UUV path planning method in complex environment based on target attraction rule adaptive particle swarm optimization algorithm
By combining the target attraction law and adaptive particle swarm optimization algorithm in UUV path planning and combining multiple environmental models for multi-threat factor evaluation, the problem of single consideration of threat factors in UUV path planning in complex marine environments is solved, and a more comprehensive approach is achieved path evaluation and optimization improve path smoothness and security.
Patent Information
- Application Number
- CN202410917404.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-12
AI Technical Summary
The existing UUV path planning methods mainly consider a single threat factor, and it is difficult to effectively evaluate multiple threat factors in complex marine environments, resulting in insufficient path planning, easy to fall into local optimality and lack sufficient exploration and utilization capabilities.
Adaptive particle swarm optimization algorithm based on the goal attraction law, combined with submarine peaks, currents and underwater sonar scanning models, multi-threat factor evaluation is carried out through environmental simulation and Bayesian network, and path planning is optimized to ensure path smoothness. Safety and efficiency.
Find the optimal path for UUVs that meet yaw angle, pitch angle constraints, shortest path length and least threat in complex marine environments, effectively avoid obstacles, improve the comprehensiveness and efficiency of path planning, and avoid local optimality question.
Smart Images

Figure CN118936468A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of underwater unmanned vehicle path planning, and in particular relates to a UUV multi-objective optimization path planning method in a complex environment. Background Art
[0002] Underwater unmanned vehicles (UUVs) are widely used in marine research because of their energy independence and highly autonomous mission execution capabilities. UUV path planning is an important technology to improve its intelligence and autonomy level, and it is also a prerequisite for its successful completion of various operational missions. However, UUVs are often affected by complex and multiple threat factors during navigation, making it difficult to achieve fully autonomous navigation. Therefore, it is of great significance to study threat assessment and path planning strategies based on multiple threat factors in complex marine environments.
[0003] At present, in the existing research at home and abroad, more attention is paid to the impact of a single threat factor on UUV navigation, while there are fewer UUV path planning problems that consider multiple threat factors and conduct threat assessment. Therefore, an effective solution is needed for the UUV path planning problem in complex environments. Summary of the invention
[0004] The main purpose of the present invention is to solve the problem that the traditional UUV path planning method only considers a single threat factor.
[0005] The UUV path planning method in complex environment based on the target attraction law adaptive particle swarm optimization algorithm includes:
[0006] S100, obtaining an environment simulation model based on ocean environment modeling; the environment simulation model at least includes a submarine mountain model and an ocean current model;
[0007] Based on the ocean environment, the threat level corresponding to the environmental simulation model is obtained:
[0008] The danger distance is determined according to the distance between the UUV and the seabed mountain and used as the threat degree of the seabed mountain obstacle, that is, the threat degree of the seabed mountain obstacle:
[0009]
[0010] In the submarine mountain model, each mountain is equivalent to a cylinder, d is the distance from the path point to the center of the cylinder, and R i′ represents the radius of the i′th mountain obstacle, D is the radius of the UUV; m is the distance threshold of the threat area; f(s i,j ) represents the jth path point s of the i-th path i,j The threat level of submarine mountain obstacles;
[0011] Determine the ocean current velocity V based on the ocean current model c (s i,j ) and the ocean current velocity V c (s i,j ) corresponding to the velocity range as the threat degree;
[0012] S200, determining the corresponding path point set p on each path from the path starting point to the path end point i =(s i,1 ,s i,2 , ..., s i,n ), s i,1 ,s i,2 , ..., s i,n represents the path points constituting the i-th path; based on the environmental simulation model and its threat degree obtained in S100, the specific threat degree corresponding to the environmental threat factor of each path point is determined, and then the threat degree cost at the path point is obtained, and then the total path cost of the smooth path is determined by combining the yaw angle and pitch angle costs and the path length cost;
[0013] S300, solving the minimum optimization problem of the total path cost in step S200 based on the target attraction law adaptive particle swarm optimization algorithm to obtain the optimal path.
[0014] Furthermore, the submarine mountain model is as follows:
[0015]
[0016] Where Z(x, y) represents the seabed height at the UUV position vector (x, y) in geodetic coordinates. i′ ,y i′ ) represents the center coordinate of the i′th submarine mountain model; h i′ Represents the terrain parameter information that controls the height of the submarine mountain model; x si′ and si′ They represent the attenuation degree of the i′th submarine mountain model along the x and y directions, and also control the slope of the submarine mountain model; n m Represents the number of peaks in the entire submarine mountain model.
[0017] Furthermore, the ocean current model is obtained by superimposing multiple Lamb vortices; the Lamb vortex model is as follows:
[0018]
[0019] in, and Respectively represent the ocean current speed in the x direction, y direction and vertical direction in the geodetic coordinate system, V c (s i,j) represents the jth path point s of the i-th path i,j The current speed at represents the current position of the geodetic coordinate system UUV, (x, y) represents the position vector of the UUV in the geodetic coordinate system XOY plane; (x0, y0) represents the position vector of the ocean current in the geodetic coordinate system XOY plane, represents the center of the ocean current, Γ and δ represent the intensity of the ocean current and the radius of the ocean current respectively.
[0020] Furthermore, the environment simulation model also includes an underwater sonar scanning model. Based on the underwater sonar scanning model, according to the path point s i,j The shortest distance to the sonar scanning area determines the threat level of the underwater sonar scanning model;
[0021] The underwater sonar scanning model is as follows:
[0022]
[0023] Among them, S th represents the sonar scanning area, P(X, Y, Z) represents a point in the three-dimensional space area S, that is, any point in the sonar scanning area; X, Y, Z represent the coordinates in the three-dimensional space in the three-dimensional plane, (X0, Y0, Z0) represent the center coordinates of the ellipsoid, a, b represent the lengths of the major axis and minor axis of the ellipse, and R represents the radius of the sonar scanning area;
[0024] The threat level of underwater sonar scanning model is as follows:
[0025]
[0026] Among them, d s Represents the path point s i,j The shortest distance to the sonar scanning area, (x i,j ,y i,j , z i,j ) represents the path point s i,j Position coordinates in the geodetic coordinate system, R t represents the maximum threat threshold of the sonar scanning area, D represents the radius of the UUV, and ε represents the threat threshold of the dangerous area; p(s i,j ) represents the jth path point s of the i-th path i,j Underwater sonar scanning model threat level.
[0027] Furthermore, the specific process of S200 includes the following steps:
[0028] S2001, in the geodetic coordinate system, the path starting point s start The coordinates are (x0, y0, z0), and the end point of the path is s end The coordinates are (xe ,y e , z e ), generate the path from the starting point s start To the end point of the path end A path consists of a set of path points p i =(s i,1 ,s i,2 , ..., s i,n )composition;
[0029] S2003. Connect all the path points in the geodetic coordinate system to form a navigation path for the UUV.
[0030] S2004, using cubic spline fitting scatter points to smooth the path in S2003 to obtain a smooth path;
[0031] S2005. Optimize the UUV smooth path obtained in S2004:
[0032] 1) Calculate the yaw and pitch angle costs F1(p i );
[0033] 2) Calculate the path length cost F2(p i );
[0034] 3) For the path points on the smooth path, the specific threat level corresponding to the environmental threat factor of each path point is determined based on the environmental simulation model obtained in S100 and its corresponding threat level, and then the path point s is obtained. i,j The threat cost TL(p i ); the environmental threat factors correspond one to one with the specific models in the environmental simulation model;
[0035] 4) Calculate the total path cost of the smoothed path obtained by S2004:
[0036] L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ) (15)
[0037] Among them, L(p i ) represents the path cost of the effective path, p i =(s i,1 ,s i,2 , ..., s i,n ) represents the i-th path of UUV; s i,1 ,s i,2 , ..., s i,n represents the path point on the i-th path, si,1 =s start Indicates the starting point of the path; s i,n =s end represents the end point of the path; α1, α2, α3 represent the weights.
[0038] Furthermore, the yaw angle and pitch angle cost F1(p i ) process includes:
[0039] s′ i,j , s′ i,j+1 and s′ i,j+2 Respectively represent the jth, (j+1)th and (j+2)th path points s of the i-th path i,j 、s i,j+1 and i,j+2 Projection point of O-xy plane in the geodetic coordinate system; yaw angle Φ i,j Indicates the path With path The angle between the projection paths of the O-xy plane in the geodetic coordinate system; the pitch angle Ψ i,j+1 Indicates the path With path The climbing angle between
[0040] Assumptions are the unit vectors in the x, y, and z directions in the geodetic coordinates respectively. From the concept of vector projection, we can know that the calculation formula of the projection path is:
[0041]
[0042] Therefore, the yaw angle Φ of the UUV during the induced path i,j It can be given by the following formula:
[0043]
[0044] The pitch angle Ψ of the UUV during the induced path i,j+1 It can be given by the following formula:
[0045]
[0046] The yaw and pitch costs for a smoothed path are:
[0047]
[0048] Where μ1 and μ2 represent the mechanism penalty constants related to the yaw angle and pitch angle, respectively, and p i =(s1, s2, ..., s n ) represents the i-th path of UUV, and n represents the path p iThe total number of path points in .
[0049] Furthermore, the path length cost F2(p i ) process includes:
[0050]
[0051] Among them, (x i,j+1 ,y i,j+1 , z i,j+1 ) represents the j+1th position of UUV; (x i,j ,y i,j , z i,j ) represents the jth position of UUV; p i =(s i,1 ,s i,2 , ..., s i,n ) represents the i-th path of UUV, and n represents the path p i The total number of path points in .
[0052] Furthermore, the specific process of step 3) includes:
[0053] First, the specific threat level value corresponding to the environmental simulation model of each path point is determined based on the threat level allocation strategy; the threat level allocation strategy includes a seabed terrain threat level allocation strategy and a current threat level allocation strategy;
[0054] Seafloor topography threat level assignment strategy:
[0055] When the waypoint s i,j The distance to the peak is d>R i′ +D+m, path point s i,j The threat level of the submarine mountain is f(s i,j )=0, indicating that the threat level of the mountain is low threat, and the threat level value is expressed as 'α1';
[0056] When the waypoint s i,j The distance to the peak is R i′ +D<d≤R i′ +D+m, path point s i,j The threat level of the submarine mountain is f(s i,j )=R i′ +D+md, indicating that the threat level of the mountain is medium threat, and the threat level value is expressed as 'α2';
[0057] When the waypoint s i,j The distance to the peak is d≤R i′ +D, waypoint s i,j The threat level of the submarine mountain is f(s i,j)=∞, indicating that the threat level of the mountain is high, and the threat level value is expressed as 'α3';
[0058] 'α1', 'α2', and 'α3' are the set values respectively;
[0059] Strategy for assigning ocean current threat levels:
[0060] When the waypoint s i,j The ocean current velocity is V c (s i,j )≤the first threshold, the path point s i,j The threat level of the current is low threat, and the threat value is expressed as 'β1';
[0061] When the waypoint s i,j The first flow velocity threshold at <V c (s i,j )≤ the second threshold, the path point s i,j The threat level of the current is medium threat, and the threat value is expressed as 'β2';
[0062] When the waypoint s i,j Flow velocity V c (s i,j )>the second threshold, the path point s i,j The threat level of the current is high, and the threat value is expressed as 'β3';
[0063] 'β1', 'β2', and 'β3' are the set values respectively;
[0064] When the environment simulation model further includes an underwater sonar scanning model, the threat level allocation strategy further includes a sonar scanning threat level allocation strategy, and the sonar scanning threat level allocation strategy is as follows:
[0065] When the waypoint s i,j The shortest distance to the sonar scanning area is d s >R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=0, indicating that the threat level of the underwater sonar scanning area is low threat, and the threat level value is expressed as 'γ1';
[0066] When the waypoint s i,j The shortest distance to the sonar scanning area is R t +D <d s <R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=R t+D, indicating that the threat level in the underwater sonar scanning area is medium, and the threat level value is expressed as 'γ2';
[0067] When the waypoint s i,j The shortest distance to the sonar scanning area is d s ≤R t +D, waypoint s i,j The threat level of the underwater sonar scanning area is p(s i,j )=∞, indicating that the threat level of the underwater sonar scanning area is high, and the threat level value is expressed as 'γ3';
[0068] 'γ1', 'γ2', and 'γ3' are the set values respectively;
[0069] Then, based on the specific threat value corresponding to the environmental simulation model of each path point, the path point s is obtained by machine learning. i,j The threat cost tl(s i,j ), and then finally determine the path point s i,j The threat cost TL(p i ):
[0070]
[0071] Among them, TL(p i ) represents the path p i The threat cost, TL(p i )0 represents the initial threat cost; tl(s i,j ) and tl′(s i,j ) represents the current path point s of the UUV i,j The threat cost of Z is β, which is a constant much larger than the theoretical path length. i,j Indicates that at the path point s i,j =(x i,j ,y i,j , z i,j ) at the UUV navigation altitude; Z(x i,j ,y i,j ) means in (x i,j ,y i,j ), n represents the height of the seabed terrain obstacle at the path p i The total number of path points in .
[0072] Furthermore, the path point s is obtained by machine learning. i,j The threat cost tl(s i,j ) is implemented based on a Bayesian network for predicting threat level cost. The Bayesian network for predicting threat level cost is pre-built. The process of building a Bayesian network for predicting threat level cost includes the following steps:
[0073] Determine threat factors according to the UUV mission execution and navigation requirements, including environmental threat factors and UUV platform status; wherein the environmental threat factors at least include seabed topography and ocean currents, and the UUV platform status at least includes UUV energy and UUV platform posture; when the threat level allocation strategy also includes a sonar scanning threat level allocation strategy, the environmental threat factors also include the opponent's sonar scanning area; determine the environmental threat factors based on the one-to-one correspondence of each specific model in the environmental simulation model, and determine the specific threat level value corresponding to the environmental threat factor according to the threat level allocation strategy; and obtain the specific threat level value corresponding to the UUV platform factor; and assign the path point s i,j The threat degree values of the threat factors at the locations are used as evidence of the Bayesian network model to construct the Bayesian network model.
[0074] Furthermore, the specific process of step S300 includes the following steps:
[0075] S3001. Initialize algorithm parameters: population size N, maximum number of iterations iterl max , maximum inertia weight ω max , minimum inertia weight ω min ;
[0076] The initial global optimal cost in the initial population is inf;
[0077] Obtain individual and global optimal solutions;
[0078] S3002, initialize the path point s of each individual in the population ij , generate a set of valid path points p′ i ; Let Q S′ Search space for the environment, if the path point s ij ∈Q S′ , then s ij is a valid waypoint; if the waypoint Then ij is an invalid waypoint;
[0079] The set of valid path points from the path starting point to the path ending point is denoted as p′ i :
[0080] p′ i =(s i1 ,s i2 , ..., s ij , ..., s in )#(16)
[0081] Among them, p′ i Represents the set of valid path points, that is, the valid path; s ijRepresents a valid path point; i represents the i-th path in the UUV path planning, and also represents the i-th individual of the particle swarm in the adaptive particle swarm optimization algorithm based on the target attraction law; j represents the j-th dimension, corresponding to the j-th path point;
[0082] S3003, using the formula L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ), the valid path point set p′ i Instead of p i , calculate the path cost of the effective path generated by the ith individual after the effective path point set is optimized; compare the path cost of the effective path corresponding to the ith individual with the global optimal cost in the population, which is inf:
[0083] If the path cost of the effective path corresponding to the i-th individual is less than the global optimal cost in the population, which is inf, then the effective path corresponding to the i-th individual is selected as the global optimal path, and the path cost of the effective path corresponding to the i-th individual is recorded;
[0084] Traverse each individual in the population to find the effective path corresponding to the individual minimum path cost ParticlesBest as the global optimal path, and update the global optimal solution GlobalBest;
[0085] S3004, set the number of iterations to iterl, and set iterl = 1;
[0086] Update the speed and position of the UUV at each moment, that is, update the valid path point set p′ i :
[0087]
[0088] Among them, V ij (k′+1) represents the speed of UUV at the k′+1th moment; V ij (k′) represents the speed of the UUV at the k′th moment; ωα represents the nonlinear adaptive weight, which is used to inherit the speed of the UUV at the previous moment and balance the global and local search capabilities of the algorithm; c1 represents the individual cognitive learning factor, which is used to balance the learning factor of the individual optimal value; c2 represents the social cognitive learning factor, which is used to balance the learning factor of the global optimal value; c3 represents the attraction learning factor, which is used to balance the learning factor of the attraction of the target endpoint; r1, r2 and r3 represent random numbers between 0 and 1; P besti (k′) represents the optimal position of the UUV in a moving direction at the k′th moment; X ij (k′) represents the position of the UUV at the k′th moment; g best(k′) represents the optimal position of the UUV in all movement directions at the k′th moment; X ij (k′+1) represents the position of UUV at the k′+1th moment; f attract represents the target attraction weight; represents the attractiveness of the target terminal position; j represents the dimension; i represents the i-th individual;
[0089] Optionally, the inertia weight factor is expressed as:
[0090]
[0091] Among them, ω max represents the maximum inertia weight; ω min It represents the minimum inertia weight; iterl represents the current number of iterations; iterl max Indicates the maximum number of iterations; δ = θ * rand represents the nonlinear radian value that changes with time, rand represents a random number from 0 to 1; θ = π / 2;
[0092] The expression of the target attraction weight is:
[0093]
[0094] Where A represents the magnitude of the target attraction weight; ξ represents the standard deviation of the target attraction weight; It represents the attraction between the current position of the UUV and the target end position;
[0095] S3005, based on the updating strategy of the UUV speed and position at each moment in S3004, the updated valid path point set in S3004 replaces the original valid path point set;
[0096] S3006, using the formula L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ), the new valid path point set p′ in S3005 i Instead of p i , calculate the path cost of the new set of valid path points;
[0097] If the new valid path point set p′ in S3005 i If the path cost is less than the individual optimal path cost in S3003, the individual optimal path and path cost are updated; at the same time, if the updated individual optimal path cost is less than the global optimal path cost in S3003, the path of the new valid path point set is selected as the global optimal path GlobalBest, and the path cost of the new valid path point set is recorded;
[0098] If the path cost of the new valid path point set in S3005 is greater than or equal to the individual optimal path cost in S3003, the global optimal solution GlobalBest determined in S3003 continues to be used as the new global optimal solution GlobalBest;
[0099] S3007, determine whether the number of iterations iterl is greater than the maximum number of iterations iterl max ;
[0100] If the number of iterations iterl is less than the maximum number of iterations iterl max , let iterl=iterl+1, and execute S3004;
[0101] If the number of iterations is iterl greater than the maximum number of iterations iterl max , output the global optimal path.
[0102] Beneficial effects:
[0103] The purpose of the present invention is to design an optimal path for a UUV that satisfies the constraints of yaw angle and pitch angle, has the shortest path length, minimizes the threat index and can effectively avoid obstacles under known target point coordinates and complex ocean environment information.
[0104] The present invention firstly considers the threat of seabed mountain obstacles and the threat of ocean current model into the environmental simulation model, so the present invention considers the seabed environment more comprehensively. On this basis, the threat of sonar scanning model can also be taken into account, and then the path planning is carried out for the various threat factors of UUV, and a more comprehensive evaluation is carried out on the various threat factors affecting the navigation of UUV, so as to ensure that the planned path has the least threat. At the same time, the present invention also uses the adaptive particle swarm optimization algorithm based on the target attraction law to optimize the path planning problem, and the nonlinear adaptive weight effectively improves the exploration and utilization ability of the algorithm and avoids the algorithm from falling into the local optimum; considering the target attraction law of the target end point, the attraction of the target end point is added to the speed and position update of UUV, which improves the working efficiency of UUV. The present invention solves the problem of single threat factor consideration in the process of UUV path planning in complex marine environment, and can also solve the problems that the existing path planning is easy to fall into the local optimum and lacks sufficient exploration and utilization ability. BRIEF DESCRIPTION OF THE DRAWINGS
[0105] Figure 1 This is a schematic diagram of the UUV path planning process in a complex environment based on an adaptive particle swarm optimization algorithm based on the target attraction law.
[0106] Figure 2Schematic diagram of UUV yaw angle and pitch angle.
[0107] Figure 3 It is a threat assessment model based on Bayesian network.
[0108] Figure 4 It is a 3D path planning graph. DETAILED DESCRIPTION
[0109] The present invention establishes a threat assessment model for threat factors that affect the safe navigation of UUVs based on Bayesian networks, so as to achieve the purpose of comprehensively considering threat factors; an adaptive particle swarm optimization algorithm based on the target attraction law is used to optimize the constraints of the yaw angle and pitch angle of the UUV, the path length, and the threat index. In the process of using the adaptive particle swarm optimization algorithm based on the target attraction law, nonlinear adaptive weights are used to achieve the ability to balance global search and local search; the target attraction law is used to improve the working efficiency of the algorithm, and then the optimal solution for the path is obtained. The following is an explanation in conjunction with a specific implementation method.
[0110] Specific implementation method 1: Combination Figure 1 To explain this embodiment,
[0111] This implementation is a UUV path planning method in a complex environment based on a target attraction law adaptive particle swarm optimization algorithm. The specific process includes:
[0112] Step 1: According to the complex marine environment in which the UUV is located when performing the mission, the marine environment is modeled to obtain an environmental simulation model; the specific process includes the following steps:
[0113] Step 1.1, according to the ocean environment in which the UUV is performing its mission, use the exponential function method to model the seabed terrain obstacles and determine the seabed mountain model and threat level;
[0114] Undersea mountain model:
[0115]
[0116] Where Z(x, y) represents the seabed height at the UUV position vector (x, y) in geodetic coordinates. i′ ,y i′ ) represents the center coordinate of the i′th submarine mountain model; h i′ Represents the terrain parameter information that controls the height of the submarine mountain model; x si′ and si′ They represent the attenuation degree of the i′th submarine mountain model along the x and y directions, and also control the slope of the submarine mountain model; n m Represents the number of peaks in the entire submarine mountain model.
[0117] Each mountain in the submarine mountain model can be approximately equivalent to a cylinder. The distance to the submarine mountain (collision area) is defined as the danger distance m, and the jth path point s of the i-th path is i,j The threat level of submarine mountain obstacles can be defined as follows:
[0118]
[0119] In the submarine mountain model, each mountain can be approximately equivalent to a cylinder, d is the distance from the path point to the center of the cylinder, R i′ represents the radius of the i′th mountain obstacle, D is the radius of the UUV, and m represents the distance threshold of the danger zone;
[0120] Step 1.2, using multiple Lamb eddies to superimpose an ocean current model;
[0121] The mathematical model of Lamb vortex is:
[0122]
[0123] in, and Respectively represent the ocean current speed in the x direction, y direction and vertical direction in the geodetic coordinate system, V c (s i,j ) represents the jth path point s of the i-th path i,j The current speed at represents the current position of the geodetic coordinate system UUV, (x, y) represents the position vector of the UUV in the geodetic coordinate system XOY plane; (x0, y0) represents the position vector of the ocean current in the geodetic coordinate system XOY plane, represents the current center position (current center coordinates), Γ and δ represent the current intensity and current radius respectively. When considering the current velocity, the vertical component can be ignored, so the vertical velocity component is set to 0.
[0124] The threat level is determined by the velocity range corresponding to the ocean current velocity;
[0125] Step 1.3, using an ellipsoid to approximate the sonar scanning area to determine the underwater sonar scanning model and threat level;
[0126] The mathematical model is expressed as:
[0127]
[0128] Among them, S threpresents the sonar scanning area, P(X, Y, Z) represents a point in the three-dimensional space area S, that is, any point in the sonar scanning area. X, Y, Z represent the coordinates in the three-dimensional space in the three-dimensional plane, (X0, Y0, Z0) represent the center coordinates of the ellipsoid, a, b represent the lengths of the major and minor axes of the ellipse, and R represents the radius of the sonar scanning area.
[0129] The jth path point s of the i-th path i,j The threat level of the underwater sonar scanning model can be defined as follows:
[0130]
[0131] Among them, d s Represents the path point s i,j The shortest distance to the sonar scanning area, (x i,j ,y i,j , z i,j ) represents the path point s i,j Position coordinates in the geodetic coordinate system, R t represents the maximum threat threshold of the sonar scanning area, D represents the radius of the UUV, and ε represents the threat threshold of the dangerous area.
[0132] It should be noted that in the process of constructing the environmental simulation model, the present embodiment constructs the submarine mountain model, the ocean current model and the underwater sonar scanning model, but in reality, only the submarine mountain model and the ocean current model can be constructed, and the subsequent steps can be performed. In the process of calculating the total path cost, the Bayesian network model can be used, and the threat cost of the underwater sonar scanning model can be ignored when constructing the Bayesian network model. Of course, in the process of constructing the environmental simulation model, other factors can be considered according to the actual situation to construct the corresponding model, determine the corresponding threat cost and perform the subsequent steps. In the process of calculating the total path cost, the Bayesian network model can be used, and the threat cost of the corresponding model can be considered when constructing the Bayesian network model.
[0133] Step 2: Determine the corresponding path point set p on each path from the path starting point to the path end point i =(s i,1 ,s i,2 , ..., s i,n ), s i,1 ,s i,2 , ..., s i,n represents the path points constituting the i-th path; based on the environment simulation model obtained in S100 and its corresponding threat degree, the specific threat degree corresponding to the environment simulation model of each path point is determined, and then the threat degree cost at the path point is obtained, and then the total path cost of the smooth path is determined by combining the yaw angle and pitch angle costs and the path length cost; the specific process includes the following steps:
[0134] Step 2.1: Known path starting point s start , path end s end , UUV starts from the path starting point s start Start from, go to the end of the path end End. In the geodetic coordinate system, the path starting point s start The coordinates are (X0, y0, z0), the end point of the path is s end The coordinates are (x e ,y e , z e ), generate the path from the starting point s start To the end point of the path end A path path consists of a set of path points p i =(s i,1 ,s i,2 , ..., s i,n )composition.
[0135] Step 2.3, in the geodetic coordinate system, connect all the path points and form the navigation path of the UUV;
[0136] Step 2.4, use cubic spline fitting scatter points to smooth the path in step 2.3 to obtain a smooth path;
[0137] Step 2.5: Optimize the UUV smooth path obtained in step 2.4 so that the UUV path satisfies the path cost constraint. The specific process is as follows:
[0138] 1) Calculate the yaw and pitch angle costs F1(p i ):
[0139] like Figure 2 As shown, s′ i,j , s′ i,j+1 and s′ i,j+2 Respectively represent the jth, (j+1)th and (j+2)th path points s of the i-th path i,j , S i,j+1 and S i,j+2 Projection point of O-xy plane in the geodetic coordinate system; yaw angle Φ i,j Indicates the path With path The angle between the projection paths of the O-xy plane in the geodetic coordinate system; the pitch angle Ψ i,j+1 Indicates the path With path The climbing angle between.
[0140] Assumptions are the unit vectors in the x, y, and z directions in the geodetic coordinates respectively. From the concept of vector projection, we can know that the calculation formula of the projection path is:
[0141]
[0142] Therefore, the yaw angle Φ of the UUV during the induced path i,j It can be given by the following formula:
[0143]
[0144] The pitch angle Ψ of the UUV during the induced path i,j+1 It can be given by the following formula:
[0145]
[0146] The path smoothness constraint cost function is:
[0147]
[0148] Where μ1 and μ2 represent the mechanism penalty constants related to the yaw angle and pitch angle, respectively, and p i =(s1, s2, ..., s n ) represents the i-th path of UUV, and n represents the path p i The total number of path points in .
[0149] 2) Calculate the path length cost F2(p2) of the smooth path obtained in step 2.4 i ):
[0150]
[0151] Among them, (x i,j+1 ,y i,j+1 , z i,j+1 ) represents the j+1th position of UUV; (x i,j ,y i,j , z i,j ) represents the jth position of UUV; p i =(S i,1 , S i,2 , ..., s i,n ) represents the i-th path of UUV, and n represents the path p i The total number of path points in .
[0152] 3) Calculate the path threat cost TL(p i ):
[0153] A Bayesian network for predicting threat cost is constructed in advance: threat factors are determined according to the UUV mission execution and navigation requirements, including environmental threat factors and platform status; environmental threat factors include seabed topography, ocean currents, and the opponent's sonar scanning area; UUV platform status includes energy and platform attitude; it is assumed that the UUV has sufficient energy and the platform attitude is stable. Based on environmental threat factors and UUV platform status, a Bayesian network is constructed as follows: Figure 3 The Bayesian network model shown;
[0154] Based on the environmental model in step 1, a threat level allocation strategy in the Bayesian network model is constructed. The threat level status is set to low threat, medium threat, and high threat. The threat level cost is numerically allocated to 1, 2, and 3 respectively.
[0155] Seafloor topography threat level assignment strategy:
[0156] From formula (2), we can see that when the path point s i,j The distance to the peak is d>R i′ +D+m, path point s i,j The threat level of the submarine mountain is f(s i,j )=0, indicating that the threat level of the mountain is low threat, and the threat level value is represented by '1'.
[0157] When the waypoint s i,j The distance to the peak is R i′ +D <d≤R i′ +D+m, path point s i,j The threat level of the submarine mountain is f(s i,j )=R i′ +D+md means that the threat level of the mountain is medium threat, and the threat level value is '2'.
[0158] When the waypoint s i,j The distance to the peak is d≤R i′ +D, waypoint s i,j The threat level of the submarine mountain is f(s i,j )=∞, indicating that the threat level of the mountain is high, and the threat level value is represented as '3'.
[0159] Strategy for assigning ocean current threat levels:
[0160] From formula (4), we can see that when the path point s i,j The ocean current velocity is V c (s i,j )≤0.2m / s, path point s i,j The threat level of the current is low threat, and the threat value is represented by '1'.
[0161] When the waypoint s i,jFlow velocity at 0.2m / s <V c (s i,j )≤0.7m / s, path point s i,j The current threat level is medium threat, and the threat value is '2'.
[0162] When the waypoint s i,j Flow velocity V c (s i,j )>0.7m / s, path point s i,j The threat level of the ocean current is high, and the threat value is '3'.
[0163] Sonar scan threat level assignment strategy:
[0164] From formula (6), we can see that when the path point s i,j The shortest distance to the sonar scanning area is d s >R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=0, indicating that the threat level of the underwater sonar scanning area is low threat, and the threat level value is represented by '1'.
[0165] When the waypoint s i,j The shortest distance to the sonar scanning area is R t +D <d s <R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=R t +D means that the threat level in the underwater sonar scanning area is medium, and the threat level value is '2'.
[0166] When the waypoint s i,j The shortest distance to the sonar scanning area is d s ≤R t +D, waypoint s i,j The threat level of the underwater sonar scanning area is p(s i,j )=∞, indicating that the threat level of the underwater sonar scanning area is high threat, and the threat level value is represented as '3'.
[0167] Based on the above threat level allocation strategy, calculate the path p i Each path point s i,j The threat level of environmental threat factors and calculate the path p i Each path point s i, j's UUV platform factor measurement value, in this implementation, it is assumed that the UUV has sufficient energy and a stable platform posture, which is used as a benchmark, and then the current UUV energy and UUV platform posture are compared with the benchmark as their respective corresponding threat values. In other implementations, other methods can also be used to directly use the UUV platform factor measurement value; the path point s i,j The threat degree values of the threat factors at the locations are used as evidence of the Bayesian network model to construct the Bayesian network model.
[0168] Based on the Bayesian network model, threat assessment is performed to solve the path point s i,j Threat cost at:
[0169]
[0170] Among them, s i,j represents the current path point of the UUV, T t represents the t=(1, 2, 3, 4, 5)th threat factor node in the Bayesian network model, t represents the tth specific factor in the threat factor, and in this implementation mode, corresponds to the environmental threat factor and the specific factor index in the UUV platform state. In this implementation mode, t=1, 2, 3, 4, 5, that is, corresponds to the seabed topography, ocean current, the opponent's sonar scanning area and energy, and the platform posture. It should be noted that in other implementation modes or embodiments, t may include other specific factors; Pa(T t ) represents the prior probability of the threat factor node, P(T t |Pa(T t )) represents the conditional probability of the threat factor.
[0171]
[0172] Among them, TL(p i ) represents the path p i The threat cost, TL(p i )0 represents the initial threat cost; tl(s i,j ) and tl′(s i,j ) represents the current path point s of the UUV i,j The threat cost of Z is β, which is a constant much larger than the theoretical path length. i,j Indicates that at the path point s i,j =(x i,j ,y i,j , z i,j ) is the UUV navigation altitude value at Z(x i,j ,y i,j ) means in (x i,j ,y i,j ), n represents the height of the seabed terrain obstacle at the path p iThe total number of path points in the . In the geodetic coordinate system, x i,j, y i,j The height of the seabed terrain obstacle at Z should be lower than the UUV navigation altitude; otherwise, it will cause a collision. i,j <Z(x i,j ,y i,j ), that is, when the UUV collides with the seabed terrain obstacle, the path point s is increased i,j The threat cost tl′(s i,j ); On the contrary, the path point s is obtained according to the evaluation of the Bayesian network model i,j The threat cost tl′(s i,j ).
[0173] 4) Calculate the total path cost of the smoothed path obtained in step 2.4:
[0174] L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ) (15)
[0175] Among them, L(p i ) represents the path cost of the effective path, p i =(s i,1 ,s i,2 , ..., s i,n ) represents the i-th path of UUV; s i,1 ,s i,2 , ..., s i,n represents the path point on the i-th path, s i,1 =s start Indicates the starting point of the path; s i,n =s end Indicates the end point of the path; F1(p i ) represents the yaw angle and pitch angle cost of the path; F2(p i ) represents the path length cost; TL(p i ) represents the path threat cost; α1, α2, α3 represent weights.
[0176] Step 3: Based on the target attraction law, the adaptive particle swarm optimization algorithm calculates the total path cost L(p i ) is solved, the optimal path obtained satisfies the constraint cost of yaw angle and pitch angle, the path length cost is the shortest, the threat cost is the smallest, and the optimal path is output; the specific process includes the following steps:
[0177] Step 3.1, initialize algorithm parameters: population size N, maximum number of iterations iterl max, maximum inertia weight ω max , minimum inertia weight ω min wait;
[0178] The initial global optimal cost in the initial population is inf;
[0179] Obtain individual and global optimal solutions;
[0180] Step 3.2: Initialize the path point s of each individual in the population ij , generate a set of valid path points p′ i . Let Q S′ Search space for the environment, if the path point s ij ∈Q S′ , then s ij is a valid waypoint; if the waypoint Then ij is an invalid waypoint;
[0181] The set of valid path points from the path starting point to the path ending point is denoted as p′ i :
[0182] p′ i =(s i1 ,s i2 ,...,s ij , ..., s in ) (16)
[0183] Among them, p′ i Represents a set of valid path points, i.e., a valid path. ij Represents a valid path point. i represents the i-th path in the UUV path planning, and also represents the i-th individual (i=1, 2, ..., N) of the particle swarm in the adaptive particle swarm optimization algorithm based on the target attraction law. j represents the j-th dimension (j=1, 2, ..., n), and n represents the total number of path points.
[0184] Step 3.3, using the formula L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ), the valid path point set p′ i Instead of p i , calculate the path cost of the effective path generated by the ith individual after the effective path point set is optimized. Compare the path cost of the effective path corresponding to the ith individual with the global optimal cost in the population, which is inf:
[0185] If the path cost of the effective path corresponding to the i-th individual is less than the global optimal cost in the population, which is inf, then the effective path corresponding to the i-th individual is selected as the global optimal path, and the path cost of the effective path corresponding to the i-th individual is recorded;
[0186] Traverse each individual in the population to find the effective path corresponding to the individual minimum path cost ParticlesBest as the global optimal path, and update the global optimal solution GlobalBest;
[0187] Step 3.4, set the number of iterations to iterl, and set iterl = 1;
[0188] Update the speed and position of the UUV at each moment, that is, update the valid path point set p′ i :
[0189]
[0190] Among them, V ij (k′+1) represents the speed of UUV at the k′+1th moment; V ij (k′) represents the speed of the UUV at the k′th moment; ωα represents the nonlinear adaptive weight, which is used to inherit the speed of the UUV at the previous moment and balance the global and local search capabilities of the algorithm; c1 represents the individual cognitive learning factor, which is used to balance the learning factor of the individual optimal value; c2 represents the social cognitive learning factor, which is used to balance the learning factor of the global optimal value; c3 represents the attraction learning factor, which is used to balance the learning factor of the attraction of the target endpoint; r1, r2 and r3 represent random numbers between 0 and 1; P besti (k′) represents the optimal position of the UUV in a moving direction at the k′th moment; X ij (k′) represents the position of the UUV at the k′th moment; g best (k′) represents the optimal position of the UUV in all movement directions at the k′th moment; X ij (k′+1) represents the position of UUV at the k′+1th moment; f attract represents the target attraction weight; represents the attractiveness of the target terminal position; j represents the dimension; i represents the i-th individual.
[0191] Optionally, the inertia weight factor is expressed as:
[0192]
[0193] Among them, ω max represents the maximum inertia weight; ω min It represents the minimum inertia weight; iterl represents the current number of iterations; iterl maxrepresents the maximum number of iterations; δ=θ*rand represents the nonlinear radian value that changes with time, rand represents a random number from 0 to 1; θ=π / 2.
[0194] The expression of the target attraction weight is:
[0195]
[0196] Where A represents the magnitude of the target attraction weight; ξ represents the standard deviation of the target attraction weight; Represents the attraction between the UUV's current position and the target end position.
[0197] Step 3.5: Based on the update strategy of the UUV speed and position at each moment in step 3.4, the updated valid path point set in step 3.4 replaces the original valid path point set;
[0198] Step 3.6: Use the formula L(p i )=α1F1(p i )+a2F2(p i )+α3TL(p i ), the new valid path point set p′ in step 3.5 i Instead of p i , calculate the path cost of the new set of valid path points;
[0199] If the new valid path point set p′ in step 3.5 i If the path cost is less than the individual optimal path cost in step 3.3, the individual optimal path and path cost are updated; at the same time, if the updated individual optimal path cost is less than the global optimal path cost in step 3.3, the path of the new valid path point set is selected as the global optimal path GlobalBest, and the path cost of the new valid path point set is recorded.
[0200] If the path cost of the new valid path point set in step 3.5 is greater than or equal to the individual optimal path cost in step 3.3, continue to use the global optimal solution GlobalBest determined in step 3.3 as the new global optimal solution GlobalBest.
[0201] Step 3.7: Determine whether the number of iterations iterl is greater than the maximum number of iterations iterl max ;
[0202] If the number of iterations iterl is less than the maximum number of iterations iterl max , let iterl=iterl+1, and execute step 3.4;
[0203] If the number of iterations is iterl greater than the maximum number of iterations iterl max , output the global optimal path.
[0204] The present invention is further explained by taking a UUV path planning problem in a complex marine environment as an example.
[0205] Figure 2 Schematic diagram of UUV yaw angle and pitch angle. Figure 3 is a threat assessment model based on Bayesian network. Figure 4 It is a 3D path planning diagram. The parameters of the adaptive particle swarm optimization algorithm based on the target attraction law are as follows:
[0206] Maximum number of iterations iterl max =100, population size N = 50, minimum inertia weight ω min =0.8, maximum inertia weight ω max =1.1, individual cognitive learning factor c1=1.7, social cognitive learning factor c2=1.7, attraction learning factor c3=1.6, amplitude of target attraction weight A=1, standard deviation of target attraction weight σ=1.
[0207] The simulation results are as follows Figure 4 As shown in the figure, it can be seen that the present invention can obtain an optimal path that meets the constraints of the yaw angle and the pitch angle, has the shortest path length and the smallest threat index; at the same time, in a complex marine environment, the UUV has a strong obstacle avoidance capability and passability.
[0208] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.
Claims
1. A UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm is characterized by: include: S100, obtaining an environmental simulation model based on ocean environment modeling; The environmental simulation model at least includes a submarine mountain model and an ocean current model; Based on the ocean environment modeling, the threat level corresponding to the environmental simulation model is obtained: The danger distance is determined based on the distance between the UUV and the submarine mountain and is used as the submarine mountain obstacle threat degree, that is, the submarine mountain obstacle threat degree: In the submarine mountain model, each mountain is equivalent to a cylinder, d is the distance from the path point to the center of the cylinder, R i' represents the radius of the i'th mountain obstacle, D is the radius of the UUV; m is the distance threshold of the dangerous area; f(s i,j ) represents the j-th path point s of the i-th path i,j The threat level of submarine mountain obstacles; Determine the ocean current velocity V based on the ocean current model c (s i,j ) and the ocean current velocity V c (s i,j ) The corresponding velocity range is used as the threat level; S200: Determine the corresponding path point set p on each path from the path starting point to the path end point. i =(s i,1 , s i,2 ,…,s i,n ), s i,1 , s i,2 ,…,s i,n represents the path points constituting the i-th path; based on the environmental simulation model and its threat degree obtained in S100, the specific threat degree corresponding to the environmental threat factor of each path point is determined, and then the threat degree cost at the path point is obtained, and then the total path cost of the smoothed path is determined by combining the yaw angle and pitch angle costs and the path length cost; S300 , solving the problem of minimizing the total cost of the path in step S200 using an adaptive particle swarm optimization algorithm based on the target attraction rule to obtain the optimal path.
2. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 1 is characterized in that: The submarine mountain model is as follows: Where Z(x, y) represents the seabed height at the UUV position vector (x, y) in geodetic coordinates, (x i' ,y i' ) represents the center coordinate of the i'th submarine mountain model; h i' Represents the terrain parameter information that controls the height of the submarine mountain model; x si' and y si' Respectively represent the attenuation degree of the i'th submarine mountain model along the x direction and the y direction, and also control the slope of the submarine mountain model; n m Indicates the number of peaks in the entire submarine mountain model.
3. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 1 is characterized in that: The ocean current model is obtained by superimposing multiple Lamb vortices; the Lamb vortex model is as follows: in, and Respectively represent the ocean current speed in the x-direction, y-direction and vertical direction in the geodetic coordinate system, V c (s i,j ) represents the j-th path point s of the i-th path i,j Current speed at Indicates the current position of the geodetic coordinate system UUV, (x, y) indicates the position vector of UUV in the geodetic coordinate system XOY plane; (x0, y0) indicates the position vector of the ocean current in the geodetic coordinate system XOY plane, represents the center of the ocean current, Γ and δ represent the ocean current intensity and the ocean current radius, respectively.
4. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to any one of claims 1 to 3, characterized in that: The environment simulation model also includes an underwater sonar scanning model. Based on the underwater sonar scanning model, according to the path point s i,j The shortest distance to the sonar scanning area determines the threat level of the underwater sonar scanning model; The underwater sonar scanning model is as follows: Among them, S th represents the sonar scanning area, P(X, Y, Z) represents a point in the three-dimensional space area S, that is, any point in the sonar scanning area; X, Y, Z represent the coordinates in the three-dimensional space of the three-dimensional plane, (X0, Y0, Z0) represent the center coordinates of the ellipsoid, a, b represent the lengths of the major and minor axes of the ellipse, and R represents the radius of the sonar scanning area; The threat level of the underwater sonar scanning model is as follows: Among them, d s Represents the path point s i,j The shortest distance to the sonar scanning area, (x i,j ,y i,j ,z i,j ) represents the path point s i,j Position coordinates in the geodetic coordinate system, R t represents the maximum threat threshold of the sonar scanning area, D represents the radius of the UUV, and ε represents the threat threshold of the dangerous area; p(s i,j ) represents the j-th path point s of the i-th path i,j Underwater sonar scanning model threat level.
5. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 4 is characterized in that: The specific process of S200 includes the following steps: S2001. In the geodetic coordinate system, the path starting point s start The coordinates are (x0, y0, z0), the end point of the path is s end The coordinates are (x e ,y e , z e ), generate the path from the starting point s start To the end point of the path s end A path consists of a set of path points p i =(s i,1 , s i,2 ,…,s i,n )composition; S2003. Connect all the path points in the geodetic coordinate system to form a navigation path for the UUV. S2004, using cubic spline fitting scatter points to smooth the path in S2003 to obtain a smooth path; S2005. Optimize the UUV smooth path obtained in S2004: 1) Calculate the yaw and pitch angle costs F1(p i ); 2) Calculate the path length cost F2(p i ); 3) For the path points on the smooth path, the specific threat level of the environmental threat factors corresponding to each path point is determined based on the environmental simulation model obtained in S100 and its corresponding threat level, and then the path point s is obtained. i,j The threat cost TL(p i ); The environmental threat factors correspond one to one with the specific models in the environmental simulation model; 4) Calculate the total path cost of the smoothed path obtained through S2004: L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ) (15) Among them, L(p i ) represents the path cost of the effective path, p i =(s i,1 , s i,2 ,…,s i,n ) represents the i-th path of UUV; s i,1 , s i,2 ,…,s i,n represents the path point on the i-th path, s i,1 =s start Indicates the starting point of the path; s i,n =s end represents the end point of the path; α1, α2, α3 represent the weights.
6. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 5 is characterized in that: The yaw angle and pitch angle cost F1 (p i ) process include: s' i,j 、s' i,j+1 and s' i,j+2 They represent the jth, (j+1)th and (j+2)th path points s of the i-th path respectively. i,j 、s i,j+1 and s i,j+2 Projection point on the O-xy plane in the geodetic coordinate system; yaw angle Φ i,j Indicates the path With path The angle between the projection paths of the O-xy plane in the geodetic coordinate system; the pitch angle Ψ i,j+1 Indicates the path With path The climb angle between Assumptions are the unit vectors in the x, y, and z directions in the geodetic coordinates respectively. According to the concept of vector projection, the calculation formula of the projection path is: Therefore, the yaw angle Φ of the UUV during the induced path i , j It can be given by the following formula: The pitch angle Ψ of the UUV during the induced path i , j+1 It can be given by the following formula: The yaw and pitch costs for a smoothed path are: where μ1 and μ2 represent the mechanism penalty constants related to the yaw angle and pitch angle, respectively, and p i =(s1, s2, ..., s n ) represents the i-th path of UUV, and n represents the path p i The total number of path points in .
7. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 5 is characterized in that: The path length cost F2(p i ) process includes: Among them, (x i,j+1 ,y i,j+1 , z i,j+1 ) represents the j+1th position of UUV; (x i,j ,y i,j , z i,j ) represents the j-th position of UUV; p i =(s i,1 , s i,2 ,…,s i,n ) represents the i-th path of UUV, n represents the path p i The total number of path points in the.
8. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 5 is characterized in that: The specific process of step 3) includes: First, a specific threat level value corresponding to the environmental simulation model of each path point is determined based on a threat level allocation strategy; the threat level allocation strategy includes a seabed terrain threat level allocation strategy and an ocean current threat level allocation strategy; Seabed topography threat level assignment strategy: When the path point s i,j The distance to the peak is d>R i' +D+m, the path point s i,j The threat level of the submarine mountain is f(s i,j )=0, indicating that the threat level of the mountain is low, and the threat level value is expressed as 'α1'; When the path point s i,j The distance to the peak is R i' +D <d≤R i' +D+m, the path point s i,j The threat level of the submarine mountain is f(s i,j )=R i' +D+md, indicating that the threat level of the mountain is medium threat, and the threat level value is expressed as 'α2'; When the path point s i,j The distance to the peak is d≤Ri ' +D, waypoint s i,j The threat level of the submarine mountain is f(s i,j )=∞, indicating that the threat level of the mountain is high, and the threat level value is expressed as 'α3'; 'α1', 'α2', and 'α3' are the set values respectively; Strategy for allocating ocean current threat levels: When the path point s i,j The ocean current velocity is V c (s i,j )≤the first threshold, the path point s i,j The threat level of the ocean current is low threat, and the threat value is expressed as 'β1'; When the path point s i,j The first flow velocity threshold at <V c (s i,j )≤the second threshold, the path point s i,j The threat level of the ocean current is medium threat, and the threat value is expressed as 'β2'; When the path point s i,j Flow velocity V at c (s i,j )> the second threshold, the path point s i,j The threat level of the ocean current is high, and the threat value is expressed as 'β3'; 'β1', 'β2', and 'β3' are the set values respectively; When the environment simulation model further includes an underwater sonar scanning model, the threat level allocation strategy further includes a sonar scanning threat level allocation strategy, and the sonar scanning threat level allocation strategy is as follows: When the path point s i,j The shortest distance to the sonar scanning area is d s >R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=0, indicating that the threat level of the underwater sonar scanning area is low threat, and the threat level value is expressed as 'γ1'; When the path point s i,j The shortest distance to the sonar scanning area is R t +D <d s <R t +D+ε, the path point s i,j The threat level of the underwater sonar scanning area is p(s i,j )=R t +D, indicating that the threat level in the underwater sonar scanning area is medium, and the threat level value is expressed as 'γ2'; When the path point s i,j The shortest distance to the sonar scanning area is d s ≤R t +D, waypoint s i,j The threat level of the underwater sonar scanning area is p(s i,j ) = ∞, indicating that the threat level of the underwater sonar scanning area is high, and the threat level value is expressed as 'γ3'; 'γ1', 'γ2', and 'γ3' are the set values respectively; Then, based on the specific threat value corresponding to the environmental simulation model of each path point, the path point s is obtained by machine learning. i,j The threat cost tl(s i,j ); then finally determine the path point s i,j The threat cost TL(p i ): Among them, TL(p i ) represents the path p i The threat cost, TL(p i )0 represents the initial threat cost; tl(s i,j ) and tl'(s i,j ) represents the UUV current path point s i,j The threat cost, β represents a constant that is much larger than the theoretical path length; Z i,j Indicates that at the path point s i,j =(x i,j ,y i,j , z i,j ) at the UUV navigation altitude; Z(x i,j ,y i,j ) means in (x i,j ,y i,j ), n represents the height of the seabed terrain obstacle at the path p i The total number of path points in .
9. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 8 is characterized in that: Use machine learning to obtain the path point s i,j The threat cost tl(s i,j The process of ) is based on a Bayesian network for predicting threat cost. The Bayesian network for predicting threat cost is pre-built. The process of constructing a Bayesian network for predicting threat cost includes the following steps: Determine threat factors based on UUV mission execution and navigation requirements, including environmental threat factors and UUV platform status; wherein environmental threat factors include at least seabed topography and ocean currents, and UUV platform status includes at least UUV energy and UUV platform posture; when the threat level allocation strategy also includes a sonar scanning threat level allocation strategy, the environmental threat factors also include the opponent's sonar scanning area; determine the environmental threat factors based on a one-to-one correspondence between each specific model in the environmental simulation model, and determine the specific threat level value corresponding to the environmental threat factor according to the threat level allocation strategy; and simultaneously obtain the specific threat level value corresponding to the UUV platform factor; and assign the path point s i,j The threat degree value of the threat factor at the location is used as evidence to construct the Bayesian network model.
10. The UUV path planning method in complex environments based on the target attraction law adaptive particle swarm optimization algorithm according to claim 1 is characterized in that: The specific process of step S300 includes the following steps: S3001. Initialize algorithm parameters: population size N, maximum number of iterations iterl max , maximum inertia weight ω max , minimum inertia weight ω min ; The initial global optimal cost in the initialized population is inf; Obtain individual and global optimal solutions; S3002, initialize the path point s of each individual in the population ij , generate a valid path point set p' i ; Let Q S' For the environment search space, if the path point s ij ∈Q S' , then s ij is a valid waypoint; if the waypoint Then s ij is an invalid waypoint; The set of valid path points from the path starting point to the path ending point is denoted as p' i : p′ i =(s i1 ,s i2 ,…,s ij ,…,s in ) (16) Among them, p' i Represents the set of valid path points, that is, the valid path; s ij Represents a valid path point; i represents the i-th path in the UUV path planning, and also represents the i-th individual in the particle swarm in the adaptive particle swarm optimization algorithm based on the target attraction law; j represents the j-th dimension, corresponding to the j-th path point; S3003, using the formula L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ), the valid path point set p' i Instead of p i , calculate the path cost of the effective path generated by the i-th individual after the effective path point set is optimized; compare the path cost of the effective path corresponding to the i-th individual with the global optimal cost in the population, which is inf: If the path cost of the effective path corresponding to the i-th individual is less than the global optimal cost ijf in the population, then the effective path corresponding to the i-th individual is selected as the global optimal path, and the path cost of the effective path corresponding to the i-th individual is recorded; Traverse each individual in the population to find the effective path corresponding to the individual minimum path cost ParticlesBest as the global optimal path, and update the global optimal solution GlobalBest; S3004, set the number of iterations to iterl, and set iterl = 1; Update the speed and position of the UUV at each moment, that is, update the valid path point set p' i : Among them, V ij (k'+1) represents the speed of UUV at the k'+1th moment; V ij (k') represents the speed of the UUV at the k'th moment; ω α represents the nonlinear adaptive weight, which is used to inherit the UUV's previous speed and balance the global and local search capabilities of the algorithm; c1 represents the individual cognitive learning factor, which is used to balance the learning factor of the individual optimal value; c2 represents the social cognitive learning factor, which is used to balance the learning factor of the global optimal value; c3 represents the attraction learning factor, which is used to balance the learning factor of the target endpoint attraction; r1, r2 and r3 represent random numbers between 0 and 1; P besti (k') represents the optimal position of the UUV in a moving direction at the k'th moment; X ij (k') represents the position of the UUV at the k'th moment; g best (k') represents the optimal position of the UUV in all movement directions at the k'th moment; X ij (k'+1) represents the position of UUV at the k'+1th moment; f attract represents the target attraction weight; represents the attraction of the target terminal position; j represents the dimension; i represents the i-th individual; Optionally, the inertia weight factor is expressed as: Among them, ω max represents the maximum inertia weight; ω min It represents the minimum inertia weight; iterl represents the current number of iterations; iterl max Indicates the maximum number of iterations; δ = θ * rand represents the nonlinear radian value that changes with time, rand represents a random number from 0 to 1; θ = π / 2; The expression of the target attraction weight is: Where A represents the magnitude of the target attraction weight; ξ represents the standard deviation of the target attraction weight; Indicates the attraction between the UUV’s current position and the target end position; S3005. Based on the updating strategy of the UUV speed and position at each moment in S3004, the updated valid path point set in S3004 is used to replace the original valid path point set. S3006, using the formula L(p i )=α1F1(p i )+α2F2(p i )+α3TL(p i ), the new valid path point set p' in S3005 i Instead of p i , calculate the path cost of the new set of valid path points; If the new valid path point set p' in S3005 i If the path cost of the individual optimal path is less than the individual optimal path cost in S3003, the individual optimal path and path cost are updated; at the same time, if the updated individual optimal path cost is less than the global optimal path cost in S3003, the path of the new valid path point set is selected as the global optimal path GlobalBest, and the path cost of the new valid path point set is recorded; If the path cost of the new valid path point set in S3005 is greater than or equal to the individual optimal path cost in S3003, then continue to use the global optimal solution GlobalBest determined in S3003 as the new global optimal solution GlobalBest; S3007, determine whether the number of iterations iterl is greater than the maximum number of iterations iterl max ; If the number of iterations iterl is less than the maximum number of iterations iterl max , set iterl=iterl+1, and execute S3004; If the number of iterations is iterl greater than the maximum number of iterations iterl max , output the global optimal path.
Citation Information
Cited By
Design optimization method for unmanned underwater vehicle under size and weight constraints
CN119623336A
