Intelligent Smoke Covering Method and Device Based on Multi-UAV Collaboration

Abstract

This invention belongs to the field of UAV control and optimization technology. Addressing the problems of short single-blanking time, low efficiency, lack of multi-unit collaborative optimization, and poor stability and adaptability in existing UAV smoke screen masking methods, this invention proposes an intelligent smoke screen masking method and device based on multi-UAV collaboration. The method includes: establishing a basic masking model for a single UAV using a single smoke grenade against a single target using a hierarchical and progressive collaborative optimization strategy; extending the basic masking model to the sequential deployment optimization of multiple smoke grenades from a single UAV; further developing it to relay masking of a single target by multiple UAVs; and achieving integrated optimization of task allocation, trajectory planning, and deployment timing for complex scenarios involving multiple UAVs and multiple smoke grenades against multiple targets. By constructing a mixed-integer programming model or combining it with intelligent optimization algorithms, the overall effective masking time can be maximized while satisfying dynamics and collision avoidance constraints, significantly improving the persistence, coverage, and adaptability of smoke screen masking.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control and optimization technology, specifically relating to an intelligent smoke screen concealment method and device based on multi-UAV collaboration. Background Technology

[0002] In regional security protection, critical facility security, or specific experimental scenarios, there is a need for continuous and effective optical and infrared shielding of high-speed moving targets or key areas. Smoke grenades, as a cost-effective optoelectronic countermeasure, can effectively attenuate or block optical and infrared signals through the smoke cloud they form, thereby disrupting observation, identification, or tracking systems that rely on such signals. Modern unmanned aerial vehicle (UAV) platforms, with their flexible maneuverability and precise delivery control capabilities, provide an ideal platform for achieving remote, planned, and multi-point smoke grenade deployment. By utilizing multiple UAVs working collaboratively and optimizing the swarm's flight paths, smoke grenade deployment locations, and activation timing (the time sequence of smoke cloud generation), it is theoretically possible to construct a collaborative, continuous, and efficient regional shielding network in dynamic and complex multi-threat scenarios, thereby significantly improving overall protection effectiveness and resource utilization.

[0003] However, existing smoke masking methods based on unmanned platforms are typically designed and optimized independently for simple scenarios involving a single platform and a single target. They lack a general collaborative decision-making and optimization framework capable of systematically handling task relationships of varying scales (e.g., one-to-one, many-to-one, many-to-many). When faced with complex task scenarios where multiple mobile platforms need to coordinate to address multiple dynamic targets, existing technologies exhibit significant bottlenecks in global resource scheduling, spatiotemporal task coordination, and overall performance optimization. Specifically: The limited number of masking units carried by a single platform and the limited effective duration of the masking area formed in a single instance make it difficult to achieve long-term, continuous coverage of dynamic targets. In multi-target scenarios, resource allocation often lacks systematic optimization, which can easily lead to wasted masking resources (such as excessive overlap of areas) or insufficient coverage of some targets, resulting in low overall resource utilization efficiency.

[0004] Most existing methods construct isolated mathematical models for specific scenarios (such as specific integer programming models), making it difficult to naturally and uniformly extend from basic single-entity decision-making models to complex group collaborative decision-making. When the task scale, number of platforms, or objectives change dynamically, the model needs to be significantly adjusted or even reconstructed, resulting in low reusability of the algorithm framework and difficulty in adapting to dynamically changing task requirements.

[0005] In collaborative scenarios involving multiple platforms, units, and objectives, the joint optimization problem of task allocation, resource scheduling (such as platform paths and unit deployment timing), and spatiotemporal coordination is highly complex. Existing optimization methods often have high computational complexity, making it difficult to meet the near real-time decision-making requirements in dynamic environments. Furthermore, these methods do not adequately consider uncertainties such as platform state deviations and environmental parameter disturbances, leading to the theoretically optimal strategy being susceptible to performance interference and exhibiting weak robustness in actual execution.

[0006] Therefore, current technologies face a core challenge in complex collaborative task scenarios characterized by multiple threats, high dynamism, and limited resources: how to design an intelligent collaborative decision-making and optimization method that is highly scalable, adaptable to different task scales, maximizes global occlusion effectiveness, and possesses strong robustness has become the key to promoting the development of this technology towards intelligence and clustering. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention proposes an intelligent smokescreen concealment method and device based on multi-UAV collaboration. The method optimizes the flight strategy of the UAV swarm, the timing of smoke grenade deployment and activation, and forms a collaborative, continuous, and efficient optical concealment zone in dynamic and complex multi-threat scenarios, thereby significantly improving the overall effectiveness and resource utilization of the protection system.

[0008] This invention provides an intelligent smoke masking method based on multi-UAV collaboration, comprising: Step 110: Obtain the motion status information of the threat source, the real-time status information of your own UAV, and environmental constraint information; Step 120: Establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking; Step 130: Construct a capability affinity matrix based on the number of threat sources and the number of drone swarms to determine the task matching relationship between drones and threat sources; the elements of the capability affinity matrix represent the maximum potential masking duration of each threat source for the drone; the task matching relationship includes at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. Step 140: Construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer; the strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model; the system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. Step 150: Based on the task matching relationship and hierarchical optimization framework, adopt the hierarchical optimization strategy corresponding to the task matching relationship to generate the optimal smoke grenade deployment scheme with the optimization objective of maximizing the overall effective shielding time; Step 160: The optimal smoke grenade deployment scheme is calculated into flight control commands and smoke grenade launch commands for each UAV, and the commands are executed.

[0009] Preferably, the method further includes step 170: using Monte Carlo simulation to perform a robustness test on the generated optimal smoke grenade deployment scheme and evaluate the stability of the shielding performance under parameter perturbation.

[0010] On the other hand, the present invention provides an intelligent smoke screen concealment device based on multi-drone collaboration, comprising: The first module is used to acquire information on the movement status of the threat source, the real-time status of our own drones, and environmental constraints. The second module is used to establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking. The third module is used to construct a capability affinity matrix based on the number of threat sources and the number of drone swarms, in order to determine the task matching relationship between drones and threat sources; the elements of the capability affinity matrix represent the maximum potential masking duration of each threat source for the drone; the task matching relationship includes at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. The fourth module is used to construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer. The strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model. The system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. The fifth module is used to generate an optimal smoke grenade deployment scheme with the goal of maximizing the overall effective shielding time, based on the task matching relationship and the hierarchical optimization framework, and by adopting a hierarchical optimization strategy corresponding to the task matching relationship. The sixth module is used to calculate the optimal smoke grenade deployment scheme into flight control commands and smoke grenade launch commands for each UAV, and execute the commands.

[0011] Compared with the prior art, the beneficial effects of the solution of the present invention include: (1) By constructing a capability affinity matrix, the system can quickly assess and establish the optimal task matching relationship between UAVs and threat sources. Combined with the hierarchical optimization framework (strategy optimization layer and system optimization layer), the corresponding optimization mechanism can be adaptively invoked according to the actual task scale (one-to-one, one-to-many, many-to-many), thereby decomposing the complex collaborative problem into sub-problems that can be solved efficiently, which significantly enhances the adaptability and scalability of the method to different scenarios.

[0012] (2) By adopting a hierarchical optimization strategy corresponding to the task matching relationship, the optimal smoke grenade deployment scheme is generated, which significantly improves the masking duration and resource utilization efficiency. Specifically, for one-to-one relationships, the strategy optimization layer is based on an accurate geometric masking model and a numerical root-finding time series analysis method based on geometric relationships, which can efficiently optimize the heading, speed, deployment time and fuse delay of a single UAV to maximize the masking duration. For complex many-to-many relationships, the system optimization layer integrates resource allocation, platform collaboration and cluster scheduling strategies, and introduces a fairness optimization term based on the bottleneck effect. While pursuing the maximum total masking duration, it ensures the balance of masking effects of each threat source, thereby achieving efficient utilization of global spatiotemporal resources.

[0013] (3) By solving the optimal smoke grenade deployment scheme into directly executable flight control commands (UAV heading angle and flight speed) and smoke grenade launch commands (smoke grenade deployment time and fuse delay), a closed loop from decision-making to control is formed; further, the robustness of the scheme is tested through Monte Carlo simulation to verify its performance stability under parameter disturbances, ensuring that the generated strategy has high reliability and practicality in actual deployment. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the steps of an intelligent smoke masking method based on multi-UAV collaboration in one embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the geometric relationship between the mission scenario, threat source, and initial position of the drone in one embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the spatiotemporal relationship between smoke grenade deployment, activation, and cloud formation descent in one embodiment of the present invention. Figure 4 This is a schematic diagram of a hierarchical optimization framework structure in one embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the calculation result of the single smoke grenade concealment duration in one embodiment of the present invention, wherein, Figure 5 (a) is a schematic diagram of the calculation results of the shading duration. Figure 5 (b) is a schematic diagram of the unit test results. Figure 5 (c) is the duration of occlusion. Following the heading angle A schematic diagram of the change curve, Figure 5 (d) is the duration of occlusion. With the effective time A schematic diagram of the sensitivity analysis curve. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0017] In one embodiment, such as Figure 1 As shown, this invention provides an intelligent smoke masking method based on multi-UAV collaboration, comprising: Step 110: Obtain the motion status information of the threat source, the real-time status information of your own UAV, and environmental constraint information; Step 120: Establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking; Step 130: Construct a capability affinity matrix based on the number of threat sources and the number of drone swarms to determine the task matching relationship between drones and threat sources; the elements of the capability affinity matrix represent the maximum potential masking duration of each threat source for the drone; the task matching relationship includes at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. Step 140: Construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer; the strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model; the system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. Step 150: Based on the task matching relationship and hierarchical optimization framework, adopt the hierarchical optimization strategy corresponding to the task matching relationship to generate the optimal smoke grenade deployment scheme with the optimization objective of maximizing the overall effective shielding time; Step 160: The optimal smoke grenade deployment scheme is calculated into flight control commands and smoke grenade launch commands for each UAV, and the commands are executed.

[0018] Specifically, in step 110, the motion state information of the threat source is obtained by analyzing the initial state and constant motion characteristics of the threat source, including: Initial position: The starting coordinates of the threat source in three-dimensional space. .

[0019] In one embodiment, the initial locations of the three threat sources M-1, M-2, and M-3 are as follows: , , All units are meters; Motion velocity (vector): using This indicates the speed and magnitude of the threat source's movement. and direction.

[0020] In one embodiment, all threat sources are... A motion at a constant speed of m / s, moving at a uniform linear velocity along a straight line pointing from its initial position toward a pre-set dummy target. m / s represents the unit: meters per second.

[0021] Motion Model: Based on the above information, determine the threat source at any given time. The position vector.

[0022] The trajectory of a threat source can be represented in a general form: (1) in, It is the position vector of the threat source at time t. A position (vector) function describing the spatial location of a threat source as it changes over time can be obtained through sensor detection, trajectory prediction, or prior intelligence.

[0023] In one embodiment, the threat source moves toward the dummy target in a straight line at a constant speed, and the equation of its trajectory is: (2) in, As a source of threat at all times The position vector, The speed of the threat source's movement. The time it takes for the threat source to reach the false target. .

[0024] In one embodiment, the motion of the threat source may include variable speed motion, curvilinear maneuvering, or other non-uniform linear motion, corresponding to a position function. It can be obtained through real-time detection and status estimation of threat sources.

[0025] In one embodiment, the three threat sources (M-1, M-2, M-3) all move at a constant speed along a straight line toward the decoy target (e.g., the origin, reference point, or decoy position of the coordinate system). 300 m / s. In this embodiment, the time orders of arrival of the three threat sources at the decoy target are as follows: 67 s, 63.8 s, 60.4s. This time window represents the upper limit of the scheduling of UAV maneuvers and smoke grenade "deployment-effectiveness-effective concealment", providing a baseline for subsequent model building and solution.

[0026] The real-time status information of the user's drones includes the initial deployment status and basic performance constraints of the user's drone cluster.

[0027] Specifically, the initial deployment state of the self-propelled drone swarm includes the three-dimensional spatial coordinates of each drone at the start of the mission. The drones in the swarm are numbered: FY-1, FY-2, ..., FY- ; The total number of drones in the drone swarm; the threat source IDs: M-1, M-2, ..., M- ; The total number of threat sources, the It is a type of data that requires dynamic statistical analysis on-site.

[0028] To illustrate the threat source and the initial position of the drone, in a typical example, take... , The initial coordinates of the five drones are: FY-1 FY-2 FY-3 FY-4 FY-5 The unit is meters. For example... Figure 2 As shown, the geometric relationship between the initial positions of the threat source and the UAV in the task scenario and coordinate system in a typical example is given. The false target is located at the origin of the coordinate system, and the real target to be protected is located at (0, 200, 0). The aim is to induce the threat source to move towards the false target under the concealment effect of the smoke grenade.

[0029] Furthermore, in multi-drone collaborative scenarios, the initial coordinates of all drones are used as known inputs.

[0030] In one embodiment, the basic performance constraints include at least platform performance constraints and load operation constraints.

[0031] The platform performance constraints include: Speed ​​range: Let the speed range of the drone be... .

[0032] In one embodiment, the drone can be adjusted within the range of 70 m / s to 140 m / s and maintain a constant speed of flight.

[0033] Flight mode: The drone maintains a constant flight altitude and flies in a straight line in the horizontal plane during the mission.

[0034] In one embodiment, the payload operation constraint includes: when two payloads are dropped consecutively in a single mission by each UAV, the minimum time interval between the dropping actions must be greater than or equal to 1 second.

[0035] The environmental constraint information includes fixed parameters or limitations such as physical environment, platform operation rules, and load operation parameters.

[0036] The physical environment includes natural physical parameters such as gravitational acceleration and atmospheric conditions (e.g., wind speed and wind direction); The platform's operational rules include the maximum number of payloads that can be dropped in a single UAV mission, the minimum time interval for continuous drops, the flight speed range, and heading adjustment constraints, etc. The load operation parameters include the effective shielding radius of the smoke grenade cloud, formation time, duration, diffusion speed or cloud sinking speed, and other inherent characteristic parameters.

[0037] Furthermore, in step 120, a basic masking model is established to illustrate the spatiotemporal relationship between smoke grenade deployment and threat source masking, including: Based on the geometric-kinematic equations, the criteria for determining the effectiveness of smoke grenade concealment are given.

[0038] Figure 3 The spatiotemporal relationship of the smoke grenade deployment, activation, and cloud descent is presented. After leaving the drone, the smoke grenade experiences downward gravitational acceleration due to gravity. However, since the drone is flying horizontally, due to inertia and according to Newton's First Law (that all objects remain at rest or in uniform rectilinear motion), the smoke grenade also possesses an initial horizontal velocity. Therefore, the smoke grenade moves forward and downward, a process that can be considered a special case of projectile motion, i.e., horizontal projectile motion. Thus, the trajectory equation of the smoke grenade from deployment to activation is obtained: (3) ; Among them, time , It is the time for drone deployment. It refers to the location of the drone at the time of deployment. The drone's velocity vector (corresponding to a velocity magnitude of) ), That's when the smoke grenade takes effect. , It's the fuse delay of a smoke grenade; Represents the gravitational acceleration vector. It is the gravitational acceleration constant. Substituting into the above motion trajectory equation, we obtain the position of the effective point. A spherical cloud forms instantly upon activation; the effective radius of the cloud is denoted as . (For example = 10 m).

[0039] To effectively characterize the cloud shading effect, it is assumed that the center of the cloud sinks at a constant speed after the effect takes effect, and that the time interval is... If the cloud has effective shielding capabilities, then the trajectory equation of the cloud center is: , (4) in, It is the center of the cloud that is constantly Location; It is the sinking speed of the cloud center; in The cloud cluster is considered invalid.

[0040] Assuming that successful cloud cover requires the cloud to come into contact with the threat source, therefore, when a moment exists... The occlusion is deemed effective if the following constraints are satisfied: , (5) in, It is the effective shielding radius of the cloud cluster; while the shielding duration of the smoke grenade cloud cluster is a time set measure that satisfies the above inequality. .

[0041] The masking effectiveness condition given by formula (5) is used to construct the basic masking model for masking duration optimization under the support of the geometric-kinematic equations of formulas (1)-(4). The basic masking model is directly applicable to the masking duration optimization of a single smoke grenade (single UAV) against a single target, and can also be extended to the total masking duration optimization applicable to multiple smoke grenades (single UAV or multiple UAVs) deployed in a time slot manner after parallel accumulation.

[0042] Based on this, a three-dimensional spatial geometric relationship model is used to establish a rigorous occlusion criterion for determining any given moment. Whether it is effective. The logic for obtaining the criterion is based on the spatial intersection relationship between the line of sight of the threat source-target and the effective sphere of the smoke grenade cloud. Specifically, it includes: The target that needs protection (the real target) is simplified into a cylinder. Indicates the coordinates of the center of the lower base of the target. Indicates time The location of the threat source and its trajectory equation are given by the following formula: (6) (7) in, It is the initial location of the threat source. It is the orientation of the source of the threat. It is the time when the threat source collides with the decoy target.

[0043] Then we can obtain the threat source-target line of sight in vector form. : ; Set time The coordinates of the center of the cloud cluster are Then, by projecting the point onto the line containing the line segment and constraining the projection parameters within... Within the interval, find the solution point. The shortest distance to the line of sight between the threat source and the target : ; (8) in, It is a constraint Projection parameters within the interval; time interval The coordinates of the center of the cloud cluster are It is given by the following formula: (9) In the above formula, This is the effective point of the smoke grenade.

[0044] This leads to a strict occlusion criterion if and only if the following condition is met: Time validity conditions: ; Space availability conditions: .

[0045] The criterion constrains the projection parameters to the line segment through the clip function, avoiding the "false obscuration" illusion caused by the extension line of the threat source passing through the cloud sphere, and only outputs the effective obscuration window within the actual flight segment.

[0046] The conditions of the strict shielding criterion, combined with the geometric-physical fundamental equations given by the aforementioned formulas (1)-(9) and the constraints of all parameters, construct a basic shielding model with the objective function of maximizing the total shielding time of the smoke cloud, which is used to predict the dynamic shielding area formed under any smoke grenade deployment scheme.

[0047] For ease of explanation, let A represent the number of drones currently deployed in our drone swarm, B represent the number of smoke grenades dropped by each drone, and C represent the number of threat sources to be dealt with. Based on this ABC configuration, the task matching relationship between drones and threat sources is determined so as to design corresponding smoke grenade deployment schemes.

[0048] Specifically, in step 130, a capability affinity matrix is ​​constructed based on the number of threat sources and the number of drone swarms to determine the task matching relationship between drones and threat sources, including: Step 131: To efficiently schedule multiple drones to implement optical and infrared masking against multiple high-speed mobile threat sources, a clustering strategy is used for dimensionality reduction and initial screening to construct a capability affinity matrix. ; in, , ; elements in Indicates drone Threat source The affinity is calculated based on the threat source under a single optimal deployment strategy. Maximum potential effective shielding duration The effective masking duration was obtained by solving a simplified basic masking model under the single UAV-single smoke grenade-single threat source scenario.

[0049] Ability Affinity Matrix This study quantitatively describes the masking capabilities of each drone relative to each threat source, providing a data foundation for subsequent swarm task allocation. The elements of the capability affinity matrix characterize the drone's masking capabilities against each threat source, enabling reasonable task allocation based on objective capability assessments when assigning multi-drone collaborative tasks.

[0050] In one embodiment, the element data in the constructed capability affinity matrix is ​​shown in Table 1. In the column containing M-1, FY-1 and FY-4 (to M-1) have the highest affinity (4.73s and 4.60s, respectively), indicating that these two drones perform best when dealing with M-1. In the column containing M-2, FY-2 and FY-5 have the highest affinity (4.40s and 4.10s, respectively), making them more advantageous choices when dealing with M-2. For example, for a single drone FY-3, its affinity value to M-3 (3.45s) is much higher than its values ​​to the other two threat sources, exhibiting a sharp expertise peak. This indicates that FY-3's value is highly dependent on its initial geometric position matching, meaning it has the best geometric fit with M-3. However, this does not mean that FY-3 is the best choice to deal with M-3, as FY-2 and FY-4 have higher affinity to M-3.

[0051] Table 1. Example of Affinity Matrix (Unit: s)

[0052] Step 132: Identify and classify task matching relationships: Based on the number of threat sources to be addressed (C), the number of drones currently deployed (A), and the number of smoke grenades that each drone can deploy (B), the task of drones deploying smoke grenades for coordinated cover is formally summarized into a specific ABC configuration, and the following four basic task matching relationships are specifically configured: One-to-one relationship means that a single drone is dealing with a single threat source. Based on the high-speed movement mode of the threat source, the default configuration is ABC = 1-1-1, which means that a single drone uses a single smoke grenade to cover a single threat source. One-to-many relationship means that a single drone is dealing with multiple threat sources. In this case, based on the actual mission considerations, a single drone needs to carry multiple smoke grenades. The default configuration is 1-BC, where B>1 and C>1. In this case, completing the masking mission involves resource allocation and mission scheduling. Many-to-single relationship means that multiple drones deal with a single threat source. Multiple drones cooperate, each dropping a single smoke grenade they carry, to jointly block the single threat source. The default configuration is A-1-1, where A>1, and spatiotemporal coordination is required. Many-to-many relationships, multi-drone swarm collaboration, combating multiple threat sources, default configuration ABC, A>1 and C>1, B=1 or>1, is the most complex joint task allocation and global scheduling problem.

[0053] Once the specific task matching relationship is identified, the corresponding optimization layer in the hierarchical optimization framework can be activated to solve the problem.

[0054] Furthermore, in step 140, to achieve full-scene coverage from simple single-point occlusion to complex cluster collaboration, a layered optimization framework including a strategy optimization layer and a system optimization layer is constructed, such as... Figure 4 As shown, the hierarchical optimization framework adopts a bottom-up (strategy optimization layer is the bottom layer) and simple-to-complex design, which clearly shows the hierarchical structure from the basic occlusion model to each optimization layer and the corresponding core algorithm, ensuring that each task matching relationship can call the most suitable hierarchical optimization strategy.

[0055] The strategy optimization layer, as the underlying foundation and core of the entire optimization system, is responsible for handling the modeling and accurate calculation of the most basic occlusion unit in the one-to-one relationship configuration.

[0056] The strategy optimization layer adopts a numerical root-finding time series analysis method based on geometric relationships. It calculates the occlusion duration through the line segment-sphere intersection criterion and uses iterative approximation to accurately solve the occlusion time window.

[0057] The calculation of the shielding duration using the line segment-sphere intersection criterion refers to the calculation of the shielding duration using the spatial intersection criterion between the line segment of the threat source-target line of sight and the effective sphere of the smoke grenade cloud.

[0058] In one embodiment, the geometrically based numerical root-finding time-series analysis method for solving the occlusion time window includes: Step 210: Within the effective time window after taking effect Inside, with step length Uniform discrete sampling is performed on s to generate a time series. And calculate the shortest distance from the threat source-target line segment to the center of the cloud cluster for each sampling point according to formula (8). .

[0059] Step 220: Scan the distance sequence and identify all that satisfy the following conditions. The continuous time intervals are merged into candidate occlusion segments. This coarse screening process effectively avoids the blindness of global binary search and can handle complex situations with multiple occlusion segments.

[0060] Step 230: Near the start and end boundaries of each candidate shading segment, refine the equation characterizing the critical shading state. An iterative approximation using numerical root-finding is performed to obtain the accurate occlusion entry time. and exit time .

[0061] In one embodiment, the iterative approximation of numerical root finding in step 230 employs a bisection method, using the start and end boundaries of the candidate occlusion segment as the initial interval. Using the dichotomy method to The solution is obtained through iterative steps. In each iteration, the midpoint of the interval is taken. ,calculate ;like and If the signs are the same, the interval shrinks to... Otherwise shrink to Set the convergence accuracy. for On the order of s, iterated to The root value is output at time s. The computational complexity of the binary search method is O(s). It exhibits robustness with global convergence.

[0062] like Figure 5 The calculation results for the duration of coverage by a single smoke grenade are shown in the figure. Among them, Figure 5 (a) is the curve showing the calculated duration of shading, with the horizontal axis representing time. The vertical axis represents the shortest distance from the threat source-target line segment to the center of the cloud cluster. The horizontal dashed line represents the effective shading radius. The two vertical dashed lines indicate the time of entry of the occlusion. and exit time ; Figure 5(b) provides the unit test verification results, with analytical comparisons given respectively. , ) and numerical convergence test conditions (e.g., step size) The entry time error and exit time error (on the order of μs) are defined as follows: the entry time error is the absolute deviation between the numerical solution and the analytical solution at the start of the occlusion, and the exit time error is the absolute deviation between the numerical solution and the analytical solution at the end of the occlusion. Figure 5 (c) is the duration of occlusion. Following the heading angle The curve showing the change in heading angle is used to mark the optimal heading angle. ; Figure 5 (d) is the duration of occlusion. With the effective time Sensitivity analysis curves, indicating the optimal effective time. .

[0063] In this example, At time s, s, The effective shielding duration is approximately 1.435082 s.

[0064] In one embodiment, the iterative approximation of numerical root finding in step 230 employs the Newton-Raphson method, using the coarse-screen boundary points as initial guesses. The Newton-Raphson method was used to... The solution is obtained through iterative steps. The iterative formula is as follows: , where the derivative Numerical approximation is performed using the central difference method, i.e. Time increment Pick On the order of s. The iteration termination condition is set to In this example, the Newton-Raphson method typically converges within 3-5 iterations, yielding the solution. s, The result s is consistent with the result of the bisection method, but the derivative information is used to accelerate convergence.

[0065] The numerical root-finding time series analysis method based on geometric relationships, which employs the bisection method, has a total computational complexity of O(n). This method combines the efficiency of global scanning with the high accuracy of the bisection method. The iterative approximation of the numerical root-finding can also be achieved using other numerical root-finding techniques, including but not limited to the secant method and the golden section search.

[0066] The strategy optimization layer is integrated into an occlusion duration estimator. For any candidate deployment scheme generated by any upper-layer optimization algorithm in the system optimization layer, this estimator performs fast and high-precision occlusion effect calculation to obtain the effective occlusion duration or the union of time intervals. The occlusion duration estimator is the foundation for the calculation of all upper-layer optimization objective functions.

[0067] The system optimization layer, building upon the precise evaluation capabilities provided by the strategy optimization layer, addresses more complex task matching relationships and solves collaborative optimization and resource scheduling problems involving multiple variables and constraints. The system optimization layer is further subdivided into three sub-layers: resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization.

[0068] Resource allocation optimization (sub-layer): mainly deals with one-to-many relationships, that is, the allocation of single-platform resources (multiple smoke grenades, adjustable heading and speed) under multiple tasks (multiple threat sources).

[0069] The resource allocation optimization adopts a global-local hybrid heuristic search method to perform collaborative global optimization of the UAV's flight speed, smoke grenade deployment time, fuse delay, and heading angle in order to allocate limited smoke grenade resources and determine the optimal deployment time and spatial orientation of each smoke grenade in order to maximize the cumulative masking effect on the target threat source.

[0070] In one embodiment, the hybrid heuristic search method is implemented by combining Latin hypercube sampling with an improved particle swarm optimization algorithm: First, a global exploration is performed using Latin hypercube sampling (LHS). Each decision variable (velocity) is then analyzed. Release time , fuse delay Heading angle The feasible interval is divided into equal parts. Layers, within each layer, randomly sample and randomly combine across dimensions to generate We have a set of sample points uniformly distributed in the decision space. We calculate the objective function value for each sample point and select the top-performing sample points. 10 (e.g., the first 6-10) are used as high-quality initial solutions. LHS ensures unbiased coverage of the high-dimensional space with fewer samples.

[0071] Secondly, using the high-quality solutions selected by LHS as the initial population, an improved Particle Swarm Optimization (PSO) algorithm is initiated for fine development. The improvement strategies include: linearly decreasing inertia weights to balance exploration and development; velocity truncation to prevent oscillations; feasible region projection to handle out-of-bounds particles; and a stagnation perturbation mechanism, which applies small mutations to some particles to escape local optima if the global optimum has not improved for several consecutive generations.

[0072] Finally, using the global optimal solution obtained by PSO as the starting point, a derivativeless local search algorithm (such as the Nelder-Mead simplex method) is called to "polish" and refine the solution, resulting in the final optimal decision vector.

[0073] In this embodiment, the optimal parameters are obtained by solving: , , , The effective masking duration was increased to approximately 4.73 seconds, an improvement of over 200% compared to the baseline strategy.

[0074] The global-local hybrid heuristic search method is not limited to the LHS+PSO combination described above. In the global exploration phase, it can be replaced by uniform design, orthogonal design, Monte Carlo sampling, etc.; in the optimization phase, it can be replaced by derivative-free optimization methods such as differential evolution, genetic algorithms, and simulated annealing. Any search method employing a hybrid architecture of "wide-area sampling exploration + fine-grained optimization" falls within the protection scope of this invention.

[0075] Platform Collaborative Optimization (Sub-layer): Primarily handles many-to-one relationships, i.e., multiple heterogeneous drone platforms collaboratively addressing a single threat source.

[0076] The platform collaborative optimization adopts a collaborative optimization method based on swarm intelligence or evolutionary computation to collaboratively optimize the smoke grenade deployment parameters of multiple drones in order to maximize the union of the masking time for the same threat source and find a global strategy that enables the masking time windows generated by each drone platform to form optimal complementary (continuous or nearly continuous) global strategies on the time axis.

[0077] In one embodiment, the collaborative optimization method based on swarm intelligence or evolutionary computation is implemented using the differential evolution (DE) algorithm. The decision vectors of the three UAVs are encoded into an individual... The optimal cooperative strategy is obtained through iterative evolution via mutation, crossover (binomial crossover), and greedy selection. The mutation operation is represented as: ,in For the mutation vector, , and This represents three distinct individuals randomly selected from the current population. This is the scaling factor, usually set to a constant between [0,2], for example, 0.5.

[0078] In another embodiment, the collaborative optimization method based on swarm intelligence or evolutionary computation is implemented using an adaptive genetic algorithm (AGA). AGA can dynamically adjust the crossover probability based on the population fitness variance and evolutionary stagnation state. and mutation probability When population diversity declines or the optimal solution stagnates, increase and reduce To enhance global exploration; as the population converges and continues to improve, reduce and maintain a high This promotes localized development. At the same time, an elite retention strategy is introduced to ensure the algorithm does not degrade.

[0079] In one embodiment, the coordinated masking results of the three UAVs exhibit clear temporal complementarity: FY-1 forms an early long masking window [2.51, 7.25] s, FY-2 covers the interruption window [41.92, 43.47] s, and FY-3 is responsible for the later masking [47.65, 51.11] s, with a total masking union duration of approximately 9.745 s.

[0080] In addition, in other embodiments, the collaborative optimization method based on swarm intelligence or evolutionary computation includes, but is not limited to: particle swarm optimization and its variant algorithms, genetic algorithms and their variant algorithms, differential evolution variant algorithms, ant colony algorithms, artificial bee colony algorithms, and hybrid or other adaptive variant algorithms of the above methods.

[0081] Cluster Collaborative Optimization (Sub-layer): Handles many-to-many relationships, i.e., the large-scale collaborative problem of multiple drones dealing with multiple threat sources.

[0082] The cluster collaborative optimization first performs initial task allocation based on the capability affinity matrix and cluster analysis. Then, by combining a heuristic scheduling strategy, a candidate action library is generated to achieve joint task allocation and global scheduling of multiple UAVs to multiple threat sources. The heuristic scheduling strategy refers to a rule-based scheduling method that gradually constructs a global scheduling scheme based on local decision information. The joint task allocation includes task allocation, resource scheduling, and spatiotemporal coordination. Simultaneously, by introducing a fairness optimization term based on the weakest link effect, it balances multiple objectives such as overall efficiency and fairness.

[0083] In one embodiment, a greedy scheduling algorithm is used as a heuristic scheduling strategy to generate a candidate action library and output a scheduling scheme, including: Step S310: Pre-set 5 representative parameters for each drone. Flight configuration combinations, targeting each combination and each potential source of threat that may be obstructed, in Within the two-dimensional decision space, a batch of high-performance candidate solutions (e.g., the top 10) are generated using LHS combined with derivativeless local search (Pattern Search) to form a candidate action library. Each candidate action is accurately evaluated, including the time interval during which it obscures all threat sources.

[0084] Step S320: Initialize the scheduling scheme to be empty. In each iteration, select the action that maximizes the marginal gain of the overall objective function from the candidate action library. If the action does not violate constraints such as the same-machine deployment interval (≥1s) and the maximum number of deployments per machine (≤2), then add it to the scheduling scheme and update the union of the masking intervals of each threat source.

[0085] In one embodiment, the overall objective function is the sum of the total occlusion duration and the minimum occlusion duration. Multiplier, It is the coefficient for minimum occlusion duration.

[0086] Step S330: Repeat step S320 until the candidate action library is exhausted or no action can further improve the objective function, and output the final scheduling scheme.

[0087] In one embodiment, the results of the coordinated masking of three threat sources by five drones are shown in Table 2. The total masking time reached 31.94 s, and the minimum masking time among all threat sources was 8.50 s, which takes into account both overall effectiveness and fairness.

[0088] In another embodiment, a heuristic scheduling strategy based on a market auction mechanism is employed to generate a candidate action library and output a scheduling scheme, including: (1) Candidate action library generation: Multiple flight configuration combinations are preset for each UAV. High-performance candidate schemes are generated using LHS combined with Pattern Search. Each candidate scheme includes the masking time interval for three threat sources.

[0089] (2) Auction initialization: Each candidate solution is treated as a "resource-task" auction item. During initialization, the concealment requirements of all threat sources are issued as "tasks", and all candidate actions participate in the auction as "bidders".

[0090] (3) Multi-round combined auction mechanism: In each round of auction, the bidding price for each candidate action is calculated based on its marginal contribution to the duration of threat source obscuring. Specifically, for threat sources... Candidate actions bid Defined as this action against the threat source Currently, there is already a masking and aggregation. Marginal increment: ; in, Candidate actions The resulting shielding area, Denotes the measure function, here This indicates that the total effective occlusion duration is calculated using a measure function. This indicates the duration for which the union of existing occlusions is calculated using a measure function. A higher bid indicates a greater contribution of the action to filling the current occlusion gap.

[0091] (4) Winning bid determination and constraint check: In each round, the action with the highest bid that meets the constraints (interval between releases of the same drone ≥ 1s, maximum release of 2 drones per drone, etc.) is selected as the winning bid for that round and assigned to the masking schedule of the corresponding threat source. After the winning bid is determined, the masking union of each threat source is updated, and candidate actions that conflict with the winning bid (such as the same drone in the same time period) are removed.

[0092] (5) Termination and Output: The auction terminates when the masking window for all threat sources reaches the predetermined target duration or when there is no new effective marginal contribution. The final multi-target allocation strategy and masking scheme for each threat source are output.

[0093] The cooperative occlusion scheme obtained through this strategy has a total occlusion duration. This is comparable to the result of the greedy scheduling (total occlusion duration of 31.94s), but in terms of fairness... The slight improvement (minimum occlusion duration increased to 8.7s) indicates that the auction mechanism allocates occlusion resources more evenly through global bidding.

[0094] The heuristic scheduling strategy is not limited to greedy scheduling algorithms or market auction-based mechanisms. Priority scheduling, round-robin scheduling, simulated annealing scheduling, tabu search, rule-based expert system scheduling, or combinations thereof can also be used. Any rule-based method that progressively constructs a global scheduling scheme based on local decision information falls within the scope of this invention.

[0095] In step 150, based on the aforementioned determined task matching relationship and the constructed hierarchical optimization framework, a hierarchical optimization strategy is used to dynamically call or combine corresponding optimization levels (strategy optimization layer or system optimization layer) to generate the final optimal smoke bomb deployment scheme, including: The first step is to activate the corresponding optimization layer combination from the hierarchical optimization framework based on the ABC configuration of the task matching relationship.

[0096] In one embodiment, for a one-to-one relationship configured in 1-1-1, the geometric relationship-based numerical root-finding time series analysis evaluator of the strategy optimization layer is invoked for accurate calculation or as the core of univariate optimization. If it is a one-to-many relationship configured as 1-3-1, then activate the strategy optimization layer (evaluator) and the resource allocation optimization in the system optimization layer (e.g., the sub-layer uses the LHS-PSO algorithm). If it is a 3-1-1 configuration of multiple pairs of relationships, then activate the policy optimization layer (evaluator) and the platform co-optimization in the system optimization layer (e.g., the sub-layer uses the DE / AGA algorithm). If it is a many-to-many relationship configured in 5-3-3, then activate the strategy optimization layer (evaluator) and the cluster collaborative optimization in the system optimization layer (the sub-layer adopts a heuristic scheduling strategy).

[0097] The second step, based on the basic occlusion model, is to formally define the optimization model corresponding to the current task matching relationship within each activated optimization layer, including: Define decision variables: Identify all parameters that need to be optimized, such as the set of flight speeds for each drone. Heading Angle Set Collection of times for each smoke grenade deployment and fuse delay set ; Establish the objective function for optimization: With maximizing the overall effective occlusion duration as the core, the objective function of the optimization model is constructed as follows: ; in, It is a source of threat (Corresponding number is M-) The total duration of shading, which is the union of the shading times, is given by the following formula: ; In the above formula, It is a source of threat The union of the effective occlusion time intervals It is a drone Threat source The effective concealment time range of smoke bombs.

[0098] For the objective function of cluster collaborative optimization, a fairness optimization term based on the weakest link effect is introduced. This is used to maximize the minimum masking duration among all threat sources to avoid uneven masking effects. The objective function of the optimization model is modified using a fairness optimization term. ; in, It is the weight of the fairness optimization term, used to balance the performance and fairness characterized by the total duration.

[0099] In one embodiment, This allows for a good balance between total duration and fairness.

[0100] The third step is to combine the parameter constraints and strict occlusion conditions in the aforementioned basic occlusion model to obtain the constraints of the optimized model.

[0101] The fourth step involves using the algorithm corresponding to the strategy optimization layer / system optimization layer to solve the optimization model corresponding to the current task matching relationship, driving the decision variables to be optimized, and repeatedly calling the occlusion duration evaluator of the underlying strategy optimization layer to calculate the objective function value until convergence. Finally, a set of decision variable values ​​that make the objective function optimal is output, which is the optimal smoke grenade deployment scheme.

[0102] The optimal smoke grenade deployment scheme includes the flight speed and heading angle that each UAV should adopt, as well as the deployment time and fuse delay setting of each smoke grenade it carries, which together constitute a collaborative action plan aimed at maximizing global shielding effectiveness.

[0103] In step 160, the optimal smoke grenade deployment scheme generated in step 150 is transformed into low-level control commands that can be directly executed by the UAV platform and mission payload, completing the final closed loop from decision-making to execution.

[0104] The heading angle and flight speed of the UAV in the optimal smoke grenade deployment scheme are calculated into flight control commands; the smoke grenade deployment time and fuse delay are calculated into smoke grenade launch commands; The flight control commands are sent to the UAV's flight control system to guide the UAV to fly along a predetermined course and at a specified speed to the mission area or maintain a specific flight state. The smoke grenade launch commands are sent to the UAV's mission management system or mission payload controller to precisely control the timing of the smoke grenade release and its effective time in the air.

[0105] In one embodiment, for a many-to-many relationship configured in 5-3-3, the optimal smoke grenade deployment scheme is shown in Table 2, where each row of data directly corresponds to a set of control instructions: The FY-1 UAV, in order to perform the mission of shielding the M-1, received the following instructions: fly at a speed of 71.18 m / s and a heading of 176.82°, and release the first smoke grenade 0.00s after the mission begins, with the fuse delay set to 2.48s. This process is repeated for all drones and all their deployment actions, ultimately forming a set of collaborative control instructions that covers the entire cluster and spans the mission timeline.

[0106] Table 2. Examples of optimal smoke grenade deployment schemes for many-to-many relationships configured in 5-3-3.

[0107] The total masking duration can be calculated from the data in Table 2 as J = 31.94 s, and the minimum masking duration among all threat sources is 8.50 s.

[0108] The generated instruction set (flight control instructions and smoke grenade launch instructions) is sent to each UAV via data link. The UAV flight control system and mission management system receive and execute the corresponding instructions, thereby triggering the deployment and activation of smoke grenades at predetermined time and space points, accurately reproducing and optimizing the planned cooperative masking scenario.

[0109] Finally, the intelligent smoke masking method based on multi-UAV collaboration proposed in this invention also includes step 170: Monte Carlo simulation is used to test the robustness of the generated optimal deployment scheme, evaluate its performance stability under real-world uncertainties, and ensure the practicality and reliability of the scheme.

[0110] First, the disturbance factors were modeled. Simulation verification considered three main types of random disturbances and modeled them as random variables that conform to a certain probability distribution (such as uniform distribution) around the nominal value: Environmental parameter disturbances: Simulating the impact of changes in wind speed and direction on the movement of smoke clouds, mainly manifested as the cloud's sinking speed. (e.g., ±0.2 m / s) and horizontal drift; System performance error: Accuracy limitations of analog sensors and actuators, including the effective shielding radius of the smoke cloud. Fluctuations (e.g. ±1 m); Control and timing errors: Delays and jitters in the issuance and execution of simulated commands, including the drone's flight speed. (e.g., ±1 m / s), heading angle (e.g., ±1°) control deviation, and the timing of smoke grenade deployment. and fuse delay Timing error (e.g., ±0.1 s).

[0111] Then, the optimal smoke grenade deployment plan was validated using Monte Carlo simulation, including multiple independent simulations: In each simulation, a set of random perturbation values ​​is generated for all parameters based on the aforementioned perturbation model. These values ​​are then superimposed on the nominal parameters of the scheme to form a perturbation instance. Under these perturbation parameters, the actual masking duration of all threat sources is recalculated strictly according to the basic masking model. .

[0112] Perform statistical evaluation by repeating the above process hundreds to thousands of times (e.g., 1000 times in the 1-3-1 configuration and 200 times in the 5-3-3 configuration) to collect a large sample size; calculate the statistical characteristics of key performance indicators, mainly including: Solution compliance rate: The percentage of solutions that still meet all hard constraints (e.g., occlusion before arrival) across all simulations.

[0113] Performance statistics: Total shading duration The mean, standard deviation, and specific quantiles (e.g., the worst 5th percentile).

[0114] Fairness statistics: Minimum occlusion duration The mean and fluctuation of.

[0115] Finally, the test conclusions and application results are obtained: The simulation results are applied to quantitatively evaluate the robustness of the scheme. For example, for a large-scale cooperative optimal smoke grenade deployment scheme with a 5-3-3 configuration and many-to-many relationships, the Monte Carlo test shows that: Within the set disturbance range, the compliance rate of the solution is as high as 96.5%.

[0116] The mean total shading duration was 30.8s (slightly lower than the ideal value of 31.94s), with a standard deviation of 1.7s, indicating that the performance fluctuation was small.

[0117] The total masking duration under the worst case (5th percentile) is still 27.9s, proving that the scheme can still maintain effective masking capability under extreme disturbances.

[0118] Step 170 provides a key decision-making basis for the engineering application of the solution: if the test results show that the solution is not robust enough, it can be fed back to the optimization stage, and the objective function can be modified (such as adding a robustness penalty term) to perform optimization iteration again until a solution that meets the stability requirements is obtained.

[0119] In summary, this invention addresses the problems of rigid strategies, low resource utilization efficiency, and poor environmental adaptability in existing smokescreen masking methods under complex collaborative scenarios. It proposes a systematic intelligent collaborative decision-making and optimization method, achieving significant technical results. Firstly, regarding the intelligence and adaptability of collaborative decision-making, this invention constructs a capability affinity matrix and a hierarchical optimization framework, enabling the system to automatically identify task scale (one-to-one, one-to-many, many-to-many) and call the corresponding optimization modules for processing. This achieves seamless expansion from individual to group collaborative strategies, overcoming the bottleneck of poor scalability and difficulty in handling dynamic tasks inherent in traditional models. Simultaneously, by introducing a fairness optimization term based on the weakest link effect, the invention ensures the balance of masking effects on various threat sources while maximizing the total masking duration. This effectively avoids protection vulnerabilities caused by uneven resource allocation and improves the reliability of the overall protection system. The intelligent features of the method in this invention include: automatically identifying task scale and relationships; adaptively invoking matching hierarchical optimization strategies; and automatically balancing overall effectiveness with masking fairness among threat sources during the optimization decision-making process. Ultimately, this achieves closed-loop automation from collaborative decision-making to platform control without requiring manual intervention for model reconstruction or strategy switching. These intelligent features are achieved collaboratively through the construction of a capability affinity matrix, the design of a hierarchical optimization framework, hierarchical optimization strategies corresponding to task matching relationships, and the introduction of fairness optimization terms.

[0120] Secondly, regarding masking effectiveness and resource utilization efficiency, this invention establishes a precise basic masking geometric model and employs a geometrically based numerical root-finding time-series analysis method / evaluator to achieve rapid and accurate evaluation of the masking duration for any deployment strategy. Based on this, it comprehensively utilizes a global-local hybrid heuristic search method, a collaborative optimization method based on swarm intelligence or evolutionary computation, and a heuristic scheduling strategy to globally optimize the UAV's heading, speed, deployment sequence, and fuse delay, significantly improving the single-attack masking duration (e.g., over 200% improvement compared to the baseline strategy in a single-UAV, single-missile scenario). Facing large-scale collaborative scenarios, a collaborative framework integrating Latin hypercube sampling and greedy scheduling is adopted, successfully solving the resource scheduling problem for multiple UAVs and multiple missiles. In a complex scenario where five UAVs deal with three threat sources, it achieves an excellent total masking duration of 31.94 seconds while maintaining fairness, significantly improving the utilization rate of spatiotemporal resources.

[0121] Furthermore, regarding the system's engineering practicality and robustness, this invention particularly emphasizes the stability of the strategy under actual disturbance environments. Through Monte Carlo simulation of the system, the optimized strategy was thoroughly tested under multiple disturbances such as wind speed changes, parameter drift, and execution errors, verifying its compliance rate of up to 96.5% and good performance stability (performance standard deviation less than 1.7s). Further sensitivity analysis clarified the degree of influence of different parameters on the masking effect, providing crucial basis for precision control and system design in engineering practice.

[0122] Finally, this invention forms a complete decision-making and control closed loop. The optimized deployment scheme can be directly calculated into the UAV's flight control commands (heading angle, speed) and smoke grenade launch commands (deployment time, fuse delay), realizing the connection from intelligent decision-making to precise execution and enhancing the system's deployability and automation level. In summary, this invention constructs a technical system integrating accurate evaluation, intelligent optimization, collaborative scheduling, and robust verification, ultimately achieving a comprehensive breakthrough in the scalability, efficiency, robustness, and executability of multi-UAV smoke grenade collaborative masking.

[0123] In one embodiment, the present invention provides an intelligent smoke screen concealment device based on multi-drone collaboration, comprising: The first module is used to acquire information on the movement status of the threat source, the real-time status of our own drones, and environmental constraints. The second module is used to establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking. The third module is used to construct a capability affinity matrix based on the number of threat sources and the number of drone swarms, in order to determine the task matching relationship between drones and threat sources; the elements of the capability affinity matrix represent the maximum potential masking duration of each threat source for the drone; the task matching relationship includes at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. The fourth module is used to construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer. The strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model. The system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. The fifth module is used to generate an optimal smoke grenade deployment scheme with the goal of maximizing the overall effective shielding time, based on the task matching relationship and the hierarchical optimization framework, and by adopting a hierarchical optimization strategy corresponding to the task matching relationship. The sixth module is used to calculate the optimal smoke grenade deployment scheme into flight control commands and smoke grenade launch commands for each UAV, and execute the commands.

[0124] In one embodiment, the apparatus further includes a seventh module for using Monte Carlo simulation to robustly test the generated optimal smoke grenade deployment scheme and evaluate the stability of the shielding performance under parameter perturbations.

[0125] Furthermore, the present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the intelligent smokescreen concealment method based on multi-UAV collaboration provided in any of the above embodiments. The computer device may be a server. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store sample data. The network interface of the computer device is used for communication with an external PC via a network connection.

[0126] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the intelligent smoke masking method based on multi-UAV collaboration provided in any of the above embodiments.

[0127] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by hardware related to computer program instructions. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0128] Matters not covered in this invention are common knowledge.

[0129] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0131] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart smoke masking method based on multi-UAV collaboration, characterized in that, Includes the following steps: Step 110: Obtain the motion status information of the threat source, the real-time status information of your own UAV, and environmental constraint information; Step 120: Establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking; Step 130: Construct a capability affinity matrix based on the number of threat sources and the number of drone swarms to determine the task matching relationship between drones and threat sources; The elements of the capability affinity matrix represent the maximum potential masking duration of the UAV for each threat source; The task matching relationships include at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. Step 140: Construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer; the strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model; the system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. Step 150: Based on the task matching relationship and hierarchical optimization framework, adopt the hierarchical optimization strategy corresponding to the task matching relationship to generate the optimal smoke grenade deployment scheme with the optimization objective of maximizing the overall effective shielding time; Step 160: The optimal smoke grenade deployment scheme is calculated into flight control commands and smoke grenade launch commands for each UAV, and the commands are executed.

2. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The basic shielding model comprehensively considers the trajectory of the smoke grenade, the diffusion law of the smoke grenade based on the trajectory of the cloud center, the effective shielding radius of the cloud, and the trajectory of the threat source. Based on strict shielding criteria, it predicts the dynamic shielding area formed under any smoke grenade deployment scheme. The strict occlusion criteria include: Time validity conditions: ; Space availability conditions: .

3. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The hierarchical optimization strategy corresponding to the task matching relationship includes: For one-to-one relationships, based on the basic masking model, the strategy optimization layer is invoked to solve for the optimal smoke grenade deployment scheme for a single UAV against a single threat source; For one-to-many relationships, the resource allocation optimization in the system optimization layer is invoked to solve the optimal smoke grenade deployment scheme for a single UAV against multiple threat sources; For many-to-single relationships, the platform collaborative optimization in the system optimization layer is invoked to solve the optimal smoke grenade deployment scheme for multiple drones against a single threat source; For many-to-many relationships, the cluster collaborative optimization in the system optimization layer is invoked. Through joint task allocation and global scheduling, the optimal smoke grenade deployment scheme for multiple drones against multiple threat sources is solved. The cluster collaborative optimization integrates the resource allocation optimization and platform collaborative optimization strategies.

4. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The strategy optimization layer adopts a numerical root-finding time series analysis method based on geometric relationships. It calculates the occlusion duration through the line segment-sphere intersection criterion and uses iterative approximation to accurately solve the occlusion time window.

5. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The resource allocation optimization is used to handle the task matching relationship between a single UAV and multiple threat sources. The resource allocation optimization adopts a global-local hybrid heuristic search method to perform collaborative global optimization of the UAV's flight speed, smoke grenade release time, fuse delay and heading angle in order to allocate limited smoke grenade resources.

6. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The platform collaborative optimization is used to handle the task matching relationship between multiple UAVs and a single threat source. The platform collaborative optimization adopts a collaborative optimization method based on swarm intelligence or evolutionary computation to collaboratively optimize the smoke grenade deployment parameters of multiple UAVs in order to maximize the union of the masking time for the same threat source.

7. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, The cluster collaborative optimization is used to handle the task matching relationship between multiple UAVs and multiple threat sources. The cluster collaborative optimization first performs initial task allocation based on the capability affinity matrix and cluster analysis, and then generates a candidate action library by combining heuristic scheduling strategies to realize joint task allocation and global scheduling of multiple UAVs to multiple threat sources.

8. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, In step 150, for the optimization objective of generating the optimal smoke grenade deployment scheme through cluster collaborative optimization, a fairness optimization term based on the weakest link effect is introduced to maximize the minimum masking duration among all threat sources in order to avoid uneven masking effects. In step 160, the flight control command includes the UAV's heading angle and flight speed; the smoke grenade launch command includes the smoke grenade release time and fuse delay.

9. The intelligent smoke masking method based on multi-UAV collaboration according to claim 1, characterized in that, It also includes step 170: using Monte Carlo simulation to test the robustness of the generated optimal smoke grenade deployment scheme and evaluate the stability of the shielding performance under parameter perturbation.

10. An intelligent smoke screen concealment device based on multi-UAV collaboration, characterized in that, The device includes: The first module is used to acquire information on the movement status of the threat source, the real-time status of our own drones, and environmental constraints. The second module is used to establish a basic masking model for the spatiotemporal relationship between smoke grenade deployment and threat source masking. The third module is used to construct a capability affinity matrix based on the number of threat sources and the number of drone swarms, in order to determine the task matching relationship between drones and threat sources; the elements of the capability affinity matrix represent the maximum potential masking duration of each threat source for the drone; the task matching relationship includes at least one-to-one, one-to-many, many-to-one, and many-to-many relationships. The fourth module is used to construct a hierarchical optimization framework, including a strategy optimization layer and a system optimization layer. The strategy optimization layer is used to optimize the smoke grenade deployment scheme under one-to-one relationships, and performs accurate solutions based on the basic occlusion model. The system optimization layer is used to handle collaborative optimization under one-to-many, many-to-one, and many-to-many relationships, including resource allocation optimization, platform collaborative optimization, and cluster collaborative optimization. The fifth module is used to generate an optimal smoke grenade deployment scheme with the goal of maximizing the overall effective shielding time, based on the task matching relationship and the hierarchical optimization framework, and by adopting a hierarchical optimization strategy corresponding to the task matching relationship. The sixth module is used to calculate the optimal smoke grenade deployment scheme into flight control commands and smoke grenade launch commands for each UAV, and execute the commands.

Images