Signal feature extraction method and device, terminal equipment and storage medium

Abstract

The invention provides a signal feature extraction method and device, terminal equipment and a storage medium, and is suitable for the technical field of acoustic signal processing, and the method comprises the steps: obtaining an underwater sound signal of a to-be-detected target; according to a preset time sequence step length, a preset state variable, a preset gating weight matrix, a preset gating bias variable and a preset standardization function, performing time sequence feature extraction on the underwater sound signal to generate an initial signal feature; according to a plurality of preset linear transformation matrixes, performing enhancement processing on the initial signal features to generate intermediate signal features; and according to a plurality of randomly generated set center points and a preset set space displacement function, performing classification calculation on the intermediate signal features to generate target signal features. According to the invention, preliminary feature extraction is automatically carried out on the underwater target sound signals according to the time sequence correlation of the sound signals, noise in the signal features is filtered through enhancement and classification processing, and the comprehensiveness and accuracy of signal feature extraction are improved.

Description

Technical Field

[0001] The present application belongs to the technical field of acoustic signal processing, and in particular, relates to a signal feature extraction method, apparatus, terminal equipment and storage medium. Background Art

[0002] At present, the demand for exploration, development and maintenance of underwater resources is increasing. As an important part of underwater operations, underwater target detection through acoustic technology has broad application prospects. For underwater target detection, the step of underwater target sound signal feature extraction is indispensable. However, the complex and changeable water environment, including low visibility, frequent biological activities, and extreme changes in temperature and pressure, poses severe challenges to the extraction of underwater target signal features.

[0003] In the prior art, features of acoustic signals are usually extracted based on the time domain or frequency domain. For example, the acoustic signal is divided into multiple short-time frames, and each frame is processed by pre-emphasis, Fourier transform, Mel filtering and other processing steps, and then the Mel spectrum cepstrum coefficients are calculated; or the time domain waveform of the acoustic signal is directly visualized, and features such as short-time energy, short-time zero-crossing rate or autocorrelation are obtained through the time domain waveform.

[0004] However, the sound signal extraction methods in the prior art usually do not consider the correlation between sound signals and are very sensitive to noise, making it difficult to comprehensively and effectively extract sound signals in complex water environments. Summary of the invention

[0005] In view of this, the embodiments of the present application provide a signal feature extraction method, apparatus, terminal device and storage medium, aiming to solve the problem in the prior art that the correlation between sound signals is ignored, the extraction process cannot eliminate noise, and the accuracy and effectiveness of the sound signal feature extraction results are low.

[0006] A first aspect of an embodiment of the present application provides a signal feature extraction method, comprising:

[0007] Acquire underwater sound signals of the target to be detected;

[0008] Extracting time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrices, preset gating bias variables and preset normalization functions to generate initial signal features;

[0009] According to a plurality of preset linear transformation matrices, the initial signal features are enhanced to generate intermediate signal features;

[0010] According to a plurality of randomly generated set center points and a preset set space displacement function, the intermediate signal features are classified and calculated to generate target signal features.

[0011] A second aspect of an embodiment of the present application provides a signal feature extraction device, including:

[0012] An underwater sound signal acquisition module is used to acquire the underwater sound signal of the target to be detected;

[0013] An initial signal feature generation module, used to extract the time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrix, preset gating bias variables and preset normalization functions, and generate initial signal features;

[0014] an intermediate signal feature generation module, configured to enhance the initial signal features according to a plurality of preset linear transformation matrices to generate intermediate signal features; and

[0015] The target signal feature generation module is used to classify and calculate the intermediate signal features according to a plurality of randomly generated set center points and a preset set space displacement function to generate target signal features.

[0016] A third aspect of an embodiment of the present application provides a terminal device, which includes a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, it implements the steps of the signal feature extraction method described in the first aspect above.

[0017] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, comprising: a computer program stored therein, wherein when the computer program is executed by a processor, the steps of the signal feature extraction method described in the first aspect above are implemented.

[0018] The beneficial effects of the embodiments of the present application compared with the prior art are: taking into account the autocorrelation and cross-correlation of each sound signal sequence, performing preliminary feature extraction on the underwater sound information of the target to be detected, and capturing the time domain, frequency domain and timing characteristics of the sound signal at the same time, and then performing linear timing feature enhancement and classification processing on the preliminary extracted features, filtering out noise information and redundant information in the signal features, and improving the comprehensiveness, effectiveness and accuracy of signal feature extraction, thereby achieving comprehensive and effective extraction of sound signals in complex water environments, so as to improve the accuracy and robustness of underwater target detection through the extracted sound signal features. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0020] Figure 1 It is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 1 of the present application;

[0021] Figure 2 It is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 2 of the present application;

[0022] Figure 3 It is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 3 of the present application;

[0023] Figure 4 It is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 4 of the present application;

[0024] Figure 5 This is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 5 of the present application;

[0025] Figure 6 This is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 6 of the present application;

[0026] Figure 7 It is a schematic diagram of the implementation flow of the signal feature extraction method provided in Example 7 of the present application;

[0027] Figure 8 is a structural schematic diagram of a signal feature extraction device provided in an embodiment of the present application;

[0028] Fig. 9 It is a schematic diagram of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0030] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.

[0031] Figure 1The following is a flowchart of the signal feature extraction method according to the first embodiment of the present invention.

[0032] Step S101, acquiring an underwater sound signal of a target to be detected.

[0033] In this embodiment, underwater sound signals can be collected by a sonar device placed in an underwater environment. The underwater sound information can be acquired by emitting sound waves and receiving reflected signals of the target to be detected. The sonar device can send the reflected signals to a server, thereby acquiring underwater sound signals of the target to be detected.

[0034] Step S102, extracting the time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrices, preset gating bias variables and preset normalization functions, and generating initial signal features.

[0035] In this embodiment, the preset timing step information, the preset state variables, the preset gating weight matrix, and the preset gating bias variables can all be manually set, wherein the preset state variables, the preset gating weight matrix, and the preset gating bias variables can also be obtained by respectively extracting the parameters in the trained specific computing model after training the specific computing model. The preset normalization function can be designed based on an exponential function, and is used to normalize the calculation result to a value within the interval (0, 1), so as to avoid the value being invalid due to the value exceeding the computing dimension of the server during the calculation process. It can be that the different data segments of the underwater sound signal are first segmented and extracted through the timing step information, and then the extracted multiple data segments are formatted and converted into multiple data matrices, and then the multiple data matrices are transformed and calculated multiple times through the state variables, the gating weight matrix, and the gating bias variables, and then the calculation result is used as the independent variable of the normalization function, and the function value of the normalization function is calculated as the initial signal feature, thereby realizing the preliminary extraction of the underwater sound signal feature.

[0036] Step S103: Perform enhancement processing on the initial signal features according to a plurality of preset linear transformation matrices to generate intermediate signal features.

[0037] In this embodiment, the preset linear transformation matrix can be artificially set. It can be that the initial signal features are transformed in different dimensions through multiple preset linear transformation matrices, thereby increasing the logical distance between multiple values ​​of the initial signal features and amplifying the difference between the initial signal feature values ​​to achieve feature enhancement processing. The enhanced features are used as intermediate signal features to facilitate further screening in the future to filter out noise components and redundant components in the extracted sound signal.

[0038] Step S104, classifying and calculating the intermediate signal features according to a plurality of randomly generated set center points and a preset set space displacement function, to generate target signal features.

[0039] In this embodiment, the preset set space displacement function can be artificially set, or it can be a Griewank function, or it can be a Rastrigin function, or it can be a Schaffer function, or it can be an Ackley function, or it can be a Rosenbrock function. The value and number of the set center point can be randomly generated, and used to classify the intermediate signal features. It can be understood that the intermediate signal features will be divided into as many categories as the number of set center points. It can be achieved by calculating the logical distance between each set center point and each value in the intermediate signal feature, and preliminarily dividing each value in the intermediate signal feature into different sets through the logical distance to achieve preliminary classification, and then further calculating the logical distance through the set space displacement function, adjusting the set to which each value in the intermediate signal feature belongs, so as to achieve deep classification calculation, which is used to distinguish the effective feature value and the noise component value and the redundant component value in the intermediate signal feature, and output the set to which the effective feature value belongs as the target signal feature, which is used to identify and detect the target to be detected in the underwater environment.

[0040] The signal feature extraction method provided in the embodiment of the present application takes into account the autocorrelation and cross-correlation of each sound signal sequence, performs preliminary feature extraction on the underwater sound information of the target to be detected, and captures the time domain, frequency domain and timing characteristics of the sound signal at the same time, and then performs linear timing feature enhancement and classification processing on the preliminary extracted features, filters out noise information and redundant information in the signal features, and improves the comprehensiveness, effectiveness and accuracy of signal feature extraction, thereby achieving comprehensive and effective extraction of sound signals in complex water environments, so as to improve the accuracy and robustness of underwater target detection through the extracted sound signal features.

[0041] Figure 2 The following is a flowchart of the signal feature extraction method provided in the second embodiment of the present application, which differs from the first embodiment in that:

[0042] The preset state variables include preset hidden state variables and preset update state variables;

[0043] The preset gating weight matrix includes a preset front gating weight matrix, a preset first intermediate gating weight matrix, a preset second intermediate gating weight matrix and a preset rear gating weight matrix;

[0044] The preset gate bias variables include a preset pre-gate bias variable, a preset first intermediate gate bias variable, a preset second intermediate gate bias variable, and a preset post-gate bias variable;

[0045] The step S102 specifically includes:

[0046] Step S201, cutting and splicing the underwater sound signal according to a plurality of preset timing step information to obtain a plurality of signal timing segment information.

[0047] In this embodiment, the preset timing step information can be set manually, and is used to perform segmentation operations on the numerical sequence composed of underwater sound signals at different intervals, and splice the segmented data segments to obtain multiple signal timing segment information.

[0048] Step S202, mapping and normalizing the plurality of signal timing segment information according to a preset pre-gating weight matrix, a preset hidden state variable, a preset pre-gating bias variable and a preset normalization function to obtain a pre-gating signal timing feature value.

[0049] In this embodiment, the preset pre-gating weight matrix, the preset hidden state variable, and the preset pre-gating bias variable can all be manually set in advance. The preset normalization function can be a hyperbolic tangent function or a logistic function. The hidden state variable can be combined with the signal timing segment information to generate a specific matrix, and then the matrix is ​​multiplied with the pre-gating weight matrix, the multiplication result is added to the pre-gating bias variable, and then the addition result is used as the independent variable of the normalization function, and the function value of the calculated normalization function is output as the pre-gating signal timing characteristic value.

[0050] Step S203, mapping and standardizing the plurality of signal timing segment information according to a preset first intermediate gating weight matrix, a preset hidden state variable, a preset first intermediate gating bias variable and a preset normalization function to obtain a first intermediate gating signal timing feature value.

[0051] In this embodiment, the preset first intermediate gating weight matrix and the preset first intermediate gating bias variable can be manually set. It can be that the hidden state variable is first combined with the signal timing segment information to generate a specific matrix, and then the matrix is ​​multiplied with the first intermediate gating weight matrix, the multiplication result is added with the first intermediate gating bias variable, and then the addition result is used as the independent variable of the normalization function, and the function value of the calculated normalization function is output as the first intermediate gating signal timing feature value.

[0052] Step S204, mapping and standardizing the signal timing segment information according to a preset second intermediate gating weight matrix, a preset hidden state function and a preset second intermediate gating bias variable to obtain an intermediate gating signal timing feature variable.

[0053] In this embodiment, the preset second intermediate gating weight matrix and the preset second intermediate gating bias variable can be manually set. It can be that the hidden state variable is first combined with the signal timing segment information to generate a specific matrix, and then the matrix is ​​multiplied with the second intermediate gating weight matrix, the multiplication result is added with the second intermediate gating bias variable, and then the addition result is used as the independent variable of the normalization function, and the function value of the calculated normalization function is output as the second intermediate gating signal timing feature value.

[0054] Step S205 , performing a hyperbolic tangent transformation on the intermediate gating signal timing characteristic variable to obtain a second intermediate gating signal timing characteristic value.

[0055] In this embodiment, the intermediate gating signal timing characteristic variable may be used as an independent variable of the hyperbolic tangent function, and the calculated function value may be output as the second intermediate gating signal timing characteristic value.

[0056] Step S206, mapping and standardizing the plurality of signal timing segment information according to a preset post-gating weight matrix, a preset hidden state variable, a preset post-gating bias variable and a preset normalization function to obtain a post-gating signal timing feature value.

[0057] In this embodiment, the preset post-gating weight matrix and the preset post-gating bias variable can be manually set. It can be that the hidden state variable is first combined with the signal timing segment information to generate a specific matrix, and then the matrix is ​​multiplied with the post-gating weight matrix, the multiplication result is added to the post-gating bias variable, and then the addition result is used as the independent variable of the normalization function, and the function value of the calculated normalization function is output as the post-gating signal timing feature value.

[0058] Step S207 , generating an initial signal feature according to the timing feature value of the pre-gating signal, the timing feature value of the first intermediate gating signal, the timing feature value of the second intermediate gating signal, and the timing feature value of the post-gating signal.

[0059] In this embodiment, the pre-gating signal timing characteristic value is multiplied by the first intermediate gating signal timing characteristic value, and the post-gating signal timing characteristic value is multiplied by the second intermediate gating signal timing characteristic value, and the two multiplication results are summed to obtain the initial signal characteristic.

[0060] The signal feature extraction method provided in the embodiment of the present application is based on multiple preset timing step information, and uses multiple preset gated weight matrices to extract features of underwater sound signals in different dimensions, so as to fully extract the timing features of underwater sound signals with nonlinear change laws in different dimensions, facilitate the analysis of long-range relationships of underwater sound signals, and improve the accuracy and robustness of underwater target detection.

[0061] Figure 3 The flowchart of the signal feature extraction method provided in the third embodiment of the present application is shown. The difference between the third embodiment and the second embodiment is that the step S207 specifically includes:

[0062] Step S301 , calculating a state characterizing timing characteristic variable according to the timing characteristic value of the pre-gating signal, the timing characteristic value of the first intermediate gating signal, the timing characteristic value of the second intermediate gating signal and a preset updating state variable.

[0063] In this embodiment, the preset update state variable may be manually set, which may be to first multiply the pre-gating signal timing characteristic value by the first intermediate gating signal timing characteristic value, then multiply the second intermediate gating signal timing characteristic value by the update state variable, and then sum the results of the two multiplications, and output the sum result as the state characterization timing characteristic variable.

[0064] Step S302, performing a hyperbolic tangent transformation on the state characterization time series characteristic variable to obtain a state transformation time series characteristic variable.

[0065] In this embodiment, the state characterization time series characteristic variable can be used as the independent variable of the hyperbolic tangent function, and the calculated function value can be output as the state transformation time series characteristic variable to achieve a hyperbolic tangent transformation of the state transformation time series characteristic variable, thereby limiting the state transformation time series characteristic variable between 0 and 1, and realizing the normalization operation of the data while retaining the data information, so as to avoid the problem of exceeding the dimension and making the numerical value invalid in the subsequent calculation process.

[0066] Step S303, calculating a plurality of sound signal timing feature representation variables according to the post-gating signal timing feature value and the state change timing feature variable.

[0067] In this embodiment, the post-gating signal timing characteristic value and the state change timing characteristic variable may be multiplied, and the multiplication result may be output as a variable representing the sound signal timing characteristic. It can be understood that the post-gating signal timing characteristic value and the state change timing characteristic variable may be presented in a matrix form, and the multiplication calculation may be an inner product or a dot product calculation.

[0068] Step S304: generating an initial signal feature according to a plurality of the sound signal time series feature representation variables.

[0069] In this embodiment, it can be understood that the sound signal time series feature representation variables are presented in the form of multiple matrices, and the matrices need to be formatted and converted into a sequence of multiple discrete values ​​before they can be output as initial signal features.

[0070] The signal feature extraction method provided in the embodiment of the present application effectively captures the complex features and long-range relationships in the sound signal by combining and transforming the timing feature value of the pre-gated signal, the timing feature value of the first intermediate gating signal, the timing feature value of the second intermediate gating signal and the preset update state variable, enhances the feature representation capability of the sound signal, and avoids the invalidation of values ​​due to values ​​exceeding the numerical dimension processed by the computer during the calculation process, thereby facilitating the subsequent enhancement and screening of the sound feature signal, thereby improving the accuracy and effectiveness of the extraction of the sound signal.

[0071] Figure 4 The following is a flowchart of the signal feature extraction method provided in the fourth embodiment of the present application, which differs from the first embodiment in that:

[0072] The preset linear transformation matrix includes a preset first linear layer transformation matrix, a preset second linear layer transformation matrix, and a preset third linear layer transformation matrix;

[0073] The step S103 specifically includes:

[0074] Step S401, calculating a first linear layer feature representation variable, a second linear layer feature representation variable, and a third linear layer feature representation variable according to the initial signal feature, a preset first linear layer transformation matrix, a preset second linear layer transformation matrix, and a preset third linear layer transformation matrix.

[0075] In this embodiment, the preset first linear layer transformation matrix, the preset second linear layer transformation matrix, and the preset third linear layer transformation matrix can all be manually set, and are used to perform multiplication operations with the initial signal features respectively, so as to widen the gap between the values ​​in the initial signal features from multiple different dimensions, so as to facilitate the enhancement processing. Specifically, the initial signal features can be multiplied or convolved with the first linear layer transformation matrix, the second linear layer transformation matrix, and the third linear layer transformation matrix, respectively, and the calculation results are output as the first linear layer feature representation variable, the second linear layer feature representation variable, and the third linear layer feature representation variable.

[0076] Step S402: Obtain an intermediate linear representation feature variable according to the first linear layer feature representation variable and the second linear layer feature representation variable.

[0077] In this embodiment, a dot product operation may be performed on the first linear layer feature representation variable and the second linear layer feature representation variable, and the operation result is used as an intermediate linear representation feature variable to achieve a linear transformation of the sound signal characteristics and increase the logical distance between the feature values ​​of each signal on the linear level.

[0078] Step S403: multiply the intermediate linear representation feature variable and the third linear layer feature representation variable to obtain a signal feature linear representation variable.

[0079] In this embodiment, the intermediate linear representation feature variable and the third linear layer feature representation variable are multiplied, and the multiplied result is output as the signal feature linear representation variable.

[0080] Step S404: performing weighted summation on the signal feature linear representation variables according to a preset linear layer weight matrix to generate an intermediate signal feature.

[0081] In this embodiment, the preset linear layer weight matrix can be manually set. The values ​​in the preset linear layer weight matrix are used as weights, and the values ​​in the linear representation variables of the signal features are weighted and summed, and the result of the weighted summation is output as the intermediate signal feature.

[0082] The signal feature extraction method provided in the embodiment of the present application performs multi-dimensional linear transformation on the initial signal features through a preset first linear layer transformation matrix, a second linear layer transformation matrix, and a third linear layer transformation matrix, thereby increasing the logical distance between each numerical value in the initial signal features, making the contrast of each sound signal feature initially extracted more obvious, thereby achieving enhanced processing of the initial signal features, effectively enhancing the feature representation capability of the sound signal features, and facilitating subsequent filtering of noise components and redundant components in the sound signal features.

[0083] Figure 5 The flowchart of the signal feature extraction method provided in the fifth embodiment of the present application is shown. The difference between the fifth embodiment and the first embodiment is that the step S104 specifically includes:

[0084] Step S501, calculating the feature set representation distance between the intermediate signal feature and a plurality of set center points.

[0085] In this embodiment, the feature set representation distance may be the Euclidean distance, and the logical distance between the intermediate signal feature and multiple set center points is quantified to facilitate subsequent set division of each numerical point of the intermediate signal feature.

[0086] Step S502: According to the feature set representation distance, the intermediate signal features are divided into sets corresponding to a plurality of set center points to generate a plurality of signal feature sets.

[0087] In this embodiment, a logical distance interval may be set. When the feature set representation distance falls within the logical distance interval, the intermediate signal features corresponding to the feature set representation distance are divided into the set to which the set center point corresponding to the feature set representation distance belongs, thereby generating multiple signal feature sets to achieve preliminary classification of signal features.

[0088] Step S503: Calculate the set space displacement of each intermediate signal feature according to the signal feature set and a preset set space displacement function.

[0089] In this embodiment, the preset set space displacement function may be artificially set, or may be a Griewank function, or a Rastrigin function, or a Schaffer function, or an Ackley function, or a Rosenbrock function. The signal feature set may be used as an independent variable of the set space displacement function, and the set space displacement function may be calculated as the set space displacement amount.

[0090] Step S504: determining the displacement center point of each of the signal feature sets according to the set spatial displacement.

[0091] In this embodiment, the value with the largest set space displacement in the signal feature set can be used as the displacement center point of the signal feature set, which is used to reclassify the intermediate signal features so that the classification results approach the global optimal solution, and is used to judge whether the classification calculation has converged in the subsequent process.

[0092] Step S505, determining whether the displacement center point is the same as the set center point; if so, proceeding to step S506; if not, proceeding to step S507.

[0093] In this embodiment, when the displacement center point is the same as the set center point, it indicates that the current classification calculation process for the intermediate signal feature has converged, and there is no need to further divide the set to which the intermediate signal feature belongs, so there is no need to further calculate the displacement center point. At this time, the result can be directly output according to the set obtained by dividing based on the set center point, and the signal feature set with no less than a specific number of set elements is used as the target signal feature. When the displacement center point is different from the set center point, it indicates that the current classification calculation for the intermediate signal feature has not converged, and the set to which the intermediate signal feature belongs needs to be further divided, and the displacement center point needs to be further calculated so that the calculation result tends to the global optimal solution. It can be understood that the intermediate signal feature contains noise components and redundant components in the sound signal feature. The noise components and redundant components are both random and have diverse sources. It is impossible to divide various noise and redundant components into a certain category or several categories. Therefore, the values ​​of the noise and redundant components will have a discrete characteristic and cannot form a set with more elements; while the sound signal belonging to the target to be detected has a certain linear law, so it can be divided into one or several sets, thereby forming multiple sets with more elements.

[0094] Step S506: Generate target signal features based on the multiple signal feature sets.

[0095] In this embodiment, when the displacement center point is the same as the set center point, it indicates that the current classification calculation process for the intermediate signal feature has converged, and there is no need to further divide the set to which the intermediate signal feature belongs. Therefore, there is no need to further calculate the displacement center point. At this time, the result output can be directly based on the set obtained by dividing based on the set center point, and the signal feature set with no less than a specific number of set elements can be used as the target signal feature for identifying and processing the target to be detected in the underwater environment.

[0096] Step S507, taking the displacement center point as the set center point, and returning to step S502.

[0097] In this embodiment, when the displacement center point is different from the set center point, it indicates that the current classification calculation for the intermediate signal features has not converged, and the effective signal features are still not distinguished from the noise components and redundant components in each of the current sets. It is necessary to further divide the sets to which the intermediate signal features belong, and it is necessary to further calculate the displacement center point so that the calculation result tends to the global optimal solution.

[0098] The signal feature extraction method provided in the embodiment of the present application performs deep classification calculation on the intermediate signal features through randomly generated set center points and preset set space displacement functions, and effectively distinguishes the effective features from the noise components and redundant components in the intermediate signal features, thereby screening out the effective components in the intermediate signal features for underwater target detection, so as to filter out the noise components and redundant components, improve the effectiveness and accuracy of sound signal feature extraction, and improve the accuracy and robustness of target detection in complex and changeable underwater environments.

[0099] Figure 6 The flowchart of the signal feature extraction method provided in the sixth embodiment of the present application is shown. The difference between the sixth embodiment and the first embodiment is that after the step S104, the method further includes:

[0100] Step S601, extracting the timing step information and the number of set center points corresponding to the plurality of target signal features.

[0101] In this embodiment, it can be understood that there are multiple timing step information for extracting the timing features of the sound signal, and there are multiple set center points for classifying the intermediate signal features. The possible values ​​of all timing steps and the number of set center points are extracted, and then the optimal configuration values ​​of the timing steps and the number of set center points are screened to screen out the combination of timing steps and the number of set center points that makes the extracted signal features have the best quality.

[0102] Step S602, generating a primary population according to the target signal characteristics, the timing step information and the number of set center points; the primary population includes a plurality of primary individuals.

[0103] In this embodiment, multiple first-generation individuals may be generated by extracting target signal features and the length value of their time sequence steps and the number of set center points as the gene values ​​of the first-generation individuals, and the multiple first-generation individuals constitute a first-generation population.

[0104] Step S603: Calculate the signal-to-noise ratio according to the intermediate signal characteristics and the target signal characteristics.

[0105] In this embodiment, it can be understood that the target signal feature is obtained by screening the intermediate signal feature, which is used to screen out the noise information and redundant information from the intermediate signal feature, thereby obtaining the target signal feature without the noise information and redundant information, so the intermediate signal feature contains the noise information and redundant information. The signal-to-noise ratio can be calculated by dividing the target signal feature with the difference between the target signal feature and the intermediate signal feature, and then taking the logarithm of the division result.

[0106] Step S604: According to the target signal characteristics, a preset underwater target detection function and a preset underwater target detection vector, the target to be detected is identified and analyzed, and a target recognition rate is calculated.

[0107] In this embodiment, the preset underwater target detection function and the preset underwater target detection vector may be artificially set. The target signal feature may be mapped to a high-dimensional nonlinear space by the preset underwater target detection function, and then the sound signal feature in the high-dimensional nonlinear space is nonlinearly transformed by the preset underwater target detection vector, and then the sound signal feature after the nonlinear transformation is compared with the preset target signal nonlinear feature, so as to calculate the target recognition rate.

[0108] Step S605, calculating the fitness of each first-generation individual according to the signal-to-noise ratio, target recognition rate and preset weight coefficient.

[0109] In this embodiment, the preset weight coefficient may be artificially set. The signal-to-noise ratio and the target recognition rate may be weighted and summed according to the preset weight coefficient, and the sum result is the fitness of each first-generation individual.

[0110] Step S606, screening the individuals of the first-generation population according to the fitness of the first-generation individuals to generate a second-generation population.

[0111] In this embodiment, the selection probability of each first-generation individual can be calculated according to the fitness of each first-generation individual, and then the first-generation population is screened by the selection probability, and the screened individuals are used as the second-generation individuals to generate the second-generation population. It can be understood that the higher the fitness of the individual, the higher the probability of being selected, and the higher the probability of being the second-generation individual.

[0112] Step S607, based on a preset crossover operator and a preset mutation operator, iteratively calculate the next generation population to obtain a final generation population.

[0113] In this embodiment, the preset crossover operator can be artificially set, and is used to select individuals that are paired in pairs in the population according to a certain probability, and randomly select certain gene values ​​for exchange. The preset mutation operator can be artificially set, and is used to transform the gene value of a single individual according to a certain probability, and can be based on the Craig code to perform an inversion operation on the binary bits in the gene value. It can be that the sub-generation population is continuously subjected to crossover operations and mutation operations to achieve iteration of the sub-generation population, so as to optimize the individuals of the sub-generation population, and the optimized individuals will recalculate the fitness, and the individuals with high fitness will be selected as the last generation individuals to generate the last generation population.

[0114] Step S608, determining the target time sequence step information and the target set center point quantity information according to the last generation population.

[0115] In this embodiment, the timing step and the number of set center points corresponding to the individual with the highest fitness in the last generation population may be determined and output as the target timing step information and the target set center point number information.

[0116] The signal feature extraction method provided in the embodiment of the present application uses each target signal feature and the corresponding timing step and the number of set center points as initial individuals to generate an initial population, selects, crosses and mutates the individuals of the initial population, thereby screening and optimizing the timing step and the number of set center points in each sound feature extraction scheme. By calculating the fitness of each individual and then screening the population individuals according to the fitness, it is possible to avoid falling into a local optimal solution during the iteration process, and the iteration process is not disturbed by the initial population, so that the global optimal solution can be quickly solved, which is used as the value of the timing step and the number of set center points in the optimal sound feature extraction scheme for output, so as to facilitate the subsequent adjustment of the timing step and the number of set center points, thereby improving the effectiveness and accuracy of the sound feature extraction.

[0117] Figure 7 The flowchart of the signal feature extraction method provided in the seventh embodiment of the present application is shown, which differs from the sixth embodiment in that:

[0118] The preset underwater target detection function includes a preset feature space mapping function and a preset Lagrange multiplier;

[0119] The preset underwater target detection vector includes a preset normal vector and a preset translation vector;

[0120] The step S604 specifically includes:

[0121] Step S701, labeling the discrete feature values ​​of the target signal features to obtain a plurality of signal feature labels; the signal feature labels correspond one to one to the discrete feature values.

[0122] In this embodiment, it can be understood that the target signal feature can be presented in the form of an array or sequence composed of multiple discrete numerical values ​​with time series correlation. All numerical values ​​are extracted and their order in the array or sequence is marked to obtain multiple signal feature labels.

[0123] Step S702, calculating the detection feature characterization variable of each discrete feature value according to the signal feature label, discrete feature value, preset feature space mapping function, preset normal vector, preset translation vector and preset Lagrange multiplier.

[0124] In this embodiment, the preset feature space mapping function, the preset normal vector, and the preset translation vector can be artificially set, and the preset Lagrange multiplier can be designed based on the Lagrange function. Among them, the preset feature space mapping function can be a Gaussian radial basis function, which is used to map the target signal feature to a high-dimensional feature space, so that the target signal feature value with a nonlinear law becomes linearly separable; the preset normal vector and the preset translation vector are used to form a logical hyperplane to divide the target signal feature into a detection target feature value and a non-detection target feature value, so as to characterize whether the target signal feature can accurately identify the target to be detected, the preset normal vector is used to characterize the direction of the logical hyperplane, and the preset translation vector is used to limit the boundary of the logical hyperplane. Specifically, each discrete feature value can be used as an independent variable of the feature space mapping function, and the feature space mapping function value is calculated, and then the feature space mapping function is multiplied by the signal feature label and the normal vector and the translation vector, and then the Lagrange multiplier is used for calculation and processing, and the calculation result is the detection feature characterization variable, which is used for subsequent calculation of whether the sound signal feature can accurately detect the target to be detected.

[0125] In this embodiment, preferably, the preset normal vector and the preset translation vector can be obtained by continuous optimization of the calculation process in the past research work on target detection, or can be obtained by continuous training of a specific calculation model, and are used to limit the generation range of the logical hyperplane, thereby achieving effective distinction between target feature values ​​and non-detected target feature values.

[0126] Step S703, determine whether the detection feature representation variable is greater than zero; if so, proceed to step S704; if not, proceed to step S705.

[0127] In this embodiment, when the detection feature characterization variable is greater than zero, it means that the sound signal feature can effectively characterize the features of the target to be detected, indicating that the sound signal feature is a valid feature. When the detection feature characterization variable is less than or equal to zero, it means that the sound signal feature cannot effectively characterize the features of the target to be detected, indicating that the sound signal feature is an invalid feature. It can be understood that even if the noise or redundant information is filtered out, the sound signal feature still has signal features that can effectively characterize the information of the target to be detected and signal features that cannot effectively characterize the information of the target to be detected. Therefore, it is necessary to calculate the target recognition rate to quantify the effectiveness and availability of the extracted sound signal features, and evaluate the applicability of the sound signal extraction scheme through the target recognition rate, so as to screen out the sound signal extraction scheme that can extract more effective features.

[0128] Step S704: determine the discrete feature value corresponding to the detection feature representation variable as the detection target feature value.

[0129] In this embodiment, when the detection feature characterization variable is greater than zero, it means that the sound signal feature can effectively characterize the characteristics of the target to be detected, so the discrete feature value corresponding to the detection feature characterization variable is determined as the detection target feature value, which is used for subsequent calculation of the target recognition rate.

[0130] Step S705: determining the discrete feature value corresponding to the detection feature characterization variable as a non-detection target feature value.

[0131] In this embodiment, when the detection feature characterization variable is less than or equal to zero, it means that the sound signal feature cannot effectively characterize the features of the target to be detected, so the discrete feature value corresponding to the detection feature characterization variable is determined as the non-detection target feature value for subsequent calculation of the target recognition rate.

[0132] Step S706, calculating the target recognition rate according to the detected target feature value and the non-detected target feature value.

[0133] In this embodiment, the target recognition rate may be obtained by dividing the number of detected target feature values ​​by the sum of the number of target feature values ​​and the number of non-detected target feature values, and is used to quantify the effectiveness of each sound signal feature.

[0134] The signal feature extraction method provided in the embodiment of the present application determines the validity of the extracted sound signal. Specifically, the optimal decision boundary is calculated through preset normal vectors and translation vectors and Lagrange multipliers to maximize the classification interval, and the target feature values ​​and non-detected target feature values ​​are accurately distinguished, so as to obtain the target recognition rate through calculation, so as to facilitate the validity of the extracted sound signal features through the target recognition rate detection, so as to optimize the sound signal extraction scheme and screen the optimal sound signal extraction scheme to ensure the accuracy, validity and robustness of the sound signal extraction in underwater target detection work.

[0135] Corresponding to the method of the above embodiment, Figure 8 A structural block diagram of a signal feature extraction device provided in an embodiment of the present application is shown. For ease of explanation, only the parts related to the embodiment of the present application are shown. Figure 8 The exemplary signal feature extraction device may be an execution subject of the signal feature extraction method provided in the aforementioned first embodiment.

[0136] Reference Figure 8 , the signal feature extraction device comprises:

[0137] An underwater sound signal acquisition module 810 is used to acquire an underwater sound signal of a target to be detected;

[0138] An initial signal feature generation module 820 is used to extract the time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrix, preset gating bias variables and preset normalization functions, and generate initial signal features;

[0139] An intermediate signal feature generating module 830, configured to enhance the initial signal features according to a plurality of preset linear transformation matrices to generate intermediate signal features; and

[0140] The target signal feature generation module 840 is used to classify and calculate the intermediate signal features according to a plurality of randomly generated set center points and a preset set space displacement function to generate target signal features.

[0141] The process of each module in the signal feature extraction device provided in the embodiment of the present application realizing its own function can be specifically referred to the aforementioned Figure 1 The description of the first embodiment is not repeated here.

[0142] The method implemented by the signal feature extraction device provided in the embodiment of the present application takes into account the autocorrelation and cross-correlation of each sound signal sequence, performs preliminary feature extraction on the underwater sound information of the target to be detected, and captures the time domain, frequency domain and timing characteristics of the sound signal at the same time, and then performs linear timing feature enhancement and classification on the preliminary extracted features, filters out noise information and redundant information in the signal features, and improves the comprehensiveness, effectiveness and accuracy of signal feature extraction, thereby achieving comprehensive and effective extraction of sound signals in complex water environments, so as to improve the accuracy and robustness of underwater target detection through the extracted sound signal features.

[0143] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0144] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0145] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0146] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0147] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish descriptions, and cannot be understood as indicating or suggesting relative importance. It should also be understood that although the terms "first", "second", etc. are used to describe various elements in some embodiments of the present application in the text, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, the first table can be named as the second table, and similarly, the second table can be named as the first table without departing from the scope of the various described embodiments. The first table and the second table are both tables, but they are not the same table.

[0148] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0149] The signal feature extraction method provided in the embodiments of the present application can be applied to terminal devices such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPC), netbooks, personal digital assistants (PDA), etc. The embodiments of the present application do not impose any restrictions on the specific types of terminal devices.

[0150] For example, the terminal device can be a station (STAION, ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a vehicle networking terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a TV set top box (STB), a customer premises equipment (CPE) and / or other devices for communicating on a wireless system and a next-generation communication system, such as a mobile terminal in a 5G network or a mobile terminal in a future evolved Public Land Mobile Network (PLMN) network, etc.

[0151] As an example but not limitation, when the terminal device is a wearable device, the wearable device can also be a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not just hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0152] Fig. 9 Schematic diagram of the structure of a terminal device provided by an embodiment of the present application. Fig. 9 As shown, the terminal device 9 of this embodiment includes: at least one processor 90 ( Fig. 9 Only one is shown in the figure), a memory 91, wherein the memory 91 stores a computer program 92 that can be run on the processor 90. When the processor 90 executes the computer program 92, the steps in the above-mentioned signal feature extraction method embodiments are implemented, such as Figure 1 Alternatively, when the processor 90 executes the computer program 92, the functions of each module / unit in the above-mentioned device embodiments are realized, for example Figure 8Functions of modules 810 to 840 are shown.

[0153] The terminal device 9 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The terminal device may include, but is not limited to, a processor 90 and a memory 91. Those skilled in the art will appreciate that Fig. 9 It is only an example of the terminal device 9 and does not constitute a limitation on the terminal device 9. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include an input sending device, a network access device, a bus, etc.

[0154] The processor 90 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0155] In some embodiments, the memory 91 may be an internal storage unit of the terminal device 9, such as a hard disk or memory of the terminal device 9. The memory 91 may also be an external storage device of the terminal device 9, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 9. Further, the memory 91 may also include both an internal storage unit of the terminal device 9 and an external storage device. The memory 91 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 91 may also be used to temporarily store data that has been sent or is to be sent.

[0156] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0157] An embodiment of the present application also provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor, wherein when the processor executes the computer program, the terminal device implements the steps in any of the above-mentioned method embodiments.

[0158] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0159] An embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0160] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0161] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0162] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0163] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0164] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A signal feature extraction method, characterized in that: include: Acquire underwater sound signals of the target to be detected; Extracting time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrices, preset gating bias variables and preset normalization functions to generate initial signal features; According to a plurality of preset linear transformation matrices, the initial signal features are enhanced to generate intermediate signal features; According to a plurality of randomly generated set center points and a preset set space displacement function, the intermediate signal features are classified and calculated to generate target signal features.

2. The signal feature extraction method according to claim 1, characterized in that: The preset state variables include preset hidden state variables and preset update state variables; The preset gating weight matrix includes a preset front gating weight matrix, a preset first intermediate gating weight matrix, a preset second intermediate gating weight matrix and a preset rear gating weight matrix; The preset gate bias variables include a preset pre-gate bias variable, a preset first intermediate gate bias variable, a preset second intermediate gate bias variable, and a preset post-gate bias variable; The step of extracting the time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrices, preset gating bias variables and preset normalization functions to generate initial signal features specifically includes: According to a plurality of preset timing step information, the underwater sound signal is cut and spliced ​​to obtain a plurality of signal timing segment information; According to a preset pre-gating weight matrix, a preset hidden state variable, a preset pre-gating bias variable and a preset normalization function, a plurality of signal timing segment information are mapped, transformed and normalized to obtain a pre-gating signal timing feature value; According to a preset first intermediate gating weight matrix, a preset hidden state variable, a preset first intermediate gating bias variable and a preset normalization function, a plurality of signal timing segment information are mapped and normalized to obtain a first intermediate gating signal timing feature value; According to a preset second intermediate gating weight matrix, a preset hidden state function and a preset second intermediate gating bias variable, mapping transformation and standardization processing are performed on the signal timing segment information to obtain an intermediate gating signal timing feature variable; Performing a hyperbolic tangent transformation on the intermediate gating signal timing characteristic variable to obtain a second intermediate gating signal timing characteristic value; According to a preset post-gating weight matrix, a preset hidden state variable, a preset post-gating bias variable and a preset normalization function, a plurality of signal timing segment information are mapped, transformed and normalized to obtain a post-gating signal timing feature value; An initial signal feature is generated according to the pre-gating signal timing feature value, the first intermediate gating signal timing feature value, the second intermediate gating signal timing feature value and the post-gating signal timing feature value.

3. The signal feature extraction method according to claim 2, characterized in that: The step of generating the initial signal feature according to the pre-gating signal timing feature value, the first intermediate gating signal timing feature value, the second intermediate gating signal timing feature value and the post-gating signal timing feature value specifically comprises: Calculating a state characterization timing characteristic variable according to the pre-gating signal timing characteristic value, the first intermediate gating signal timing characteristic value, the second intermediate gating signal timing characteristic value and a preset update state variable; Performing a hyperbolic tangent transformation on the state characterization time series characteristic variable to obtain a state transformation time series characteristic variable; Calculating a plurality of sound signal timing feature representation variables according to the post-gating signal timing feature value and the state change timing feature variable; An initial signal feature is generated based on a plurality of the sound signal time series feature representation variables.

4. The signal feature extraction method according to claim 1, characterized in that: The preset linear transformation matrix includes a preset first linear layer transformation matrix, a preset second linear layer transformation matrix, and a preset third linear layer transformation matrix; The step of enhancing the initial signal features according to a plurality of preset linear transformation matrices to generate intermediate signal features specifically includes: Calculate a first linear layer feature representation variable, a second linear layer feature representation variable, and a third linear layer feature representation variable according to the initial signal feature, a preset first linear layer transformation matrix, a preset second linear layer transformation matrix, and a preset third linear layer transformation matrix; Obtaining an intermediate linear representation feature variable according to the first linear layer feature representation variable and the second linear layer feature representation variable; Multiplying the intermediate linear representation feature variable and the third linear layer feature representation variable to obtain a signal feature linear representation variable; According to a preset linear layer weight matrix, weighted summation is performed on the linear representation variables of the signal features to generate intermediate signal features.

5. The signal feature extraction method according to claim 1, characterized in that: The step of classifying and calculating the intermediate signal features according to the randomly generated multiple set center points and the preset set space displacement function to generate the target signal features specifically includes: Calculate the feature set representation distance between the intermediate signal feature and multiple set center points; According to the feature set representation distance, the intermediate signal features are divided into sets corresponding to a plurality of set center points to generate a plurality of signal feature sets; Calculating the set space displacement of each intermediate signal feature according to the signal feature set and a preset set space displacement function; Determining the displacement center point of each of the signal feature sets according to the set spatial displacement; Determine whether the displacement center point is the same as the set center point; If yes, generating a target signal feature according to a plurality of said signal feature sets; If not, the displacement center point is used as the set center point, and the process returns to the step of dividing the intermediate signal features into sets corresponding to multiple set center points according to the feature set characterization distance to generate multiple signal feature sets.

6. The signal feature extraction method according to claim 1, characterized in that: After the step of classifying and calculating the intermediate signal features according to the randomly generated multiple set center points and the preset set space displacement function to generate the target signal features, the method further includes: Extracting the timing step information and the number of set center points corresponding to the plurality of target signal features; Generate a primary population according to the target signal characteristics, the time sequence step information and the number of set center points; the primary population includes a plurality of primary individuals; Calculating a signal-to-noise ratio according to the intermediate signal characteristics and the target signal characteristics; According to the target signal characteristics, the preset underwater target detection function and the preset underwater target detection vector, the target to be detected is identified and analyzed, and the target recognition rate is calculated; Calculate the fitness of each first-generation individual according to the signal-to-noise ratio, target recognition rate and preset weight coefficient; According to the fitness of the individuals of the first generation, the individuals of the first generation are screened to generate a second generation population; Based on a preset crossover operator and a preset mutation operator, iteratively calculate the next generation population to obtain a final generation population; According to the last generation population, the target time sequence step information and the target set center point quantity information are determined.

7. The signal feature extraction method according to claim 6, characterized in that: The preset underwater target detection function includes a preset feature space mapping function and a preset Lagrange multiplier; The preset underwater target detection vector includes a preset normal vector and a preset translation vector; The step of identifying and analyzing the target to be detected according to the target signal characteristics, the preset underwater target detection function and the preset underwater target detection vector, and calculating the target recognition rate specifically includes: The discrete feature values ​​of the target signal feature are marked to obtain a plurality of signal feature labels; the signal feature labels correspond one to one to the discrete feature values; Calculate the detection feature characterization variable of each discrete feature value according to the signal feature label, the discrete feature value, the preset feature space mapping function, the preset normal vector, the preset translation vector and the preset Lagrange multiplier; Determining whether the detection feature representation variable is greater than zero; If yes, the discrete feature value corresponding to the detection feature characterization variable is determined as the detection target feature value; If not, the discrete feature value corresponding to the detection feature characterization variable is determined as a non-detection target feature value; The target recognition rate is calculated according to the detected target feature value and the non-detected target feature value.

8. A signal feature extraction device, characterized in that: include: An underwater sound signal acquisition module is used to acquire the underwater sound signal of the target to be detected; An initial signal feature generation module, used to extract the time series features of the underwater sound signal according to a plurality of preset time series step information, preset state variables, preset gating weight matrix, preset gating bias variables and preset normalization functions, and generate initial signal features; An intermediate signal feature generation module, used to enhance the initial signal features according to a plurality of preset linear transformation matrices to generate intermediate signal features; as well as The target signal feature generation module is used to classify and calculate the intermediate signal features according to a plurality of randomly generated set center points and a preset set space displacement function to generate target signal features.

9. A terminal device, characterized in that: The terminal device includes a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images