Sound velocity profile acquisition method based on self-organizing neural network and fuzzy physical matching

Abstract

The invention provides a sound velocity profile acquisition method based on a self-organizing neural network and fuzzy physical matching, and relates to the technical field of underwater sonar, the sound velocity profile acquisition method comprises the following steps: acquiring historical profile information and historical remote sensing information of a domain sea area, and constructing a data set; obtaining a data processing model based on the data set; wherein the data processing model is a model obtained by combining training of a self-organizing neural network and a fuzzy physical relational expression, and finally, the data processing model is used for determining and obtaining sound velocity profile information corresponding to the target sea area, so that the change characteristics of the seawater sound velocity value along with the depth are determined through the sound velocity profile information. According to the method, the self-organizing neural network is matched with the fuzzy physical expression related to physical driving, neurons which do not conform to the physical law are filtered, the technical problems that in the prior art, precision is low, and errors are large are solved, and the precision of sound velocity profile obtaining is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater sonar, and in particular, to a method for obtaining a sound speed profile based on a self-organizing neural network and fuzzy physical matching. Background Art

[0002] The sound speed profile information generally refers to the variation of the seawater sound speed value with depth. In the prior art, the sound speed profile information is often used as a data reference for the high-performance operation of a sonar system. Generally, the methods for obtaining the sound speed profile information include: direct measurement method, acoustic inversion method, or statistical regression method. Among them, the direct measurement method can directly perform in-situ measurement at sea through a sound speed profile instrument or a CTD (Conductivity, Temperature, Depth) instrument. Although it has the highest accuracy, it will cost higher time and labor costs compared with the acoustic inversion method or the statistical regression method. The acoustic inversion method utilizes the effect that the sound wave is refracted due to the influence of the sound speed profile during the process of passing through the water body, and this method has high requirements for the scene conditions. Based on this, most technicians obtain the sound speed profile information through the statistical regression method.

[0003] However, although the statistical regression method can save labor costs and can obtain relatively accurate and real-time sound speed profile inversion results through publicly available data without any experimental scene restrictions, the way of establishing the regression relationship under the traditional statistical regression method cannot truly simulate the actual data of the ocean. At the same time, the measurement error also causes a certain randomness between the input parameters and the output parameters, resulting in a decrease in the accuracy of the sound speed profile information.

[0004] Based on this, there is an urgent need for a sound speed profile acquisition solution that can overcome the technical problems of low accuracy and large error in the prior art and improve the accuracy of sound speed profile acquisition. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method for obtaining a sound speed profile based on a self-organizing neural network and fuzzy physical matching. By matching the self-organizing neural network with a fuzzy physical expression related to physical driving, neurons that do not conform to physical laws are filtered, so as to overcome the technical problems of low accuracy and large error in the prior art and improve the accuracy of sound speed profile acquisition.

[0006] To achieve the above purpose, the technical solution adopted in the embodiment of the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for obtaining a sound speed profile based on a self-organizing neural network and fuzzy physical matching. The sound speed profile acquisition method includes the following steps:

[0008] Obtain the historical profile information and historical remote sensing information of the target sea area, and construct a data set;

[0009] A data processing model is obtained based on the data set; wherein, the data processing model is a model trained by combining a self-organizing neural network and a fuzzy physical relationship formula;

[0010] The data processing model is used to obtain the sound speed profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound speed value with depth in the target sea area through the sound speed profile information.

[0011] Optionally, the steps of obtaining the historical profile information and historical remote sensing information of the domain sea area and constructing the data set include:

[0012] Preprocess the historical profile information to obtain an initial function corresponding to the historical profile information;

[0013] Screen the remote sensing information with the same time resolution as the historical profile information and the closest data grid to the longitude and latitude of the corresponding historical profile information from the historical remote sensing information to obtain the final historical remote sensing information; wherein, the final historical remote sensing information includes sea surface height parameters and sea surface temperature parameters;

[0014] Take the final historical remote sensing information and the sound speed profile coefficients of each order under the initial function as a set of data samples to obtain a data set containing multiple sets of data samples.

[0015] Optionally, the calculation formula of the initial function is expressed as:

[0016] ;

[0017] Wherein, is the steady-state background profile; is the profile corresponding to the stable part in the current sound speed profile; is the sound speed profile coefficient corresponding to the nth order; is the basis function.

[0018] Optionally, the steps of obtaining the data processing model based on the data set include:

[0019] Determine the fuzzy physical relationship formula between the data samples in the data set, and the fuzzy physical relationship formula is used to characterize the mapping relationship between the basis function and the sea surface height parameter and the sea surface temperature parameter;

[0020] Based on each data sample in the data set, use the self-organizing neural network algorithm to construct a self-organizing neural network to obtain an initial data processing model, and use the multiple output results output by the initial data processing model as data-driven parameters; wherein, the initial data processing model correspondingly includes multiple neurons, and each neuron corresponds to an output result;

[0021] Data matching is performed using a fuzzy physical relationship formula based on data-driven parameters to obtain a matching result; the matching result is used to characterize the physical driving parameters related to the fuzzy physical relationship formula.

[0022] Construct a digital-analog mismatch function, and train an initial self-organizing neural network based on the matching result and the digital-analog mismatch function to obtain a data processing model.

[0023] Optionally, the steps of training an initial self-organizing neural network based on the matching result and the digital-analog mismatch function to obtain a data processing model include:

[0024] For any output result, use the digital-analog mismatch function to calculate the error between the physical driving parameter and the data-driven parameter.

[0025] Calculate the mean error between the output parameters to evaluate the matching degree between the physical driving parameter and the data-driven parameter.

[0026] Judge whether the mean error meets the convergence condition; if not, update the neurons in the initial self-organizing neural network according to the data sample to obtain the trained self-organizing neural network, and return to execute the step of obtaining the sound velocity profile information; if not, use the current initial self-organizing neural network as the final data processing model.

[0027] Optionally, the steps of using the data processing model to obtain the sound velocity profile information corresponding to the target sea area further include:

[0028] Obtain the preset data-driven parameters corresponding to the target sea area.

[0029] Calculate the matching degree between each output result under the data-driven parameter and the preset data-driven parameter, and use the output result corresponding to the maximum matching degree value as the target neuron, and update the initial function corresponding to the historical profile information with the preset order corresponding to the target neuron to obtain the sound velocity profile information.

[0030] Optionally, the preset data-driven parameters include a preset sea surface height parameter and a preset sea surface temperature parameter, and the calculation formula for calculating the matching degree between each output result under the data-driven parameter and the preset data-driven parameter is expressed as:

[0031] ;

[0032] Where is the matching value corresponding to any output result, is the sea surface temperature parameter in the current output result; is the sea surface height parameter in the current output result; is the preset sea surface temperature parameter; is the preset sea surface height parameter.

[0033] Optionally, the step of performing data matching based on data-driven parameters using a fuzzy physical relationship includes:

[0034] For any output result, use the fuzzy physical relationship to combine the sea surface height parameter and the sea surface temperature parameter under the output result to obtain the physical driving parameter related to the fuzzy physical relationship.

[0035] Optionally, the fuzzy physical relationship is expressed as:

[0036] ;

[0037] where is the sound speed profile coefficient corresponding to the nth order; is the sea surface temperature parameter; is the sea surface height parameter; , and are the first fuzzy coefficient, the second fuzzy coefficient, and the third fuzzy coefficient corresponding to the nth order, respectively.

[0038] Optionally, the calculation formula of the digital-analog mismatch function is expressed as:

[0039] ;

[0040] where is the root mean square error of the matching result; is the sound speed profile corresponding to the data-driven parameter at the th depth point; is the sound speed profile corresponding to the physical driving parameter related to the fuzzy physical relationship at the th depth point; is the total number of depth points.

[0041] A method for obtaining a sound speed profile based on a self-organizing neural network and fuzzy physical matching provided by the present invention includes the following steps: obtaining historical profile information and historical remote sensing information of a domain sea area, constructing a data set; then obtaining a data processing model based on the data set to obtain sound speed profile information; where the data processing model is a model trained by combining a self-organizing neural network and a fuzzy physical relationship, and finally using the data processing model to determine the sound speed profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound speed value with depth through the sound speed profile information. The present invention matches through a self-organizing neural network and a fuzzy physical expression related to physical driving, filters neurons that do not conform to physical laws, so as to overcome the technical problems of low accuracy and large error in the prior art, and improve the accuracy of sound speed profile acquisition.

[0042] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following specifically presents preferred embodiments and, in conjunction with the accompanying drawings, provides a detailed description as follows. Description of the Drawings

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as a limitation of the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0044] Figure 1 Shows the step flowchart of the sound velocity profile acquisition method provided by the embodiments of the present invention;

[0045] Figure 2 Shows the sub-step flowchart of step 100 in the embodiments of the present invention;

[0046] Figure 3 Shows the sub-step flowchart of step 200 in the embodiments of the present invention;

[0047] Figure 4 Shows the structural schematic diagram of the self-organizing neural network provided by the embodiments of the present invention;

[0048] Figure 5 Shows the sub-step flowchart of step 204 in the embodiments of the present invention;

[0049] Figure 6 Shows the sub-step flowchart of step 300 in the embodiments of the present invention;

[0050] Figure 7 Shows the block diagram of the server provided by this embodiment.

[0051] Icons: 10 - Self-organizing neural network; 20 - Server. Detailed Embodiments

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0053] Accordingly, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0054] It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0055] As described in the background art, in the existing sound speed profile acquisition scheme, the physical-driven method uses physical relationships to establish regression relationships. However, the real ocean system is extremely complex and cannot be perfectly described by a simple function expression. At the same time, measurement errors also introduce a certain degree of randomness between input and output parameters. Therefore, using the fixed physical-driven relationship will inevitably introduce these errors, resulting in a decrease in accuracy. Although the data-driven method avoids the constraints of fixed relationships and helps to improve the generalization ability of the inversion method, the pure data-driven method is too dependent on data, has high requirements for data, is easily interfered by some data noises, is difficult to use when the samples are insufficient, and sometimes obtains abnormal sound speed profile results that do not conform to physical laws because there is no support from physical mechanisms.

[0056] Based on this, the present application provides a sound speed profile optimization scheme, which can overcome the above technical problems and achieve the purpose of improving the accuracy of sound speed profile acquisition.

[0057] The above technical solutions will be introduced in detail below.

[0058] Please refer to Figure 1 , Figure 1 which shows a flowchart of the steps of the sound speed profile acquisition method based on self-organizing neural network and fuzzy physical matching in this embodiment. Among them, the sound speed profile acquisition method includes steps 100 to 300.

[0059] Step 100: Obtain the historical profile information and historical remote sensing information of the target sea area, and construct a data set.

[0060] Step 200: Obtain a data processing model based on the data set.

[0061] The data processing model is a model obtained by jointly training a self-organizing neural network and a fuzzy physical relational expression.

[0062] Step 300: Use the trained data processing model to obtain the sound speed profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound speed value with depth in the target sea area through the sound speed profile information.

[0063] In this embodiment, a self-organizing neural network is formed by data driving and matched with the corresponding fuzzy physical expression driven by physics. For example, parameter relationships related to physical driving are established using historical sea surface height, sea surface temperature, and sound speed profile. Subsequently, the sea surface height and sea surface temperature are used as inputs to the self-organizing neural network, and neurons that do not conform to physical laws are filtered out, and then the sound speed profile is solved. For example, in this embodiment, parameter relationships are established using historical sea surface height parameters, sea surface temperature parameters, and sound speed profile, and the corresponding sound speed profile information is solved by using preset sea surface height parameters and preset sea surface temperature parameters as inputs.

[0064] Based on this, this embodiment not only makes full use of the powerful generalization ability of the neural network but also ensures the rationality of the results through physical constraints, achieving the purpose of improving the accuracy of obtaining the sound speed profile.

[0065] In this embodiment, please refer to Figure 2 , Figure 2 which shows the sub-step flow chart of step 100 in this embodiment. Among them, step 100 of obtaining the historical profile information and historical remote sensing information of the domain sea area and constructing the data set includes steps 101 to 103.

[0066] Step 101: Preprocess the historical profile information to obtain the initial function corresponding to the historical profile information.

[0067] Step 102: Screen out the remote sensing information from the historical remote sensing information that has the same time resolution as the historical profile information and the data grid is closest to the longitude and latitude of the corresponding historical profile information to obtain the final historical remote sensing information.

[0068] The final historical remote sensing information includes sea surface height parameters and sea surface temperature parameters.

[0069] Step 103: Use the final historical remote sensing information and the sound speed profile coefficients corresponding to each order under the initial function as a set of data samples to obtain a data set containing multiple sets of data samples.

[0070] In this embodiment, the method for obtaining the historical profile information of the target sea area may be as follows: collect the sound speed profiles in the adjacent sea area over the years, or convert the temperature and salinity profile samples into sound speed profiles through the sound speed empirical formula. The method for obtaining the historical remote sensing information may be determined through a large number of publicly available data products. Among them, the historical remote sensing information includes sea surface height parameters and sea surface temperature parameters. Usually, the sea surface height parameters and sea surface temperature parameters need to have a daily time resolution and a spatial resolution as high as possible.

[0071] To improve the accuracy of the data and further enhance the accuracy of obtaining the sound speed profile, the historical profile information in this embodiment needs to correspond to the data of the historical remote sensing information. For example, the sea surface remote sensing data on the same day and with the data grid closest to the longitude and latitude of the sound speed profile can be selected from a large number of remote sensing information as the remote sensing information corresponding to the sound speed profile, so as to obtain the historical profile information.

[0072] It should be noted that to avoid excessive data volume of the historical profile information, increase the processing cost, and improve the data accuracy, this embodiment can perform dimensionality reduction processing on the historical profile information. That is, in this embodiment, the calculation formula of the initial function corresponding to the historical profile information is expressed as:

[0073] ;

[0074] Among them, is the steady-state background profile; is the profile corresponding to the stable part in the current sound speed profile, which can be obtained by averaging all samples; is the sound speed profile coefficient corresponding to the nth order, preferably 5th order; is the basis function, which can preferably be an orthogonal empirical function.

[0075] Among them, the product part of the basis function and the sound speed profile coefficients of the corresponding multiple orders is used to characterize the perturbed part of the sound speed profile. In this embodiment, to effectively describe the perturbed information and avoid large error orders introduced by high-order perturbations, the order can be set to 5. Among them, the order can be determined through the following regression calculation.

[0076] It should be noted that the data set in this embodiment contains multiple groups of data samples. Among them, any one data sample includes at least 7 data elements. For example, the above-mentioned sea surface height parameters, sea surface temperature parameters, and at least fifth-order coefficients.

[0077] Please, on the basis of Figure 1 , refer to Figure 3 , Figure 3 which shows the sub-step flow chart of step 200 in this embodiment. Step 200 in this embodiment includes steps 201 to 204.

[0078] Step 201: Determine the fuzzy physical relationship between data samples in the data set.

[0079] Among them, the fuzzy physical relationship is used to characterize the mapping relationship between the basis function and the sea surface height parameter and the sea surface temperature parameter.

[0080] Step 202: Based on each data sample in the data set, use the self-organizing neural network algorithm to construct a self-organizing neural network, obtain an initial data processing model, and use the multiple output results output by the initial data processing model as data-driven parameters.

[0081] Among them, the initial data processing model correspondingly includes multiple neurons, and each neuron corresponds to an output result.

[0082] Step 203: Use the fuzzy physical relationship to perform data matching based on the data-driven parameters to obtain a matching result.

[0083] Among them, the matching result is used to characterize the physical driving parameters related to the fuzzy physical relationship;

[0084] Step 204: Construct a digital-analog mismatch function, and train the initial self-organizing neural network based on the matching result and the digital-analog mismatch function to obtain a data processing model.

[0085] The following will describe in detail the above method for establishing a parameter relationship using historical sea surface height parameters, sea surface temperature parameters, and sound speed profile parameters, that is, the process of determining the trained self-organizing neural network.

[0086] In this embodiment, the fuzzy physical relationship is expressed as:

[0087] ;

[0088] Among them, is the sound speed profile coefficient corresponding to the nth order; is the sea surface temperature parameter; is the sea surface height parameter; , and are the first fuzzy coefficient, the second fuzzy coefficient, and the third fuzzy coefficient corresponding to the nth order respectively.

[0089] Among them, the above fuzzy physical relationship is used to characterize a fuzzy physical conclusion among the relationship between data sets, the sea-air interaction relationship, and the positive baroclinic mode energy and material transport relationship, so as to quantify the above basis function and the near-linear relationship between the sea surface height parameter and the sea surface temperature parameter.

[0090] Based on this, in this embodiment, regression analysis can be performed on the sound velocity profile coefficients, sea surface temperature parameters, and sea surface height parameters of multiple data samples, so as to obtain the first fuzzy coefficient in the above relationship formula. , the second fuzzy coefficient , the third fuzzy coefficient The corresponding coefficient values at each order respectively.

[0091] In this embodiment, the self-organizing neural network algorithm can be preferentially used to construct a self-organizing neural network to obtain an initial data processing model, so as to establish an implicit relationship between the input and output of the sea area through a data-driven method.

[0092] Please refer to Figure 4 , Figure 4 , which shows the structural schematic diagram of the self-organizing neural network in this embodiment. The self-organizing neural network 10 includes an input layer and an output layer; in this embodiment, the above 7 data elements, including the above sea surface height parameter, sea surface temperature parameter, and at least fifth-order coefficients, these 7 data elements correspond to an input neuron in the self-organizing neural network, and each data sample forms an original neuron in a training layer, and then the self-organizing neural network algorithm is used for training, such as the SOM algorithm, to obtain an input layer, and further obtain a neural network driven by generated data. In this embodiment, the input layer essentially generalizes the implicit relationship between the 7 data elements of the training samples by the neural network algorithm to generate more samples with similar rules.

[0093] Among them, the key hyperparameter of the self-organizing neural network is the number of neurons in the generated network. In this embodiment, the number of neurons in the generated network can be set to 8 times the number of data samples to ensure that the method achieves better results.

[0094] After obtaining the initial data processing model, the multiple output results output by the initial data processing model can be used as data-driven parameters.

[0095] In this embodiment, by adding physical mechanism constraints to the data-driven solution through the fuzzy physical relationship formula, the above data-driven parameters can be input into the fuzzy physical relationship formula, and the accuracy of the data-driven parameters can be judged by using the fuzzy physical relationship formula.

[0096] In this embodiment, step 203 is specifically as follows:

[0097] For any output result, the physical driving parameters related to the fuzzy physical relationship formula are obtained by using the fuzzy physical relationship formula in combination with the sea surface height parameter and sea surface temperature parameter under the output result.

[0098] After obtaining the physical driving parameters and data-driven parameters, refer to Figure 5 , Figure 5The sub - step flowchart of step 204 in this embodiment is shown. In this embodiment, step 204 includes steps 2041 to 2043.

[0099] Step 2041: For any output result, calculate the error between the physical driving parameter and the data - driving parameter using the digital - analog mismatch function.

[0100] Step 2042: Calculate the mean error between the output parameters to evaluate the matching degree between the physical driving parameter and the data - driving parameter.

[0101] Step 2043: Determine whether the mean error satisfies the convergence condition;

[0102] If not, update the number of neurons in the initial self - organizing neural network according to the data sample, and return to execute the step of obtaining the sound - speed profile information.

[0103] If satisfied, use the current corresponding initial self - organizing neural network as the final data - processing model.

[0104] In this embodiment, the digital - analog mismatch function can be constructed by the root - mean - square error. The calculation formula of the corresponding digital - analog mismatch function is expressed as:

[0105] ;

[0106] Where, is the root - mean - square error of the matching result; is the sound - speed profile corresponding to the data - driving parameter at the sound - speed value at the th depth point; is the sound - speed profile corresponding to the physical driving parameter related to the fuzzy physical relationship formula at the sound - speed value at the th depth point; is the total number of depth points.

[0107] In this embodiment, based on the digital - analog mismatch function, the degree of mismatch between the data - driving method and the physical - driving method in processing the same set of sea - surface height and sea - surface temperature can be obtained. Among them, the larger the root - mean - square error E, the more mismatched the two methods are, and the greater the difference between the data - driving method and the physical - driving method.

[0108] To improve the differences among the above methods, the average of the digital-analog mismatch function values calculated by all neurons can be further used, that is, the average root mean square error of the current neural network, to evaluate the degree of mismatch between the data-driven parameters and the physically-driven parameters. And the convergence condition is used to adjust the degree of mismatch. For example, the convergence condition is set as: the average root mean square error of the self-organizing neural network in the current iteration is less than one percent of the result of the average root mean square error of the self-organizing neural network in the previous iteration or the result that the average root mean square error of the current self-organizing neural network is greater than the average root mean square error of the previous self-organizing neural network. If this convergence condition is not met, it means that the current self-organizing neural network does not match the physical mechanism well enough, and further training of the self-organizing neural network is required. Otherwise, the current self-organizing neural network matches the physical mechanism up to the preset target, and the current initial self-organizing neural network can be used as the final data processing model.

[0109] This embodiment can filter data samples to establish a self-organizing neural network that is more in line with data-driven and physically-driven. In a possible implementation manner, the step of updating the number of neurons in the initial self-organizing neural network according to the data samples in this embodiment is as follows:

[0110] Sort the digital-analog mismatch function values corresponding to each neuron in the output layer of the current initial self-organizing neural network from largest to smallest, select one-eighth of the total number of neurons in the current output layer, the neuron with the smallest digital-analog mismatch function value, and replace the original training sample with the neuron that meets the condition, and return to the step of obtaining the sound speed profile information.

[0111] This embodiment refers to Figure 6 , Figure 6 FIG. shows the sub-step flowchart of step 300 in this embodiment. Step 300 includes steps 301 to 302.

[0112] Step 301: Obtain the preset data-driven parameters corresponding to the target sea area.

[0113] Step 302: Calculate the matching degree between each output result under the data-driven parameters and the preset data-driven parameters, and use the output result corresponding to the maximum matching degree value as the target neuron, and update the initial function corresponding to the historical profile information with the preset order corresponding to the target neuron to obtain the sound speed profile information.

[0114] In this embodiment, the preset data-driven parameters include the preset sea surface height parameter and the preset sea surface temperature parameter. Then, the calculation formula for calculating the matching degree between each output result under the data-driven parameters and the preset data-driven parameters is expressed as:

[0115] ;

[0116] Where, is the matching value corresponding to any output result, is the sea surface temperature parameter in the current output result; is the sea surface height parameter in the current output result; is the preset sea surface temperature parameter; is the preset sea surface height parameter.

[0117] Specifically, in this embodiment, the neuron corresponding to the minimum matching degree value can be calculated based on the matching degree as the solution neuron, that is, the above-mentioned target neuron, and then the initial function corresponding to the historical profile information of the fifth-order coefficient updated by the target neuron is used, that is, the above-mentioned fifth-order coefficient is brought into the expression of the initial function as the final sound speed profile information.

[0118] Based on this, a method for obtaining a sound speed profile based on a self-organizing neural network and fuzzy physical matching provided by the present invention obtains historical profile information and historical remote sensing information of a domain sea area and constructs a data set; then a data processing model is obtained based on the data set; wherein, the data processing model is a model trained by combining a self-organizing neural network and a fuzzy physical relationship, and finally the data processing model is used to determine the sound speed profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound speed value with depth through the sound speed profile information.

[0119] The present invention matches through a fuzzy physical expression related to a self-organizing neural network and physical driving, and filters neurons that do not conform to physical laws, so as to overcome the technical problems of low accuracy and large error in the prior art and improve the accuracy of obtaining the sound speed profile.

[0120] Please refer to Figure 7 , Figure 7 which shows a block diagram of the server provided in this embodiment. The server 20 includes a memory, a processor, and a communication module. Each element of the memory, the processor, and the communication module is directly or indirectly electrically connected to each other to realize data transmission or interaction. For example, these elements can be electrically connected to each other through one or more communication buses or signal lines.

[0121] Among them, the memory is used to store programs or data. The memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electric Erasable Programmable Read-Only Memory (EEPROM), etc.

[0122] The processor is used to read / write the data or programs stored in the memory and execute corresponding functions, that is, to obtain the historical profile information and historical remote sensing information of the target sea area and construct a data set; then obtain a data processing model based on the data set; among them, the data processing model is a model trained by combining a self-organizing neural network and a fuzzy physical relationship formula; finally, use the data processing model to determine the sound velocity profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound velocity value with depth through the sound velocity profile information.

[0123] The communication module is used to establish a communication connection between the server and other communication terminals through the network and is used to send and receive data through the network.

[0124] It should be understood that Figure 7 The structure shown is only a schematic diagram of the server structure, and the server may also include more or fewer components than those shown Figure 7 in the figure, or have a different configuration from that shown Figure 6 in the figure. Figure 7 Each component shown in the figure can be implemented by hardware, software, or a combination thereof.

[0125] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0126] In addition, each functional module in various embodiments of the present invention can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0127] If the above-mentioned functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0128] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for acquiring a sound velocity profile based on a self-organizing neural network and fuzzy physical matching, characterized in that: The method for acquiring the sound velocity profile comprises the following steps: Obtain historical profile information and historical remote sensing information of the domain sea area and build a data set; A data processing model is obtained based on the data set; wherein the data processing model is a model obtained by combining a self-organizing neural network and a fuzzy physical relationship training; The data processing model is used to obtain the sound velocity profile information corresponding to the target sea area, so as to determine the variation characteristics of the seawater sound velocity value with depth in the target sea area through the sound velocity profile information.

2. The method for obtaining a sound velocity profile according to claim 1, characterized in that: The steps of obtaining historical profile information and historical remote sensing information of the sea area and constructing a data set include: Preprocessing the historical section information to obtain an initial function corresponding to the historical section information; Filtering remote sensing information with the same time resolution as the historical profile information and with a data grid closest to the longitude and latitude of the corresponding historical profile information from the historical remote sensing information to obtain final historical remote sensing information; wherein the final historical remote sensing information includes sea surface height parameters and sea surface temperature parameters; The final historical remote sensing information and the corresponding coefficients of each order under the initial function are taken as a group of data samples to obtain a data set containing multiple groups of data samples.

3. The method for acquiring the sound velocity profile according to claim 2, characterized in that: The calculation formula of the initial function is expressed as: ； in, is the steady-state background profile; It is the section corresponding to the stable part of the current sound velocity section; is the sound velocity profile coefficient corresponding to the nth order; is the basis function.

4. The method for acquiring a sound velocity profile according to claim 1, characterized in that: The step of obtaining a data processing model based on the data set comprises: Determine a fuzzy physical relationship between each data sample in the data set, wherein the fuzzy physical relationship is used to characterize a mapping relationship between a basis function and a sea surface height parameter and a sea surface temperature parameter; Based on each data sample in the data set, a self-organizing neural network is constructed using a self-organizing neural network algorithm to obtain an initial data processing model, and multiple output results output by the initial data processing model are used as data driving parameters; wherein the initial data processing model includes multiple neurons, each of which corresponds to an output result; Using the fuzzy physical relationship to perform data matching based on the data driving parameters to obtain a matching result; the matching result is used to characterize the physical driving parameters related to the fuzzy physical relationship; A digital-analog mismatch function is constructed, and the initial self-organizing neural network is trained based on the matching result and the digital-analog mismatch function to obtain a data processing model.

5. The method for acquiring the sound velocity profile according to claim 4, characterized in that: The step of training the initial self-organizing neural network based on the matching result and the digital-analog mismatch function to obtain a data processing model comprises: For any output result, using the digital-analog mismatch function to calculate the error between the physical driving parameter and the data driving parameter; Calculating the mean error between the output parameters to evaluate the matching degree between the physical driving parameters and the data driving parameters; Determine whether the error mean satisfies the convergence condition; if not, update the neurons under the initial self-organizing neural network according to the data sample, and return to execute the step of obtaining the sound speed profile information; if not, use the current initial self-organizing neural network as the final data processing model.

6. The method for acquiring the sound velocity profile according to claim 5, characterized in that: The step of using the data processing model to obtain the sound speed profile information corresponding to the target sea area includes: Obtain preset data-driven parameters corresponding to the target sea area; The matching degree between each output result under the data-driven parameters and the preset data-driven parameters is calculated, and the output result corresponding to the maximum matching degree value is used as the target neuron. The initial function corresponding to the historical profile information is updated with the preset order corresponding to the target neuron to obtain the sound speed profile information.

7. The method for acquiring the sound velocity profile according to claim 6, characterized in that: The preset data driving parameters include a preset sea surface height parameter and a preset sea surface temperature parameter. The calculation formula for calculating the matching degree between each output result under the data driving parameters and the preset data driving parameters is expressed as: ； in, is the matching value corresponding to any output result, is the sea surface temperature parameter in the current output result; It is the sea surface height parameter in the current output result; To preset the sea surface temperature parameters; It is the preset sea surface height parameter.

8. The method for acquiring a sound velocity profile according to claim 4, characterized in that: The step of performing data matching based on the data-driven parameters using the fuzzy physical relationship includes: For any output result, the fuzzy physical relationship is used in combination with the sea surface height parameter and the sea surface temperature parameter under the output result to obtain the physical driving parameters related to the fuzzy physical relationship.

9. The method for acquiring a sound velocity profile according to claim 4, characterized in that: The fuzzy physical relational expression is expressed as: ； in, is the sound velocity profile coefficient corresponding to the nth order; is the sea surface temperature parameter; is the sea surface height parameter; , as well as They are the first fuzzy coefficient, the second fuzzy coefficient, and the third fuzzy coefficient corresponding to the nth order respectively.

10. The method for acquiring a sound velocity profile according to claim 4, characterized in that: The calculation formula of the digital-analog mismatch function is expressed as: ； in, is the root mean square error of the matching result; The sound velocity profile corresponding to the data-driven parameter In the The sound velocity value at each depth point; is the sound velocity profile corresponding to the physical driving parameter related to the fuzzy physical relationship In the The sound velocity value at each depth point; is the total number of depth points.

Images