An Improved Navigation-Following-Escort Method Based on Intelligent Interception and Fault-Tolerant Reconfiguration

Abstract

This invention discloses an improved navigation and escort method based on intelligent interception and fault-tolerant reconfiguration, comprising the following steps: Step 1, constructing a control framework for the leading unmanned surface vessel (USV) following the formation, and constructing a sensing function for each following USV; Step 2, when a friendly USV senses an enemy USV, triggering an attack USV trajectory prediction algorithm to predict the trajectory of the attack USV; Step 3, the following USV uses the predicted trajectory of the attack USV in Step 2 as a look-ahead point and intercepts in the direction of the look-ahead point; Step 4, after observing the attack USV, the following USV sequentially executes a series of interception actions: sensing, prediction, and interception. This invention addresses the escort and defense scenario of USV swarms against high-value targets by using a combination of multiple algorithms to construct a multi-layered rapid interception and defense strategy, achieving collaborative control in maritime affairs, multi-vessel coordination, and other escort and patrol scenarios.

Description

Technical Field

[0001] This invention belongs to the field of escort patrol cooperative control, specifically involving an improved navigation and following escort method based on intelligent interception and fault-tolerant reconstruction. Background Technology

[0002] The methods currently used in the field of high-target escort and defense include:

[0003] Improved Extended Kalman Filter Algorithm: This method improves upon the traditional extended Kalman filter by adopting an adaptive prediction time step strategy. Based on the dynamic monitoring of parameters such as the speed and heading rate of change of the attacking unmanned surface vessel, the prediction time step is flexibly adjusted to enhance the flexibility and accuracy of interception.

[0004] An improved line-of-sight (LOS) algorithm based on Extreme Learning Machines (ELM): By combining ELM with a LOS algorithm, the gain required by the LLM is utilized to allow the unmanned surface vessel (USV) to adjust its course more flexibly, fully leveraging the advantages of the LOS algorithm. This combination not only improves the tracking accuracy and adaptability of the LOS algorithm but also enhances the USV's ability to efficiently complete interception missions in highly dynamic environments.

[0005] Fault-tolerant reconfiguration task switching method for friendly unmanned surface vessels (USVs): When a friendly USV malfunctions and is unable to complete the integrated task of prediction-interception-return to formation, a backup plan will be adopted to conduct a suicide interception of the attacking USV. After the interception mission is completed, the information will be transmitted to other USVs. The remaining USVs will then reorganize their formation to ensure even distribution and continue to surround the high-value target to ensure effective defense. When the friendly USVs suffer losses to a certain extent, the remaining USVs will proactively approach the high-value target to strengthen the defensive formation and further improve the robustness of the defense.

[0006] The existing technology has the following problems and shortcomings:

[0007] Existing trajectory prediction algorithms typically employ methods such as Kalman filtering and particle filtering, but these methods are unsuitable for naval battlefields requiring rapid decision-making and maneuverability by unmanned surface vessels (USVs). With the development of artificial intelligence, neural networks are increasingly being applied to trajectory prediction algorithms. However, methods like Kalman filtering can only handle linear systems; while methods like particle filtering can handle nonlinear problems, their computational complexity is high, especially in real-time applications, making them unsuitable for naval battlefields requiring rapid decision-making and maneuverability by USVs; and although neural network trajectory prediction is flexible, it involves long training times and is overly dependent on data quality.

[0008] Many existing trajectory tracking algorithms typically employ methods such as proportional-integral-derivative control, sliding mode control, model predictive control, and deep reinforcement learning to guide agents in completing corresponding control tasks. However, these algorithms suffer from drawbacks such as poor adaptability, susceptibility to control chattering leading to system instability, excessive computational resources required, and high training time costs. They are insufficient to meet the requirements of strong adaptability, high accuracy, rapid response, and ease of adjustment required in maritime battlefields. Therefore, a simple, accurate algorithm that can be easily transferred to embedded control systems is needed as a trajectory tracking method for unmanned surface vessels.

[0009] Many existing trajectory tracking algorithms are inadequate. Current interception strategies only consider two scenarios: successful interception or exiting the formation after failure. They lack sufficient solutions for handling individual nodes that fail and for the entire formation to respond to faulty nodes. During operations, various forms of active or passive interference can prevent friendly vessels from effectively implementing interception strategies. Therefore, a fault-tolerant reconfiguration and task switching strategy is needed to adapt to unforeseen circumstances and ensure the safety of high-value targets. Summary of the Invention

[0010] To address the aforementioned technical issues, this invention provides an improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction. Targeting the escort and defense scenario of unmanned surface vessel swarms against high-value targets, it constructs a multi-layered rapid interception and defense strategy by combining multiple algorithms, thereby achieving collaborative control in maritime affairs, multi-vessel coordination, and other scenarios requiring escort and patrol operations.

[0011] The objective of this invention is achieved through the following technical solution: an improved navigation and escort method based on intelligent interception and fault-tolerant reconfiguration, comprising the following steps:

[0012] Step 1: Construct a control framework for the navigating unmanned surface vessel (USV) to follow the formation, and construct a sensing function for each USV following it.

[0013] Step 2: When our unmanned surface vessel (USV) senses an enemy USV, the attack USV trajectory prediction algorithm is triggered to predict the trajectory of the attack USV.

[0014] Step 3: Follow the unmanned surface vessel and use the trajectory of the attacking unmanned surface vessel predicted in Step 2 as the look-ahead point to intercept it in the direction of the look-ahead point.

[0015] Step 4: After the unmanned surface vessel observes the attacking unmanned surface vessel, it sequentially performs a series of interception actions, including sensing, prediction, and interception.

[0016] Preferably, in step 1, the control framework for the navigating unmanned surface vessel (USV) following the formation includes:

[0017] The equations of motion for the lead and follower unmanned surface vessels (USVs), the error vector of each follower USV at the desired position, the heading angle of each follower USV, and the velocity formula for the follower USVs to maintain their formation position quickly in response to disturbances.

[0018] Preferably, in step 1, the equation of motion for the pilot unmanned surface vessel is as follows:

[0019]

[0020] In the formula, For the navigator's velocity component along the x-axis, v Leader For the navigator's speed, cos(θ) Leader ) is the projection of the navigator's heading angle onto the x-axis; correspondingly, v represents the navigator's velocity component along the y-axis. Leader For the speed of the navigator, sin(θ) Leader () is the projection of the navigator's heading angle onto the y-axis;

[0021] The equations of motion for following the unmanned surface vessel are as follows:

[0022]

[0023] in To track the unmanned surface vessel's velocity component along the x-axis, v Follower To follow the speed of the unmanned surface vessel, cos(θ) Follower ) represents the projection of the heading angle of the unmanned surface vessel onto the x-axis; correspondingly, To follow the unmanned surface vessel's velocity component along the y-axis, v Follower To follow the speed of the unmanned surface vessel, sin(θ) Follower () is the projection of the heading angle of the unmanned surface vessel onto the y-axis.

[0024] Preferably, in step 1, the error vector of each following unmanned surface vessel at the desired position is represented by the following formula:

[0025]

[0026] In the formula, x Leader y Leader and x Follower y Follower Let be the positions of the navigator and the follower UAVs, respectively; d be the expected distance between the navigator and the follower UAVs; and φ be the expected deviation of the heading angle of each follower UAV relative to the navigator. For multiple follower UAVs, the distance is calculated as π / N. follower Different angles can be obtained.

[0027] Preferably, step 2 includes at least the following steps:

[0028] Step 2.1: When our unmanned surface vessel senses an enemy unmanned surface vessel, we use an improved extended Kalman filter for trajectory prediction and adopt an adaptive time step as the update frequency.

[0029] Step 2.2: After setting the time step, predict the state of the attacking unmanned surface vessel;

[0030] Step 2.3: Predict the covariance to describe the level of confidence in the predicted trajectory;

[0031] Step 2.4: Calculate the Kalman gain to obtain the Kalman gain matrix, and weigh the predicted values ​​against the actual observed values.

[0032] Step 2.5: After obtaining the Kalman gain matrix, perform state update and use the Kalman gain to form the final value after fusing the predicted value and the observed value as the starting point for the next prediction.

[0033] Step 2.6: Calculate the covariance matrix of the fused trajectory points to measure the degree of trust in the fused attack unmanned surface vessel trajectory points.

[0034] Preferably, in step 2.1, the adaptive time step is used as the update frequency formula as shown below:

[0035]

[0036] Where, Δt Attacker To determine the adaptive step size for the attacking unmanned surface vessel's trajectory, R is the set maximum sensing distance for following the unmanned surface vessel, and θ is... max ω is the preset maximum range of heading angle variation. Attacker The observed rate of change of the heading angle of the attacking unmanned surface vessel.

[0037] Preferably, in step 2.6, the covariance matrix of the fused trajectory points is as follows:

[0038]

[0039] In the formula, K t The obtained Kalman gain weight matrix, H represents the covariance matrix calculated in the previous stage. t It represents the Jacobian matrix of the observation matrix.

[0040] Preferably, step 3 includes at least the following steps:

[0041] Step 3.1: After sensing and predicting the trajectory of the attacking unmanned surface vessel, the following unmanned surface vessel uses the line-of-sight algorithm of the Extreme Learning Machine as a control method, and uses the predicted and fused trajectory points as look-ahead points to prompt the following unmanned surface vessel to intercept in the direction of the look-ahead points.

[0042] Step 3.2: By combining the hidden layer output weights obtained after training with the Extreme Learning Machine, the randomly generated input weights, and the bias, the gain required by the line-of-sight algorithm is obtained;

[0043] Step 3.3: The line-of-sight algorithm guides the unmanned surface vessel (USV) to maneuver and turn by using the angle error. Based on the predicted trajectory points obtained in Step 2, the errors involved in the line-of-sight algorithm are changed, and finally the rate of change of the heading angle of the USV is changed.

[0044] Preferably, in step 3.3, the error involved in changing the line-of-sight algorithm is calculated using the following formula:

[0045]

[0046] In the formula, and These represent the y and x coordinates of the final fused predicted trajectory point of the attack unmanned surface vessel.

[0047] Preferably, in step 4, if the unmanned surface vessel is unable to intercept and return to its original position due to external or internal interference, the embedded fault node fault-tolerant reconstruction task switching method is triggered, and the current task is switched to suicide interception.

[0048] Compared with the prior art, the present invention has the following advantages:

[0049] This invention provides an improved navigation and escort method based on intelligent interception and fault-tolerant reconfiguration. It is proposed for the escort defense scenario. With the widespread application of unmanned systems such as drones and unmanned surface vessels that are inexpensive and easy to mass-produce, the cost-effectiveness of traditional kill interception is usually not ideal. Therefore, the method of using unmanned systems to counter unmanned systems can effectively intercept enemy targets in order to protect friendly targets.

[0050] This invention uses extended Kalman filtering to predict the trajectory of attacking unmanned surface vessels (USVs), which has the advantages of low computational resource consumption and strong robustness. It is particularly suitable for nonlinear system environments such as naval battlefields. By improving the step size of the extended Kalman filter to an adaptive time step size, the prediction frequency of the attacking USV trajectory will be accelerated, which will facilitate the rapid maneuvering response of friendly USVs to the displacement of the enemy.

[0051] This invention introduces a line-of-sight algorithm based on Extreme Learning Machine (ELM). Compared to other algorithms, the line-of-sight algorithm has advantages such as real-time response, smooth control output, and suitability for distributed node control. By utilizing the dynamic output gain of the ELM, the environmental adaptability and control accuracy of the line-of-sight algorithm can be enhanced.

[0052] This invention employs a predictive and interception-integrated escort formation defense strategy. By using the predicted trajectory points of the improved extended Kalman filter as look-ahead points for the line-of-sight algorithm based on the extreme learning machine neural network, the following unmanned surface vessel adjusts its own state to move toward the predicted trajectory point of the attacking unmanned surface vessel for efficient interception.

[0053] This invention adds a fault-tolerant reconfiguration task conversion method for fault nodes. When the fault diagnosis module determines that the following unmanned surface vessel cannot perform the complete integrated interception task of prediction-follow-return formation, it will adjust the task to take a suicide interception attack on the unmanned surface vessel. At the same time, it will send the node loss information to the original formation to help the original formation dynamically adjust the formation position to ensure that the target is always in the protected formation. Attached Figure Description

[0054] Figure 1 This is a flowchart of an improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction in an embodiment of the present invention;

[0055] Figure 2 This is a schematic diagram of the Extreme Learning Machine neural network architecture in an embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram illustrating the switching of interception strategies in an embodiment of the present invention;

[0057] Figure 4 This is a module relationship diagram of the improved navigation, following, and escorting method based on intelligent interception and fault-tolerant reconstruction in an embodiment of the present invention;

[0058] Figure 5 This is a schematic diagram illustrating a scenario where the interception strategy is triggered in an embodiment of the present invention;

[0059] Figure 6 This is a schematic diagram of a fault-tolerant reconstruction scenario for a fault node in an embodiment of the present invention. Detailed Implementation

[0060] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0061] like Figure 1 As shown, the technical solution of the present invention provides an improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction, including the following steps:

[0062] Step 1: Construct a control framework for the navigating unmanned surface vessel (USV) to follow the formation, and construct a sensing function for each USV following it.

[0063] Step 2: When our unmanned surface vessel (USV) senses an enemy USV, the attack USV trajectory prediction algorithm is triggered to predict the trajectory of the attack USV.

[0064] Step 3: Follow the unmanned surface vessel and use the trajectory of the attacking unmanned surface vessel predicted in Step 2 as the look-ahead point to intercept it in the direction of the look-ahead point.

[0065] Step 4: After the unmanned surface vessel observes the attacking unmanned surface vessel, it sequentially performs a series of interception actions, including sensing, prediction, and interception.

[0066] In one embodiment of the present invention, a method for constructing a navigating unmanned surface vessel (USV) formation control framework and an USV sensing function for step 1 is provided, as follows:

[0067] The unmanned surface vessels (USVs) follow the protected target in a circular pattern, with their overall numbers evenly distributed at angles. The motion equation of the navigator can be simplified to the following equation:

[0068]

[0069] in, For the navigator's velocity component along the x-axis, v Leader For the navigator's speed, cos(θ) Leader ) is the projection of the navigator's heading angle onto the x-axis; correspondingly, v represents the navigator's velocity component along the y-axis. Leader For the speed of the navigator, sin(θ) Leader The y-axis represents the projection of the navigator's heading angle. The equation of motion for the following unmanned surface vessel can also be simplified to the following formula:

[0070]

[0071] in To track the unmanned surface vessel's velocity component along the x-axis, v Follower To follow the speed of the unmanned surface vessel, cos(θ) Follower ) represents the projection of the heading angle of the unmanned surface vessel onto the x-axis; correspondingly, To follow the unmanned surface vessel's velocity component along the y-axis, v Follower To follow the speed of the unmanned surface vessel, sin(θ) Follower () is the projection of the heading angle of the unmanned surface vessel onto the y-axis.

[0072] To ensure that the follower UAVs are evenly distributed around the navigator in a circular formation, an error vector needs to be defined. The desired position of each follower UAV is specified by ensuring that the error vector is zero. The formula for the error vector is as follows:

[0073]

[0074] In the above formula, x Leader y Leader and x Follower y Follower Here, φ represents the positions of the navigator and the follower UAVs, respectively; d is the expected distance between the navigator and the follower UAVs; and φ is the expected deviation of the heading angle of each follower UAV relative to the navigator. For multiple follower UAVs, this can be represented by π / N. follower To obtain different angles. Furthermore, to ensure that each follower UAV at a designated location can follow the leader in the same direction, the heading angle of each follower UAV needs to be calculated in advance using the following formula:

[0075]

[0076] Regarding the speed of the unmanned surface vessel (USV), the USV needs to have the ability to quickly maintain its formation position in the face of disturbances. The speed formula is as follows:

[0077]

[0078] Where k v As a speed gain parameter, adjusting this parameter can determine the speed at which the unmanned surface vessel recovers its formation position and angle.

[0079] After refining the navigator-follower formation control framework, a sensing function needs to be introduced for each following unmanned surface vessel (USV). When a USV senses an attack target entering its sensing range, it will trigger the next step of predicting the trajectory of the attacking USV and tracking and intercepting it. If no attacking USV is sensed, it will continue to follow the navigator along its original formation position. The specific logic formula is as follows:

[0080]

[0081] The specific meaning of the sensing function is as follows: when the distance between the attacking unmanned surface vessel and the following unmanned surface vessel reaches a certain level, the next operation of the following unmanned surface vessel will be triggered, which is represented as 1 in the sensing function; if no attacking unmanned surface vessel is sensed, it is represented as 0, and the following unmanned surface vessel continues to maintain its navigation attitude.

[0082] In one embodiment of the present invention, a specific implementation method for step 2 is provided as follows:

[0083] When our unmanned surface vessel (USV) senses an enemy USV, it will first trigger an attack USV trajectory prediction algorithm. This solution specifically employs an improved extended Kalman filter (EPF) for trajectory prediction. Typically, EPF uses a fixed time step Δt as the update frequency for state prediction. Here, we improve upon the fixed time step by using an adaptive time step as the update frequency, as shown in the following formula:

[0084]

[0085] In this formula, Δt Attacker It is an adaptive step size for the trajectory of the attacking unmanned surface vessel, where R is the maximum sensing distance for following the unmanned surface vessel, and θ is the adaptive step size for the trajectory of the attacking unmanned surface vessel. max ω is the preset maximum range of heading angle variation. Attacker This represents the observed rate of change in the heading angle of the attacking unmanned surface vessel (USV). If the USV's speed is high or its heading angle fluctuates drastically, the corresponding time step will also become smaller, and the predicted trajectory point will change more dramatically, facilitating rapid maneuvering and interception of the attacking USV by friendly USVs.

[0086] After setting the time step, the state of the attacking unmanned surface vessel is predicted. The state transition process is as follows:

[0087]

[0088] parameter Represents the time interval t-Δt from the previous time. Attacker The predicted state information of the attacking unmanned surface vessel at time t is given by f, which is the state transition function. The parameters in f are the position information of the attacking unmanned surface vessel at the previous time step, the control input of the attacking unmanned surface vessel at the previous time step, and the time step.

[0089] After predicting the location of the attacking unmanned surface vessel, the covariance is then predicted to describe the degree of confidence in the predicted trajectory. The specific formula is as follows:

[0090]

[0091] parameter The covariance matrix represents the degree of confidence in the current state estimation error. The smaller the covariance matrix, the more confident the predicted state is; the larger the covariance matrix, the greater the error and the less confident the predicted state is. Q is the Jacobian matrix. A filter is used to linearize the state transition equations for computation. t (Δt Attacker ) is the process noise covariance matrix, which describes the random noise or uncertainty introduced into the system during state transition.

[0092] Next, a trade-off needs to be made between the predicted and actual observed values. This is achieved by calculating the Kalman gain, as shown in the following formula:

[0093]

[0094] Where K t This is the resulting Kalman gain weight matrix, which determines the level of confidence in the observed and predicted values. The parameters are explained as follows: H represents the covariance matrix calculated in the previous stage. t The Jacobian matrix, representing the observation matrix, is used to linearize the observation equation. R t This represents the observation noise covariance matrix, which, together with the covariance matrix, affects the degree of confidence in the predicted state and the actual observed state. The final gain will comprehensively judge the covariance matrix and the observation noise covariance matrix, and the final fused state will tend to choose the state with the smaller covariance matrix.

[0095] After obtaining the Kalman gain matrix, a state update is performed. The final value obtained by fusing the predicted and observed values ​​using the Kalman gain is used as the starting point for the next prediction. The state update formula is as follows:

[0096]

[0097] here The final trajectory point of the attacking unmanned surface vessel, z, is obtained by combining predicted and observed values. t It observes the trajectory of the attacking unmanned surface vessel, and h is a nonlinear function of the observation model.

[0098] Finally, the covariance matrix of the fused trajectory points is calculated to measure the degree of trust in the fused attack unmanned surface vessel trajectory points; the larger the error, the lower the trust, and vice versa. The formula is shown below:

[0099]

[0100] In the formula, parameter P t|t This represents the covariance matrix of the fused trajectory points.

[0101] In one embodiment of the present invention, a specific implementation method for step 3 is provided as follows:

[0102] After sensing and predicting the trajectory of the attacking unmanned surface vessel (USV), the following USV uses a line-of-sight algorithm combined with an Extreme Learning Machine (ELM) as its control method. The predicted and fused trajectory points serve as look-ahead points, prompting the following USV to intercept the attacking USV in the direction of these look-ahead points. An ELM is a lightweight neural network with a single hidden layer feedforward structure. However, unlike multi-layer feedforward neural networks, it does not require backpropagation to update the parameters of the hidden layers, allowing for rapid training. The neural network structure diagram is shown below. Figure 2 As shown.

[0103] Where x is the neural network input, the blue part represents the input layer, the green part represents a single hidden layer, the red part is the output layer, and the output is represented by y. The basic principle of the Extreme Learning Machine is as follows:

[0104]

[0105] Where h(x) i Let β be the output of the neural network, g be the activation function, L be the number of nodes in the hidden layer, and β be the output weights of the hidden layer. In an extreme learning machine, only the output weights from the hidden layer to the output layer are calculated. The input weights w and bias b in the neural network are randomly generated. During training, a random matrix and bias are first generated, and then the hidden layer output matrix is ​​calculated as shown below:

[0106]

[0107] After obtaining the output matrix, the hidden layer output weights β are obtained using the least squares method with the formula Hβ = T, where T is the true value to be approximated. In this method, the gain k required by the line-of-sight algorithm can be obtained by combining the hidden layer output weights obtained after training, the randomly generated input weights, and the bias.

[0108] The line-of-sight algorithm has the advantages of fast and accurate response. By combining it with an extreme learning machine neural network, it can improve the interception performance of unmanned surface vessels. The core formula of the line-of-sight algorithm is as follows:

[0109]

[0110] The line-of-sight algorithm guides the unmanned surface vessel (USV) to maneuver and turn by using angular errors. Based on the predicted trajectory points obtained in step 2, the error formula involved in the line-of-sight algorithm is modified as follows:

[0111]

[0112] in and These represent the y and x coordinates of the final fused predicted trajectory of the attack unmanned surface vessel. The gain is then calculated as follows:

[0113]

[0114] The final rate of change in the heading angle of the following unmanned surface vessel is:

[0115]

[0116] In one embodiment of the present invention, a specific implementation method for step 4 is provided as follows:

[0117] Under normal operating conditions, the follower UAV will sequentially execute a series of interception actions—sensing, prediction, and interception—upon detecting an attacking UAV. If the follower UAV malfunctions due to external or internal interference and is unable to intercept normally and return to its original position within the formation, the embedded fault-tolerant refactoring task switching method will be triggered, switching the current task to a suicide interception. The interception effect is illustrated in the diagram below. Figure 3 As shown.

[0118] Figure 3 In the diagram, blue arrows represent the formation's navigation direction, red nodes represent follower UAVs, red dashed boxes represent the blast radius, blue nodes represent high-value protected targets, and green nodes represent attacking UAVs. When a follower UAV detects a malfunction while initiating an interception strategy, it will employ a suicide interception protection strategy to ram the attacking UAV and transmit the strategy switch information to the original formation. Upon receiving information about the losses at the malfunctioning nodes, the formation will further reconstruct its formation, recalculating the angles by calculating the remaining number of follower UAVs to ensure that the follower UAVs are evenly distributed around the high-value protected targets.

[0119] In this embodiment, the complete mission flow is as follows: After the following unmanned surface vessel (USV) senses the attacking USV through a sensing function, it predicts the trajectory of the attacking USV using an extended Kalman filter with an adaptive time step. Then, it uses a line-of-sight algorithm combining the output gain of an extreme learning machine neural network to guide the following USV to complete the final interception mission. If the following USV detects a malfunction during the interception process and cannot complete the integrated interception mission of prediction-interception-return to formation, it will flexibly switch its current mission to a suicide interception strategy. Simultaneously with the strategy switch, it will send information to the original formation, which will recalculate angles and readjust its formation to ensure optimal defense. The specific algorithm pseudocode structure is as follows:

[0120]

[0121]

[0122] The formation defense strategy provided by this invention is based on the leader-follower method, where a high-value protected target acts as the leader, and a swarm of unmanned surface vessels (USVs) are evenly distributed around the protected target to ensure the target successfully reaches its designated objective point. During the escort defense process, some attacking USVs may attack the leader. In this case, the method triggers a sensing function through distance calculation to mobilize the corresponding follower USVs to intercept the attacking USVs. During the interception process, the module first diagnoses its own faults. If its performance is good, it first predicts the trajectory points of the attacking USVs using an extended Kalman filter based on an adaptive time step. Then, the predicted look-ahead points are output as a gain for the line-of-sight algorithm via an extreme learning machine neural network and transmitted to the follower USVs to help them make timely adjustments and accurately intercept the attacking USVs. If the fault diagnosis module finds that it cannot complete the prediction-interception-return to formation operation, it flexibly switches its mission mode to a suicide interception of the attacking USVs, while simultaneously transmitting node loss information back to the original formation. The formation recalculates its angles and adjusts its position accordingly, moving closer to the protected target when necessary to ensure effective defense. The module relationship diagram is shown below. Figure 4 As shown.

[0123] The following two specific scenarios further illustrate the formation defense strategy based on the leader-follower method provided in this embodiment of the invention:

[0124] Scenario 1: The navigator is surrounded by a platoon of unmanned surface vessels (USVs). These USVs accompany the navigator, protecting it until it reaches the designated target. Upon detecting an attacking USV during the journey, they execute a strategy of predicting the trajectory, actively intercepting, and returning to the platoon. A detailed MATLAB simulation is shown below. Figure 5 As shown.

[0125] Scenario 2: If the unmanned surface vessel (USV) detects a malfunction and cannot execute the predicted trajectory-active interception-return to formation strategy, it will switch to a suicide interception mission. After completing the mission, the remaining members of the formation will adjust their positions in real time based on the information, as detailed below. Figure 6 As shown.

[0126] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An improved navigation and escort method based on intelligent interception and fault-tolerant reconfiguration, characterized in that: The method includes the following steps: Step 1: Construct a control framework for the navigating unmanned surface vessel (USV) to follow the formation, and construct a sensing function for each USV following it. Step 2: When our unmanned surface vessel (USV) senses an enemy USV, the attack USV trajectory prediction algorithm is triggered to predict the trajectory of the attack USV. Step 3: Follow the unmanned surface vessel and use the trajectory of the attacking unmanned surface vessel predicted in Step 2 as the look-ahead point to intercept it in the direction of the look-ahead point. Step 3 includes at least the following steps: Step 3.1: After sensing and predicting the trajectory of the attacking unmanned surface vessel, the following unmanned surface vessel uses the line-of-sight algorithm of the Extreme Learning Machine as a control method, and uses the predicted and fused trajectory points as look-ahead points to prompt the following unmanned surface vessel to intercept in the direction of the look-ahead points. Step 3.2: By combining the hidden layer output weights obtained after training with the Extreme Learning Machine, the randomly generated input weights, and the bias, the gain required by the line-of-sight algorithm is obtained; Step 3.3: The line-of-sight algorithm guides the unmanned surface vessel (USV) to maneuver and turn by using angular error. Based on the predicted trajectory points obtained in Step 2, the errors involved in the line-of-sight algorithm are changed, and finally the rate of change of the heading angle of the USV is changed. Step 4: After the unmanned surface vessel observes the attacking unmanned surface vessel, it sequentially performs a series of interception actions, including sensing, prediction, and interception.

2. The improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 1, characterized in that: In step 1, the control framework for the navigating unmanned surface vessel (USV) following the formation includes: The equations of motion for the lead and follower unmanned surface vessels (USVs), the error vector of each follower USV at the desired position, the heading angle of each follower USV, and the velocity formula for the follower USVs to maintain their formation position quickly in response to disturbances.

3. The improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 2, characterized in that: In step 1, the equation of motion for the pilot unmanned surface vessel is as follows: In the formula, For the navigator The velocity components of the shaft, For the speed of the leader, For the navigator's heading angle The projection of the axis; correspondingly, For the navigator The velocity components of the shaft, For the speed of the leader, For the navigator's heading angle at Projection of the axis; The equations of motion for following the unmanned surface vessel are as follows: in To follow the unmanned surface vessel The velocity components of the shaft, In order to keep up with the speed of the unmanned boat, To follow the heading angle of the unmanned surface vessel The projection of the axis; correspondingly, To follow the unmanned surface vessel The velocity components of the shaft, In order to keep up with the speed of the unmanned boat, To follow the unmanned surface vessel's heading angle Projection of the axis.

4. The improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 3, characterized in that: In step 1, the error vector of each following unmanned surface vessel at the desired position is represented by the following formula: In the formula, , and , These represent the positions of the navigator and the follower unmanned surface vessel, respectively. The expected distance between the navigator and the following unmanned surface vessel. The expected deviation of the heading angle of each following unmanned surface vessel relative to the leader, for multiple following unmanned surface vessels passing through... Different angles can be obtained.

5. An improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 4, characterized in that: Step 2 includes at least the following steps: Step 2.1: When our unmanned surface vessel senses an enemy unmanned surface vessel, we use an improved extended Kalman filter for trajectory prediction and adopt an adaptive time step as the update frequency. Step 2.2: After setting the time step, predict the state of the attacking unmanned surface vessel; Step 2.3: Predict the covariance to describe the level of confidence in the predicted trajectory; Step 2.4: Calculate the Kalman gain to obtain the Kalman gain matrix, and weigh the predicted values ​​against the actual observed values. Step 2.5: After obtaining the Kalman gain matrix, perform state update and use the Kalman gain to form the final value after fusing the predicted value and the observed value as the starting point for the next prediction. Step 2.6: Calculate the covariance matrix of the fused trajectory points to measure the degree of trust in the fused attack unmanned surface vessel trajectory points.

6. The improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 5, characterized in that: In step 2.1, the adaptive time step size is used as the update frequency formula as shown below: in, To adapt the step size to the trajectory of the attacking unmanned surface vessel, It is the set maximum sensing distance for following the unmanned surface vessel. The preset maximum heading angle variation range, The observed rate of change of the heading angle of the attacking unmanned surface vessel.

7. An improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 6, characterized in that: In step 2.6, the covariance matrix of the fused trajectory points is shown below: In the formula, where The obtained Kalman gain weight matrix, This represents the covariance matrix calculated in the previous stage. It represents the Jacobian matrix of the observation matrix.

8. An improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 7, characterized in that: In step 3.3, the error involved in changing the line-of-sight algorithm is calculated using the following formula: In the formula, and These represent the final fusion prediction of the attack unmanned surface vessel's trajectory points. and coordinate.

9. An improved navigation and escort method based on intelligent interception and fault-tolerant reconstruction as described in claim 8, characterized in that: In step 4, if the unmanned surface vessel malfunctions due to external or internal interference and is unable to intercept and return to its original position in the formation, the embedded fault node fault-tolerant reconstruction task switching method is triggered, and the current task is switched to suicide interception.

Images