AUV (Autonomous Underwater Vehicle) guiding method based on active sound source searching

Abstract

According to the AUV guiding method based on active sound source searching, in order to solve the problems existing in the black box searching process of the AUV in a complex marine environment, dynamic guiding planning is carried out based on an improved differential evolution algorithm in a real-time guiding searching stage; meanwhile, the influence of sound source signal loss caused by entering a sound source sound shadow area during guidance is considered, and the searching efficiency of the AUV in the process of searching the active sound source in a deep sea in a large range is improved. The method comprises the following steps: firstly, defining a fitness function as a target function optimized by a differential evolution algorithm; secondly, generating an initial AUV path planning population which comprises a starting point, a target point and a path search space, selecting an optimal individual or group by adopting a roulette strategy, and randomly mutating some individuals in the group in the search space to generate new individuals; performing crossover operation on two individuals in the parent group to generate a new offspring; an optimal individual in the current population is reserved by adopting an Elitism strategy and the like; finally, the optimal path of the AUV is output, the purpose of AUV guiding path planning is achieved, and the obtained guiding path result has certain reference significance for black box searching work of the AUV.

Description

Technical Field

[0001] This invention relates to the field of underwater robot path planning, specifically a research on a method that enables AUVs to perform underwater guidance based on active sound source search to complete predetermined target tasks. Background Technology

[0002] Autonomous underwater vehicles (AUVs) have broad application prospects in marine resource exploration, marine scientific research, and marine environmental monitoring. AUV path planning is a crucial step in AUV mission execution, and its quality and efficiency directly impact the success of the mission. Traditional AUV path planning methods suffer from problems such as long path lengths, large path curvatures, and a tendency to get trapped in local optima, affecting mission execution efficiency. Therefore, how to efficiently and optimally plan AUV paths has become a research hotspot in the field of underwater robotics. The improved differential evolution algorithm is a highly efficient global optimization algorithm that has been widely applied in AUV path planning in recent years and has achieved good results in simulation practice.

[0003] Significant progress has been made by numerous scholars both domestically and internationally in AUV path planning. In 2018, S. Mahmoud Zadeh et al. proposed a fast path planning method for autonomous underwater vehicles (AUVs) in time-varying underwater environments using Differential Evolution (DE). This method utilizes DE to solve the AUV path planning problem in complex 3D underwater environments. In 2017, Yang An et al. proposed a two-step method using a greedy strategy to plan the AUV mission process, completing the AUV survey task at minimum cost by iteratively executing the K-means algorithm and ant colony algorithm. In 2020, Xu Ruilian et al. set constraint functions and fitness functions based on the actual operating environment of the UAV, and used an improved differential evolution algorithm to divide the population into two subpopulations based on the fitness of individuals, selecting different mutation strategies to achieve the UAV path planning task. In 2019, Sui Haichen proposed a three-step strategy for the search of deep-sea active acoustic sources, which involves a rapid search approach that proceeds from far to near and gradually advances, making the search work more scientific and rational.

[0004] This invention primarily addresses the task requirements of AUVs (Autonomous Vehicles) for active sound source searching by improving the differential evolution algorithm. First, considering the impact of sound field intensity on AUV guidance, the fitness function is improved. Simultaneously, through two differential calculations, the accuracy of AUV guidance path calculation is enhanced. Furthermore, the simulation environment is modeled to account for the influence of the sound shadow zone generated when the active sound source encounters obstacles. This solves the problems of inaccurate calculations and slow speed in AUV path guidance during active sound source searching. Summary of the Invention

[0005] To address the shortcomings of existing technologies and specific task requirements, this invention provides an AUV black box search technology based on an improved differential evolution algorithm. The proposed method is more suitable for AUV black box search in deep-sea environments, and the obtained AUV path planning results have better reference value for AUV black box search.

[0006] The technical solution adopted by this invention to achieve the above objectives is: an AUV target search method based on an improved differential evolution algorithm, comprising:

[0007] The search tasks for AUVs in the deep-sea environment are divided into large-scale multi-target search and guided search;

[0008] During the large-scale multi-target search phase, the sonar array payload carried by the AUV is used to perform scanning operations to detect underwater acoustic signals emitted by acoustic beacons;

[0009] During the guided search phase, the reachable area of ​​the AUV is divided into grid positions; a fitness function representing the path length between grid positions is defined as the objective function for optimization by the differential evolution algorithm; the search path of the AUV is planned by the improved differential evolution algorithm, and the search link is output.

[0010] During the guided search phase, the grid positions for dividing the AUV reachable area are as follows:

[0011]

[0012] In the formula, Let represent the i-th grid, and D represent the dimension of the grid position; for the i-th grid.

[0013] The fitness function F(X) i Using Euclidean distance as the metric for path length, calculate the path length from the i-th grid to the next grid:

[0014]

[0015] In the formula, x start Represents the initial position in the x-direction within two-dimensional space, y start Represents the initial position in the y-direction within two-dimensional space, x i1 This indicates the position of the next location in the x-direction within two-dimensional space. i1 The next position represents the position in the x-direction within two-dimensional space, and the y-direction represents the position in the x-direction. i1 This indicates the position of the next location in the y-direction within two-dimensional space. This represents the sound field intensity parameter of an active sound source.

[0016] The process of planning the search path for the AUV using the improved differential evolution algorithm includes:

[0017] S1. Generate the initial AUV path planning population, including the starting point, the target point, and the path search space. Use random numbers to generate the initial population.

[0018] S2. Use a roulette wheel to select the best individual or group as the parent of the next generation.

[0019] S3. Randomly mutate individuals in the population within the search space to generate new individuals;

[0020] S4. Perform crossover operations on individuals in the parent group to produce new offspring;

[0021] S5. Use the Elitism strategy to retain the best individuals in the current population and replace some individuals in the parent population with newly generated offspring individuals to form a new population.

[0022] Once the required number of iterations is reached, the path planning for the AUV is completed.

[0023] The generation of the initial AUV path planning population includes:

[0024] S101. Determine the number and range of independent variables in path planning;

[0025] S102. Randomly generate the initial population P0, with the independent variable of each individual being randomly generated within the range of values;

[0026] S103. Calculate the fitness of each individual in the initial population P0;

[0027] S104. Using individuals with high fitness as parents, generate new individuals according to the differential evolution algorithm formula, and add them to the population P1.

[0028] The formula for generating new individuals using the differential evolution algorithm:

[0029]

[0030] in, Let x represent the value of the independent variable for the newly generated i-th individual. r1 x r2 x r3 These are the independent variable values ​​of three randomly selected different parent individuals, where r1, r2, and r3 are the number of parent individuals, and F is the scaling factor of the differential evolution algorithm, which is set to F = 0.5 here.

[0031] Step 5 specifically includes the following steps:

[0032] S201. Use roulette wheel selection to choose the optimal individual or population, and calculate the fitness probability distribution:

[0033]

[0034] Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size;

[0035] S201. Calculate the cumulative probability distribution:

[0036]

[0037] Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size;

[0038] S201. Generate random numbers r∈[0,1), and select the i-th individual. i ;

[0039] Select i =x j Q i -1≤r≤Q j

[0040] The random mutation generates new individuals as follows:

[0041] V i =x r1 +F×(x r2 -x r3 )

[0042] Where V i For newly generated individuals, x r1 x r2 x r3 r1, r2, and r3 are the values ​​of the independent variables for three randomly selected different parent individuals, where r1, r2, and r3 are the number of parent individuals, and F is the scaling factor for the differential evolution algorithm.

[0043] The crossover operation on individuals in the parent group includes:

[0044] Binary crossover is used to perform the crossover operation, selecting two parent individuals for crossover and combination to obtain a new individual:

[0045]

[0046] Among them, u i,j Let x represent a random difference vector of individual i. i,jv represents the information of the j-th dimension of individual i. i,j This represents the information of the j-th dimension of the newly generated individual; CR is the crossover probability, and rand... j A random number that is uniformly distributed in the range [0,1], j r It is a random number.

[0047] The method of employing the Elitism strategy to retain the best individuals in the current population and replacing some individuals in the parent population with newly generated offspring individuals to form a new population includes the following steps:

[0048] S501. Select the best individual x in the current population. best Then, for each new individual x′, compare it with the original individuals to obtain a new population P′;

[0049]

[0050] Among them, x i Let x represent the i-th individual in the original population. i ' represents the individual corresponding to it in the new population, f(x) i ) represents x i fitness, f(x) i ) represents x i The fitness of x′, if x′ i The fitness of ′ is greater than that of x i Okay, then use x i ′ replace x i Otherwise, keep x. i .

[0051] The present invention has the following beneficial effects and advantages:

[0052] 1. Improved search efficiency: Traditional search methods require a large investment of manpower and resources and are inefficient, while differential evolution algorithms can significantly improve search efficiency and save costs through global optimization and adaptive parameter adjustment.

[0053] 2. Improved search accuracy: The improved differential evolution algorithm can explore a wider search space through random search, mutation, and two crossover operations, avoiding getting trapped in local optima and improving search accuracy.

[0054] 3. High adaptability: The improved differential evolution algorithm can adaptively adjust the algorithm parameters according to the characteristics of the problem and the optimization requirements, making it more adaptable. Attached Figure Description

[0055] Figure 1 This is a flowchart of the method of the present invention;

[0056] Figure 2 This is a model diagram of an AUV searching for a black box in a deep-sea environment;

[0057] Figure 3 This is a schematic diagram of the characteristics of sound wave diffraction;

[0058] Figure 4 This is a schematic diagram of a simulation map;

[0059] Figure 5 This is a diagram illustrating the guided search process of the method of the present invention; Detailed Implementation

[0060] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0061] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0063] like Figure 1 As shown, the AUV guidance method based on active sound source search includes the following steps:

[0064] Step 1: Based on the workflow of AUVs in deep-sea environments, the search phase is divided into two stages;

[0065] from Figure 2 The mission model diagram for AUV searching for the black box in a deep-sea environment shows that the process can be divided into two parts: a large-scale multi-target search phase and a guided search phase. This invention mainly studies the path planning problem of the AUV during the guided search phase. After completing the large-scale multi-target search phase, the AUV enters the guided search phase. During this phase, the AUV needs to receive the 37.5kHz frequency acoustic signal emitted by the black box using its onboard hydrophone array.

[0066] Step 2: Study the AUV path planning problem in the linear guidance stage, and plan the search path of the AUV by using the differential evolution algorithm; at the same time, consider the impact of the sound shadow zone generated when the active sound source encounters obstacles;

[0067] like Figure 3 As shown, the AUV's mission in the deep-sea environment is planned according to the theoretical model of the differential evolution algorithm. First, the working environment of the AUV is regarded as a two-dimensional plane, and the two-dimensional plane of the search space is divided into grid regions, with each grid corresponding to a state position.

[0068]

[0069] In the formula, t represents the time series. Let represent the i-th grid state position in the time series t, and D represent the dimension of the state. For the i-th state position, a fitness function F(X) can be defined. i ), which represents the path length from the current position to the state of that position.

[0070] Step 3: Define the fitness function. Input the coordinates of the AUV at a given location, the starting point, and the sound field intensity of the active sound source into the fitness function as the initial function for the differential evolution algorithm. Use Euclidean distance as the measure of path length to calculate the path length at the i-th state position. The fitness function F(X) is then obtained through this calculation. i The value provides basic data for subsequent population individual evolution and variation.

[0071] The fitness formula is:

[0072]

[0073] Among them, x start Represents the initial position in the x-direction within two-dimensional space, y start Represents the initial position in the y-direction within two-dimensional space, x i1 This indicates the position of the next location in the x-direction within two-dimensional space. i1 The next position represents the position in the x-direction within two-dimensional space, and the y-direction represents the position in the x-direction. i1 The x-axis represents the position in the y-direction within two-dimensional space of the next position. i2 This indicates the position in the x-direction within the next two-dimensional space, and the y-direction... i2 This indicates the position in the y-direction within the next two-dimensional space. This represents the sound field intensity parameter of an active sound source.

[0074] Step 4: Generate the initial AUV path planning population, including the starting point, the target point, and the path search space. Use random numbers to generate the initial population.

[0075] Step 4.1: Determine the number and range of independent variables for individual x in the path planning;

[0076] Step 4.2: Randomly generate the initial population P0, where the independent variable of each individual x is randomly generated within the range of values;

[0077] Step 4.3: Calculate the fitness of each individual x in the initial population P0 according to the fitness formula;

[0078] Step 4.4: Select the top three individuals with the highest fitness as parents, generate new individuals according to the differential evolution algorithm formula, and add them to the population P1. The formula for generating new individuals using the differential evolution algorithm is as follows:

[0079]

[0080] in, Let x represent the value of the independent variable for the newly generated i-th individual. r1 x r2 x r3 These are the independent variable values ​​of three randomly selected different parent individuals, where r1, r2, and r3 are the number of parent individuals, and F is the scaling factor of the differential evolution algorithm, which is set to F = 0.5 here.

[0081] Step 5: Use a roulette wheel to select the best individuals or groups so that they can serve as parents for the next generation of the population;

[0082] Step 5.1: Select the optimal individual or population using a roulette wheel selection method, and calculate the fitness probability distribution:

[0083]

[0084] Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size.

[0085] Step 5.2: Calculate the cumulative probability distribution, then we have:

[0086]

[0087] Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size.

[0088] Step 5.3: Generate random numbers r∈[0,1).

[0089] Select i =x j Q i -1≤r≤Q j

[0090] Select iIt is the i-th selected individual. Based on the calculation results, the optimal individual or population is selected as the parent of the next population, preparing for the next stage of individual variation.

[0091] Step 6 specifically includes the following steps:

[0092] Step 6: Randomly mutate some individuals in the population within the search space to generate new individuals;

[0093] Randomly mutate certain individuals in the population within the search space to generate new individuals:

[0094] V i =x r1 +F×(x r2 -x r3 )

[0095] Where V i For newly created individuals. r1 x r2 x r3 Let r1, r2, and r3 be the values ​​of the independent variables from three randomly selected different parent individuals, where r1, r2, and r3 represent the number of parent individuals, and F is the scaling factor for the differential evolution algorithm. By calculating the difference using the scaling factor, the newly generated individuals can be kept within the individual interval; here, F = 0.5 is chosen. The mutated individuals are then used for the next crossover operation.

[0096] Step 7: Perform crossover on two individuals from the parent population to produce new offspring. Crossover involves combining certain gene segments from two individuals to generate a new individual.

[0097] A binary crossover operation is used to generate new individuals. The crossover operation selects two individuals (parent individuals), and the information of the two individuals is crossed and combined with a certain probability to obtain a new individual.

[0098]

[0099] Among them, u i,j Let x represent a random difference vector of individual i. i,j v represents the information of the j-th dimension of individual i. i,j This represents the information of the j-th dimension of the newly generated individual. CR is the crossover probability, and rand is the crossover probability. j A random number that is uniformly distributed in the range [0,1], j r It is a random number used to ensure that the new individual will have at least one dimension that is the same as the parent individual, thereby avoiding the complete randomization of the information of the new individual and improving search efficiency.

[0100] For each dimension j, u is determined according to the cross probability CR. i,j The value assigned to v i,j , or x i,j The value assigned to v i,j Meanwhile, when j = j r At that time, u must be i,j The value assigned to v i,j This operation ensures that each new individual has at least one dimension that is the same as the parent individual, thus avoiding excessive randomization.

[0101] Step 8: Use the Elitism strategy to retain the best individuals in the current population and replace some individuals in the parent population with the newly generated offspring individuals to form a new population;

[0102] Following the steps above, the optimal individual is selected, and the Elitism strategy is used to form a new population. For the current population P, the optimal individual x in the current population is first selected. best Then, for each new individual x′, it is compared with the original individuals. If the fitness of x′ is better than that of x, then x′ replaces x as an individual in the new population, ultimately resulting in the new population P′. The specific formula is as follows:

[0103]

[0104] Among them, x i Let x represent the i-th individual in the original population. i ' represents the individual corresponding to it in the new population, f(x) i ) represents x i fitness, f(x) i ′) represents x i The fitness of x′, if x′ i The fitness of ′ is greater than that of x i Okay, then use x i ′ replace x i Otherwise, keep x. i This ensures that at least one individual in the new population is identical to the current best individual, thus avoiding the shortcomings of genetic algorithms.

[0105] Finally, the detection path of the AUV during the linear guidance phase was obtained, providing a solution for subsequent AUV missions to search for the black box in the deep sea environment. After completing the guidance and search phase, the AUV determined the location of the black box.

[0106] Once the required number of iterations is reached, the AUV's guided path planning is complete.

[0107] To verify the effectiveness and stability of the method of this invention for AUV path planning, a schematic diagram of environmental modeling is provided, considering the influence of the sound shadow zone generated when an active sound source encounters an obstacle. Figure 4 As shown. The path planning results obtained by using the method of this invention in a simulated environment are as follows. Figure 5 As shown.

[0108] In summary, through the analysis of the active acoustic source search mission in the deep-sea environment, the mission is divided into two stages: a large-scale multi-target search stage in the suspected sea area and a dynamic guidance search stage. Once the hydrophone array on the AUV receives the acoustic signal from the active acoustic source, it enters the dynamic guidance search stage. In this stage, a direct guidance method is used to guide the AUV, and an improved differential evolution algorithm is employed for path guidance. Based on the characteristics of the active acoustic source, an underwater environment simulation model is performed. The improved DE (Differential Evolution) allows the AUV to more quickly determine the search path during the dynamic guidance stage, thus completing the AVU's task of searching for the active acoustic source.

[0109] It should be emphasized that the embodiments described in this invention are illustrative rather than limiting. Therefore, this invention includes, but is not limited to, the embodiments described in the specific implementation. Any other implementations derived by those skilled in the art based on the technical solutions of this invention are also within the scope of protection of this invention.

Claims

1. A target-searching method for AUVs based on an improved differential evolution algorithm, characterized in that, include: The search tasks for AUVs in the deep-sea environment are divided into large-scale multi-target search and guided search; During the large-scale multi-target search phase, the sonar array payload carried by the AUV is used to perform scanning operations to detect underwater acoustic signals emitted by acoustic beacons; During the guided search phase, the reachable area of ​​the AUV is divided into grid positions; a fitness function representing the path length between grid positions is defined as the objective function for optimization of the differential evolution algorithm; the search path of the AUV is planned through the improved differential evolution algorithm, and the search link is output.

2. The AUV target search method based on an improved differential evolution algorithm according to claim 1, characterized in that, During the guided search phase, the grid positions for dividing the AUV reachable area are as follows: In the formula, Let represent the i-th grid, and D represent the dimension of the grid position; for the i-th grid.

3. The AUV guidance method based on active sound source searching according to claim 1, characterized in that, The fitness function F(X) i Using Euclidean distance as the metric for path length, calculate the path length from the i-th grid to the next grid: In the formula, x start Represents the initial position in the x-direction within two-dimensional space, y start Represents the initial position in the y-direction within two-dimensional space, x i1 This indicates the position of the next location in the x-direction within two-dimensional space. i1 The next position represents the position in the x-direction within two-dimensional space, and the y-direction represents the position in the x-direction. i1 This indicates the position of the next location in the y-direction within two-dimensional space. This represents the sound field intensity parameter of an active sound source.

4. The AUV guidance method based on active sound source search according to claim 1, characterized in that, The process of planning the search path for the AUV using the improved differential evolution algorithm includes: S1. Generate the initial AUV path planning population, including the starting point, the target point, and the path search space. Use random numbers to generate the initial population. S2. Use a roulette wheel to select the best individual or group as the parent of the next generation. S3. Randomly mutate individuals in the population within the search space to generate new individuals; S4. Perform crossover operations on individuals in the parent group to produce new offspring; S5. Use the Elitism strategy to retain the best individuals in the current population and replace some individuals in the parent population with newly generated offspring individuals to form a new population. Once the required number of iterations is reached, the path planning for the AUV is completed.

5. The AUV guidance method based on active sound source search according to claim 1, characterized in that, The generation of the initial AUV path planning population includes: S101. Determine the number and range of independent variables in path planning; S102. Randomly generate the initial population P0, with the independent variable of each individual being randomly generated within the range of values; S103. Calculate the fitness of each individual in the initial population P0; S104. Using individuals with high fitness as parents, generate new individuals according to the differential evolution algorithm formula, and add them to the population P1. The formula for generating new individuals using the differential evolution algorithm: in, Let x represent the value of the independent variable for the newly generated i-th individual. r1 x r2 x r3 These are the independent variable values ​​of three randomly selected different parent individuals, where r1, r2, and r3 are the number of parent individuals, and F is the scaling factor of the differential evolution algorithm, which is set to F = 0.5 here.

6. The AUV guidance method based on active sound source search according to claim 1, characterized in that, Step 5 specifically includes the following steps: S201. Use roulette wheel selection to choose the optimal individual or population, and calculate the fitness probability distribution: Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size; S201. Calculate the cumulative probability distribution: Among them, P i Let f(x) represent the probability that the i-th individual is selected. i ) represents the fitness value of the i-th individual, and N represents the population size; S201. Generate random numbers r∈[0,1), and select the i-th individual. i ; Select i =x j ,Q i -1≤r≤Q j 。 7. The AUV guidance method based on active sound source search according to claim 1, characterized in that, The random mutation generates new individuals as follows: V i =x r1 +F×(x r2 -x r3 ) Where V i For newly generated individuals, x r1 x r2 x r3 r1, r2, and r3 are the values ​​of three randomly selected different parent individuals, where r1, r2, and r3 are the number of parent individuals, and F is the scaling factor of the differential evolution algorithm.

8. The AUV guidance method based on active sound source search according to claim 1, characterized in that, The crossover operation on individuals in the parent group includes: Binary crossover is used to perform the crossover operation, selecting two parent individuals for crossover and combination to obtain a new individual: Among them, u i,j Let x represent a random difference vector of individual i. i,j v represents the information of the j-th dimension of individual i. i,j This represents the information of the j-th dimension of the newly generated individual; CR is the crossover probability, and rand... j A random number that is uniformly distributed in the range [0,1], j r It is a random number.

9. The AUV guidance method based on active sound source search according to claim 1, characterized in that, The method of employing the Elitism strategy to retain the best individuals in the current population and replacing some individuals in the parent population with newly generated offspring individuals to form a new population includes the following steps: S501. Select the best individual x in the current population. best Then, for each new individual x′, compare it with the original individuals to obtain a new population P′; Where, x i Let x′ represent the i-th individual in the original population. i f(x) represents the individual corresponding to it in the new population. i ) represents x i fitness, f(x) i ) represents x′ i The fitness if x′ i fitness ratio x i Okay, then use x′ i Replace x i Otherwise, keep x. i .

Images