Underwater target cross-domain collaborative searching and tracking method based on AUV and USV
By establishing kinematic models of underwater targets, USV and AUV and adopting improved particle swarm optimization algorithm, the track planning problem of cross-domain collaborative search and tracking of underwater targets is solved, and the efficiency of underwater target search and tracking tasks is improved, and the realization of the realization of the task of underwater targets is achieved. The efficiency of cross-domain collaborative search tracking is achieved.
Patent Information
- Application Number
- CN202411214300.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-31
- Publication Date
- 2024-12-06
AI Technical Summary
The existing technology has not yet solved the problem of track planning for cross-domain collaborative search and tracking of underwater targets. Especially when unmanned platforms work in collaboratively in different spatial domains, path planning is inefficient and it is difficult to achieve efficient underwater target search and tracking.
A cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is proposed. By establishing kinematic models of underwater targets, USV and AUV, and using improved particle swarm optimization algorithm, combining inertial weights and adaptive learning factor strategies , realize collaborative search and tracking of cross-domain heterogeneous systems.
Improve the efficiency of underwater target search and tracking tasks, avoid local optimal problems, meet unmanned platform constraints, and realize more efficient cross-domain collaborative search and tracking. Simulation experiments show that this method can complete water faster and more accurately Search and track the target.
Smart Images

Figure CN119088024A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of underwater target search, and in particular to a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV. Background Art
[0002] Due to the natural shielding and opacity of the marine environment, ships are vulnerable to threats from unknown underwater equipment, so it is necessary to search and track underwater targets. With the development of unmanned system technology, it has gradually become a consensus to use unmanned swarm platforms with high mission efficiency to perform marine missions. During the search and tracking process, the path planning of the unmanned swarm platform has an important impact on the search and tracking efficiency.
[0003] At present, the research on path planning of unmanned cluster platforms in marine environments is mainly aimed at the research of cluster platforms in the same spatial domain. For example, for the surface unmanned boat USV cluster, there is a solution in the prior art that proposes a method of combining the improved artificial potential field method with the Bezier curve to increase the self-repulsion field to solve the problem of USV cluster collaborative obstacle avoidance trajectory planning, so that the USV can effectively avoid frequent large turns and sharp turns during obstacle avoidance navigation. There is also prior art that studies the path planning problem of unmanned boat formation cruise detection in shallow water areas, integrating the Boustrophedon algorithm, wildfire method and A* algorithm to solve the problem of detection area coverage of unmanned boat formations, and improving the path obstacle avoidance function during the unmanned boat formation cruise path.
[0004] In the underwater field, for underwater autonomous vehicle AUV clusters, there is a method to simplify the path planning of multiple AUVs that reach the destination at the same time in the shortest time into two independent modules: the shortest time path planning of a single platform and the path replanning of multiple platforms under given time. The obstacle-free nodes are determined, the adjacent nodes, the corresponding state values and edge weights are set, and finally the two modules are solved through the BMC protocol. There is also a solution for the three-dimensional underwater working environment with variable ocean currents. Combined with the improved self-organizing graph neural network, a multi-AUV dynamic task allocation and path planning algorithm is proposed, and the goal is set to control a group of AUVs to reach all designated target locations at once under the premise of workload balance and sufficient energy, while ensuring the minimum consumption in the variable ocean currents.
[0005] Currently, there is no trajectory planning method for collaborative search and tracking of underwater targets by unmanned platforms in different spatial domains. Summary of the invention
[0006] In view of the problems existing in the prior art, the present invention proposes a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV, which combines the characteristics of AUV's long underwater endurance time and high underwater mission accuracy with the characteristics of USV's high speed and strong flexibility.
[0007] The technical solution of the present invention is:
[0008] A cross-domain collaborative search and tracking method for underwater targets based on AUV and USV includes the following steps:
[0009] Step 1: Establish kinematic models of underwater targets, USVs and AUVs:
[0010] The kinematic model of underwater target is:
[0011]
[0012] The position and attitude vector u of the underwater target T and the velocity vector V T for:
[0013]
[0014] where x T ,y T ,z T is the coordinate of the target in the x, y, and z axes of the spatial coordinate system, θ T is the pitch angle of the target in the vertical plane; is the yaw angle of the target in the horizontal plane; v T is the translation speed of the target in the horizontal plane, p T is the angular velocity of the target around the y-axis, q T is the angular velocity of the target around the z axis, a T Indicates the acceleration of the target;
[0015] The USV kinematic model is:
[0016]
[0017] USV position and attitude vector u S and the velocity vector V S for:
[0018]
[0019] where x S ,y S ,z S is the coordinate of USV in the x, y, z axis direction of the spatial coordinate system, z S =0; is the yaw angle of USV in the horizontal plane; v S is the translation speed of USV in the horizontal plane, q S is the angular velocity of the USV around the z axis, a S is the acceleration of USV;
[0020] The AUV kinematic model is:
[0021]
[0022] The position and attitude vector u of the AUV A and the velocity vector V A for:
[0023]
[0024] where x A ,y A ,z A is the coordinate of AUV in the x, y, z axis direction of the spatial coordinate system, θ A is the pitch angle of the AUV in the longitudinal plane; is the yaw angle of the AUV in the horizontal plane; v A is the translation speed of AUV in the horizontal plane, p A is the angular velocity of the AUV around the y-axis, q A is the angular velocity of the AUV around the z axis, a A represents the acceleration of the AUV;
[0025] Step 2: The cross-domain heterogeneous system composed of USV and AUV searches and tracks underwater targets: When the USV or AUV detects the position of the underwater target, the search phase ends; the position information of the searched underwater target is transmitted to the AUV, and the AUV completes the subsequent tracking task; during the search and tracking process, the USV and AUV share information;
[0026] In the process of searching and tracking, based on the kinematic models of underwater targets, USVs and AUVs established in step 1, the motion parameters of the cross-domain heterogeneous system composed of AUVs and USVs for searching and tracking underwater targets are optimized by using an improved particle swarm optimization algorithm to achieve cross-domain collaborative search and tracking of underwater targets:
[0027] In the target search phase, the particle parameters are The superscript j indicates the jth particle;
[0028] In the target tracking stage, the particle parameters are set as the particles of the USV platform and AUV platforms
[0029] The particle velocity update formula and the corresponding particle position update formula of the particle swarm optimization algorithm are:
[0030]
[0031] Where ω is the inertia weight, c1 is the individual learning factor, c2 is the social learning factor, rand1 and rand2 are random real numbers uniformly distributed in the range [0,1]; is the optimal solution for individual particles; X best The optimal solution for the group;
[0032] During the target search phase,
[0033]
[0034] Where a1 and a2 are both non-negative constants, ranging from (0,1); G max is the upper limit of the working cycle; c 1max 、c 2max To set an upper limit;
[0035] In the target tracking phase,
[0036]
[0037] The fitness function of the particle swarm optimization algorithm is:
[0038] During the target search phase,
[0039] F=max(S(d S ' -A ),f2,f search )
[0040] Where S(d′ S-A ) is the optimization index of the distance between USV and AUV, f2 is the optimization index of the set search space, and f search It is the basis for judging whether the heterogeneous system has searched for the target;
[0041] In the target tracking phase,
[0042] The fitness function of AUV is:
[0043] F A (k) = min(d A-T (k))
[0044] The fitness function of USV is:
[0045] F S (k) = min(μd S-A (k)+d S-T (k))
[0046] Among them, d A-T (k) is the distance between the AUV and the underwater target, d S-T (k) is the distance between AUV and USV, d S-T(k) is the distance between the USV and the underwater target; μ is the weight value added according to whether the AUV can currently search for the target; it is a constant less than 1 when the AUV can search for the target, and it is a constant greater than 1 when the AUV cannot search for the target.
[0047] Furthermore, in the kinematic model of step 1, the model constraints of the underwater target are:
[0048] Movement space constraints: underwater targets move within a set sea area, and a maximum diving depth range is set according to their own pressure bearing capacity;
[0049]
[0050] z Tmin The maximum operating depth of the target underwater;
[0051] Performance constraints: The target sets speed and acceleration limits based on its own performance range:
[0052]
[0053] The model constraints of USV are:
[0054] Working space constraints: USV searches for underwater targets in bounded sea areas where underwater targets have appeared. The spatial constraints are:
[0055] x min ≤x S ≤x max ,y min ≤y S ≤y max
[0056] Self-performance constraints: During navigation, the USV sets its speed and acceleration limits according to its own physical properties as follows:
[0057]
[0058] Detection capability constraints: In the cross-domain collaborative search and tracking task, the search range of the USV is in the shape of a hemisphere with the USV mass point as the center, and the radius of the search area is r det,S , when the underwater target appears in the search area of the USV, it is considered that the USV has successfully searched for the underwater target, that is:
[0059]
[0060] Communication constraints: During the search and tracking mission, the USV communicates with the underwater AUV. Assume that the communication range of the USV is a hemispherical shape with the center of mass of the USV as the center of the sphere, and the communication radius is d com(S-A) ; When the distance between USV and AUV does not exceed dcom(S-A) , they thought that the two could communicate with each other;
[0061] The model constraints of AUV are:
[0062] Workspace constraints: According to the AUV's own pressure resistance and search range, its workspace constraints are as follows:
[0063]
[0064] Performance constraints: Due to its own physical performance limitations, the AUV has speed and acceleration constraints during navigation as follows:
[0065]
[0066] Detection capability constraint: AUV works underwater. Assume that the search area of AUV is a sphere with the center of mass of AUV as the center of the search area, and the search radius is r. det,A , when the distance between the underwater target and the AUV is less than r det,A When , it is considered that the AUV has successfully searched for the underwater target, that is
[0067]
[0068] Target tracking constraint: The task requirement of the AUV in the underwater target tracking phase is to successfully track the underwater target. When the distance between the AUV and the underwater target is less than a certain threshold d(t E ) A-T When , it is considered that the target tracking task is completed at this time, that is,
[0069]
[0070] where t E is the end time of the tracking phase;
[0071] Communication constraint: When the distance between USV and AUV does not exceed d com(S-A) , they thought the two could communicate with each other.
[0072] Furthermore, in step 2, the process of searching and tracking underwater targets by the cross-domain heterogeneous system composed of USV and AUV is divided into two cases:
[0073] Case 1: The USV detects the underwater target first and transmits the target's real-time location information to the AUV until the AUV detects the target and completes the tracking task;
[0074] Case 2: The AUV detects the underwater target first, transmits the target's real-time location information to the USV, and continuously shortens the distance to the target to complete the tracking task. After receiving the information from the AUV, the USV moves closer to the target. The USV fuses the target information detected by both and transmits more accurate underwater target location information to the AUV in real time until the AUV completes the tracking task.
[0075] Furthermore, in step 2, during the process of searching and tracking underwater targets by the cross-domain heterogeneous system composed of USV and AUV, USV also predicts the future position of the target through the following process and transmits it to AUV; the prediction process is:
[0076] Assuming that the current period is the kth period, the coordinate information of the target in the current period is ρ T (k) = [x T (k),y T (k),z T (k)] T , the state information in the k-1 cycle is The target information obtained by USV is only the coordinate information ρ T ;
[0077] Then the USV predicts the state of the target as
[0078]
[0079] The target approximate predicted position in the k+Δkth period should be
[0080]
[0081] Where T hunt It is the tracking cycle of the tracking phase.
[0082] Furthermore, T hunt The value of is related to the AUV search radius and the distance between the AUV and the target, and is specifically set as follows:
[0083]
[0084] Where, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV;
[0085] Furthermore, the dynamic value rule of Δk is as follows:
[0086]
[0087] Among them, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV.
[0088] Furthermore, in the target search phase, the particles The parameter selection range is:
[0089] Acceleration and Respectively in [0, a Smax ] and [0, a Amax ] is selected according to the probability of Gaussian distribution, and the angular velocity and In [-q Smax ,q Smax ] and [-q Amax ,q Amax ] is uniformly distributed within the range; the angular velocity In [-p Amax , p Amax ] and take values uniformly distributed within the range;
[0090] During the target tracking phase, the particles of the USV platform and particles from AUV platforms The parameter value selection range is:
[0091] USV particles Acceleration In [-a Smax , a Smax ] is selected according to Gaussian distribution, and the angular velocity In [-0.5·q Smax , 0.5·q Smax ] randomly selects values according to uniform distribution within the range; the particles of AUV The acceleration in In [-a Amax , a Amax ] is taken in the positive range according to Gaussian distribution, and the angular velocity In [-q Amax ,q Amax ] according to uniform distribution.
[0092] Furthermore, in the fitness function, the optimization index S(d S ' -A )for
[0093]
[0094] r com,A is the connection radius;
[0095]
[0096] dS-A (k) is the distance between AUV and USV; d com,A It is the threshold set according to the connection radius.
[0097] Furthermore, in the AUV fitness function, when the current distance between the AUV and the target is greater than the AUV search radius r det,A When , the mathematical expression of the distance between the two is as follows:
[0098]
[0099] When the target is already in the AUV’s search domain, the distance between the AUV and the predicted position of the target is used to evaluate the performance of the path.
[0100]
[0101] where k pred =k+Δk is the k+Δkth period in the prediction.
[0102] Beneficial Effects
[0103] The present invention adopts a heterogeneous system of underwater autonomous vehicles (AUVs) and unmanned vehicles (USVs) to perform underwater target search and tracking tasks. The path planning problem of the AUV&USV system when performing cross-domain collaborative search and tracking tasks for underwater targets is explored, and improvements are made to the problems that the PSO algorithm is prone to falling into local optimality and lacks adaptability. Under the requirements of unmanned platform constraints, the cross-domain collaborative search and tracking tasks for underwater targets are completed more efficiently.
[0104] 1. The present invention divides the task into the target search phase and the target tracking phase, establishes a collaborative working mode of the USV&AUV heterogeneous system for underwater targets in the two phases, designs an information communication mechanism between AUV&USV, and proposes a method for predicting the target position in the target tracking phase.
[0105] 2. The present invention improves the traditional particle swarm optimization algorithm based on the online path planning problem of USV&AUV heterogeneous systems. The improved algorithm adopts centralized path planning and distributed path planning according to different mission stages, and uses the acceleration and angular velocity of the platform as particles; the algorithm uses an elite preservation strategy to retain the optimal parent generation of the particle group, introduces inertia weights to control particle velocity properties, and proposes an adaptive learning factor strategy to control the evolution speed of particle positions at different mission stages.
[0106] 3. The present invention establishes algorithm fitness functions for different mission stages. The fitness function of the target search stage comprehensively considers the working platform spacing of the heterogeneous system, the degree of search for new areas with low safety, and whether the target position can be searched; the fitness functions of AUV and USV are set separately in the target tracking stage, where the fitness function of AUV is the distance between itself and the target.
[0107] Simulation experiments show that the AUV&USV system using the improved particle swarm algorithm proposed in the present invention can complete underwater target search and tracking tasks more efficiently.
[0108] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0109] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0110] Figure 1 :Schematic diagram of USV and AUV models;
[0111] Figure 2 : Improved particle swarm optimization algorithm flow chart;
[0112] Figure 3 :Sigmoid function diagram of the distance of USV&AUV;
[0113] Figure 4 : Two-dimensional path diagram of the whole task process based on IPSO algorithm; (a) t = 5s, (b) t = 50s, (c) t = 90s, (d) t = 164s;
[0114] Figure 5 : A three-dimensional path diagram of the entire task process based on the IPSO algorithm;
[0115] Figure 6 :Unmanned platform spacing change diagram of the whole process based on IPSO algorithm;
[0116] Figure 7 : A deep change diagram of the whole process of collaborative path planning for heterogeneous systems;
[0117] Figure 8 : Speed variation diagram of the whole process of collaborative path planning of heterogeneous systems;
[0118] Fig. 9 : Yaw angle variation diagram for collaborative path planning of heterogeneous systems;
[0119] Fig.10 : Pitch angle variation diagram for collaborative path planning of heterogeneous systems. DETAILED DESCRIPTION
[0120] Embodiments of the present invention are described in detail below. The embodiments are exemplary and intended to be used to explain the present invention, but should not be construed as limiting the present invention.
[0121] 1. AUV&USV system search and tracking mission
[0122] The method proposed in this embodiment is mainly used for underwater target search and tracking in offshore waters. The USV and AUV work on the surface and underwater respectively, and search for underwater targets across domains in a collaborative manner. When the USV or AUV detects the position of the underwater target, the search phase ends. The position information of the searched underwater target is transmitted to the AUV, and the AUV completes the subsequent tracking task. During the execution of the search and tracking task, in order to maintain the real-time nature of the USV and AUV track planning, the two need to share information. Therefore, during the target search phase, the distance between the USV and the AUV is always kept within the maximum allowable communication range.
[0123] In the target search phase, the AUV moves before the USV. The USV decides its next action based on the AUV's action to ensure normal communication between the two during the search process. This process continues until the underwater target is detected by the USV or AUV. The USV has a faster search speed than the AUV, so usually the USV detects the underwater target first. However, since the position and movement of the underwater target are unknown, the AUV still has the possibility of discovering the underwater target first. The cross-domain collaborative operation of the two can effectively improve the completion efficiency of the search phase. When the target is detected, the AUV sails towards the target based on the real-time position information of the target obtained and successfully completes the tracking task. For convenience, the situation where the target is detected is divided into situation 1 and situation 2.
[0124] Case 1: The USV detects the underwater target first, transmits the target's real-time position information to the AUV, reduces its own speed, follows the target from a distance, and tries not to be discovered by the target while ensuring that the target is not lost, until the AUV detects the target and completes the tracking task.
[0125] Case 2: The AUV detects the underwater target first, transmits the target's real-time location information to the USV, and continuously shortens the distance to the target to complete the tracking task. After receiving the information from the AUV, the USV quickly moves towards the target until it detects the target, then reduces its own speed and follows the target from a distance, trying not to be discovered by the target as much as possible while ensuring that the target is not lost. At this time, the role of the USV is to transmit more accurate underwater target location information to the AUV in real time by fusing the target information detected by both, until the AUV completes the tracking task.
[0126] 2. System Model
[0127] 1. Underwater target modeling
[0128] The position and attitude vector u of the underwater target T and the velocity vector V T for:
[0129]
[0130] where x T ,y T ,z T is the coordinate of the target in the x, y, and z axes of the spatial coordinate system, θ T is the pitch angle of the target in the vertical plane; is the yaw angle of the target in the horizontal plane; v T is the translation speed of the target in the horizontal plane, p T is the angular velocity of the target around the y-axis, q T is the angular velocity of the target around the z axis, a T Indicates the acceleration of the target.
[0131] According to the conversion relationship between coordinates and speed, the position and attitude vector u of the underwater target can be obtained: T and the velocity vector V T Relationship:
[0132]
[0133] The constraints of underwater targets mainly consider two aspects: the motion space and motion performance of underwater targets.
[0134] Motion space constraints: The underwater target moves in a certain sea area, and a maximum diving depth range is set according to its own pressure bearing capacity.
[0135]
[0136] In this embodiment, z Tmin =-250m is the maximum working depth of the target underwater.
[0137] Performance constraints: The target sets speed and acceleration upper limits based on its own performance range.
[0138]
[0139] The motion state of underwater targets is unknown. During the underwater target tracking phase, the USV can only obtain the current information of the target through its own detection equipment. If the USV only transmits the current position information of the target to the AUV through the towed sonar, the target information grasped by the AUV will lag behind the actual movement of the target. Therefore, the AUV needs to use a reliable method to predict the future position of the underwater target, so that the AUV can go to the position where the target will arrive in advance, so as to complete the task of tracking the underwater target faster.
[0140] The position prediction of underwater targets is completed by USV with stronger computing power. The specific prediction method is as follows:
[0141] Assuming that the current period is the kth period, the coordinate information of the target in the current period is
[0142] ρ T (k) = [x T (k),y T (k),z T (k)] T , the state information in the k-1 cycle is
[0143] The target information obtained by USV is only the coordinate information ρ T .
[0144] Therefore, the USV predicts the state of the target as
[0145]
[0146] Similarly, the target approximate predicted position in the k+Δkth period should be
[0147]
[0148] Where T hunt is the tracking period of the tracking phase. The path planning of the USV&AUV heterogeneous system for the search and tracking of underwater moving targets is a dynamic path planning problem. During the target search process, the USV and AUV move at a fixed time interval T search =T periodically plans the platform trajectory. Since the AUV tracking task requires that the final time interval with the target be less than the set threshold, during the target tracking process, the AUV needs to flexibly adjust its own heading and attitude according to the target's motion trajectory to ensure that the AUV can successfully capture the target under the limitation of its own navigation performance. Continue to maintain the time interval of T search The fixed cycle of planning the AUV path cannot meet the complex requirements of dynamic tracking, so it is necessary to adjust the tracking cycle length T of the target tracking phase. hunt As the distance between the AUV and the target continues to shorten, T huntThe duration of T should also be gradually reduced to match the AUV's approach to the target. hunt The value of is related to the AUV search radius and the distance between the AUV and the target, and is specifically set as follows:
[0149]
[0150] Where, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV.
[0151] In the AUV tracking process, in order to better complete the tracking task, it is necessary to dynamically set the value of Δk according to the distance between the AUV and the current position of the target, so as to quickly change the movement direction of the AUV. And the larger the prediction period, the lower the accuracy of the target prediction position, so it is also necessary to set the threshold of Δk.
[0152] The dynamic value rule of Δk in this embodiment is as follows:
[0153]
[0154] Among them, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV. A-T >r det,A When the AUV is in the process of detecting underwater targets, it can only rely on the information transmitted by the USV to understand the location of the target. At this time, the AUV's task of including the underwater target in its own detectable area has a higher priority than the need to track the underwater target, so it only moves according to the current location information of the underwater target.
[0155] 2. USV modeling
[0156] According to the basic assumptions of USV, USV moves in a two-dimensional plane on the sea surface, and its model is described as follows:
[0157]
[0158] The meanings of the parameters in this motion model are consistent with those described in the previous section. The subscripts represent USV, z S =0m constant; V S is the velocity vector of USV.
[0159] Where the position and attitude vector u of the USV S and the velocity vector V S Relationship:
[0160]
[0161] Constraints:
[0162] In the cross-domain collaborative path planning problem, the constraints of USV are analyzed from three aspects: motion space, own performance and USV&AUV cross-domain collaborative constraints:
[0163] Working space constraint: USV searches for underwater targets in bounded sea areas where underwater targets have appeared. The spatial constraint is:
[0164] x min ≤x S ≤x max ,y min ≤y S ≤y max
[0165] Own performance constraints: Due to the physical property limitations of the USV itself, the USV has a certain speed and acceleration range during navigation. The upper limits of its speed and acceleration are set as follows:
[0166]
[0167] Detection capability constraint: In the cross-domain collaborative search and tracking task, it is assumed that the search range of the USV is a hemispherical shape with the USV mass point as the center, and the radius of the search area is r det,S , when the underwater target appears in the search area of the USV, it is considered that the USV has successfully searched for the underwater target, that is:
[0168]
[0169] Communication constraints: During the search and tracking mission, the USV needs to communicate with the underwater AUV, using a towed communication buoy as a communication relay. It is assumed that the communication range of the USV is a hemispherical shape with the center of mass of the USV as the center of the sphere, and the communication radius is d com(S-A) When the distance between USV and AUV does not exceed d com(S-A) , it is assumed that the two can communicate with each other. The distance formula between USV and AUV is:
[0170]
[0171] 3. AUV Modeling
[0172] When performing the search and tracking mission of underwater targets, the AUV performs three-dimensional non-fixed depth motion in the underwater space. Therefore, the kinematic model of the AUV is the same as the three-dimensional motion model of the underwater target, and its mathematical form is as follows:
[0173]
[0174] V A =[v A ,p A ,q A ,aA ] T
[0175] Constraints
[0176] The constraints on AUVs should be analyzed from the aspects of AUV’s workspace, its own performance and mission completion signs:
[0177] Workspace constraints: According to the AUV's own pressure resistance and search range, its workspace constraints are as follows:
[0178]
[0179] Performance constraints: Due to its own physical performance limitations, the AUV has speed and acceleration constraints during navigation as follows:
[0180]
[0181] Detection capability constraint: AUV works underwater. Assume that the search area of AUV is a sphere with the center of mass of AUV as the center of the search area, and the search radius is r mentioned above. det,A , when the distance between the underwater target and the AUV is less than r det,A When , it is considered that the AUV has successfully searched for the underwater target, that is
[0182]
[0183] Target tracking constraint: The task requirement of the AUV in the underwater target tracking phase is to successfully track the underwater target. When the distance between the AUV and the underwater target is less than a certain threshold d(t E ) A-T When , it is considered that the target tracking task is completed at this time, that is,
[0184]
[0185] where t E This is the end time of the tracking phase.
[0186] Communication constraints: The information interaction formula between AUV and USV is the same as that of USV modeling.
[0187] (III) Optimization indicators
[0188] The goals of the two stages are different. In the underwater target search stage, the total search space should be expanded as much as possible to increase the probability of detecting underwater targets, while the USV and AUV maintain communication. In the underwater target tracking stage, the AUV should minimize the distance to the underwater target to achieve real-time tracking of the underwater target, while the USV needs to take into account the distance to both the AUV and the underwater target.
[0189] The optimization indicators in the target search phase are as follows:
[0190] F = max(S(d′ S-A ),f2,f search )
[0191] Where S(d′ S-A ) is the optimization index of the distance between USV and AUV, f2 is the optimization index of the set search space, and f search It is the basis for judging whether a heterogeneous system has searched for a target.
[0192] In the target tracking phase, the AUV only focuses on the distance between itself and the underwater target, while the USV also needs to consider maintaining communication with the AUV. Therefore, the optimization indicators of the two in the current cycle of the target tracking phase are given separately.
[0193] The optimization indicators of AUV are as follows:
[0194] F A (k) = min(d A-T (k))
[0195] The optimization indicators of USV are as follows:
[0196] F S (k) = min(μd S-A (k)+d S-T (k))
[0197] Among them, d A-T (k) is the distance between the AUV and the underwater target, d S-T (k) is the distance between AUV and USV, d S-T (k) is the distance between the USV and the underwater target. μ is the weight value added according to whether the AUV can currently search for the target. It is a constant less than 1 when the AUV can search for the target, and a constant greater than 1 when the AUV cannot search for the target.
[0198] (IV) Cross-domain collaborative search and tracking path planning algorithm
[0199] 1. Improved particle swarm optimization algorithm
[0200] Compared with traditional optimization algorithms (such as gradient descent), particle swarm optimization can simultaneously evaluate the fitness of multiple particles in the group, has a wide solution coverage, and has good parallel computing capabilities. However, the traditional PSO algorithm also has some problems: the algorithm lacks dynamic adjustment of particle velocity properties, which makes it difficult for particles to control the convergence speed when approaching the optimal solution. If the convergence speed is too fast, the algorithm will easily fall into the trap of local optimality, resulting in low convergence accuracy.
[0201] To solve this problem, this embodiment designs an improved particle swarm optimization algorithm (IPSO), such as Figure 2 As shown in the figure. The algorithm adds an adaptive learning factor control strategy in the iteration, and refers to the optimization idea of the genetic algorithm, and adds an elite preservation strategy in the particle update process. While maintaining the high convergence speed of the PSO algorithm, it ensures that a feasible optimal online path is generated in real time and stably.
[0202] Since the tasks undertaken by the platforms within the USV&AUV heterogeneous system are different during the target search phase and the target tracking phase, the two phases perform path planning independently.
[0203] In the target search phase, the particle swarm size of the USV&AUV heterogeneous system is set to N. Under the premise of information interconnection, the two platforms perform centralized particle swarm algorithm path planning; in the target tracking phase, due to the different task divisions of USV and AUV, in order to facilitate AUV to perform target tracking, the heterogeneous system adopts distributed path planning. In the case of distributed path planning, the particle swarm size of the USV platform and the AUV platform is set to N.
[0204] Accordingly, the particle swarm record of the k-th period heterogeneous system in the particle swarm optimization algorithm is
[0205] X(k)={X 1 (k),…,X j (k),…,X N (k)},j=1,2,…,N
[0206] Where X j (k) represents the jth particle in the kth period particle group.
[0207] Since the periodic path solutions of the USV and AUV platforms are determined by the initial state values, acceleration and angular velocity of the two platforms in this period, the motion control quantities of the two platforms are taken as parameters of the algorithm.
[0208] (1) Target search phase
[0209] Since the USV and AUV are connected most of the time in this stage, the USV&AUV system mainly adopts the overall path planning. The particle dimension of this stage is set to 5, and the parameters are set to
[0210]
[0211] Considering that USV and AUV need to fully search the sea area, the greater the platform speed, the larger the single-cycle search range, so particle X j Linear acceleration in (k=1) and Respectively in [0, a Smax ] and [0, a Amax] is taken according to the probability of Gaussian distribution within the positive range, and the angular velocity and Respectively and [-q Amax ,q Amax ] is uniformly distributed within the range; the angular velocity In [-p Amax , p Amax ] and take values uniformly distributed within the range of .
[0212] (2) Target tracking stage
[0213] The particle parameters of this stage are set as the particles of the USV platform and AUV platforms The underwater target moves at a fixed depth, and the particle dimension of USV and AUV is 2.
[0214] Since the USV and AUV have their initial state values at the beginning of this stage, and considering the requirements of underwater tracking of AUV, the particle swarm of USV Linear acceleration In [-a Smax , a Smax ] is selected according to Gaussian distribution, and the angular velocity In [-0.5·q Smax , 0.5·q Smax ] is randomly selected according to uniform distribution. Linear acceleration in In [-a Amax , a Amax ] is taken in the positive range according to Gaussian distribution, and the angular velocity In [-q Amax ,q Amax ] according to uniform distribution.
[0215] After initializing the particle swarm, calculate the fitness function value corresponding to the current particle swarm and record the fitness function value F(X j (k=1)),j=1,2,…,N, and take it as the best individual in history And record the corresponding particle information At the same time, save the optimal fitness g in the particle swarm best And the corresponding optimal particle X best .
[0216] 2. Improved content
[0217] In the kth cycle, the particle velocity update formula and the corresponding particle position update formula of the particle swarm optimization algorithm designed in this embodiment are as follows:
[0218]
[0219] Where ω is the inertia weight, c1 is the individual learning factor, c2 is the social learning factor, and rand1 and rand2 are random real numbers uniformly distributed in the range [0,1]. is the optimal solution for individual particles; X best The optimal solution for the group.
[0220] When X j When the size of a control quantity in (k) exceeds the performance threshold of the working platform, it is reset to the critical value closest to this value.
[0221] The improvements to the PSO algorithm in this embodiment are as follows:
[0222] (1) Adaptive learning factor control strategy
[0223] The improved particle swarm algorithm introduces the individual learning factor c1 and the social learning factor c2 into the particle speed update formula, where c1 represents the ability of the particle to learn its own historical experience, and c1 can be used to regulate the learning weight of the particle's own experience in the evolution process; c2 represents the ability of the particle to learn in the group, and can be used to regulate the learning weight of the group experience in the evolution process of a single particle. It should be pointed out that if c1 = 0, the algorithm will lose group diversity due to the single reference direction, and may even fall into a local optimum and be unable to jump out; if c2 = 0, the algorithm will converge slowly because the particles cannot share the optimal evolution reference direction, so it is necessary to reasonably adjust the assignment of the two.
[0224] In order to achieve a balance between c1 and c2, the IPSO algorithm adopts an adaptive learning factor control strategy, and the control strategies adopted in the two stages are different.
[0225] Target search phase
[0226] The mathematical expression of the control strategy used in this stage is:
[0227]
[0228] Where a1 and a2 are both non-negative constants, with values between (0,1). max is the upper limit of the duty cycle.
[0229] This control strategy allows c2 to play a leading role in particle updating in the early stage of the IPSO algorithm's search and when the evolution effect of the early group particles is not obvious, promoting the evolution of particles towards the optimal direction of the group and accelerating the group evolution speed; in the middle and late stages of the group search, since the search task is time-consuming but has not yet succeeded, it means that the evolution effect of the social learning factor has decreased. At this time, the influence of c2(k) is reduced, and c1 is allowed to play a leading role in particle updating. Increasing c1(k) increases the diversity of feasible solutions for the group, expands the exploration interval, and enhances the algorithm's optimization performance to explore better feasible solutions.
[0230] Target tracking phase
[0231] In this stage, the position of the target point is known, dynamic tracking path planning has stronger requirements on the convergence of the optimization algorithm, and the population particles already have a certain initial state. The expression of adjusting the learning factor control strategy is as follows:
[0232]
[0233] Compared with the control strategy in the target search phase, c1(k) is adjusted in the target tracking phase, and c1(k) adopts an open incremental weight strategy. This is because the tracking path of the AUV in each cycle has higher requirements for the diversity of the particle group. Therefore, on the basis of the dynamic growth of c1(k) with the cycle, its growth rate is further adjusted, so that each path solution can further enhance its own learning degree in the evolution process, and promote the diversification of group particles under the premise that the path solution in each cycle has the same initial value of the motion state, so as to achieve the effect of balancing diversity and convergence.
[0234] (2) Elite preservation strategy
[0235] The IPSO algorithm refers to the optimization idea of the genetic algorithm and adds an elite preservation strategy to the link of updating the group particles. After every Δk1 cycles, if the heterogeneous system has not completed the current cycle task, the particle with the worst fitness in the group will be replaced by the group's optimal solution to assist in generating a better collaborative path planning solution in the next cycle, thereby enhancing the particle swarm optimization algorithm's optimization ability.
[0236] After the particle group in the kth period completes the update of the velocity and position attributes, the fitness function value of each new particle is calculated. If the fitness value of the contemporary particle is better than the individual historical optimal Then update the particle And update the corresponding individual optimal solution of the particle If a fitness value appears in the contemporary world that is better than the optimal g of the population best The particle swarm optimal g is updated best , and update the corresponding particle to the group optimal solution X bestThe above process is repeated until the heterogeneous system path planning task is completed or the path planning reaches the upper limit of the working cycle.
[0237] 3. Fitness function setting
[0238] (1) Fitness function of heterogeneous systems in the target search phase
[0239] The task of the AUV&USV heterogeneous system in this stage is to successfully search for the target, with the USV as the main search force and the AUV assisting in the search. For the convenience of calculation, the fitness functions of the two are set centrally in this stage. The fitness function mainly considers three aspects: the distance between the AUV and USV platforms, the degree of search of the working area by the heterogeneous system, and whether the target position can be searched.
[0240] Distance between USV & AUV:
[0241] The basis for judging whether the two working platforms of the heterogeneous system are connected in the kth cycle is whether the real-time distance between the two platforms is less than the connection radius r com,A ,Right now
[0242]
[0243] If the distance between the two platforms is greater than r com,A , the two platforms cannot sense each other’s specific location information and lose communication. In this case, the USV will call the control center to find the closest cycle k to the kth cycle time. rec AUV position record A (k rec ), and use it as the reference position of the AUV to estimate the reference distance between the USV and the AUV, and use the reference distance to replace the actual distance between the AUV and the USV in this cycle. As shown in the following formula:
[0244]
[0245] where x A (k rec ),y A (k rec ),z A (k rec ) is the kth rec The position coordinates of the AUV during each cycle.
[0246] When the reference distance value is greater than a certain range, it is considered that the kth rec The AUV record of the cycle loses its reference function. In this case, d S-A (k) is a fixed value significantly larger than the connection radius to illustrate the disconnection state of the two platforms. The selection of the critical value and the assignment value are related to the connection radius of the USV. In this embodiment, the value is taken as dcom,A =2·r com,A . The comprehensive expression of the periodic distance between the two platforms is obtained as follows:
[0247]
[0248] Since the smaller the distance between USV and AUV, the larger the overlapping range of the search domains of the two platforms is, the smaller the corresponding joint search domain space is, and the two are coupled with each other, so we directly set d S ' -A (k) as the fitness part will make the fitness function tend to continuously reduce the distance between the two, resulting in overlapping search domains. However, if the connection between USV and AUV is simply controlled by the yes / no connection state, it will be detrimental to the subsequent tracking phase. Therefore, the distance between USV and AUV obtained by the above formula is processed twice.
[0249] The sigmoid function commonly used in deep learning has the characteristics of continuity and symmetry, can preserve the amplitude characteristics of data while compressing data, and has a value range of [0,1]. Therefore, this chapter introduces the sigmoid function to normalize the distance.
[0250] The sigmoid function expression is as follows:
[0251]
[0252] The sigmoid function formula for the distance of USV&AUV set in this embodiment is:
[0253]
[0254] Its function graph is as follows Figure 3 As shown in the figure, the red curve part is the corresponding function curve when the USV and AUV are in the connected range. It can be seen that as the distance between the two decreases, S(d S ' -A ) will rise rapidly at first, and then gradually slow down after reaching a certain amplitude, so that S(d S ' -A ) as much as possible within [0.5r com,A ,r com,A ] within an appropriate range, so that the connection between the two can remain stable within a certain period.
[0255] (2) Fitness function of heterogeneous systems in the target tracking phase
[0256] The task of this stage is for the AUV in the heterogeneous system to gradually reduce the distance to the target until it captures the target. In this process, the AUV is the main underwater tracking force, and the USV dynamically transmits the real-time position of the target to the AUV. Since the work tasks undertaken by the USV and AUV in this stage are inconsistent, the fitness functions of the two are set separately.
[0257] AUV platform fitness function settings:
[0258] The only task of the AUV in the tracking phase is to complete the tracking of the target, so the criterion for measuring the quality of different path solutions of the AUV in the current k-th period is the distance between the end position of the AUV path in the k-th period and the target. The smaller the distance, the better the path solution.
[0259] Therefore, the fitness function of the AUV is the same as the optimization index in the target search phase. When the current distance between the AUV and the target is greater than the search radius r of the AUV det,A When , the mathematical expression of the distance between the two is as follows:
[0260]
[0261] When the target is already within the search domain of the AUV, the distance between the predicted position of the AUV and the target is used to evaluate the performance of the path.
[0262]
[0263] where k pred =k+Δk is the k+Δkth period in the prediction.
[0264] The corresponding AUV fitness function is:
[0265] F A (k) = min(d A-T (k))
[0266] USV platform fitness function settings:
[0267] The task of the USV in the tracking phase is to transmit the target position information to the AUV throughout the process, so the evaluation criteria for different path solutions of the USV in the kth cycle are: the distance between the end of the path in the kth cycle and the AUV and the distance to the underwater target.
[0268] The terminal communication distance between USV and AUV should not exceed the distance d set in the previous article. com(S-A) , the critical value of the reference distance is adjusted, and the adjusted formula is as follows:
[0269]
[0270] The expression of the distance between the USV and the underwater target at the end of the cycle is as follows:
[0271]
[0272] When the underwater target is not within the search range of the USV, the distance between them is set to a fixed value for the convenience of calculation. The selection of the fixed value is related to the search radius. In this embodiment, the fixed value is taken as d det,S =2·r det,S , so the comprehensive mathematical expression of the distance between the two is as follows:
[0273]
[0274] Since the search radius r of the USV det,S and the communication radius d com(S-A) The sizes of the two distances are different, so in the fitness function, the calculation results of the two distances need to be normalized first to achieve a balance between evaluating the degree of connectivity and the degree of proximity to the target.
[0275] The effect of normalization is as follows:
[0276]
[0277] Easy to get D S-A (k),D S-T (k)∈[0,1], the closer its value is to 1, the closer the distance between USV and AUV is. Therefore, the general formula of USV fitness function in the target tracking stage is:
[0278] F S (k) = min(μD S-A (k)+D S-T (k))
[0279] Simulation experiment verification:
[0280] In this embodiment, a search and tracking simulation test is performed on an underwater target performing two-dimensional fixed-depth motion.
[0281] The motion parameters of the USV platform are: The connection radius of USV & AUV is d com(S-A) =500m, the search radius of USV is r det,S =500m, maximum working speed V Smax =20m / s, maximum acceleration a Smax =3m / s 2 , maximum operating angular velocity q Smax =0.2rad / s.
[0282] The motion parameters of the AUV platform are: The search radius of the AUV is r det,A =300m, maximum working speed V Amax =10m / s, maximum acceleration aAmax =2m / s 2 , maximum operating angular velocity q Amax =0.2rad / s, maximum pitch angular velocity p Amax = 0.2rad / s, AUV uses sonar to achieve real-time tracking of underwater targets, so the tracking target distance threshold d hunt =3m.
[0283] The motion parameters of the underwater target are set as follows: The maximum navigation speed of the underwater target is V Tmax =5m / s, maximum acceleration a Tmax =1m / s 2 , maximum yaw angular velocity q Tmax =0.1rad / s, maximum pitch angular velocity p Tmax =0.1rad / s.
[0284] Since the USV&AUV heterogeneous system has the real-time requirement of searching and tracking underwater moving targets, the algorithm needs to balance the requirements of real-time and search performance. Therefore, the number of particle swarms M of the algorithm is set P =20, the upper limit of the platform working cycle is also set to G max =300, the duration of each search cycle is T = 5s.
[0285] The following simulation test is carried out on the two-dimensional fixed-depth motion model of underwater targets to prove the feasibility and superiority of the unmanned platform and map model established in this paper and the improved path planning algorithm proposed.
[0286] Since the workspace of USV&AUV heterogeneous system is relatively limited, the workspace is set as
[0287] x∈[0,3000],y∈[0,3000]
[0288] The coordinates and status information of USV, AUV and underwater targets at the initial moment are shown in the following table:
[0289] Table 1 Initial state information of USV, AUV and 2D depth-fixed moving target
[0290]
[0291] After setting the above parameters, the USV&AUV heterogeneous system uses the IPSO algorithm to perform collaborative path planning for two-stage tasks. The two-dimensional top view of the target search process and tracking process is shown in the figure below. Figure 4 As shown. The path identification of USV, AUV and target is shown in the legend in the upper right corner. The blue dotted circle is the dynamic search domain of USV, the blue solid circle is the information connection domain of USV&AUV system, and the red dotted circle is the dynamic search domain of AUV.
[0292] Figure 4 (a) to (b) are process diagrams of the target search phase, and (c) to (d) are dynamic processes of the underwater target tracking phase. It can be seen that the two unmanned platforms conduct cross-domain collaborative search in a communicative state, in which the target search phase takes 50 seconds; after entering the underwater target tracking phase, the USV and AUV continue to approach the target position. When the distance between the USV and the underwater target is less than 500m, the USV will reduce its speed and track the underwater target from a distance. With the assistance of the USV's information, the AUV searches for the underwater target at 40 seconds of the mission; the cross-domain collaborative path planning of the USV&AUV heterogeneous system takes 164 seconds in total.
[0293] Figure 5 The figure shows a 3D diagram of the collaborative path planning of a single USV & AUV heterogeneous system for target search and tracking tasks. Figure 4-5 It can be concluded that through the information connection between USV&AUV, the USV&AUV heterogeneous system can quickly complete the search for the target, and the AUV can accurately track and capture the target with the assistance of USV information.
[0294] Figure 6 The figure shows the real-time change of the distance between the USV, AUV and underwater targets during the whole process of the USV&AUV heterogeneous system performing cross-domain collaborative tasks. The black dotted line in the figure is the time boundary between the underwater target search and tracking stages. In the target search stage, the USV first searches for the target and enters the target tracking stage. The USV continuously transmits the location information of the underwater target to the AUV to ensure the effective tracking of the underwater target by the AUV until the AUV successfully searches and tracks the underwater target.
[0295] Table 2 shows other data of the USV&AUV heterogeneous system during the entire process of searching and tracking underwater targets.
[0296] Table 2 Other data of tasks based on IPSO algorithm
[0297]
[0298] Figure 7 The figure shows the depth changes of the AUV and the target during the entire USV&AUV heterogeneous system mission. The magenta solid line in the figure is the depth of the AUV during the target search phase, which shows the process of the AUV gradually approaching the target from a depth perspective.
[0299] Figure 8 , Fig. 9 They are the changes in the navigation speed and yaw angle of the AUV, USV and target during the whole process of the USV&AUV heterogeneous system performing the collaborative task. Fig.10It is the change of the pitch angle of the AUV during the whole process of the collaborative task. As can be seen from the above figure, during the whole process of the task, the changes of the speed, yaw angle and pitch angle of the working platform always remain continuous without sudden changes. Therefore, it can be concluded that the single USV & AUV heterogeneous system collaborative task proposed by the present invention and the designed IPSO algorithm are feasible for practical application.
[0300] All the above experimental results and analysis show that the USV&AUV heterogeneous system can realize the information connection between USV and AUV while meeting the vehicle constraint requirements, and successfully complete the collaborative search for the target and the underwater tracking of the target by AUV in the connected state. This proves that the collaborative task model and improved particle swarm optimization algorithm established in the present invention can solve the collaborative online path planning problem of heterogeneous systems.
[0301] In summary, the USV&AUV heterogeneous system using the IPSO search algorithm can search for targets faster. That is, the improved particle swarm algorithm of the present invention can more effectively solve the cross-domain collaborative path online planning problem of the USV&AUV heterogeneous system compared with the traditional algorithm.
[0302] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and intent of the present invention.
Claims
1. A cross-domain collaborative search and tracking method for underwater targets based on AUV and USV, characterized by: The following steps are involved: Step 1: Establish kinematic models of underwater targets, USVs and AUVs: The kinematic model of underwater target is: The position and attitude vector u of the underwater target T and the velocity vector V T for: where x T ,y T ,z T is the coordinate of the target in the x, y, and z axes of the spatial coordinate system, θ T is the pitch angle of the target in the vertical plane; is the yaw angle of the target in the horizontal plane; v T is the translation speed of the target in the horizontal plane, p T is the angular velocity of the target around the y-axis, q T is the angular velocity of the target around the z axis, a T Indicates the acceleration of the target; The USV kinematic model is: USV position and attitude vector u S and the velocity vector V S for: where x S ,y S ,z S is the coordinate of USV in the x, y, z axis direction of the spatial coordinate system, z S =0; is the yaw angle of USV in the horizontal plane; v S is the translation speed of USV in the horizontal plane, q S is the angular velocity of the USV around the z axis, a S is the acceleration of USV; The AUV kinematic model is: The position and attitude vector u of the AUV A and the velocity vector V A for: where x A ,y A ,z A is the coordinate of AUV in the x, y, z axis direction of the spatial coordinate system, θ A is the pitch angle of the AUV in the longitudinal plane; is the yaw angle of the AUV in the horizontal plane; v A is the translation speed of AUV in the horizontal plane, p A is the angular velocity of the AUV around the y-axis, q A is the angular velocity of the AUV around the z axis, a A represents the acceleration of the AUV; Step 2: The cross-domain heterogeneous system composed of USV and AUV searches and tracks underwater targets: When the USV or AUV detects the position of the underwater target, the search phase ends; the position information of the searched underwater target is transmitted to the AUV, and the AUV completes the subsequent tracking task; during the search and tracking process, the USV and AUV share information; In the process of searching and tracking, based on the kinematic models of underwater targets, USVs and AUVs established in step 1, the motion parameters of the cross-domain heterogeneous system composed of AUVs and USVs for searching and tracking underwater targets are optimized by using an improved particle swarm optimization algorithm to achieve cross-domain collaborative search and tracking of underwater targets: In the target search phase, the particle parameters are The superscript j indicates the jth particle; In the target tracking stage, the particle parameters are set as the particles of the USV platform and particles from AUV platforms The particle velocity update formula and the corresponding particle position update formula of the particle swarm optimization algorithm are: Where ω is the inertia weight, c1 is the individual learning factor, c2 is the social learning factor, rand1 and rand2 are random real numbers uniformly distributed in the range [0,1]; is the optimal solution for individual particles; X best The optimal solution for the group; During the target search phase, Where a1 and a2 are both non-negative constants, ranging from (0,1); G max is the upper limit of the duty cycle; c 1max 、c 2max To set an upper limit; In the target tracking phase, The fitness function of the particle swarm optimization algorithm is: During the target search phase, F=max(S(d′ S-A ),f2,f search ) Where S(d′ S-A ) is the optimization index of the distance between USV and AUV, f2 is the optimization index of the set search space, and f search It is the basis for judging whether the heterogeneous system has searched for the target; In the target tracking phase, The fitness function of AUV is: F A (k)=min(d A-T (k)) The fitness function of USV is: F S (k)=min(μd S-A (k)+d S-T (k)) Among them, d A-T (k) is the distance between the AUV and the underwater target, d S-T (k) is the distance between AUV and USV, d S-T (k) is the distance between the USV and the underwater target; μ is the weight value added according to whether the AUV can currently search for the target; it is a constant less than 1 when the AUV can search for the target, and it is a constant greater than 1 when the AUV cannot search for the target.
2. According to claim 1, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In the kinematic model of step 1, the model constraints of the underwater target are: Movement space constraints: underwater targets move within a set sea area, and a maximum diving depth range is set according to their own pressure bearing capacity; z Tmin The maximum operating depth of the target underwater; Performance constraints: The target sets speed and acceleration limits based on its own performance range: The model constraints of USV are: Working space constraints: USV searches for underwater targets in bounded sea areas where underwater targets have appeared. The spatial constraints are: x min ≤x S ≤x max ,and min ≤y S ≤y max Self-performance constraints: During navigation, the USV sets its speed and acceleration limits according to its own physical properties as follows: Detection capability constraints: In the cross-domain collaborative search and tracking task, the search range of the USV is in the shape of a hemisphere with the USV mass point as the center, and the radius of the search area is r det,S , when the underwater target appears in the search area of the USV, it is considered that the USV has successfully searched for the underwater target, that is: Communication constraints: During the search and tracking mission, the USV communicates with the underwater AUV. Assume that the communication range of the USV is a hemispherical shape with the center of mass of the USV as the center of the sphere, and the communication radius is d com(S-A) ; When the distance between USV and AUV does not exceed d com(S-A) , they thought that the two could communicate with each other; The model constraints of AUV are: Workspace constraints: According to the AUV's own pressure resistance and search range, its workspace constraints are as follows: Performance constraints: Due to its own physical performance limitations, the AUV has speed and acceleration constraints during navigation as follows: Detection capability constraint: AUV works underwater. Assume that the search area of AUV is a sphere with the center of mass of AUV as the center of the search area, and the search radius is r. det,A , when the distance between the underwater target and the AUV is less than r det,A When , it is considered that the AUV has successfully searched for the underwater target, that is Target tracking constraint: The task requirement of the AUV in the underwater target tracking phase is to successfully track the underwater target. When the distance between the AUV and the underwater target is less than a certain threshold d(t E ) A-T When , it is considered that the target tracking task is completed at this time, that is, where t E is the end time of the tracking phase; Communication constraint: When the distance between USV and AUV does not exceed d com(S-A) , they thought the two could communicate with each other.
3. According to claim 1, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In step 2, the process of searching and tracking underwater targets by the cross-domain heterogeneous system composed of USV and AUV can be divided into two cases: Case 1: The USV detects the underwater target first and transmits the target's real-time location information to the AUV until the AUV detects the target and completes the tracking task; Case 2: The AUV detects the underwater target first, transmits the target's real-time location information to the USV, and continuously shortens the distance to the target to complete the tracking task. After receiving the information from the AUV, the USV moves closer to the target. The USV fuses the target information detected by both and transmits more accurate underwater target location information to the AUV in real time until the AUV completes the tracking task.
4. According to claim 3, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In step 2, during the process of searching and tracking underwater targets by the cross-domain heterogeneous system composed of USV and AUV, the USV also predicts the future position of the target through the following process and transmits it to the AUV; The prediction process is: Assume that the current period is the kth period, and the coordinate information of the target in the current period is ρ T (k) = [x T (k),y T (k),z T (k)] T , the state information in the k-1 cycle is The target information obtained by USV is only the coordinate information ρ T ; Then the USV predicts the state of the target as The target approximate predicted position in the k+Δkth period should be Where T hunt It is the tracking cycle of the tracking phase.
5. According to claim 4, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: T hunt The value of is related to the AUV search radius and the distance between the AUV and the target, and is specifically set as follows: Where, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV.
6. According to claim 4, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: The dynamic value rules of Δk are as follows: Among them, d A-T is the distance between AUV and underwater target, r det,A is the detection radius of AUV.
7. According to claim 1, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In the target search phase, the particles The parameter selection range is: Acceleration and Respectively in [0, a Smax ] and [0, a Amax ] is selected according to the probability of Gaussian distribution, and the angular velocity and In [-q Smax ,q Smax ] and [-q Amax ,q Amax ] is uniformly distributed within the range; the angular velocity In [-p Amax , p Amax ] and take values uniformly distributed within the range; During the target tracking phase, the particles of the USV platform and particles from AUV platforms The parameter value selection range is: USV particles Acceleration In [-a Smax , a Smax ] is selected according to Gaussian distribution, and the angular velocity In [-0.5~q Smax , 0.5~q Smax ] randomly selects values according to uniform distribution within the range; the particles of AUV The acceleration in In [-a Amax , a Amax ] is taken in the positive range according to Gaussian distribution, and the angular velocity In [-q Amax ,q Amax ] according to uniform distribution.
8. According to claim 1, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In the fitness function, the optimization index S(d S ' -A )for r com,A is the connection radius; d S-A (k) is the distance between AUV and USV; d com,A It is the threshold set according to the connection radius.
9. According to claim 1, a cross-domain collaborative search and tracking method for underwater targets based on AUV and USV is characterized in that: In the AUV fitness function, when the current distance between the AUV and the target is greater than the AUV search radius r det,A When , the mathematical expression of the distance between the two is as follows: When the target is already in the AUV’s search domain, the distance between the AUV and the target’s predicted position is used to evaluate the performance of the path. where k pred =k+Δk is the k+Δkth period in the prediction.
Citation Information
Cited By
Air-sea cross-domain unmanned intelligent cluster cooperative path planning method
CN119717888A
Communication scheduling method and device for cooperative encircling of underwater clusters
CN120075235A
Cross-domain multi-agent collaborative search method for dynamic marine environment
CN120494055A
Underwater target searching and tracking cooperative path planning method based on AUV and multiple USVs
CN122306090A
Underwater target search and tracking cooperative path planning method based on AUV and multiple USVs
CN122306090B
