Underwater intelligent matching method and device based on adaptive variable scale domain

Abstract

The invention relates to the technical field of underwater navigation, in particular to an underwater intelligent matching method and device based on a self-adaptive variable scale domain. The method comprises the following steps: selecting a local small domain center through a variable scale domain coarse positioning mechanism of an adaptive particle swarm optimization algorithm, and realizing high-precision repositioning of a real position of a submersible vehicle through a local small domain fine positioning mechanism of an optimal evaluation value. Simulation results show that the method can effectively improve the positioning efficiency and out-of-domain matching performance of underwater gravity matching navigation. In the test of four tracks, the occurrence of out-of-domain mismatching is completely avoided, and the average improvement ratios of the out-of-domain average matching precision and the out-of-domain worst matching precision are 83% and 85% respectively. The superiority of the method provided by the invention in matching performance is effectively verified, and a new method is provided for realizing long-endurance and high-precision underwater gravity matching navigation under the global ocean background.

Description

Technical Field

[0001] The present invention relates to the field of underwater navigation technology, and in particular to an underwater intelligent matching method and device based on an adaptive variable scale domain. Background Art

[0002] The two main factors that affect the performance of underwater gravity matching navigation are the low efficiency of indiscriminate traversal matching in traditional large grid domains and mismatching. Mismatching is divided into two categories: mismatching within the domain and mismatching outside the domain. Compared with mismatching within the domain, mismatching outside the domain occurs outside the matching domain, resulting in larger positioning errors, which has a more serious impact on navigation performance.

[0003] Underwater Gravity-aided Inertial Navigation (UGIN) is one of the key technologies for autonomous underwater vehicles (AUVs) to adaptively correct the inherent drift error accumulated by the inertial navigation system over time and assist in achieving long-term and high-precision navigation missions for underwater vehicles. Gravity (gradient) field, as a kind of geophysical attribute information, is less susceptible to the influence of ocean environments such as climate and waves than the seabed topography that is easily affected by crustal movement or the geomagnetic field that is easily interfered by other signals, and has a relatively more stable long-term invariant characteristic; at the same time, it can obtain the gravity digital reference map loaded by underwater submersibles through the altimetry inversion of the ocean gravity field by the Global Navigation Satellite System (GNSS) and the gravity continuation technology integrated with the water depth, and assist the autonomous navigation of underwater submersibles. Therefore, gravity-based assisted inertial navigation technology has become an important research topic for underwater passive navigation and intelligent positioning (such as underwater gravity matching navigation, adaptation area optimization, and track planning).

[0004] The three core modules of gravity-assisted inertial navigation system for underwater passive navigation are gravity digital reference map, gravity field information measuring instrument and gravity matching algorithm, which correspond to the high-precision and high-resolution gravity map drawing and encryption of satellite / aircraft / shore / ship multi-source data fusion, the development of high-precision gravimeter and the research of high-performance matching algorithm. Limited by the fact that the former two require a lot of resources (such as manpower, material resources and financial resources) and the development cycle is relatively long, how to improve the performance of underwater passive navigation under the conditions of existing gravity map and measuring instrument requires the performance enhancement research of gravity matching navigation algorithm. At present, the common underwater gravity matching algorithms are mainly single point matching method and sequence matching method. The former is mainly based on Sandia Inertial Terrain Aided Navigation, SITAN algorithm and its extended variants. Although the matching is real-time, it is weak in robustness; the latter is represented by TERCOM and ICCP algorithms. On the contrary, its matching robustness is high but the positioning efficiency is relatively low. Therefore, the two major goals of the sequence matching method to assist underwater passive navigation are to pursue high positioning efficiency while ensuring high matching reliability. As a sequence matching algorithm, TERCOM usually generates matching domain grid points with half-side length of three times the accumulated drift error of the inertial navigation system (3sigma) and performs optimal matching positioning (σ represents the standard deviation of the navigation error of the inertial navigation system since the last update). It has the advantages of being insensitive to initial position errors and having strong matching robustness, and has received continuous attention and research from scholars.

[0005] The research on TERCOM matching reliability mainly involves the improvement of positioning accuracy and the analysis of mismatching inducements and weakening / avoidance mechanisms that seriously affect matching accuracy. Mismatching is a matching failure phenomenon in which the distance between the best calibration position output by the algorithm and the actual position of the submersible exceeds the maximum allowable error limit. The multiple accumulation or single extreme value of the positioning error will affect the matching reliability of underwater navigation. In comparison, mismatches occurring in the 3sigma domain generally only have matching errors of several grid resolutions, and their single impact is relatively small, but the multiple cumulative impacts are serious, which belongs to the multiple cumulative impact type mismatch; while mismatches occurring outside the 3sigma domain are usually as high as matching errors of dozens of grid resolutions, and their single impact is large, which belongs to the single large impact type mismatch. Therefore, the former is called in-domain mismatch, and the latter is called out-of-domain mismatch. Existing research mainly focuses on in-domain mismatches, where the causes of mismatches include flat or smooth terrain, subtle feature changes (especially in areas with similar background feature changes), and unpredictable changes such as vegetation changes or wind-driven sand dunes. At the same time, mismatch diagnosis methods based on restricted spatial order constraints or similarity extreme value detection

[11] , a priori recursive iterative least squares mismatch correction methods, and relative position constraint pattern matching methods have been proposed to reduce the probability of this type of mismatches, but they have not considered out-of-domain mismatches. Out-of-domain mismatch is a new type of mismatch discovered in our research [Zhao S, Zheng W, Li Z, et al. Improving Matching Efficiency and Out-of-Domain Positioning Reliability of Underwater Gravity Matching Navigation Based on a Novel Domain-Center Adaptive-Transfer Matching Method[J], IEEE Trans. Instrum. Meas., vol. 72, pp. 1–11, Jan. 2023.], which is a link that must be overcome to promote the application of TERCOM-type matching algorithms in underwater submersibles. According to the previous exploration, the essential reason for the out-of-domain mismatch is the boundary limitation of the 3sigma inherent matching domain, which makes the generated matching grid domain unable to effectively cover the real position of the submersible outside the TERCOM matching domain. The traversal matching of all grid points in the domain is just a futile waste of time but cannot change the inevitability of out-of-domain mismatch. Therefore, it is necessary to break the fixed boundary limitation of the traditional TERCOM matching domain and adopt a dynamic domain mechanism that can adaptably cover the real position of the submersible to enhance the effective coverage and high-reliability positioning of the real position of the submersible outside the domain.

[0006] In the literature [Zhao S, Zheng W, Li Z, et al. Improving the Out-of-domain Matching Reliability and Positioning Accuracy of Underwater Gravity Matching Navigation based on a Novel Cyclic Boundary Semisquare-domain Researching Method [J]. IEEE Sensors Journal, 2023.], a cyclic boundary semisquare domain re-search method (Cyclic Boundary Semisquare-domain Researching Method, CBSR) is proposed. Through the boundary judgment of the optimal calibration position and the matching domain and the selective regeneration technology of the semisquare domain, the matching reliability and positioning accuracy of the real position of the underwater submersible are effectively improved. The effectiveness of the TERCOM out-of-domain re-matching mechanism that breaks the fixed matching domain boundary is experimentally verified. Unfortunately, the matching efficiency of this method is limited by the traversal comparative matching of more grid points in the initial matching domain and the iteratively regenerated semisquare domain. Therefore, while taking into account the enhancement of matching reliability, it is also necessary to strengthen the research on the efficiency of underwater passive navigation and intelligent positioning. Summary of the invention

[0007] In order to solve the technical problem in the prior art that the ergodic comparative matching of more grid points in the initial matching domain and the iterative regeneration semi-square domain limits the matching efficiency of the method, the embodiment of the present invention provides an underwater intelligent matching method and device based on an adaptive variable scale domain. The technical solution is as follows:

[0008] On the one hand, an underwater intelligent matching method based on an adaptive variable scale domain is provided, characterized in that the method comprises:

[0009] S1. Based on the contour distribution characteristics of underwater navigation evaluation index values, an adaptive variable scale domain matching method is designed; through the adaptive variable scale domain matching method, combined with the optimal principle of evaluation index, the best matching position in the variable scale domain is determined;

[0010] S2, introducing the particle swarm optimization algorithm into the adaptive variable scale domain method, and obtaining the variable scale domain local small domain evolutionary matching method based on particle swarm optimization;

[0011] S3, based on the variable scale domain local small domain evolutionary matching method of particle swarm optimization, roughly locate the best matching position, and obtain the current optimal solution position and evaluation index value;

[0012] S4. Based on the current optimal solution position and the evaluation index value, a local small domain fine positioning mechanism of the best evaluation value is constructed at the current optimal solution position; based on the local small domain fine positioning mechanism of the best evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

[0013] Optionally, in S1, based on the contour distribution characteristics of the underwater navigation evaluation index value, an adaptive variable scale domain matching method is designed; the adaptive variable scale domain matching method is combined with the optimal principle of the evaluation index to determine the best matching position in the variable scale domain, including:

[0014] Determine the end point of the INS indicated track The coordinates of the nearest neighbor grid point on the gravity map

[0015] Calculate is the corresponding sequence evaluation index of the end point to be calibrated

[0016] The inertial navigation drift error is taken as the domain half-side length. The domain half-side length increases with the MSD value, and the domain half-side length increases in sections. The lateral half-side length of the variable scale domain is calculated as and half length of the longitudinal side

[0017] Taking the grid resolution R of the gravity map as the domain grid interval, determine the number of horizontal grid points of the half-side length of the variable scale domain and the number of vertical grid points

[0018] Position indication by inertial navigation The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain, and the number of half-edge horizontal grid points and the number of vertical grid points Generate a grid point domain to be matched, and iteratively calculate the grid points in the scaled domain And its evaluation indicators

[0019] According to the evaluation index The optimal principle obtains the best matching position in the variable scale domain

[0020] Optionally, the horizontal half-side length of the variable scale domain is calculated and half length of the longitudinal side include:

[0021] Calculate the horizontal half-side length of the variable scale domain according to the following formula

[0022]

[0023] in, Indicates the end point of the track indicated by the inertial navigation The coordinates of the nearest neighbor grid points on the gravity map MAP The corresponding evaluation index value for the submersible calibration position;

[0024] Presets The boundary value is based on the calculation of the variable scale domain transverse half-side length formula and The horizontal half-side length of the variable scale domain is constructed by the boundary value of The relaxation rule of

[0025] The formula for calculating the horizontal half-side length of the variable scale domain is σ x Replace with σ y , calculate the longitudinal half length And the longitudinal half-side length relaxation rule.

[0026] Optionally, use inertial navigation to indicate position The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain, and the number of half-edge horizontal grid points and the number of vertical grid points Generate a grid point domain to be matched, and iteratively calculate the grid points in the scaled domain And its evaluation indicators include:

[0027] The coordinates of each grid point in the domain As the end point to be matched for underwater submersible calibration, the position of the gravity sensor N samplings corresponding to the track line L is inferred in reverse order by the speed υ and heading θ of the submersible. And calculate and extract the coordinates of its nearest neighbor grid points on the gravity map MAP And the corresponding gravity value Then get Sequence of approximate gravity values ​​for the track line of underwater vehicle calibration position

[0028] Record and output the gravity value sequence of the sequential sampling points measured by the submersible sensor Calculate the grid points in the variable scale domain Evaluation Metrics

[0029]

[0030] Optionally, in S3, based on the variable scale domain local small domain evolutionary matching method of particle swarm optimization, the best matching position is roughly located to obtain the current optimal solution position and evaluation index value, including:

[0031] In the D-dimensional search space of the problem to be optimized, there are P particles forming a population, and each particle represents a feasible solution to the problem to be optimized; calculate the speed and position update information of the k-th particle at the t+1th iteration of the T-iteration process in the particle swarm optimization algorithm;

[0032] End point indicated by inertial navigation The nearest neighbor point position on the graph The center is the number of relative grid points (m1, m2) in the horizontal and vertical directions that deviate from the center position in the adaptive variable scale domain as the optimization object; when the t∈{1,2,…,T}th iteration, the k∈{1,2,…,P}th particle is (m k,1 (t),m k,2 (t)), calculate the corresponding grid point position At this time, the fitness value of the kth particle is As an evaluation indicator of the end point to be calibrated

[0033] The current optimal solution is updated in the iterative optimization process through the particle swarm optimization algorithm. Location and evaluation index values

[0034] Construct a multi-level iterative auxiliary mechanism. When the maximum number of iterations T is reached, Greater than the preset index value The particle swarm optimization algorithm will continue to iterate until the particle swarm optimization algorithm reaches the additional maximum number of iterations T Add The iteration is terminated when .

[0035] Optionally, S2 also includes:

[0036] During the iterative evolution of the population, check whether the newly generated speed and position are out of bounds. If so, out-of-bounds processing is required. The out-of-bounds problem is processed according to the following formula:

[0037]

[0038]

[0039] Optionally, a multi-level iterative assistance mechanism, including:

[0040] The level-level iterative auxiliary mechanism is constructed according to the following formula:

[0041]

[0042] Optionally, the multi-level iterative auxiliary mechanism further includes:

[0043] Construct a triggered termination criterion for the particle swarm optimization algorithm; the termination criterion includes: when the current optimal solution after update Evaluation index value Not greater than the preset evaluation index value threshold Then the iteration is terminated and the trigger marker M is recorded. Trigger is 1; otherwise, because the termination criterion is not triggered, the particle swarm optimization algorithm continues until the maximum number of iterations T is reached. Add and terminate, at this time M Trigger =0.

[0044] Optionally, in S4, based on the current optimal solution position and the evaluation index value, a local small domain fine positioning mechanism of the best evaluation value is constructed at the current optimal solution position, and based on the local small domain fine positioning mechanism of the best evaluation value, the best position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain, including:

[0045] Based on the current optimal solution position and evaluation index value, at the optimal solution position Regenerate a small grid area locally and hope to find a better re-matching position based on probability The optimal solution position of the particle swarm optimization algorithm is the center of the local domain, with the horizontal half length and the horizontal half length and the vertical half length It is to stretch the horizontal and vertical half side lengths into local and then match small domains;

[0046] According to the optimal solution Evaluation indicators The relative size of the small domain is adaptively determined by layering to determine the horizontal half-side length of the small domain and the horizontal half length and the vertical half length

[0047] Transverse half-edge length based on small domain and the horizontal half length and the vertical half length Calculate the number of grid points on the horizontal and vertical half sides of the small domain, and obtain the position coordinates of each grid point in the local re-matching small domain;

[0048] According to the evaluation index Calculate the grid points in the local domain Evaluation Metrics And obtain the best matching position for fine positioning according to the optimal evaluation value principle That is, the best matching position finally output by PSO-VLEM.

[0050] On the other hand, an underwater intelligent matching device based on an adaptive variable scale domain is provided, and the device is applied to an underwater intelligent matching method based on an adaptive variable scale domain, and the device comprises:

[0051] The module for designing the adaptive variable scale domain matching method is used to design the adaptive variable scale domain matching method based on the contour distribution characteristics of the underwater navigation evaluation index value; the adaptive variable scale domain matching method is used to determine the best matching position in the variable scale domain in combination with the optimal principle of the evaluation index;

[0052] Algorithm optimization module, used to introduce particle swarm optimization algorithm into adaptive variable scale domain method, and obtain variable scale domain local small domain evolutionary matching method based on particle swarm optimization;

[0053] The coarse positioning module is used to roughly locate the best matching position based on the variable scale domain local small domain evolutionary matching method of particle swarm optimization to obtain the current optimal solution position and evaluation index value;

[0054] The precise positioning module is used to construct a local small domain fine positioning mechanism of the best evaluation value at the current optimal solution position based on the current optimal solution position and the evaluation index value; based on the local small domain fine positioning mechanism of the best evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

[0055] On the other hand, an underwater intelligent matching device based on an adaptive variable scale domain is provided, and the underwater intelligent matching device based on an adaptive variable scale domain comprises: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned underwater intelligent matching methods based on the adaptive variable scale domain is implemented.

[0056] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned underwater intelligent matching methods based on adaptive variable scale domain.

[0057] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0058] In the embodiment of the present invention, an adaptive variable scale domain intelligent matching method is proposed to improve the comprehensive performance of underwater gravity matching navigation, such as out-of-domain matching reliability and positioning efficiency. The adaptive variable scale domain matching method generates an initial matching domain with the neighboring grid points and their MSD values ​​of the inertial navigation indication terminal position to fully cover the real position of the underwater submersible, and calculates the MSD value of the grid points in the domain to find the best matching position. Considering the matching efficiency problem of the adaptive variable scale domain matching method, this paper proposes a variable scale domain local small domain evolutionary matching method based on particle swarm optimization. By introducing a particle swarm optimization algorithm with high efficiency and high precision optimization performance, the variable scale matching grid domain is roughly positioned for the best matching position, and then supplemented by the fine positioning of the local small domain re-matching and the final best matching position is obtained according to the optimal matching index principle, so as to calibrate the inertial navigation system sensor parameters and assist in completing the underwater navigation task. The test results show that AVDIM can effectively improve the out-of-domain matching reliability and positioning efficiency of underwater gravity matching navigation. At the same time, in the simulation test of the track in three different gravity zones, the occurrence of out-of-domain mismatching was completely avoided, and the average out-of-domain positioning accuracy of AVDIM was improved by 91%, 74% and 81% respectively compared with TERCOM. The excellent underwater navigation performance and good applicability of the new adaptive variable scale domain intelligent small domain matching method in different gravity zones were verified. In future research, other advanced intelligent optimization algorithms can be introduced into underwater gravity matching navigation to further improve its matching efficiency and positioning accuracy, thereby promoting the development of efficient and high-precision underwater autonomous navigation technology. This will provide more reliable technical support for long-term stable navigation in complex marine environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 It is a flow chart of an underwater intelligent matching method based on an adaptive variable scale domain provided by an embodiment of the present invention;

[0061] Figure 2 is a statistical characteristic diagram of the underwater navigation problem provided by an embodiment of the present invention;

[0062] Figure 3 is a schematic diagram of adaptive changes in a variable scale domain provided by an embodiment of the present invention;

[0063] Figure 4 is a comparison chart of out-of-domain mismatch test performance provided by an embodiment of the present invention;

[0064] Figure 5is a diagram of AVDM and PSO-VLEM matching results of a real position in a 3σ domain provided by an embodiment of the present invention;

[0065] Figure 6 AVDM and PSO-VLEM matching result diagram of the real position outside the 3σ domain provided by an embodiment of the present invention;

[0066] Figure 7 is a diagram of the matching result of the multi-level iterative auxiliary mechanism of PSO-VLEM provided by an embodiment of the present invention;

[0067] Figure 8 is a comparison diagram of out-of-domain matching performance of three test tracks in different gravity zones provided by an embodiment of the present invention;

[0068] Fig. 9 It is a block diagram of an underwater intelligent matching device based on an adaptive variable scale domain provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0069] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0070] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0071] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0072] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0073] The embodiment of the present invention provides an underwater intelligent matching method based on an adaptive variable scale domain, which can be implemented by an underwater intelligent matching device based on an adaptive variable scale domain, and the underwater intelligent matching device based on an adaptive variable scale domain can be a terminal or a server. Figure 1 The flowchart of the underwater intelligent matching method based on the adaptive variable scale domain is shown. The processing flow of the method may include the following steps:

[0074] S1. Based on the contour distribution characteristics of underwater navigation evaluation index values, an adaptive variable scale domain matching method is designed. The optimal matching position in the variable scale domain is determined by the adaptive variable scale domain matching method combined with the optimal principle of evaluation index.

[0075] In a feasible implementation, the positioning error of underwater gravity-assisted inertial navigation mainly comes from mismatching, among which out-of-domain mismatching is the main factor causing large positioning errors. To better understand out-of-domain mismatching, this paper analyzes and counts the statistical characteristics of underwater navigation problems, such as Figure 2 The traditional TERCOM algorithm does not fully utilize the statistical characteristics of underwater navigation problems. Figure 2 It can be clearly seen in (a) that quite a few of the real positions of underwater submersibles are not covered by the 3sigma domain, which will limit the positioning accuracy of underwater navigation to some extent. Based on the contour distribution characteristics of the evaluation index value, this paper proposes an adaptive variable-scale domain matching (AVDM) method to avoid the TERCOM algorithm matching domain being unable to fully cover the real position of the underwater submersible, thereby causing mismatches outside the domain with large positioning errors. Therefore, expanding the matching domain can improve matching performance and reduce positioning errors, but this will also increase matching time, and it is necessary to find a balance between matching accuracy and matching efficiency. That is, under the premise of ensuring high robustness of sequence matching, how to effectively improve matching efficiency is an important issue in this type of research.

[0076] The intelligent optimization algorithm not only significantly improves the positioning accuracy of the underwater gravity matching navigation algorithm, but also performs well in improving the positioning efficiency, thereby ensuring that the underwater submersible can achieve efficient and stable navigation positioning. Therefore, the present invention proposes a particle swarm optimization-based variable-scale-domain local-small-domain evolutionary matching method (PSO-VLEM), which simultaneously improves the overall matching efficiency of underwater navigation positioning after achieving the same level of matching accuracy as AVDM.

[0077] In a feasible implementation manner, the contour distribution characteristic mainly refers to the distribution law of the index value displayed by the contour (isovalue) map in different spatial regions.

[0078] In a feasible implementation, in S1, an adaptive variable scale domain matching method is designed based on the contour distribution characteristics of the underwater navigation evaluation index value; the adaptive variable scale domain matching method is combined with the optimal principle of the evaluation index to determine the best matching position in the variable scale domain, including:

[0079] Determine the end point of the INS indicated track The coordinates of the nearest neighbor grid point on the gravity map

[0080] Calculate is the corresponding sequence evaluation index of the end point to be calibrated

[0081] Taking the inertial navigation drift error as the half-side length of the domain, the half-side length of the domain increases with the MSD value, and the half-side length of the domain increases in sections. The lateral half-side length of the variable scale domain is calculated as and half length of the longitudinal side

[0082] Taking the grid resolution R of the gravity map as the domain grid interval, determine the number of horizontal grid points of the half-side length of the variable scale domain and the number of vertical grid points

[0083] Position indication by inertial navigation The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain, and the number of half-edge horizontal grid points and the number of vertical grid points Generate a grid point domain to be matched, and iteratively calculate the grid points in the scaled domain And its evaluation indicators

[0084] According to the evaluation index The optimal principle obtains the best matching position in the variable scale domain

[0085] In a feasible implementation manner, the inertial navigation here represents an inertial navigation system; the matching index value MSD reflects to a certain extent the distance between the inertial navigation indicated position and the actual position of the submersible, that is, the smaller the MSD value at the inertial navigation indicated position, the smaller the matching grid domain can cover the actual position of the submersible; conversely, a larger matching grid domain is required to cover the actual position of the submersible. Therefore, the half-side length of the matching grid domain can be adaptively changed according to the MSD value at the inertial navigation indicated position, as shown in FIG. Figure 3 shown.

[0086] In a feasible implementation manner, the present invention calculates the inertial navigation drift error based on the gyroscope drift, accelerometer zero bias, velocity error, angle error and other errors of the inertial navigation system and the time factor; in order to ensure the high reliability of the matching grid domain under a small MSD value, the inertial navigation drift error is used as the half-side length of the domain, and the half-side length of the domain increases in sections as the MSD value increases, then the horizontal half-side length of the domain The calculation formula is

[0087]

[0088] in, Indicates the end point of the track indicated by the inertial navigation The coordinates of the nearest neighbor grid points on the gravity map MAP The corresponding evaluation index value for the submersible calibration position;

[0089] In addition, combined with probability theory, to ensure the basis The MSD index value is calculated based on the position of the nearest neighboring grid point of the inertial navigation system to obtain the high reliability and robustness of the matching grid domain generated by the value and to avoid the time consumption of searching a large number of invalid grid points in the matching grid domain. The present invention takes As bounded; Based on the calculation of the variable scale domain transverse half-side length formula and The horizontal half-side length of the variable scale domain is constructed by the boundary value of The relaxation rule is:

[0090]

[0091] Similarly, with σ y Substitute σ in equation (2) x , we can get the adaptive domain variable scale longitudinal half-side length The calculation formula is

[0092]

[0093] In a feasible implementation, the grid resolution R of the gravity map is used as the domain grid point interval, and the number of horizontal grid points of the half-side length of the adaptive variable scale domain is and the number of vertical grid points The calculation formulas are:

[0094]

[0095] In a feasible implementation mode, the position is indicated by inertial navigation. The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain, and the number of half-edge horizontal grid points and the number of vertical grid points Generate the grid point domain to be matched, then the position coordinates of each grid point in the adaptive variable scale domain The calculation formula is

[0096]

[0097] in,

[0098] Iterative calculation of grid points in variable scale domain And its evaluation indicators The coordinates of each grid point in the domain As the end point to be matched for underwater submersible calibration, the position of the gravity sensor N samplings corresponding to the track line L is inferred in reverse order by the speed υ and heading θ of the submersible. And calculate and extract the coordinates of its nearest neighbor grid points on the gravity map MAP And the corresponding gravity value Then get Sequence of approximate gravity values ​​for the track line of underwater vehicle calibration position

[0099] Record and output the gravity value sequence of the sequential sampling points measured by the submersible sensor Calculate the grid points in the variable scale domain Evaluation Metrics

[0100]

[0101] By evaluation index The optimal principle obtains the best matching position in the variable scale domain To calibrate the position of the underwater submersible inertial navigation system and assist in completing the underwater diving mission. The execution pseudo code of AVDM is shown in Algorithm 1.

[0102]

[0103]

[0104] S2. Introduce the particle swarm optimization algorithm into the adaptive variable scale domain method to obtain a variable scale domain local small domain evolutionary matching method based on particle swarm optimization.

[0105] In a feasible implementation mode, the AVDM method proposed in this application indicates the neighboring grid points of the inertial navigation. The evaluation index value at When is small, the generated variable scale matching grid domain is small, which is conducive to achieving high matching efficiency and high positioning reliability of underwater gravity matching navigation; but when When the value is large, a larger variable-scale matching grid domain needs to be generated. Although it will improve the positioning reliability of AVDM, it is not conducive to maintaining high matching efficiency due to the high time consumption of traversing all grid points in the domain. Therefore, this section introduces an intelligent optimization algorithm (i.e., particle swarm optimization algorithm) with high efficiency and high precision optimization performance into the coarse positioning of the best matching position in the variable-scale matching grid domain, and then supplements it with the fine positioning of local small domain re-matching, and proposes a variable-scale domain local small domain evolutionary matching method based on particle swarm optimization (Particle Swarm Optimization-based Variable-scale-domain Local-small-domain Evolutionary Matching, PSO-VLEM).

[0106] S3, based on the variable scale domain local small domain evolutionary matching method of particle swarm optimization, roughly locate the best matching position, and obtain the current optimal solution position and evaluation index value;

[0107] In a feasible implementation, the particle swarm optimization algorithm (PSO) is a classic intelligent optimization algorithm proposed by Eberhart and Kennedy inspired by the foraging behavior of bird flocks in nature. It simulates the cooperation and information sharing mechanism between individual bird flocks to find the optimal solution to the problem through parallel iterative optimization of multiple individuals.

[0108] In a feasible implementation, in S3, based on the variable scale domain local small domain evolutionary matching method of particle swarm optimization, the best matching position is roughly located to obtain the current optimal solution position and evaluation index value, including:

[0109] In the D-dimensional search space of the problem to be optimized, there are P particles forming a population, and each particle represents a feasible solution to the problem to be optimized; if the position and velocity of the kth particle are recorded as X k =(x k1 ,x k2 ,…,x kD ) and V k =(v k1 ,v k2 ,…,v kD ), then calculate the speed and position update information of the kth particle at the t+1th iteration of the Tth iteration process in the particle swarm optimization algorithm:

[0110]

[0111] x k,j (t+1)=x k,j (t)+v k,j (t+1) (8)

[0112] Where k = 1, 2, ..., P represents the kth particle in the population, j = 1, 2, ..., D represents the jth dimension of the particle, represents the jth dimension of the individual optimal solution of the kth particle, Represents the jth dimension of the global optimal solution of the particle population; c1=c2=1.49445 is a constant set by the algorithm, and r1=r2∈[0,1] is a random number.

[0113] In order to realize the intelligent positioning of the best matching position in the adaptive variable scale domain based on the PSO algorithm, the present invention uses the inertial navigation to indicate the end point. The nearest neighbor point position on the graph The PSO particle is centered on the horizontal and vertical relative grid points (m1, m2) that deviate from the center position in the adaptive variable scale domain as the optimization object, that is, the particle dimension is 2-dimensional, that is, the dimension D = 2, where Indicates the number of lateral deviations from the grid points and when the grid point is located On the right side, the value is a positive number; Indicates the number of vertical deviation grid points and when located When the value is positive, the value is positive when the iteration is t∈{1,2,…,T}th; when the k∈{1,2,…,P}th particle is (m k,1 (t),m k,2 (t)), calculate the corresponding grid point position In the present invention, T and P are both set to 10, then the corresponding grid point position The calculation formula is:

[0114]

[0115] At this time, the fitness value of the kth particle is As an evaluation indicator of the end point to be calibrated

[0116] In addition, to adapt to the adaptive scale factor index value in the variable scale domain The difference in particle velocity V k =(v k,1 ,v k,2 )'s horizontal and vertical valid value ranges are set to and The present invention sets ξ to 0.2. Therefore, the initialization calculation formula for the position and velocity of the kth particle in the j∈{1,2}th dimension is:

[0117]

[0118]

[0119] in, and They represent the upper and lower bounds of the effective values ​​of the j-th dimension of the particle, respectively. Taking j = 1 as an example, then and Similarly, we can get and The corresponding value of .

[0120] The current optimal solution is updated in the iterative optimization process through the particle swarm optimization algorithm. Location and evaluation index values

[0121] Construct a multi-level iterative auxiliary mechanism. When the maximum number of iterations T is reached, Greater than the preset index value The particle swarm optimization algorithm will continue to iterate until the particle swarm optimization algorithm reaches the additional maximum number of iterations T Add The iteration is terminated when .

[0122] In a feasible implementation manner, S3 further includes:

[0123] When the particles iteratively update their speed and position according to equations (7) and (8), the phenomenon of exceeding the effective value range of the variables may occur, which is called the out-of-bounds problem. Therefore, during the iterative evolution of the population, it is necessary to check whether the newly generated speed and position have an out-of-bounds problem. If so, it is necessary to perform an out-of-bounds processing. The out-of-bounds problem is processed according to the following formula:

[0124]

[0125]

[0126] Obviously, formula (13) is equivalent to the particle initialization operation, which is conducive to ensuring the diversity of the population.

[0127] In a feasible implementation, the multi-level iterative auxiliary mechanism includes:

[0128] A multi-level iterative auxiliary mechanism is constructed according to the following formula:

[0129]

[0130] In one feasible implementation, to prevent the maximum number of iterations T from being reached, Larger, leading to the optimal solution The location may be far from the real location, resulting in out-of-domain mismatching. Therefore, a multi-level iterative auxiliary mechanism is designed. Greater than the preset index value The present invention sets The algorithm will continue to iterate until PSO reaches the maximum number of additional iterations T Add and terminate.

[0131] In a feasible implementation manner, the multi-level iterative auxiliary mechanism further includes:

[0132] Since the PSO algorithm updates the current optimal solution during the iterative optimization process Evaluation index value of the location is relatively small, and more iterative matching of multiple individuals may not necessarily find a better solution position. Therefore, we designed a trigger termination criterion for the PSO algorithm;

[0133] Termination criteria include: when the current optimal solution after update Evaluation index value Not greater than the preset evaluation index value threshold Then the iteration is terminated and the trigger marker M is recorded. Trigger is 1; otherwise, because the termination criterion is not triggered, the particle swarm optimization algorithm continues until the maximum number of iterations T is reached. Add and terminate, at this time M Trigger = 0. This paper refers to the literature [Zhao S, Zheng W, Li Z, et al. A Novel Cross-line Adaptive Domain Matching Algorithm for Underwater Gravity Aided Navigation [J]. IEEE Geosci. Remote Sensing Lett., vol. 21, pp. 1-5, 2024, Art no. 7503605.] In order to synchronously regulate the matching reliability and positioning efficiency of underwater gravity-assisted navigation based on the PSO-VRM mechanism.

[0134] S4. Based on the current optimal solution position and the evaluation index value, a local small-domain precise positioning mechanism (PSO Optimal Evaluation Value-based Local-small-domain Precise-positioning Mechanism, OEV-LPM) with the best evaluation value is constructed at the current optimal solution position; based on the local small-domain precise positioning mechanism with the best evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

[0135] In one feasible implementation, considering the optimal solution found by the PSO algorithm Probably the best match In order to avoid local miss of the best matching position, it is necessary to Regenerate a small grid area locally and hope to find a better re-matching position based on probability

[0136] In a feasible implementation, in S4, based on the current optimal solution position and the evaluation index value, a local small domain fine positioning mechanism of the best evaluation value is constructed at the current optimal solution position, and based on the local small domain fine positioning mechanism of the best evaluation value, the best position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain, including:

[0137] Based on the current optimal solution position and evaluation index value, at the optimal solution position A small grid is regenerated locally; since the PSO algorithm has a certain randomness, it is expected to find a better re-matching position by probability. The optimal solution position of the particle swarm optimization algorithm is the center of the local domain, with the horizontal half length and the horizontal half length and the vertical half length It is to stretch the horizontal and vertical half side lengths into local and then match small domains;

[0138] According to the optimal solution Evaluation indicators The relative size of the small domain is adaptively determined by layering to determine the horizontal half-side length of the small domain and the horizontal half length and the vertical half length To locally match the horizontal half-side length of the small domain For example, the calculation formula is:

[0139]

[0140] σ y Substitute σ in equation (14) x The longitudinal half-side length of the local re-matching domain can be obtained Furthermore and Substitute (4) and The number of grid points on the horizontal and vertical half sides of the small domain can be obtained and Then locally match the position coordinates of each grid point in the small domain The calculation formula is

[0141]

[0142] in,

[0143] Transverse half-edge length based on small domain and the horizontal half length and the vertical half length Calculate the number of grid points on the horizontal and vertical half sides of the small domain, and obtain the position coordinates of each grid point in the local re-matching small domain;

[0144] According to the evaluation index Formula (6) is derived from the context, and the grid points in the local domain are calculated Evaluation Metrics And obtain the best matching position for fine positioning according to the optimal evaluation value principle That is, the best matching position finally output by PSO-VLEM.

[0145] In a feasible implementation, PSO-VLEM uses a particle swarm optimization algorithm to achieve the optimal position rough positioning in the adaptive variable scale domain, and then uses the optimal position fine positioning in the local re-matching small domain to achieve high reliability and high efficiency positioning of the real position of the underwater gravity-assisted inertial navigation submersible. The specific implementation pseudo code is shown in Algorithm 2.

[0146]

[0147]

[0148] In a feasible implementation manner, the present invention verifies the effectiveness of AVDM and PSO-VLEM in improving the underwater navigation positioning efficiency and out-of-domain matching reliability through the following simulation experiments.

[0149] In a feasible implementation, the experimental sea area is the longitude and latitude range of the South China Sea (113°E-115°E, 17°N-19°N). The gravity data with a resolution of 1'×1' comes from the website of the University of California, San Diego (http: / / topex.ucsd.edu / ), and is encrypted into 100m×100m gravity data by bilinear interpolation method. The experimental platform is a personal computer with Windows 11 operating system and Intel(R)Core(TM)I7-13700HX CPU. The simulation parameter settings of the underwater submersible and sensor are shown in Table 1.

[0150] Table 1 Experimental parameters

[0151]

[0152]

[0153] In all tests, coverage rate (CR), matching success rate (MSR), average positioning accuracy (Ave), standard deviation (STD), worst positioning accuracy (Worst) and best positioning accuracy (Best) are used to evaluate the performance of underwater navigation algorithms. Among them, the matching success rate represents the proportion of valid matching times with matching accuracy less than the judgment threshold in the total number of tests (the judgment threshold of matching accuracy in this paper is set as the diagonal length of the unit grid resolution, i.e. ).

[0154] Coverage is the probability or proportion of evaluating whether the real position of the submersible is covered by the matching domain. Specifically, if there is a set of all tested matching domains A and a real position RP, there are n regions A1, A2, ..., A n , each area A i It has its own coverage. In this paper, n is set to 10000, and the calculation formula of CR is:

[0155]

[0156] Among them A i The value of (RP) is 1 or 0. When the real position RP is i When covered, A i (RP) = 1; otherwise, A i (RP) = 0. The value of CR is divided into three cases: when CR = 100%, it means that the real position RP is covered by all matching domains; when 0 < CR < 100%, it means that the real position RP is covered by some matching domains; when CR = 0, it means that the real position RP is not covered by any matching domain.

[0157] The starting point (SP) of the simulation experiment is the grid coordinates (300, 950) on the gravity anomaly reference map. A simulation track is generated according to the relevant parameters in Table 1. In order to ensure the same experimental conditions and fairness of the comparison results of the comparative test, the TERCOM algorithm is used to repeat the experiment 10,000 times to obtain and save the benchmark condition data of each test and use it as the import data of the proposed AVDM and PSO-VLEM. The re-experiments based on AVDM and PSO-VLEM are completed under the same test conditions. The matching results of each experiment are recorded and summarized. In particular, according to the relative position between the 3σ domain boundary of TERCOM and the actual position of the submersible and whether mismatching occurs, all experiments will be divided into two types of tests: overall performance experiments and out-of-domain performance experiments. The corresponding numbers of in-domain mismatching and out-of-domain mismatching are recorded as IMN and OMN, respectively. The corresponding comparative statistical results are shown in Tables 2 and 3, respectively.

[0158] Table 2 Comparative statistical results based on different matching algorithms (10,000 times)

[0159]

[0160] By analyzing the above Table 2, it can be seen that AVDM and PSO-VLEM are better than the comparison algorithms in most indicators in all tests. Compared with TERCOM, the worst positioning accuracy of AVDM and PSO-VLEM is improved by about 80%. And combined with the coverage and worst positioning accuracy of TERCOM, it can be seen that although out-of-domain mismatches do not occur often, once they occur, a large matching error will occur, thus affecting the subsequent continuous navigation results. The 49 out-of-domain mismatches that occurred in TERCOM were completely avoided by AVDM and PSO-VLEM, of which five mismatches were converted from out-of-domain to in-domain. The above results verify the overall matching performance of AVDM and PSO-VLEM, confirming that they have efficient and high-precision underwater navigation capabilities.

[0161] Table 3. Comparison statistics of traditional 3σ domain based on different matching algorithms

[0162]

[0163] Note: a(b) indicates the number of mismatches outside (inside) the domain of AVDM and PSO-VLEM compared with TERCOM

[0164] Further analysis of the various indicators under the out-of-domain test category of 3σ in Table 3 shows that, compared with TERCOM, the average positioning accuracy, standard deviation, worst positioning accuracy and best positioning accuracy of out-of-domain mismatching of AVDM are increased by 86%, 86%, 87% and 89% respectively. From the CR indicator, it can be seen that TERCOM has zero coverage of the true out-of-domain position, while AVDM and PSO-VLEM have complete coverage of the true out-of-domain position. Moreover, while PSO-VLEM achieves the same level of matching accuracy as AVDM, its matching efficiency is improved compared with AVDM and TERCOM. The above results show that AVDM and PSO-VLEM not only avoid the occurrence of out-of-domain mismatching, but also reduce the worst positioning accuracy from more than 20 grid resolutions to 3 and a half grid resolutions. And from the standard deviation, it can be seen that AVDM and PSO-VLEM have high stability. In summary, these results verify that AVDM has excellent out-of-domain matching accuracy in underwater gravity matching navigation, and PSO-VLEM achieves efficient navigation and positioning on the basis of AVDM's excellent out-of-domain matching accuracy. Among them, Figure 4 The high out-of-domain positioning accuracy of AVDM and PSO-VLEM is intuitively demonstrated.

[0165] In a feasible implementation, the focus is on exploring and analyzing the reasons for the excellent matching performance of AVDM and PSO-VLEM in underwater gravity matching navigation. In order to explore the high reliability of AVDM and PSO-VLEM positioning, according to the variable scale domain adaptive generation mechanism based on the inertial navigation indication underwater submersible terminal position index value and the intelligent matching positioning mechanism based on the modified particle swarm optimization algorithm, examples of the in-domain and out-of-domain matching results of the relative position between the 3σ domain boundary based on TERCOM and the actual position of the submersible are drawn, such as Figure 5 and Figure 6 shown.

[0166] By analyzing Figure 5 It can be seen that AVDM and PSO-VLEM effectively cover the real position of the underwater submersible in the traditional 3σ domain through the variable scale domain, and adaptively determine the size of the variable scale domain according to the MSD index value of the inertial navigation position to reduce the matching time, and further improve the positioning accuracy through the local small domain re-matching mechanism.

[0167] pass Figure 6 It is obvious that the real position of the underwater vehicle is not covered by the TERCOM matching domain, while the adaptive variable scale domains of AVDM and PSO-VLEM can effectively cover the real position, thereby avoiding the occurrence of mismatching outside the domain and satisfying the high-precision positioning capability of the real position outside the domain of the underwater vehicle. In addition, PSO-VLEM introduces the PSO algorithm with high efficiency and high precision optimization performance into the coarse positioning of the best matching position in the variable scale domain, avoiding the indiscriminate traversal search of all grid points in a larger variable scale domain, so that PSO-VLEM can meet the same level of positioning accuracy of AVDM while greatly improving the average positioning time through the PSO algorithm, which solves the problem of low matching efficiency of the underwater gravity navigation algorithm when ensuring high robustness of sequence matching to a certain extent.

[0168] In a feasible implementation, in most matches, the optimal position close to the true position can be found by setting the maximum number of PSO iterations to 10. However, there are still rare cases where the evaluation index value after 10 iterations is large, that is, the distance between the best matching position of PSO and the true position is far. Figure 7 It is intuitively shown that 10 PSO iterations with a local small domain may not cover the real position. However, the use of a multi-level iterative auxiliary mechanism can effectively achieve full coverage of the real position, thereby improving the matching efficiency and reliability of underwater gravity navigation.

[0169] In general, the above analysis results verify the excellent matching performance of the AVDM and PSO-VLEM proposed in this paper in underwater gravity matching navigation, such as high efficiency, high-precision positioning, and high reliability outside the domain. The next section will further verify the good applicability of these two algorithms in underwater gravity matching navigation through track lines in different gravity zones.

[0170] In order to further test the good applicability of AVDM and PSO-VLEM to the track lines in different gravity zones, this section uses the grid coordinates A (750, 950), B (850, 750), and C (250, 950) on the gravity anomaly map as the new starting points of each simulated track. The other experimental parameter settings refer to Table 1 and Section 4.1 to conduct 10,000 independent tests. The test results are shown in Tables 4 and 5, respectively.

[0171] Table 4 Comparative statistical results based on tracks in different gravity zones (10,000 times)

[0172]

[0173] From the analysis in Table 4, it can be seen that the results of AVDM and PSO-VLEM on the three simulated tracks in terms of matching accuracy statistics, matching success rate, and out-of-domain matching reliability are almost better than those of the TERCOM algorithm, especially in terms of the number of out-of-domain mismatches (OM index) and the worst matching accuracy (Worst index). In terms of the OM index, AVDM and PSO-VLEM completely avoid the out-of-domain mismatches that occur in TERCOM, and a single-digit out-of-domain mismatches are converted into in-domain mismatches. In terms of the worst matching accuracy index, compared with 1595.32m, 1013.19m, and 1533.44m of TERCOM, AVDM is reduced to 413.23m, 331.30m, and 326.35m, and PSO-VLEM is also reduced to 413.23m, 331.30m, and 326.35m. In terms of average positioning accuracy (Ave index), AVDM and PSO-VLEM are both improved compared with TERCOM. These results show that PSO-VLEM can maintain high positioning efficiency and high out-of-domain matching reliability in different gravity zones, and exhibits comprehensive performance such as high matching success rate and excellent matching accuracy. In summary, the proposed algorithm is superior to the traditional TERCOM algorithm in all key indicators and has significant application advantages.

[0174] Table 5 Comparative statistical results of traditional 3σ out-of-domain mismatch test for tracks in different gravity zones

[0175]

[0176] By analyzing Table 5, it can be seen that the proposed AVDM and PSO-VLEM have significantly improved various matching performance indicators in the 3σ out-of-domain mismatch test on three tracks. In particular, the occurrence of out-of-domain mismatch is completely avoided, which verifies that the AVDM and PSO-VLEM algorithms can still maintain efficient underwater navigation positioning and matching reliability in the out-of-domain mismatch test. In addition, in terms of matching accuracy statistical indicators, the Ave indicators of AVDM and PSO-VLEM are reduced from an average of 511m in TERCOM to an average of about 100m, and the average accuracy improvement ratio is 80%, which proves that the proposed AVDM and PSO-VLEM have high out-of-domain average positioning accuracy. The standard deviation of its out-of-domain accuracy (STD indicator) is about half the grid resolution on average, and the best positioning accuracy (Best indicator) is less than 20m, indicating that AVDM and PSO-VLEM have stronger matching robustness and optimal positioning value. In addition, the worst matching accuracy of AVDM and PSO-VLEM has also been reduced from an average of 1380m of the comparison algorithm to only about 200m, with an average reduction of about 85%. This shows that AVDM and PSO-VLEM can still find the best matching position with relatively small errors even in extremely bad situations, and maintain good potential application prospects in underwater navigation in different gravity zones. Figure 8 The high out-of-domain positioning accuracy of AVDM and PSO-VLEM in different gravity zones is intuitively demonstrated.

[0177] In a feasible implementation, the present invention integrates the variable scale domain adaptive generation mechanism based on the inertial navigation indication underwater submersible terminal position index value and the intelligent matching positioning mechanism based on the modified particle swarm optimization algorithm to propose a new underwater gravity matching algorithm. The main contributions are as follows:

[0178] Breaking the traditional TERCOM domain boundary restrictions, constructing an adaptive variable-scale domain based on the contour distribution characteristics, adaptively adjusting the size of the matching domain to achieve effective coverage of the real position of the submersible, and avoiding out-of-domain mismatching that causes large positioning errors.

[0179] Abandoning the traversal search mechanism of all grid points in the variable-scale domain, an intelligent optimization algorithm with high-efficiency and high-precision optimization performance is introduced to the coarse positioning of the best matching position in the variable-scale matching grid domain to ensure high-efficiency matching navigation when the variable-scale domain is large.

[0180] The local small domain obtained from the best matching position in the variable scale domain is re-matched for fine positioning, and the final output matching position is determined to correct the inertial navigation system so that the underwater gravity-assisted inertial navigation can maintain efficient and high-precision positioning capabilities.

[0181] Fig. 9is a block diagram of an underwater intelligent matching device 300 based on an adaptive variable scale domain according to an exemplary embodiment. The device 300 is used for an underwater intelligent matching method based on an adaptive variable scale domain. Fig. 9 The device includes an adaptive variable scale domain matching method design module 310, an algorithm optimization module 320, a coarse positioning module 330 and a fine positioning module 340. Among them:

[0182] The adaptive variable scale domain matching method design module 310 is used to design an adaptive variable scale domain matching method based on the contour distribution characteristics of the underwater navigation evaluation index value; through the adaptive variable scale domain matching method, combined with the optimal principle of the evaluation index, the best matching position in the variable scale domain is determined;

[0183] Algorithm optimization module 320, used to introduce particle swarm optimization algorithm into adaptive variable scale domain method, and obtain variable scale domain local small domain evolutionary matching method based on particle swarm optimization;

[0184] A coarse positioning module 330 is used to coarsely locate the best matching position based on a variable scale domain local small domain evolutionary matching method of particle swarm optimization to obtain the current optimal solution position and evaluation index value;

[0185] The precise positioning module 340 is used to construct a local small domain fine positioning mechanism of the best evaluation value at the current optimal solution position based on the current optimal solution position and the evaluation index value; based on the local small domain fine positioning mechanism of the best evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

[0186] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An underwater intelligent matching method based on adaptive variable scale domain, characterized in that: The method comprises: S1. Based on the contour distribution characteristics of underwater navigation evaluation index values, an adaptive variable scale domain matching method is designed; through the adaptive variable scale domain matching method, combined with the optimal principle of evaluation index, the best matching position in the variable scale domain is determined; S2, introducing the particle swarm optimization algorithm into the adaptive variable scale domain method to obtain a variable scale domain local small domain evolutionary matching method based on particle swarm optimization; S3, based on the variable scale domain local small domain evolutionary matching method based on particle swarm optimization, roughly locate the best matching position to obtain the current optimal solution position and evaluation index value; S4. Based on the current optimal solution position and the evaluation index value, a local small domain fine positioning mechanism of the optimal evaluation value is constructed at the current optimal solution position; based on the local small domain fine positioning mechanism of the optimal evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

2. The underwater intelligent matching method based on adaptive variable scale domain according to claim 1 is characterized in that: In S1, based on the contour distribution characteristics of the underwater navigation evaluation index value, an adaptive variable scale domain matching method is designed; The adaptive variable scale domain matching method is combined with the optimal principle of evaluation index to determine the best matching position in the variable scale domain, including: Determine the end point of the INS indicated track The coordinates of the nearest neighbor grid point on the gravity map Calculate the end point of the track is the corresponding sequence evaluation index of the end point to be calibrated The inertial navigation drift error is taken as the domain half-side length. The domain half-side length increases with the MSD value, and the domain half-side length increases in sections. The horizontal half-side length of the variable scale domain is calculated. and half length of the longitudinal side Taking the grid resolution R of the gravity map as the domain grid interval, determine the number of horizontal grid points of the half-side length of the variable scale domain and the number of vertical grid points Position indication by inertial navigation The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain; the number of half-edge horizontal grid points and the number of vertical grid points Generate a grid point domain to be matched, and iteratively calculate the grid points in the scaled domain And its evaluation indicators According to the evaluation index The optimal principle obtains the best matching position in the variable scale domain 3. The underwater intelligent matching method based on adaptive variable scale domain according to claim 1 is characterized in that: The calculated horizontal half-side length of the variable scale domain and half length of the longitudinal side include: Calculate the horizontal half-side length of the variable scale domain according to the following formula in, Indicates the end point of the track indicated by the inertial navigation The coordinates of the nearest neighbor grid points on the gravity map MAP The corresponding evaluation index value for the submersible calibration position; Presets The boundary value is based on the calculation formula of the variable scale domain transverse half-side length and the The horizontal half-side length of the variable scale domain is constructed by the boundary value of The relaxation rule of The formula for calculating the horizontal half-side length of the variable scale domain is σ x Replace with σ y , calculate the longitudinal half length And the longitudinal half-side length relaxation rule.

4. The underwater intelligent matching method based on adaptive variable scale domain according to claim 3 is characterized in that: The position indicated by inertial navigation The nearest neighbor point on the gravity map of is the center position of the adaptive variable scale domain, and the number of half-edge horizontal grid points and the number of vertical grid points Generate a grid point domain to be matched, and iteratively calculate the grid points in the scaled domain And its evaluation indicators include: The coordinates of each grid point in the domain As the end point to be matched for underwater submersible calibration, the position of the gravity sensor N samplings corresponding to the track line L is inferred in reverse order by the speed υ and heading θ of the submersible. And calculate and extract the coordinates of its nearest neighbor grid points on the gravity map MAP And the corresponding gravity value Then get Sequence of approximate gravity values ​​for the track line of underwater vehicle calibration position Record and output the gravity value sequence of the sequential sampling points measured by the submersible sensor Calculate the grid points in the variable scale domain Evaluation Metrics 5. The underwater intelligent matching method based on adaptive variable scale domain according to claim 1 is characterized in that: In S3, based on the variable scale domain local small domain evolutionary matching method based on particle swarm optimization, the best matching position is roughly located to obtain the current optimal solution position and evaluation index value, including: In the D-dimensional search space of the problem to be optimized, there are P particles forming a population, and each particle represents a feasible solution to the problem to be optimized; calculate the speed and position update information of the k-th particle at the t+1th iteration of the T-iteration process in the particle swarm optimization algorithm; End point indicated by inertial navigation The nearest neighbor point position on the graph Centered on the horizontal axis, the relative number of grid points (m1, m2) in the horizontal and vertical directions that deviate from the center position in the adaptive variable scale domain is optimized. When the t∈{1,2,…,T}th iteration, the k∈{1,2,…,P}th particle is (m k,1 (t),m k,2 (t)), calculate the corresponding grid point position At this time, the fitness value of the kth particle is As an evaluation indicator of the end point to be calibrated The current optimal solution is updated in the iterative optimization process through the particle swarm optimization algorithm. Location and evaluation index values Construct a multi-level iterative auxiliary mechanism. When the maximum number of iterations T is reached, Greater than the preset index value Then the particle swarm optimization algorithm will continue to iterate until the particle swarm optimization algorithm reaches the additional maximum number of iterations T Add The iteration is terminated when .

6. The underwater intelligent matching method based on adaptive variable scale domain according to claim 5 is characterized in that: The S2 also includes: During the iterative evolution of the population, check whether the newly generated speed and position are out of bounds. If so, out-of-bounds processing is required. The out-of-bounds problem is processed according to the following formula:

7. The underwater intelligent matching method based on adaptive variable scale domain according to claim 5 is characterized in that: The multi-level iterative auxiliary mechanism includes: A multi-level iterative auxiliary mechanism is constructed according to the following formula:

8. The underwater intelligent matching method based on adaptive variable scale domain according to claim 5 is characterized in that: The multi-level iterative auxiliary mechanism further includes: Construct a triggered termination criterion for the particle swarm optimization algorithm; the termination criterion includes: When the current optimal solution is updated Evaluation index value Not greater than the preset evaluation index value threshold Then the iteration is terminated and the trigger marker M is recorded. Trigger is 1; otherwise, because the termination criterion is not triggered, the particle swarm optimization algorithm continues until the maximum number of iterations T is reached. Add and terminate, at this time M Trigger =0.

9. The underwater intelligent matching method based on adaptive variable scale domain according to claim 5, characterized in that: In S4, based on the current optimal solution position and the evaluation index value, a local small domain fine positioning mechanism of the optimal evaluation value is constructed at the current optimal solution position, and based on the local small domain fine positioning mechanism of the optimal evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain, including: Based on the current optimal solution position and evaluation index value, at the optimal solution position Regenerate a small grid area locally and hope to find a better re-matching position based on probability The optimal solution position of the particle swarm optimization algorithm is the center of the local domain, with the horizontal half length and the horizontal half length and the vertical half length It is to stretch the horizontal and vertical half side lengths into local and then match small domains; According to the optimal solution Evaluation indicators The relative size of the small domain is adaptively determined by layering to determine the horizontal half-side length of the small domain and the horizontal half length and the vertical half length Based on the horizontal half-side length of the small domain and the horizontal half length and the vertical half length Calculate the number of grid points on the horizontal and vertical half sides of the small domain, and obtain the position coordinates of each grid point in the local re-matching small domain; According to the evaluation index Calculate the grid points in the local domain Evaluation Metrics And obtain the best matching position for fine positioning according to the optimal evaluation value principle That is, the best matching position finally output by PSO-VLEM.

10. An underwater intelligent matching device based on an adaptive variable scale domain, the underwater intelligent matching device based on an adaptive variable scale domain is used to implement the underwater intelligent matching method based on an adaptive variable scale domain as claimed in any one of claims 1 to 9, characterized in that: The device comprises: An adaptive variable scale domain matching method design module is used to design an adaptive variable scale domain matching method based on the contour distribution characteristics of the underwater navigation evaluation index value; through the adaptive variable scale domain matching method, combined with the optimal principle of the evaluation index, the best matching position in the variable scale domain is determined; An algorithm optimization module, used for introducing the particle swarm optimization algorithm into the adaptive variable scale domain method to obtain a variable scale domain local small domain evolutionary matching method based on particle swarm optimization; A coarse positioning module is used to coarsely locate the best matching position based on the variable scale domain local small domain evolutionary matching method based on particle swarm optimization, and obtain the current optimal solution position and evaluation index value; The precise positioning module is used to construct a local small domain fine positioning mechanism of the best evaluation value at the current optimal solution position based on the current optimal solution position and the evaluation index value; based on the local small domain fine positioning mechanism of the best evaluation value, the optimal position is precisely positioned to complete underwater intelligent matching based on the adaptive variable scale domain.

Images