Protection game method based on prospective situation awareness and region triggering adaptive defense

Abstract

The invention relates to a protection game method based on prospective situation awareness and region-triggered adaptive defense, which aims at an island protection scene, aims at intercepting a target to protect the safety of high-value fixed targets such as an island to the greatest extent as a main task, and aims at enhancing the anti-saturation strike capability as a secondary task. Firstly, a rolling time domain algorithm is optimized to improve interceptor response efficiency; on the basis, the own modeling is simplified by smoothing the enemy trajectory, and the planning and deployment time of the interception strategy is shortened; and the position of the enemy unmanned ship is pre-judged by combining a foresight situation awareness module, and a high-value important target is protected by using a region triggering type self-adaptive return interception mechanism through a tail end state. The method plays an important role in complicated and severe marine environments such as marginal-coast island reef defense and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned boat control, and in particular to a protection game method based on forward-looking situation awareness and regional triggered adaptive defense. Background Art

[0002] Regarding the application scenarios of unmanned boats on islands and reefs, the main task is to intercept targets to protect the safety of high-value fixed targets such as islands and reefs to the greatest extent, and the secondary task is to enhance the ability to resist saturation strikes.

[0003] The core of the linear quadratic rolling horizon algorithm is to dynamically calculate the time range of optimal control and implement the optimal strategy. However, only considering the attacker's current location information and ignoring the attacker's possible future action path to calculate the optimization range will lack foresight, which may cause the interceptor's interception strategy to lag and reduce the chance of effective interception.

[0004] Existing pursuit and protection game problems often ignore the situation where the attacker's trajectory fluctuates suddenly due to nonlinear or uncertain factors. When the attacker's trajectory fluctuates violently, if the interceptor is too focused on the attacker's violent state transition behavior when modeling and planning the deployment of interception strategies, it may make it very difficult for the interceptor to model and formulate strategies, resulting in the inability to make fast and effective interception actions or even the consequences of stopping. Therefore, there is an urgent need for a method that can smooth the violent changes in the attacker's trajectory to facilitate the interceptor to plan and deploy interception strategies.

[0005] In the current pursuit and protection game problems, the conventional method is to simply use the control input obtained by merging the optimal solution of the linear quadratic game to form the interception strategy of the interceptor. The game solution is more based on the current position information of the interceptor and the attacker, which is relatively passive. The interception foresight may be poor, making it difficult to formulate a more proactive interception strategy, thereby reducing the interception effect of the interceptor.

[0006] Existing protection game methods pay more attention to the attacker's position information from the interceptor's perspective, and tend to ignore the important impact of the protection target's position on the game when the attacker is at a dangerous distance from the protection target. This may expose the protection target to the attacker's threat range, making the interceptor's defense area insufficiently concentrated. Once the attacker suddenly changes his strategy at a dangerous distance, the interceptor will fail to intercept because of the lack of a multi-level defense mechanism and cannot effectively return to defense.

[0007] Among the current protection game problems, for intelligent unmanned equipment with area killing capabilities such as unmanned boats, there is a lack of an indicator that takes into account the attack range and position at the terminal defense moment to distinguish the performance of the interceptor and the safety of the protected target.

[0008] In summary, the response time of the interceptor in the current protection game model needs to be optimized. There is a problem that the optimal control time range ignores the future evolution trend of the system. It is difficult to model the sudden maneuvers of the attacker. The interceptor cannot use the attacker's future predicted state information for interception. It is also impossible to make adaptive defensive actions when the attacker is at a dangerous distance from the protected target. Summary of the invention

[0009] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a protection game method based on forward-looking situational awareness and regional triggered adaptive defense, so as to solve the problem of the current lack of game methods for pursuit and protection.

[0010] The purpose of the present invention can be achieved by the following technical solutions:

[0011] The present invention provides a protection game method based on forward-looking situational awareness and regional triggered adaptive defense, comprising the following steps:

[0012] Step S1, for the protection game scenario, construct a dynamic equation including the interceptor and the attacker;

[0013] Step S2, setting the attacker's goal to avoid the interceptor and occupy the protected target, setting the interceptor's goal to intercept the attacker before the attacker reaches the protected target, and constructing a protection game objective function based on soft constraints;

[0014] Step S3, based on the predicted position of the attacker and the position of the protected target, dynamically calculating the optimal control time range using a linear quadratic rolling horizon algorithm;

[0015] Step S4, calculating the control inputs of the interceptor and the attacker based on the dynamic equation, the protection game objective function and the optimal control time range;

[0016] Step S5, performing smoothing processing on the attacker's current sudden behavior and updating the attacker's control input;

[0017] Step S6, based on the distance between the interceptor and the attacker, adaptively adding a control input for the predicted position of the attacker to the control input of the interceptor to achieve forward-looking situational awareness;

[0018] Step S7, if the distance between the attacker and the protected target is less than the preset danger distance, a control input to move closer to the protected target is added to the control input of the interceptor to achieve regional triggered adaptive defense.

[0019] As a preferred technical solution, in step S1, the dynamic equation is:

[0020]

[0021] in, and are the speed changes of the attacker on the x-axis and y-axis, v a is the attacker's speed, and u ay are the attacker’s control inputs on the x-axis and y-axis respectively, and are the speed changes of the interceptor on the x-axis and y-axis, v i is the speed of the interceptor, and u iy are the control inputs of the interceptor on the x-axis and y-axis, respectively, x 、u y are the control inputs of the attacker or interceptor on the x-axis and y-axis respectively, and u is u x and u y The combined control input, Use a or i instead to indicate different roles.

[0022] As a preferred technical solution, in step S2, the protection game objective function is:

[0023]

[0024] Among them, γ a ,γ i Represent the feedback strategies of the attacker and the interceptor respectively, γ a (x,t)∈u a , γ i (x,t)∈u i ,u a and u i are the control inputs of the attacker and the interceptor respectively, T is the terminal state, P(x a )、P(x i )、P(x t ) are the location information of the attacker, interceptor and protected target respectively, is the distance penalty between the attacker and the protected target during the game, is the distance penalty between the interceptor and the attacker during the game, w a is the distance penalty between the terminal state attacker and the protected target, w i The distance penalty between the terminal state interceptor and the attacker.

[0025] As a preferred technical solution, in step S3, the process of calculating the optimal control time range includes the following steps:

[0026] Step S301, based on the predicted position of the attacker and the position of the protected target, an optimal control time is obtained using an optimal control time calculation formula;

[0027] Step S302, performing optimal control in a loop, during which, when the time exceeds the optimal control time, the optimal control time is updated.

[0028] As a preferred technical solution, the optimal control time calculation formula is:

[0029]

[0030] Among them, norm is the Euclidean distance, P(x t ) represent the predicted attacker's location information and the protected target's location information, respectively, a Indicates the speed of the attacker.

[0031] As a preferred technical solution, in step S4, the control inputs of the interceptor and the attacker are obtained by dynamically solving the state feedback matrix using the Riccati equation within the optimal control time range.

[0032] As a preferred technical solution, in step S5, the attacker's additional control input toward the protected target is added to the attacker's control input, thereby achieving the updating of the attacker's control input.

[0033] As a preferred technical solution, in step S6, when the distance between the interceptor and the attacker is less than a preset value, the control input for the predicted position of the attacker and the control input for the current position of the attacker are weighted based on learnable adaptive weights to obtain an updated control input of the interceptor.

[0034] As a preferred technical solution, the attacker and interceptor are different unmanned boats.

[0035] As a preferred technical solution, it also includes:

[0036] Step S8, by calculating the intersection area of ​​a circle formed by a circle with one times the attack range and the interceptor's attack range with the protected target as the center, the security of the protected target is evaluated.

[0037] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0038] (1) Improving the response time of the interceptor: Based on the predicted position of the attacker and the position of the protected target, the present invention uses a linear quadratic rolling time domain algorithm to dynamically calculate the optimal control time range, thereby improving the response time of the interceptor.

[0039] (2) Avoiding interception failures caused by paying too much attention to the details of the attacker's maneuvers: The present invention smoothes the current sudden behavior in the attacker's control input, allowing the attacker to have a stronger sense of purpose. On the basis of the interceptor's attack range, the interceptor can quickly build models and quickly plan and deploy interception strategies, avoiding interception failures caused by paying too much attention to the details of the attacker's maneuvers.

[0040] (3) Forward-looking situational awareness: Based on the distance between the interceptor and the attacker, the present invention adaptively adds control input for the attacker's predicted position to the interceptor's control input, comprehensively considers the attacker's current state and future state to help the interceptor make a better interception strategy, and introduces adaptive weights and learning rates. By judging the distance between the interceptor and the attacker, it is determined whether to move toward the predicted position for interception, thereby achieving forward-looking situational awareness.

[0041] (4) Regional triggered adaptive defense: When the Euclidean distance between the interceptor and the attacker satisfies both situational awareness and adaptive defense mechanisms, the present invention adds a control input to the interceptor's control input to move closer to the protected target. By comprehensively considering the positions of the three parties, the interceptor's next moving direction and distance are determined to ensure that the interceptor can be effectively returned to defense and blocked in an emergency, thereby improving the defect of insufficient attention to the protected target position information in the terminal defense state.

[0042] (5) Realizing the evaluation of the security of the protected target: The present invention realizes the evaluation of the security of the protected target by calculating the intersection area of ​​a circle formed by one times the attack range and the attack range of the interceptor with the protected target as the center, and combines the indicators of attack range and position information to measure the security of the protected target and its ability to resist saturation attacks. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a protection game method based on forward-looking situational awareness and regional triggered adaptive defense in an embodiment;

[0044] Figure 2 It is a module relationship diagram in the embodiment;

[0045] Figure 3 is T before improvement in the embodiment o Schematic diagram of the calculation process;

[0046] Figure 4 is the improved T in the embodiment o Schematic diagram of the calculation process;

[0047] Figure 5 This is a schematic diagram of the interception strategy before optimization in the embodiment;

[0048] Figure 6This is a schematic diagram of the optimized interception strategy in the embodiment;

[0049] Figure 7 A schematic diagram of an interception strategy that adds forward-looking situational awareness in an embodiment;

[0050] Figure 8 This is a schematic diagram of an ideal protection state in the embodiment;

[0051] Fig. 9 This is a schematic diagram of the comparison of the anti-saturation strike capability in the embodiment;

[0052] Fig.10 A schematic diagram of an interception strategy for adding trend-triggered adaptive defense in the embodiment;

[0053] Fig.11 This is a MATLAB simulation diagram. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0055] Example 1

[0056] This embodiment is aimed at the game scenario of chasing and protecting unmanned boats on islands and reefs. It takes intercepting targets to protect the safety of high-value fixed targets such as islands and reefs to the greatest extent as the main task, and enhancing the ability to resist saturation strikes as a secondary task. It provides a game method for chasing and protecting based on forward-looking situational awareness and regional triggered adaptive defense mechanism. First, the rolling time domain algorithm is optimized to improve the response efficiency of the interceptor; on this basis, the enemy trajectory is smoothed to simplify the modeling of the own side and shorten the planning and deployment time of the interception strategy; the forward-looking situational awareness module is combined to predict the position of the enemy unmanned boat and the terminal state uses the regional triggered adaptive defense interception mechanism to protect high-value important targets. This method plays an important role in complex and harsh marine environments such as border and coastal defense and island and reef defense.

[0057] The protection game based on the island and reef unmanned boat defense scenario is a special form of the pursuit game. It is modeled as a linear quadratic game problem. In order to optimize the interception performance of the interceptor in the game scenario, the relevant algorithms are optimized and new functional modules are added to improve the interception capability. For details, see Figure 1 and Figure 2 , the method comprises the following steps:

[0058] Step S1: Establish the dynamic equation in the protection game problem.

[0059] use represents the attacker's state change rate at a certain moment, x a (t) represents the state variable of the attacker at a certain moment, A a represents the attacker's state transition matrix, which represents the attacker's inherent evolution mode when there is no control input, B a represents the attacker's control input matrix, which determines how the control input acts on the attacker to change the attacker's state, u a represents the attacker's control input; the corresponding represents the rate of change of the interceptor's state at a certain moment, x i (t) represents the state variable of the interceptor at a certain moment, A i represents the state transition matrix of the interceptor, B i represents the interceptor's control input matrix, u i represents the control input of the interceptor. Assuming that the protected target is a high-value fixed target, the dynamic equation is as follows:

[0060]

[0061] Combining the dynamic equations of the attacker and the interceptor, the joint dynamic equation is as follows:

[0062]

[0063] Here x(t) is the attacker x a (t) and interceptor x i (t), the A matrix is ​​the joint state transfer matrix, and the other parameters are introduced as above. For the joint dynamic equation, x(t), A, B a , B i The matrix descriptions are shown below.

[0064]

[0065] In order to further simplify the dynamic equation, only the impact of control input on the attacker and interceptor is considered. In addition, since the game participants in the scenario of unmanned islands and reefs are assumed to be unmanned boats, the dynamics are only reflected in the x-axis and y-axis. The dynamic equation can be further refined as follows:

[0066]

[0067] in and Represents the attacker's speed change on the x-axis and y-axis, v a represents the speed of the attacker, and u ayRepresent the attacker’s control input on the x-axis and y-axis respectively; and Represents the speed change of the interceptor on the x-axis and y-axis, v i Represents the speed of the interceptor. and u iy Represents the control input of the interceptor on the x-axis and y-axis respectively.

[0068] To ensure constraints on the control input, we use To constrain the angle, in order to further ensure boundedness, the following nonlinear function is used to further constrain the control input:

[0069]

[0070] It can be seen that the control input can be better constrained by introducing the normalization restriction operation, where u x 、u y are the control inputs of the attacker or interceptor on the x-axis and y-axis respectively, and u is u x and u y The combined control input, It can be replaced by a or i to indicate different roles.

[0071] Step S2: Construct the objective function of the protection game.

[0072] Considering that hard constraints require strict satisfaction of constraints, which may lead to an unsolvable optimization problem, soft constraints are introduced in the objective function as penalty items. Soft constraints allow the system to violate constraints to a certain extent. This flexibility makes it very likely that the system will find a feasible solution when faced with difficult-to-satisfy conditions. The constructed objective function is as follows:

[0073]

[0074] The main purpose of the objective function is to balance the control input and the game objective based on position deviation to find the optimal solution. In the above formula, γ a ,γ i Represent the feedback strategies of the attacker and the interceptor respectively, specifically γ a (x,t)∈u a and γ i (x,t)∈u i ,u a and u i are the control inputs of the attacker and the interceptor respectively. In the above formula, T represents the terminal state, P(x a )、P(x i )、P(x t ) represent the location information of the attacker, interceptor and protected target respectively. It reflects the distance penalty between the attacker and the protected target during the game; represents the distance penalty between the interceptor and the attacker during the game; w a represents the distance penalty between the terminal state attacker and the protected target; w i Represents the distance penalty between the terminal state interceptor and the attacker. The objective function can be used to obtain the goals of the attacker's unmanned boat and the interceptor's unmanned boat in the game process: the enemy's unmanned boat hopes to avoid attacking the islands and reefs by its own unmanned boats, and its own unmanned boat hopes to effectively intercept the enemy's unmanned boat before it reaches the islands and reefs.

[0075] By combining the penalty weights into the process penalty matrix Q and the terminal penalty matrix Q f , in order to simplify the objective function, the simplified objective function is as follows:

[0076]

[0077] Where Q represents the penalty matrix in the game process, Q f Represents the penalty for the terminal state.

[0078] Step S3: Use the improved linear quadratic rolling horizon algorithm to dynamically solve the optimal control time range T o .

[0079] Considering that the soft constraints provide penalty terms for the optimization control process within a certain time range, the improved linear quadratic rolling horizon algorithm is used to dynamically solve the optimal control time range T o The core of the linear quadratic rolling horizon algorithm is to derive the most appropriate T o And implement the optimal control strategy. If the traditional linear quadratic rolling horizon algorithm is used in the context of protection game, it needs to use the location information of the attacker and the location information of the protected target. Although this can speed up the response time of the interceptor, in the ever-changing ocean, the defender needs to allow the defender's unmanned boat to have a faster response time.

[0080] This embodiment is based on the idea of ​​increasing the interception performance through prediction, and dynamically calculates T using the predicted attacker's position and the protected target's position. o , which can effectively shorten the interceptor's response time. The specific formula is as follows.

[0081]

[0082] The formula mentioned is the state variable of the attacker after prediction, in order to better show T o Calculation of T before and after improvemento The calculation process is visualized as follows Figure 3 and Figure 4 shown.

[0083] from Figure 3 It can be seen that the algorithm before improvement has an effect on T o The calculation process, where the red one is the attacker, the circle is the attack range, T1 is the shortest time for the attacker to attack the protected target in the initial state, and T1 is recorded as the initial T o After T1, the attacker reaches the second position. Then the shortest time for the attacker to attack the protected target is calculated again, T2, which is recorded as the new T o By continuously adjusting T o To ensure the normal introduction of soft constraints.

[0084] Figure 4 It is the improved T o Calculation process, compared with the previous time T1, the predicted attacker's location information is used to calculate the shortest time T to the protected target 1' As a new T o , which can shorten the interceptor's action reaction time and encourage the interceptor to improve his performance.

[0085] Specifically, the linear quadratic rolling horizon algorithm process of this step is shown in Table 1.

[0086] Table 1 Improved linear quadratic rolling time domain algorithm

[0087]

[0088] The improved linear quadratic rolling time domain algorithm provided in this step calculates the optimal control range by jointly predicting the attacker's position information and the protected target's position information, which enables the interceptor to have better proactive judgment to improve the accuracy of interception. Compared with the algorithm before the improvement, it can help the interceptor respond more promptly and facilitate the interceptor to find a suitable optimization interval in a dynamic environment, avoiding the disadvantages of over-planning caused by too large an optimal control range and the disadvantages of relatively short-sightedness caused by too small an optimal control range.

[0089] Step S4: Based on the dynamic equations and objective functions in steps S1-S2, the control range T is dynamically calculated using the improved linear quadratic rolling horizon algorithm. o Under the premise of dynamic in different T o The Riccati equation is used to preliminarily solve the state feedback matrix, and the formula is as follows.

[0090]

[0091] Among them, A and B a , Bi The matrix introduction is shown in step S1. The Z matrix reflects the feedback gain, Q f The present invention uses the reverse integration method to solve the optimal control input within the range, and the attacker and interceptor control inputs are finally obtained as follows:

[0092]

[0093] Step S5: The attacker's targeting is enhanced by strengthening the influence of the target location information on the attacker's control input, making it easier for the interceptor to model and plan strategies.

[0094] In the past, the control input was usually obtained by calculating the Riccati equation to obtain the solution matrix and then obtaining the optimal control input u through state feedback. a and u i However, in some scenarios, the attacker may experience some sudden fluctuations or deviations in his own trajectory due to some uncontrollable external factors such as environmental disturbances and sensor noise. At this time, if the interceptor pursues refinement too much, it may cause the interceptor to over-respond to these uncertain trajectories and cause interception failure. To address this problem, this embodiment adopts a method to enhance the attacker's control input of the protected target location information to simplify the attacker's trajectory to form an effective fuzzy prediction trajectory. This trajectory is not a prediction of the future but an effective smoothing of the current attacker's sudden behavior, so as to facilitate the interceptor to formulate interception strategies more quickly. In addition, on the basis of the interceptor's attack range, the interceptor can improve its active strike capability against the attacker. The modified attacker control input is: in Refers to additional control input from the attacker towards the protected target. Figure 5 , Figure 6 By increasing the attacker's control input on the protected target location information to simplify the interceptor modeling and strategy planning deployment, the before-and-after effects of the interceptor's implementation of the interception strategy are demonstrated.

[0095] Figure 5 This is the interception strategy of the interceptor before optimization without adding control input to the attacker. The red dots represent the trajectory of the attacker, the blue dots represent the trajectory of the interceptor, the red and blue circles represent the attack range, the blue cross represents the interception anomaly leading to mission failure, and the green represents the fixed protection target. When the attacker makes a sudden turn maneuver due to uncontrollable factors, if the modeling is too refined, it may lead to the inability to implement the interception strategy normally or even cause abnormal behavior and no maneuvering action.

[0096] Figure 6is the optimized interception strategy, where the yellow dot is the attacker's trajectory after smoothing, the yellow arrow is the initial smooth prediction stage, and the yellow circle is the attack range. Considering that the attacker may fail to intercept due to abnormal maneuvers that cause complex trajectories, this method provides a method to strengthen the control input of the target position information to obtain a smoothed attacker behavior trajectory to prevent the interceptor from falling into a local optimum and being unable to intercept normally. After obtaining the improved trajectory, the interceptor can respond quickly and plan strategies. There will be certain controllable errors in this process, but the existence of such errors does not affect the interceptor's effective interception of enemy targets.

[0097] In summary, the method of smoothing the attacker's trajectory provided in this step aims to solve the problem in the past protection game problem that the attacker may undergo sudden state transitions due to some external factors, which makes it difficult for the interceptor to model the attacker. By introducing more information about the protection target into the attacker's attack strategy, the attacker's targeting is enhanced, and a smooth fuzzy prediction trajectory is generated. This trajectory can simplify the interceptor's modeling process and facilitate the interceptor to activate the interception mechanism. Even if there are slight errors in the attacker's interception strategy, the enemy's unmanned boat can be well intercepted on the basis of having an attack range.

[0098] Step S6: Based on steps S3-S5, in order to avoid the drawback of only considering the attacker's current state information and lack of foresight, this method adds attention to the attacker's future state information on the basis of the interceptor's control input to avoid local optimality. When the interceptor reaches a certain distance from the attacker, the interceptor will activate the forward-looking situation awareness module, and the control input becomes: in and They are the interceptor's control input to the attacker's current position and the attacker's predicted position, respectively. k1 and k2 are learnable weights. The learning goal of the weights is that the greater the difference between the attacker's current state and the predicted state, the greater the weight of the interceptor's control input to the attacker's predicted position. This method optimizes decision-making planning based on the attacker's current and future state information. In addition, adaptive weights and learning rates are introduced in this module. By calculating the Euclidean distance between the attacker's current position information and the predicted position information, the interceptor can optimize the decision-making planning based on the attacker's current and future state information. When the distance is too large, the interceptor will be more inclined to move closer to the attacker's predicted position to ensure successful interception. The specific interception diagram is as follows Figure 7 shown.

[0099] from Figure 7It can be seen that after adding the situational awareness module, the interceptor will respond sensitively to the attacker by comprehensively judging the attacker's current position and future predicted position. This method enables the interceptor to better adapt to environmental changes and uncertainties and avoid being misled by the attacker's short-term sudden behavior. At the same time, the attacker can use his own attack range combined with situational awareness to better destroy the enemy target.

[0100] This step provides an interception strategy with a forward-looking situational awareness module. It relies on adaptive weights to balance the interceptor's dependence on the attacker's current state information and the attacker's predicted state information. The learning rate is introduced to dynamically adjust the adaptive weights. The Euclidean distance between the attacker's current position and the predicted position is judged to determine whether the interceptor is more concerned about the predicted attacker's position. This module will balance the interceptor's attention to the attacker's current position information and the attacker's predicted position information when the interceptor reaches a certain distance from the attacker. Compared with the interception strategy without the introduction of the prediction module, it significantly increases the dynamic adjustment and interception capabilities.

[0101] Step S7: In the unmanned defense scenario of islands and reefs based on the protection game, the core task is to ensure that high-value fixed targets such as islands and reefs are not attacked and occupied by enemy unmanned boats. Considering that the closer the enemy unmanned boat is to its own protection target, its own unmanned boat should pay more attention to the enemy unmanned boat and have a stronger confrontation with the enemy unmanned boat. However, at the end of the confrontation game, the position information of the protection target is usually not taken seriously. In some cases, the position of the interceptor cannot successfully intercept the attacker. At this time, if the attacker is continuously tracked, it will deviate from the key position of defense, resulting in defense failure. To address this problem, this embodiment allows the interceptor to move closer to the protection target based on the current and predicted position information of the attacker, and intercepts the attacker in a "shortcut" way by means of returning to defense.

[0102] For the protected target, when the interceptor is at the center of the protected target, the protected target is the safest and has the strongest ability to resist saturation attacks. See the schematic diagram. Figure 8 shown.

[0103] See also Figure 8 For an extremely ideal protection state, unmanned boats usually need to perform dynamic patrol tasks and are not suitable for long-term garrison at the protected target, which would waste military resources. In some scenarios, being close to the protected target will also enhance the target's ability to resist saturation attacks, such as Fig. 9 shown.

[0104] according to Fig. 9 ,It can be seen that after the interceptor is appropriately close to the protected target, it can better protect the high-value fixed target when facing multiple attackers.,This shows that returning to defense close to the protected target is an effective defense mechanism.

[0105] This step introduces a regional triggered adaptive defense mechanism. The idea of ​​this mechanism is as follows: once the distance between the attacker and the protected target is norm(P(x a )-P(x t )) is less than or equal to the danger distance, the adaptive defense mechanism will be triggered. On the basis of the control input optimized in step S6, some control inputs are added to guide the interceptor to move closer to the protected target to produce the effect of returning to defense. The control input after optimization is: This improvement can protect the safety of the target to the greatest extent, make the interception strategy more flexible, and avoid the target being destroyed by the enemy due to continuous over-tracking. It not only meets the requirements of the forward-looking situational awareness module, but also meets the requirements of the adaptive defense mechanism. Fig.10 shown.

[0106] Fig.10 In the figure, the red dots represent the attacker, the purple dots are the predicted trajectory of the attacker, and the blue dots represent the interceptor. Under the premise of meeting situational awareness, when the attacker is at a dangerous distance from the protected target, the interceptor will pay more attention to the protected target, and the position of the protected target will also become part of the guide for the interceptor to change its position, thereby increasing the interception probability of the interceptor.

[0107] Step S8: After the game process is completed, an indicator is used to judge the safety of the protected target after the interception. In the past, the indicators only judged the quality of the interception strategy by distance, which is not suitable for the unmanned defense scenario of the islands and reefs in the protection game. This method provides an indicator that combines distance and attack range: the protected domain is used as the evaluation standard. This indicator is the intersection area of ​​the circle formed by taking the protected target as the center and taking a circle with one times the attack range and the interceptor's attack range. The larger the intersection area, the safer the protected target is, and it also means that it is within the controllable range of the interceptor's protection.

[0108] This step introduces an indicator that can measure the security of the protected target. It comprehensively considers the location information and the attack range of both the attack and interception parties, and takes the location of the protected target as the basis for comprehensive consideration. This indicator can be used to measure whether the defense domain of the target is fully covered, and can also be used to roughly judge the ability of the protected target to resist saturation attacks.

[0109] To verify the effectiveness of this method, MATLAB software was used to simulate this method, and the application scenario was set to the protection and interception of high-value fixed targets such as islands and reefs in the border waters. It was assumed that both sides were unmanned boats, and both sides had attack ranges. The goal of the enemy unmanned boat was to quickly destroy the protected target and form a safe distance with the unmanned boat. The goal of the unmanned boat was to quickly intercept the enemy unmanned boat and try to keep the distance between the enemy unmanned boat and the protected target. During the initial interception process, the improved algorithm was used to improve the response speed of the unmanned boat. At the same time, the trajectory of the attacking unmanned boat was smoothed by more control inputs about the protected target to facilitate the unmanned boat to quickly model and plan the interception strategy. Subsequently, the unmanned boat intercepted normally, and after reaching a certain distance from the enemy unit, the situational awareness module was started to predict the enemy's position. At the same time, according to the Euclidean distance between the predicted position and the current position, it was decided whether to pay more attention to the predicted trajectory. If the enemy unit is in a dangerous distance from the protected target, the unmanned boat will start the regional triggered adaptive defense mechanism to effectively intercept the attacker before the attacker reaches the protected target by returning to defense. MATLAB simulation diagram Fig.11 As shown, green represents the protected target, red is the attacking unmanned boat, and blue is the defending unmanned boat. It can be seen that the scheme successfully intercepts the enemy unmanned boat at the end.

[0110] In summary, this embodiment addresses the problems that the interceptor response time in the current protection game model needs to be optimized, the optimal control time range ignores the future evolution trend of the system, the difficulty in modeling the attacker's sudden maneuvers, the interceptor cannot use the attacker's future predicted state information for interception, and cannot make adaptive defense actions when the attacker is at a dangerous distance from the protected target. Based on an improved linear quadratic rolling time domain algorithm framework, an interception strategy of a forward-looking situational awareness module and a regional triggered adaptive defense module is introduced, which improves the interception efficiency of enemy unmanned boats in the unmanned defense scenario of islands and reefs, and at the same time proposes an indicator to measure the interception efficiency and the safety of the protected target based on the terminal defense state.

[0111] This method has the following beneficial effects:

[0112] (1) By calculation To improve the optimal control time range T in the linear quadratic rolling horizon algorithm o Calculation to improve the response time of the interceptor

[0113] (2) Set the attacker's control input to To make the attacker more targeted, smooth the attacker's trajectory, and enable the interceptor to quickly build models and plan and deploy interception strategies based on the interceptor's attack range, avoiding interception failures caused by paying too much attention to the attacker's maneuvering details.

[0114] (3) By introducing a forward-looking situation awareness module, the optimized control input is Comprehensively consider the attacker's current state and future state to help the interceptor make a better interception strategy, and introduce adaptive weights and learning rates. The size judgment determines whether to prefer moving toward the predicted position for interception.

[0115] (4) By introducing a regionally triggered adaptive defense mechanism, the defect of insufficient attention to the location information of the protected target in the terminal defense state is improved. If the Euclidean distance between the interceptor and the attacker satisfies both situational awareness and adaptive defense mechanism, the control input is optimized as By comprehensively considering the positions of the three parties, we can determine the interceptor's next movement direction and distance, ensuring that we can effectively return to defense and block the interceptor in an emergency.

[0116] (5) By introducing an indicator that combines the attack range and location information, the security of the protected target and its ability to resist saturation attacks are measured.

[0117] Example 2

[0118] This embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the pursuit and protection game method based on forward-looking situational awareness and area-triggered adaptive defense mechanism as described in Example 1.

[0119] Example 3

[0120] This embodiment provides a computer-readable storage medium, including one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the pursuit and protection game method based on forward-looking situational awareness and area-triggered adaptive defense mechanism as described in Example 1.

[0121] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0123] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0124] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0125] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0126] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A protection game method based on forward-looking situational awareness and regional triggered adaptive defense, characterized in that: The steps include: Step S1, for the protection game scenario, construct a dynamic equation including the interceptor and the attacker; Step S2, setting the attacker's goal to avoid the interceptor and occupy the protected target, setting the interceptor's goal to intercept the attacker before the attacker reaches the protected target, and constructing a protection game objective function based on soft constraints; Step S3, based on the predicted position of the attacker and the position of the protected target, dynamically calculating the optimal control time range using a linear quadratic rolling horizon algorithm; Step S4, calculating the control inputs of the interceptor and the attacker based on the dynamic equation, the protection game objective function and the optimal control time range; Step S5, performing smoothing processing on the attacker's current sudden behavior and updating the attacker's control input; Step S6, based on the distance between the interceptor and the attacker, adaptively adding a control input for the predicted position of the attacker to the control input of the interceptor to achieve forward-looking situational awareness; Step S7, if the distance between the attacker and the protected target is less than the preset danger distance, a control input to move closer to the protected target is added to the control input of the interceptor to achieve regional triggered adaptive defense.

2. A protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1, characterized in that: In the step S1, the dynamic equation is: in, and are the speed changes of the attacker on the x-axis and y-axis, v a is the attacker's speed, and u ay are the attacker’s control inputs on the x-axis and y-axis respectively, and are the speed changes of the interceptor on the x-axis and y-axis, v i is the speed of the interceptor, and u iy are the control inputs of the interceptor on the x-axis and y-axis, respectively, x 、u y are the control inputs of the attacker or interceptor on the x-axis and y-axis respectively, and u is u x and u y The combined control input, Use a or i instead to indicate different roles.

3. A protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 2, characterized in that: In step S2, the protection game objective function is: Among them, γ a ,γ i Represent the feedback strategies of the attacker and the interceptor respectively, γ a (x,t)∈u a , γ i (x,t)∈u i ,u a and u i are the control inputs of the attacker and the interceptor respectively, T is the terminal state, P(x a )、P(x i )、P(x t ) are the location information of the attacker, interceptor and protected target respectively, is the distance penalty between the attacker and the protected target during the game, is the distance penalty between the interceptor and the attacker during the game, w a is the distance penalty between the terminal state attacker and the protected target, w i The distance penalty between the terminal state interceptor and the attacker.

4. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1 is characterized in that: In step S3, the process of calculating the optimal control time range includes the following steps: Step S301, based on the predicted position of the attacker and the position of the protected target, an optimal control time is obtained using an optimal control time calculation formula; Step S302, performing optimal control in a loop, during which, when the time exceeds the optimal control time, the optimal control time is updated.

5. A protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 4, characterized in that: The optimal control time calculation formula is: Among them, norm is the Euclidean distance, P(x t ) represent the predicted attacker's location information and the protected target's location information, respectively, a Indicates the speed of the attacker.

6. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1 is characterized in that: In the step S4, the control inputs of the interceptor and the attacker are obtained by dynamically solving the state feedback matrix using the Riccati equation within the optimal control time range.

7. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1 is characterized in that: In the step S5, the attacker's additional control input toward the protected target is added to the attacker's control input, thereby updating the attacker's control input.

8. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1 is characterized in that: In step S6, when the distance between the interceptor and the attacker is less than a preset value, the control input for the predicted position of the attacker and the control input for the current position of the attacker are weighted based on learnable adaptive weights to obtain an updated control input for the interceptor.

9. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1, characterized in that: The attacker and interceptor are different unmanned boats.

10. The protection game method based on forward-looking situational awareness and regional triggered adaptive defense according to claim 1, characterized in that: Also includes: Step S8, by calculating the intersection area of ​​a circle formed by a circle with one times the attack range and the interceptor's attack range with the protected target as the center, the security of the protected target is evaluated.

Images