equipment, obtain the unmanned ship mode and the unmanned ship's regulation output, and transmit the unmanned ship mode and the unmanned ship's regulation output to the sensor for sampling; S2: Based on the unmanned vessel's regulation output and unmanned vessel mode, an event triggering mechanism is constructed. Based on the event triggering mechanism, the regulation output and unmanned vessel mode after being deceived are constructed and transmitted to the shore-based system for observation of the deception attack to establish a deception attack detection observer. A deception attack detector is constructed based on the time window chi-square detection technology. S3: Construct a switching fuzzy wavelet network and build a security controller based on the switching fuzzy wavelet network and the detector for spoofing attacks; S4: Design a shore-based-ship channel event triggering mechanism, combine the security controller and the shore-based-ship channel event triggering mechanism to construct the security controller output and security controller mode after a deception attack, and then design a deception attack detector for the security controller; S5: The deception attack detector based on the security controller is based on the construction of a control input compensator, and on the control input compensator and the security controller, an event-triggered control input is constructed. S6: Construct an error system by combining a deception attack detection observer, a switching LPV networked unmanned surface vessel system, and an event-triggered control input; and construct a closed-loop system based on the error system and the switching LPV networked unmanned surface vessel system. S7: Based on the closed-loop system, construct the conditions for realizing the dynamic positioning of the unmanned vessel, and realize the dynamic positioning of the unmanned vessel through the closed-loop system under the constructed conditions.

2. The unmanned vessel safety dynamic positioning method based on switched fuzzy wavelet networks according to claim 1, characterized in that, Obtain the number of onboard devices and construct a switching LPV networked unmanned surface vessel system based on the number of onboard devices using a switching LPV modeling method, including: S11. Construct an unmanned vessel model, as shown in formula (1). （1） Among them, surge velocity sway speed yaw rate Forming a vector Position coordinates , and yaw angle Forming a vector Control input vector , , , These represent the sway force, the roll force, and the pitching moment, respectively. , Let be the inertia matrix, where It is hydrodynamics that increases mass. This represents the longitudinal component of the difference vector from the origin to the ship's center of gravity in the ship's fixed reference frame. It is the inertial matrix, the total mass of the unmanned vessel. ,satisfy ,in and It is the total mass of the unmanned vessel when it is empty and fully loaded. This represents the payload of the unmanned vessel as it changes over time. Indicates the surge mass as it changes over time; , Let be the damping matrix, where It is a linear damping coefficient; , This is the transformation matrix; Disturbance signal As shown in formula (2), （2） in, Here is the perturbation matrix; S12. Obtain the total number of shipborne equipment. Use the change in the total number of shipborne equipment on the unmanned vessel to reflect the change in the unmanned vessel's modes. The unmanned vessel modes are shown in formula (3). （3） in, This indicates the total number of onboard devices carried by the unmanned surface vessel. , This indicates the maximum number of devices on a fully loaded unmanned surface vessel. , Represents the smallest unit mass of shipboard equipment; the... The quantity and unit mass of the shipborne equipment are respectively and , ; ; The total mass of the unmanned vessel is redefined using the unmanned vessel modes and the total number of onboard equipment, as shown in formula (4). （4） in ; Reconstruct the matrix using unmanned vessel modes and the number of onboard devices. The matrix is ​​as follows: Among them, Reconstructed as in It is the total unloaded mass of the unmanned vessel. The corresponding hydrodynamic increase in mass, It is the total unloaded mass of the unmanned vessel. The corresponding inertia matrix, It is the total unloaded mass of the unmanned vessel. The corresponding linear damping coefficient; It is the first Unit mass of shipborne equipment The corresponding hydrodynamic increase in mass, It is the first Unit mass of shipborne equipment The corresponding inertia matrix, It is the first Unit mass of shipborne equipment The corresponding linear damping coefficient; Based on formulas (3) and (4), formulas (1) and (2) are rewritten as switching LPV networked unmanned vessel systems, as shown in formula (5). （5） in, ; Represents the state matrix; ; ; Represents the input matrix, Represents the perturbation matrix; Indicates the system status. Indicates adjusting the output. and These are the output matrix and the output perturbation matrix, respectively. It is the measurement output of the dynamic positioning of the unmanned vessel under disturbance.

3. The unmanned vessel safety dynamic positioning method based on switched fuzzy wavelet networks according to claim 2, characterized in that, Based on the unmanned surface vessel's (USV) regulation output and USV mode construction, an event-triggered mechanism is established. This mechanism is used to construct the USV's regulation output and USV mode after a deception attack, which are then transmitted to a shore-based system for deception attack observation to establish a deception attack detection observer. Furthermore, a deception attack detector is defined based on a time-window chi-square detection technique, including: S21. Use sensors to sample the switching LPV networked unmanned surface vessel system to obtain the regulated output. and Construct an event triggering mechanism, as shown in formula (6). （6） in, The trigger time of the (k+1)th event. For the (k+1)th time, the error condition is satisfied. The triggering time, For the (k+1)th time, the modal condition is satisfied. Triggering time: ; , It is the triggering time. The next One sampling point, ; Output error , This is the weight matrix to be designed. It is a given threshold parameter; for The modality of time, for Each sampling time, for Adjusting output at any time for Adjusting the output at any time; S22. Based on the event triggering mechanism, establish the regulation output and unmanned vessel mode after being tampered with by the deception attack, as shown in formulas (7) and (8). （7） （8） in, The adjusted output after being deceived and tampered with. To tamper with the output of a deceptive attack. The unmanned ship mode was altered after being deceived and attacked. For the first The unmanned ship mode at any time, Deceptive attacks aimed at altering the mode of unmanned vessels; S23. Use the adjusted output received by the shore-based system to observe the status of the unmanned vessel system and detect spoofing attacks in the hull-shore channel, and construct a spoofing attack detection observer as shown in formula (9). （9） in, These are state observations. It outputs the observed values. It is the observer gain matrix that is obtained by substitution. It is the modal estimate of the safety controller. , To adjust the output residual signal for detecting spoofing attacks in the ship-to-shore channel, Output matrix for observation; S24. Use time window chi-square detection technology to determine whether the deception attack has tampered with the regulation output and the unmanned vessel mode. The detector is shown in formula (10). （10） in, , For time windows The set detection function, For the detection threshold, When the detection function At that time, the test results At this point, it is confirmed that the regulation output and the unmanned vessel mode have been tampered with simultaneously; otherwise, it is confirmed that they have not been tampered with.

4. The unmanned vessel safety dynamic positioning method based on switched fuzzy wavelet networks according to claim 3, characterized in that, A switched fuzzy wavelet network is constructed, and a security controller is built based on the switched fuzzy wavelet network and a detector for spoofing attacks, including: S31. Establish fuzzy rules: rule :if yes , , yes The output of the fuzzy rule is shown in formula (11). （11） in, , It is the total number of fuzzy rules. It is the first The input of the fuzzy rule, , It is the first The output of the fuzzy rules, It is the weight matrix of the output layer. It is the first The first fuzzy rule Wavelet functions, For fuzzy sets, The Gaussian fuzzy membership function is shown in formula (12). （12） S32. Construct the output vector of the fuzzy wavelet network, as shown in formula (13). （13） in, These represent the state, scale, translation parameters, and weights of the fuzzy wavelet network, respectively. , , ; For fuzzy wavelet basis function vectors, ; No. The fuzzy wavelet basis functions are: ,in , ; S33. Obtain the deception attack estimate based on the output vector of the fuzzy wavelet network, as shown in formula (14). （14） in, Describes the basis functions for estimation; , , They represent respectively to , , The estimated value; , , The following switching adaptive law is satisfied: （15） in, For weight parameters, , , and For intermediate parameters; ； ； , ; , , , , , These are, respectively, state parameters, control and detection parameters, error parameters, estimation parameters, Lyapunov parameters, and time delay matrix parameters. , Let represent the basis functions for estimating the scaling and translation parameters, respectively; S34. The detection results of the deception attack detector. The deception attack estimate and the modified regulation output obtained by the deception attack are combined to obtain the security controller output and the security controller mode, as shown in formulas (16) and (17). （16） （17） When the regulation output received by the shore-based system and the unmanned vessel mode are not subjected to spoofing attacks, the safety output uses the trigger output. The safety controller mode uses a sensor-to-controller network. Otherwise, the security output uses a spoofing attack estimate to compensate for the signal transmitted through the sensor-to-controller network. The safety controller modality uses the controller modality estimate. ; S35. Construct a safety controller based on the safety output and safety controller mode, as shown in formula (18). （18） in, The controller gain matrix to be determined is the safety controller mode. Corresponding switching sequence exist Number of switches on Equal to stable subsequence Number of switching and unstable subsequences Number of switching The sum of .

5. The unmanned vessel safety dynamic positioning method based on switched fuzzy wavelet networks according to claim 4, characterized in that, Design a shore-based-ship channel event triggering mechanism, and combine the security controller and the shore-based-ship channel event triggering mechanism to construct the security controller output and security controller mode after a deception attack. Then, design a deception attack detection standard for the security controller, including: S41. Design the shore-to-ship passage event triggering mechanism, as shown in formula (19). （19） in, Indicates the first A trigger moment, It is a sufficiently small positive scalar. To trigger the counter, the counter is incremented by 1 after the event triggering mechanism is triggered; S42. Construct the security controller output and security controller mode that have been tampered with by the deception attack by combining the security controller and the shore-based-ship channel event triggering mechanism, as shown in formulas (20) and (21). （20） （21） in, For the first The safety controller output at each trigger moment For the first Safety controller modality at each trigger moment; For deception attack; S43. Compare the safety controller modes and the number of shipboard devices. To determine whether the security controller output has been tampered with by a deception attack, a detector is constructed as shown in Equation (22). （22） Among them, the test results This means that a spoofing attack was detected in the output of the security controller. If the input is tampered with, it needs to be compensated; otherwise, it will be detected that it has not been tampered with.

6. The unmanned vessel safe dynamic positioning method based on switched fuzzy wavelet networks according to claim 5, characterized in that, The deception attack detector based on the security controller is based on constructing a control input compensator, and on constructing event-triggered control inputs based on the control input compensator and the security controller, including: Based on the detection results and the output of the security controller that has been tampered with by the deception attack, a control input compensator is constructed to obtain the compensated control input, as shown in formula (23). （23） in, For the compensated control input, The maximum event trigger interval to be determined is... Previously, when a spoofing attack was detected, the control input compensator used... Compensate the safety controller output; otherwise, continue using the safety controller output. Combining formulas (7), (16), (18) and (23), the event-triggered control input for the switching LPV networked unmanned vessel system under deception attack is constructed, as shown in formula (24). （24） in, Let be the controller gain matrix to be determined; interval Divide into sub-intervals, as follows: in ; Using the input delay method on the above-described sub-intervals, the lower bound of the delay is derived. , ; Based on the divided sub-intervals, redefine the event trigger control input as shown in formula (25). （25）。 7. The method for safe dynamic positioning of unmanned vessels based on switched fuzzy wavelet networks according to claim 6, characterized in that, A deception attack detection observer, a switching LPV networked unmanned surface vessel system, and event-triggered control inputs are used to construct an error system. Based on the error system and the switching LPV networked unmanned surface vessel system, a closed-loop system is constructed, including: S61. By switching the LPV networked unmanned surface vessel system and the deception attack detection observer together, and substituting them into the redefined event-triggered control input, an error system is constructed as shown in formula (26). （26） in, Here is the observer gain matrix, and the observation error is... Let the estimation error of the deception attack be... Thus, we obtain formula (27). （27） in, , , For weighted error, It is the error basis function; To approximate the error; Linearization using Taylor series expansion techniques Expanding the equation, we get formula (28). （28） In the formula , , yes about The partial derivative, yes about The partial derivative, It is a higher-order term in a Taylor expansion. and These are scale error and translation error, respectively. Substituting equations (27) and (28) into equation (26), the error system is rewritten as shown in equation (29). （29） in, For intermediate parameters; ； S62. Based on the estimation error of the error system and the construction of the closed-loop system of the LPV networked unmanned vessel system, as shown in formula (30), （30） in, Represents the joint state of the closed-loop system, initially , ; , , , , , , These are, respectively, state parameters, control and detection parameters, error parameters, estimation parameters, Lyapunov parameters, and time delay matrix parameters. , , and These are all intermediate parameters; ； ； ； ； ， ， ； ； matrix , and To meet the requirements of formula (31), （31）。 8. The method for safe dynamic positioning of unmanned vessels based on switched fuzzy wavelet networks according to claim 7, characterized in that, Based on the aforementioned closed-loop system, conditions are established for achieving dynamic positioning of the unmanned vessel, and dynamic positioning of the unmanned vessel is achieved through the closed-loop system under these conditions, including: Under the constraints of formula (31), the conditions for realizing the dynamic positioning of unmanned ships are constructed as follows: Condition 1: There exists a matrix , , , , , , , Make （32） （33） and It is a constant; Condition 2: For known parameters , , , , There exists a matrix , , To meet the requirements of formulas (34) and (35), （34） （35） in, These are delay-free, first-order delay, and second-order delay matrices; Condition 3: Modes of the unmanned vessel The required number of stable / unstable unmanned surface vessel switching points needs to be met. Number of stable / unstable controller switching points Quantity restrictions between: （36） And the average length of stay criteria: （37） in dimensional block matrix It consists of the following block matrices, ， ， ， ， ， ， ， ， ， ， ， ， ， ， ， The block matrices at the remaining positions are zero matrices of appropriate dimension. ， ， ， ， ， ， 。

Images