A low signal-to-noise echo data de-redundancy target detection method and system

Abstract

The application relates to the technical field of echo detection, and discloses a low-signal-to-noise echo data de-redundancy target detection method and system. The method comprises the following steps: performing structural analysis on low-signal-to-noise echo data to obtain standardized echo data; performing time-frequency analysis on the standardized echo data to obtain time-frequency difference characteristics, and performing multi-dimensional screening to obtain a target waveform; performing mode deconstruction on the standardized echo data, and performing dimension reduction mapping to obtain low-dimensional embedded data; performing evolution analysis on the low-dimensional embedded data according to distribution trajectories and condensation characteristics in the low-dimensional embedded data, obtaining a mode consistency rule, and performing reverse deconstruction on the mode consistency rule to obtain a dynamic screening criterion; based on the dynamic screening criterion, mutual information optimization is performed on the low-dimensional embedded data to obtain de-redundancy data; and comprehensive research and judgment are performed on the de-redundancy data to obtain final detection data; and the application can improve the accuracy of low-signal-to-noise echo data de-redundancy target detection.

Description

Technical Field

[0001] This invention relates to the field of echo detection technology, and in particular to a method and system for redundancy removal of low signal-to-noise ratio echo data for target detection. Background Technology

[0002] In low signal-to-noise ratio echo data processing, existing technologies struggle to effectively separate noise from valid signals during data preprocessing. This results in subsequent feature extraction stages failing to accurately capture core information, leading to biases and omissions in feature representation. Furthermore, traditional modal deconstruction methods lack specificity and fail to fully uncover deep correlations and inherent patterns in the data. Consequently, decomposition coefficients fail to accurately reflect the data's essence, thus impacting the effectiveness of subsequent dimensionality reduction mapping.

[0003] Existing redundancy removal mechanisms lack dynamic adaptability, relying solely on fixed rules for data filtering. This makes it difficult to cope with dynamic changes and complex characteristics in data distribution, resulting in incomplete removal of redundant data, increasing computational load and interfering with target recognition accuracy. Furthermore, the lack of coordination among processing stages and insufficient integration between evolutionary analysis and the construction of filtering criteria lead to low overall efficiency and unreliable results in target detection. Therefore, improving the accuracy of redundancy removal and target detection from low signal-to-noise ratio echo data has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for redundancy removal target detection from low signal-to-noise ratio echo data to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for redundancy removal and target detection of low signal-to-noise ratio echo data, comprising: S1. Acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. S2. Perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform of the low signal-to-noise ratio echo data. S3. Based on the target waveform, perform modal decomposition on the standardized echo data to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimensionality reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. S4. Based on the distribution trajectory and aggregation characteristics in the low-dimensional embedded data, perform evolutionary analysis on the low-dimensional embedded data to obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. S5. Based on the dynamic screening criteria, perform mutual information optimization on the low-dimensional embedded data to obtain the deredundant data of the low signal-to-noise ratio echo data. S6. Perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

[0006] In a preferred embodiment, the step of acquiring low signal-to-noise ratio (SNR) echo data and performing structured parsing on the low SNR echo data to obtain standardized echo data of the low SNR echo data includes: Acquire low signal-to-noise ratio echo data; The low signal-to-noise ratio echo data is sparsely decoupled to obtain the primitive waveform sequence of the low signal-to-noise ratio echo data; Residual extraction is performed on the primitive waveform sequence to obtain the background noise of the low signal-to-noise ratio echo data; Based on the background noise, noise cancellation is performed on the low signal-to-noise ratio echo data to obtain clean waveform data of the low signal-to-noise ratio echo data. The clean waveform data is time-domain registered to obtain the standardized echo data of the low signal-to-noise ratio echo data.

[0007] In a preferred embodiment, the step of performing time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and then performing multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform of the low signal-to-noise ratio echo data, includes: The time-frequency characteristic distribution of the low signal-to-noise ratio echo data is obtained by performing joint time-frequency domain characterization on the standardized echo data; The time-frequency difference characteristics of the low signal-to-noise ratio echo data are obtained by extracting the difference components from the time-frequency feature distribution. The time-frequency difference features are orthogonally projected to obtain the projection components of the low signal-to-noise ratio echo data. The projection components are subjected to feature extraction to obtain the core feature components of the low signal-to-noise ratio echo data; The core feature components are reconstructed to obtain the target waveform of the low signal-to-noise ratio echo data.

[0008] In a preferred embodiment, the step of performing modal decomposition on the standardized echo data based on the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and performing dimensionality reduction mapping on the decomposition coefficients to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data, includes: The standardized echo data is interpreted in layers to obtain the structured representation components of the echo data; By correlating and coupling the structured representation components, the co-expression coefficients of the low signal-to-noise ratio echo data are obtained; Based on the collaborative expression coefficients, the structured representation components are reconstructed exponentially to obtain the decomposition coefficients of the low signal-to-noise ratio echo data; A topological graph of the decomposition coefficients is constructed using the decomposition coefficients as nodes and the correlation strength between the decomposition coefficients as edges. Embedding analysis is performed on the topological map to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data.

[0009] In a preferred embodiment, the step of performing embedding analysis on the topological map to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data includes: Calculate the adaptive weights of the feature vectors in the topological graph, wherein the formula for calculating the adaptive weights is as follows: ; in, For the first Adaptive weights for each feature vector. For the first The eigenvalues ​​of each eigenvector The total number of the feature vectors. This is an adjustment factor used to control the degree of nonlinearity in the weight distribution; Based on the adaptive weights, the feature vectors are weighted and fused to obtain the weighted feature space of the low signal-to-noise ratio echo data; The decomposition coefficients are mapped to the weighted feature space to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data.

[0010] In a preferred embodiment, the step of performing evolutionary analysis on the low-dimensional embedded data based on the distribution trajectory and aggregation characteristics in the low-dimensional embedded data to obtain the mode consistency law of the low signal-to-noise ratio echo data, and then performing inverse deconstruction on the mode consistency law to obtain the dynamic screening criteria for the low signal-to-noise ratio echo data, includes: The distribution trajectory of the low-dimensional embedded data is tracked to obtain the trajectory evolution sequence of the low signal-to-noise ratio echo data; Agglomeration characteristics are identified in the trajectory evolution sequence to obtain the evolution characteristics of the low signal-to-noise ratio echo data; Based on the evolutionary characteristics, a mode evaluation is performed on the low signal-to-noise ratio echo data to obtain the evolutionary mode of the low signal-to-noise ratio echo data. The evolutionary patterns are jointly verified to obtain the pattern consistency rules of the low signal-to-noise ratio echo data; Based on the pattern consistency rule, the low signal-to-noise ratio echo data is extracted in reverse according to the rules to obtain the dynamic screening criteria for the low signal-to-noise ratio echo data.

[0011] In a preferred embodiment, the step of performing rule-based reverse extraction on the low signal-to-noise ratio (SNR) echo data based on the pattern consistency law to obtain the dynamic screening criteria for the low SNR echo data includes: By systematically constructing the pattern consistency rules, a rule framework for the low signal-to-noise ratio echo data is obtained; The core constraints of the rule framework are extracted to obtain the key constraints of the low signal-to-noise ratio echo data. The key constraints are translated into the filtering logic for the low signal-to-noise ratio echo data. The filtering logic is criterion-based and integrated to obtain the dynamic filtering criteria for the low signal-to-noise ratio echo data.

[0012] In a preferred embodiment, the step of optimizing the mutual information of the low-dimensional embedded data based on the dynamic screening criteria to obtain the deredundant data of the low signal-to-noise ratio echo data includes: Based on the dynamic screening criteria, the low-dimensional embedded data is targeted to obtain candidate features of the low signal-to-noise ratio echo data. The candidate features are subjected to a correlation measurement to obtain a correlation measurement representation of the candidate features; Based on the aforementioned correlation metric, redundancy is removed from the low signal-to-noise ratio echo data to obtain the de-redundant data of the low signal-to-noise ratio echo data.

[0013] In a preferred embodiment, the step of comprehensively analyzing the deredundant data to obtain the final detection data of the low signal-to-noise ratio echo data includes: The performance of the deredundant data is evaluated to obtain a comprehensive evaluation characterization of the low signal-to-noise ratio echo data; The comprehensive evaluation characterization is compared with the preset evaluation benchmark value by magnitude to obtain an effective characterization of the low signal-to-noise ratio echo data. The effective characterizations are integrated at the decision level to obtain the target detection data from the low signal-to-noise ratio echo data.

[0014] To address the above problems, the present invention also provides a low signal-to-noise ratio echo data redundancy removal target detection system, the system comprising: The data processing module is used to acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. The waveform generation module is used to perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and to perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform data of the low signal-to-noise ratio echo data. The dimension reduction mapping module is used to perform modal deconstruction on the standardized echo data according to the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimension reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. The criterion generation module is used to perform evolutionary analysis on the low-dimensional embedded data based on the distribution trajectory and cohesion characteristics in the low-dimensional embedded data, obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. The mutual information optimization module is used to perform mutual information optimization on the low-dimensional embedded data based on the dynamic screening criteria to obtain redundant data of the low signal-to-noise ratio echo data. The comprehensive analysis module is used to perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This low signal-to-noise ratio echo data redundancy removal target detection method effectively extracts core data features through multiple precise processing steps such as structured analysis, time-frequency analysis, and modal deconstruction, improving the accuracy of target waveform recognition. It enables low-dimensional embedded data to better reflect the essential laws of the data, providing a high-quality data foundation for subsequent redundancy removal and detection, and significantly improving the accuracy of target detection.

[0016] 2. The construction of dynamic screening criteria and the application of mutual information optimization strategies have enabled efficient redundancy removal of low-dimensional embedded data and reduced interference from invalid data. At the same time, the comprehensive judgment process, through performance evaluation and decision-level integration, further ensures the reliability of detection results and significantly improves the efficiency and stability of target detection in low signal-to-noise ratio echo data. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for redundancy removal target detection from low signal-to-noise ratio echo data according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a low signal-to-noise ratio echo data redundancy removal target detection system provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for redundancy removal target detection using low signal-to-noise ratio (SNR) echo data. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a low signal-to-noise ratio (SNR) echo data redundancy removal target detection method according to an embodiment of the present invention. In this embodiment, the low SNR echo data redundancy removal target detection method includes: S1. Acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. In this embodiment of the invention, acquiring low signal-to-noise ratio (SNR) echo data and performing structured parsing on the low SNR echo data to obtain standardized echo data of the low SNR echo data includes: Acquire low signal-to-noise ratio echo data; The low signal-to-noise ratio echo data is sparsely decoupled to obtain the primitive waveform sequence of the low signal-to-noise ratio echo data; Residual extraction is performed on the primitive waveform sequence to obtain the background noise of the low signal-to-noise ratio echo data; Based on the background noise, noise cancellation is performed on the low signal-to-noise ratio echo data to obtain clean waveform data of the low signal-to-noise ratio echo data. The clean waveform data is time-domain registered to obtain the standardized echo data of the low signal-to-noise ratio echo data.

[0021] Low signal-to-noise ratio (SNR) echo signals from the target detection scene are received using specialized data acquisition equipment. These signals may contain useful information reflected by the target as well as irrelevant components such as environmental interference and equipment noise. The received signals are stored completely in the form of raw data to ensure the integrity and traceability of the data in subsequent processing, thereby completing the acquisition of low SNR echo data.

[0022] The acquired low signal-to-noise ratio echo data is sparsely decoupled by signal decomposition. According to the inherent characteristics of the signal such as frequency and amplitude, the complex echo data is decomposed into multiple independent basic signal units that cannot be further decomposed. Each basic signal unit has unique waveform characteristics. These independent basic signal units together constitute the primitive waveform sequence of the low signal-to-noise ratio echo data.

[0023] Using the primitive waveform sequence as the analysis object, by comparing the shape and characteristics of each primitive waveform, the signal components that do not have the characteristics of the target signal and are randomly distributed are separated. These randomly distributed signal components are the background noise of the low signal-to-noise ratio echo data. The whole process ensures the accuracy of background noise extraction by identifying and extracting the primitive waveforms one by one.

[0024] The extracted background noise is superimposed on the original low signal-to-noise ratio echo data, so that the phase of the background noise is opposite to the phase of the noise component in the original echo data and their amplitudes cancel each other out. Through this targeted signal cancellation operation, the noise interference in the original echo data is removed, and the useful signal carrying the target information is retained, finally obtaining the clean waveform data of the low signal-to-noise ratio echo data.

[0025] Using a preset standard time axis as a reference, the time dimension of the clean waveform data is adjusted and calibrated to keep the time start point, duration and other parameters of each signal component in the clean waveform data consistent with the standard time axis, eliminating the deviation of different signal components in the time dimension, so that the entire clean waveform data reaches a unified and standardized state in the time dimension, thereby obtaining standardized echo data of low signal-to-noise ratio echo data.

[0026] The beneficial effect is that through the above series of coherent and precise processing steps, the background noise is effectively separated and removed from the original low signal-to-noise ratio echo data, resulting in standardized echo data. This not only improves the cleanliness and consistency of the echo data, but also provides a high-quality and standardized data foundation for subsequent processing steps such as time-frequency analysis and target waveform extraction, ensuring the accuracy and effectiveness of the entire low signal-to-noise ratio echo data redundancy removal and target detection process.

[0027] S2. Perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform of the low signal-to-noise ratio echo data. In this embodiment of the invention, the step of performing time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and performing multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform of the low signal-to-noise ratio echo data, includes: The time-frequency characteristic distribution of the low signal-to-noise ratio echo data is obtained by performing joint time-frequency domain characterization on the standardized echo data; The time-frequency difference characteristics of the low signal-to-noise ratio echo data are obtained by extracting the difference components from the time-frequency feature distribution. The time-frequency difference features are orthogonally projected to obtain the projection components of the low signal-to-noise ratio echo data. The projection components are subjected to feature extraction to obtain the core feature components of the low signal-to-noise ratio echo data; The core feature components are reconstructed to obtain the target waveform of the low signal-to-noise ratio echo data.

[0028] The standardized echo data is subjected to feature analysis in both time and frequency dimensions. In the time dimension, the amplitude fluctuations and occurrence sequence of the data over time are recorded. In the frequency dimension, the various frequency components contained in the data and their corresponding intensity are analyzed. The analysis results of the two dimensions are correlated and integrated according to the correspondence between time and frequency to form a holistic set that can comprehensively reflect the characteristic performance of standardized echo data at different times and frequencies, thus obtaining the time-frequency feature distribution of low signal-to-noise ratio echo data.

[0029] Using the most common and stable features in the time-frequency feature distribution as a reference benchmark, each feature unit in the time-frequency feature distribution is compared one by one. Feature parts that are significantly different from the common features are identified from multiple aspects such as amplitude range, frequency interval, and duration. These distinctive feature parts are summarized and organized to obtain the time-frequency difference features of low signal-to-noise ratio echo data.

[0030] Construct a projection plane that is perpendicular to the feature space containing the time-frequency difference features. Project each feature element in the time-frequency difference features along a direction perpendicular to the projection plane, so that each feature element forms a unique corresponding mapping point on the projection plane. Then, according to the original association logic of each feature element, these mapping points are combined in an orderly manner to obtain the projection component of the low signal-to-noise ratio echo data.

[0031] All feature elements in the projection components are identified one by one. Based on the possible feature patterns of the target signal, redundant feature elements with weak amplitude, scattered distribution and irrelevant to the target signal are removed. Feature elements with stable amplitude, obvious regularity and high correlation with the target signal are retained. The retained feature elements are then structurally sorted and information is integrated to obtain the core feature components of the low signal-to-noise ratio echo data.

[0032] According to the time sequence of each feature element in the core feature component, and combined with the amplitude information of each feature element, these feature elements are connected sequentially. At the same time, reasonable transition segments are added according to the amplitude change trend of adjacent feature elements, so that each feature element forms a continuous, complete waveform structure that can accurately reflect the target information, and the target waveform of low signal-to-noise ratio echo data is obtained.

[0033] The beneficial effects are that the characteristic information of standardized echo data is fully captured by joint time-frequency domain characterization, and the core features related to the target are gradually screened out through differential composition analysis, orthogonal projection, and feature purification. Finally, the accurate target waveform is obtained through waveform reconstruction. The whole process is progressive and highly targeted, effectively removing irrelevant interference features and significantly improving the accuracy and reliability of the target waveform. This lays a solid foundation for the subsequent modal deconstruction of standardized echo data and the entire target detection process.

[0034] S3. Based on the target waveform, perform modal decomposition on the standardized echo data to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimensionality reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. In this embodiment of the invention, the step of performing modal decomposition on the standardized echo data based on the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and performing dimensionality reduction mapping on the decomposition coefficients to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data, includes: The standardized echo data is interpreted in layers to obtain the structured representation components of the echo data; By correlating and coupling the structured representation components, the co-expression coefficients of the low signal-to-noise ratio echo data are obtained; Based on the collaborative expression coefficients, the structured representation components are reconstructed exponentially to obtain the decomposition coefficients of the low signal-to-noise ratio echo data; A topological graph of the decomposition coefficients is constructed using the decomposition coefficients as nodes and the correlation strength between the decomposition coefficients as edges. Embedding analysis is performed on the topological map to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data.

[0035] The embedding analysis of the topological map yields low-dimensional embedded data of the low signal-to-noise ratio echo data, including: Calculate the adaptive weights of the feature vectors in the topological graph, wherein the formula for calculating the adaptive weights is as follows: ; in, For the first Adaptive weights for each feature vector. For the first The eigenvalues ​​of each eigenvector The total number of the feature vectors. This is an adjustment factor used to control the degree of nonlinearity in the weight distribution; Based on the adaptive weights, the feature vectors are weighted and fused to obtain the weighted feature space of the low signal-to-noise ratio echo data; The decomposition coefficients are mapped to the weighted feature space to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data.

[0036] The standardized echo data is decomposed and analyzed layer by layer according to the signal frequency level, amplitude intensity level, and time variation level. Each level focuses on the signal composition and characteristic performance of the corresponding dimension, decomposing the originally complex standardized echo data into multiple components with clear structural attributes and independent characterization. These components with clear structural attributes are the structured representation components of the echo data.

[0037] The interaction relationships between each structured representation component are analyzed one by one, including synchronous change relationships in the time dimension, mutual influence relationships in the amplitude dimension, and cooperative matching relationships in the frequency dimension. Based on these actual correlations, a unified quantization method is used to accurately characterize the degree of cooperative effect between each component. The quantization result of this characterization of the degree of cooperative effect is the cooperative expression coefficient of low signal-to-noise ratio echo data.

[0038] Based on the co-expression coefficients, the contribution ratio and priority of each structured representation component in the overall echo data are determined. The structured representation components are proportionally adjusted according to this ratio and priority. At the same time, the adjusted components are recombined and reconstructed according to the inherent logic of the signal, so that each reconstructed component can accurately reflect its core characteristics in the overall echo data. These reconstructed quantization results are the decomposition coefficients of the low signal-to-noise ratio echo data.

[0039] Each decomposition coefficient is treated as an independent node in the topological graph. The degree of correlation between any two decomposition coefficients is analyzed one by one. The correlation strength is determined by judging the degree of mutual dependence between the two in terms of signal transmission, feature representation, etc. The nodes with correlation are connected by lines as edges. The correlation strength is reflected by the changes in the attributes of the lines. The closer the correlation, the more prominent the line attributes. According to this definition of nodes and edges, all decomposition coefficients and their correlation relationships are fully presented in graphical form, and the topological graph of the decomposition coefficients is constructed.

[0040] A comprehensive analysis is conducted to determine the actual value of the signal features corresponding to each feature vector in the topological graph. By combining the distribution position of the feature vector in the topological graph, the frequency of connection with other feature vectors, and the correlation depth, the importance of each feature vector to the overall signal representation is comprehensively judged. Based on this comprehensive judgment result, each feature vector is assigned a unique importance metric value. This unique importance metric value is the adaptive weight of the feature vector in the topological graph.

[0041] No. The eigenvalues ​​of each eigenvector come from the corresponding eigenvectors in the topological graph. The total number of eigenvectors is the result of statistical analysis of all eigenvectors in the topological graph. The adjustment factor that controls the nonlinearity of the weight distribution is a fixed value that is preset to adapt to different data feature distributions.

[0042] This calculation process is achieved by analyzing the first... The eigenvalues ​​of the nth eigenvectors are subjected to a specified power operation, then the eigenvalues ​​of all eigenvectors are subjected to the same power operation and summed. Finally, the result of the first operation is divided by the result of the second operation to obtain the nth eigenvector. The adaptive weights of each feature vector are used to quantify the importance of each feature vector in the topological graph.

[0043] When the When the eigenvalues ​​of the eigenvectors increase, the result after the specified power operation will increase accordingly. If the sum of the power operations on the eigenvalues ​​of all eigenvectors remains unchanged, then the eigenvector with the eigenvalues ​​as the power of the eigenvectors will increase accordingly. The adaptive weight of the i-th feature vector will increase accordingly; when the i-th feature vector... When the eigenvalues ​​of an eigenvector decrease, its adaptive weights will also decrease. When the adjustment factor controlling the nonlinearity of the weight distribution is fixed, the influence of the eigenvalues ​​on the adaptive weights remains stable, and the eigenvalues ​​and adaptive weights are positively correlated.

[0044] According to the adaptive weights corresponding to each feature vector, different feature vectors are assigned corresponding levels of importance. Feature vectors with high weight values ​​are given higher influence in the fusion process, while feature vectors with low weight values ​​have their role in the fusion process reduced. All feature vectors are fully integrated according to this weight allocation principle to form a unified feature set that can highlight the core important features and reasonably weaken the secondary features. This unified feature set is the weighted feature space of low signal-to-noise ratio echo data.

[0045] The quantization information of each decomposition coefficient is matched one by one with each feature dimension in the weighted feature space. Based on the attribute features of the decomposition coefficient and the dimensional characteristics of the weighted feature space, the originally complex high-dimensional information of the decomposition coefficient is transformed into a concise low-dimensional representation in the feature space. This ensures that each decomposition coefficient can find a unique and accurate corresponding low-dimensional expression in the weighted feature space. All these low-dimensional expressions are integrated together to form the low-dimensional embedded data of the low signal-to-noise ratio echo data.

[0046] The beneficial effects are that, through a series of coherent and precise processing steps such as hierarchical interpretation, correlation coupling, and exponential reconstruction, standardized echo data is gradually transformed into low-dimensional embedded data. This not only fully preserves the key feature information related to target detection, but also effectively reduces the dimensionality complexity of the data and the computational load of subsequent data processing. At the same time, the reasonable setting of adaptive weights allows important features to be fully highlighted, providing high-quality and efficient data support for subsequent evolutionary analysis and dynamic screening criterion generation based on low-dimensional embedded data. This significantly improves the accuracy and operational efficiency of the entire low signal-to-noise ratio echo data redundancy removal target detection process.

[0047] S4. Based on the distribution trajectory and aggregation characteristics in the low-dimensional embedded data, perform evolutionary analysis on the low-dimensional embedded data to obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. In this embodiment of the invention, the step of performing evolutionary analysis on the low-dimensional embedded data based on the distribution trajectory and aggregation characteristics in the low-dimensional embedded data to obtain the mode consistency law of the low signal-to-noise ratio echo data, and then performing inverse deconstruction on the mode consistency law to obtain the dynamic screening criteria for the low signal-to-noise ratio echo data, includes: The distribution trajectory of the low-dimensional embedded data is tracked to obtain the trajectory evolution sequence of the low signal-to-noise ratio echo data; Agglomeration characteristics are identified in the trajectory evolution sequence to obtain the evolution characteristics of the low signal-to-noise ratio echo data; Based on the evolutionary characteristics, a mode evaluation is performed on the low signal-to-noise ratio echo data to obtain the evolutionary mode of the low signal-to-noise ratio echo data. The evolutionary patterns are jointly verified to obtain the pattern consistency rules of the low signal-to-noise ratio echo data; Based on the pattern consistency rule, the low signal-to-noise ratio echo data is extracted in reverse according to the rules to obtain the dynamic screening criteria for the low signal-to-noise ratio echo data.

[0048] In this embodiment of the invention, the step of performing rule-based reverse extraction on the low signal-to-noise ratio (SNR) echo data based on the pattern consistency law to obtain the dynamic screening criteria for the low SNR echo data includes: By systematically constructing the pattern consistency rules, a rule framework for the low signal-to-noise ratio echo data is obtained; The core constraints of the rule framework are extracted to obtain the key constraints of the low signal-to-noise ratio echo data. The key constraints are translated into the filtering logic for the low signal-to-noise ratio echo data. The filtering logic is criterion-based and integrated to obtain the dynamic filtering criteria for the low signal-to-noise ratio echo data.

[0049] Each independent data point in the low-dimensional embedded data is fully tracked. According to the chronological order of the generation of each data point, its specific location information in the low-dimensional space is recorded one by one to ensure that the timestamp of each data point corresponds accurately to its spatial location. Then, these location information arranged in chronological order are connected to form a continuous record that can completely show the movement path and change trend of the data points in the low-dimensional space. This continuous record is the trajectory evolution sequence of the low signal-to-noise ratio echo data.

[0050] Carefully observe the spatial distribution of all data points in the trajectory evolution sequence, focusing on whether data points cluster towards specific areas. Accurately identify the specific spatial intervals where data points are concentrated, and statistically analyze the proportion of data points within these intervals to the total number of data points in the trajectory evolution sequence. Analyze the density of data point distribution within these concentrated intervals, and simultaneously record the entire process of data points moving from an initial dispersed state to convergence within a concentrated interval, or from a concentrated interval to disperse outwards. Systematically extract and integrate these spatial distribution characteristics, proportions, density, and trends related to data point clustering to obtain the evolution characteristics of low signal-to-noise ratio echo data.

[0051] Based on the extracted evolutionary features, a comprehensive analysis is conducted on key indicators such as the range of concentration intervals, the changing trend of data point density, the duration of aggregation, and the data point movement rate. The intrinsic relationships and changing patterns among these indicators are explored in depth, and repeatable and stable feature combinations and manifestations are summarized. For example, some evolutionary features show that data points maintain stable aggregation in a fixed low-dimensional space interval for a long time, while others show that the concentration interval of data points gradually shrinks and the density continuously increases. These different types of manifestations with clear patterns are clearly defined and classified to obtain the evolutionary pattern of low signal-to-noise ratio echo data.

[0052] Each initially obtained evolutionary pattern is comprehensively cross-compared with the target waveform, decomposition coefficients, original features of low-dimensional embedded data, and other relevant data obtained in the previous processing. The logical consistency between each evolutionary pattern and these relevant data is verified one by one. It is checked whether the laws reflected by the evolutionary pattern are mutually corroborated with the features of the relevant data without contradictions or conflicts. Those evolutionary patterns that do not match the features of the relevant data and cannot form effective corroboration are resolutely eliminated. Only evolutionary patterns that can fully match the features of the relevant data, exist stably in multiple comparisons, and have no logical loopholes are retained. Furthermore, the common and essential laws of these stable matching evolutionary patterns are summarized to obtain the pattern consistency law of low signal-to-noise ratio echo data.

[0053] Based on the pattern consistency law, the data is systematically classified and organized from multiple perspectives, including feature dimensions, scope of application, evolutionary stages, and applicable scenarios. Laws belonging to the same feature dimension are grouped into the same category. The core description, specific applicable scenarios, and prerequisites for effectiveness of each category of law are clearly defined. Then, following the logical order from basic feature descriptions to complex evolutionary laws, and from universally applicable laws to laws for specific scenarios, the various laws are arranged hierarchically to construct a complete, hierarchical, and logically clear framework system. This ensures that each type of law has a clear location and association within the framework, resulting in a rule framework for low signal-to-noise ratio echo data.

[0054] A deep analysis of the constructed rule framework is conducted, systematically outlining the specific content of each type of rule. The core requirements that decisively influence subsequent data filtering results are precisely identified within each rule. For each core requirement, specific definition criteria and judgment bases are clarified. For example, the core requirement that "data points must be clustered within a specific low-dimensional spatial interval" requires a clear definition of the specific boundary range of that low-dimensional spatial interval. This ensures that each core requirement has clear and quantifiable judgment criteria. All these decisive core requirements and their corresponding definition criteria are extracted to obtain the key constraints for low signal-to-noise ratio echo data.

[0055] For each extracted key constraint, it is transformed into a specific, clear, and directly executable judgment logic statement, detailing how to make judgments based on this constraint during the actual data screening process. For example, if the key constraint is "data points must be clustered within a specific low-dimensional spatial interval and the density must meet the requirements," the corresponding screening logic is "first determine whether the data points in the low-dimensional embedded data are located within this specific low-dimensional spatial interval; if they are located within this interval, then further determine whether the density of the data points within this interval meets the set standard." The judgment logic corresponding to all key constraints is integrated to form a complete, coherent, and conflict-free logical system, resulting in the screening logic for low signal-to-noise ratio echo data.

[0056] The integrated filtering logics were comprehensively reviewed, and each logic was ranked according to its importance in the data filtering process. The core filtering logics to be executed first and the auxiliary filtering logics to be executed later were determined. At the same time, the connection and interaction between different filtering logics were clarified. For example, the filtering logic based on spatial location was used for preliminary filtering to select candidate data that met the spatial requirements. Then, the filtering logic based on density was used to further refine the candidate data. Specific execution standards, judgment thresholds and operation steps were defined for each filtering logic. These ranked filtering logics, clear execution standards, reasonable correlations and operation steps were unified and integrated to form a standardized, systematic and directly implementable criterion system, resulting in a dynamic filtering criterion for low signal-to-noise ratio echo data.

[0057] The beneficial effects are as follows: through a series of progressive processing steps, including tracking the distribution trajectory of low-dimensional embedded data, identifying agglomeration characteristics, evaluating evolution patterns, and co-validating, core constraints are accurately extracted from the data evolution patterns and transformed into standardized and executable screening criteria. The entire process fully integrates various relevant information such as the original characteristics, target waveform, and decomposition coefficients of the low signal-to-noise ratio echo data, ensuring that the dynamic screening criteria can deeply fit the actual characteristics and evolution patterns of the data. It has strong pertinence, dynamic adaptability, and reliability, providing a precise and effective execution basis for subsequent mutual information optimization and redundancy removal of low-dimensional embedded data. It can not only efficiently remove redundant components in the data, but also retain the accuracy and efficiency of target detection to the greatest extent.

[0058] S5. Based on the dynamic screening criteria, perform mutual information optimization on the low-dimensional embedded data to obtain the deredundant data of the low signal-to-noise ratio echo data. In this embodiment of the invention, the step of optimizing the mutual information of the low-dimensional embedded data based on the dynamic screening criteria to obtain the deredundant data of the low signal-to-noise ratio echo data includes: Based on the dynamic screening criteria, the low-dimensional embedded data is targeted to obtain candidate features of the low signal-to-noise ratio echo data. The candidate features are subjected to a correlation measurement to obtain a correlation measurement representation of the candidate features; Based on the aforementioned correlation metric, redundancy is removed from the low signal-to-noise ratio echo data to obtain the de-redundant data of the low signal-to-noise ratio echo data.

[0059] Based on the specific requirements defined in the dynamic screening criteria, including the spatial distribution range, feature attribute matching standards, and evolutionary law fit conditions that low-dimensional embedded data must meet, each feature in the low-dimensional embedded data is checked and verified one by one and comprehensively. Each feature's spatial location is checked against the criteria to ensure it falls within the specified range, its attributes match the criteria, and its evolutionary trend aligns with the evolutionary laws defined by the criteria. This ensures that each check is strictly performed according to the dynamic screening criteria, retaining all features that fully meet the criteria and effectively reflect target-related information. These rigorously selected features together constitute the candidate features for low signal-to-noise ratio echo data.

[0060] A comprehensive correlation measurement analysis is conducted on the selected candidate features. This analysis meticulously breaks down and analyzes the interrelationships between each pair of candidate features from multiple key perspectives, including feature amplitude trends, temporal synchronicity, frequency range matching, and the degree of information overlap. The correlation performance of each candidate feature with other candidate features across different dimensions is recorded in detail. For example, whether there are similar fluctuation patterns in amplitude changes, synchronous increases or decreases over time, highly overlapping frequency ranges, and content duplication in information transmission. These correlation performances are systematically summarized and quantitatively described to form a quantitative result that clearly reflects the degree, type, and scope of correlation between candidate features. This quantitative result constitutes the correlation measurement representation of the candidate features.

[0061] Using correlation metrics as the core criterion, all candidate features are individually screened and analyzed, with a focus on determining the uniqueness and irreplaceability of each feature. When correlation metrics show a high degree of overlap in the core information carried by two or more candidate features, and the effective information contained in some features can be completely covered by other features, without any additional, irreplaceable supplementary effective information value, these repetitive and unnecessary features are identified as redundant features. Following this standard, redundant features among all candidate features are comprehensively screened and eliminated, retaining only those candidate features that carry unique and effective information and cannot be replaced by other features in terms of information expression. All the retained features are integrated together to form the deredundant data of the low signal-to-noise ratio echo data.

[0062] The beneficial effects are as follows: by using dynamic screening criteria for targeted screening, candidate features highly relevant to target detection are accurately identified; then, the relationships between candidate features are comprehensively clarified through multi-dimensional correlation measurements; and finally, redundant features are efficiently eliminated based on correlation measurement representation. The entire process ensures that the retained de-redundant data can comprehensively and accurately cover the key effective information required for target detection, while minimizing the interference of useless redundant information on subsequent processing. This significantly reduces the complexity and computational load of subsequent data processing, providing high-quality and high-purity data support for the accurate acquisition of final target detection data, and effectively improving the efficiency and accuracy of target detection in low signal-to-noise ratio echo data.

[0063] S6. Perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

[0064] In this embodiment of the invention, the step of comprehensively analyzing the deredundant data to obtain the final detection data of the low signal-to-noise ratio echo data includes: The performance of the deredundant data is evaluated to obtain a comprehensive evaluation characterization of the low signal-to-noise ratio echo data; The comprehensive evaluation characterization is compared with the preset evaluation benchmark value by magnitude to obtain an effective characterization of the low signal-to-noise ratio echo data. The effective characterizations are integrated at the decision level to obtain the target detection data from the low signal-to-noise ratio echo data.

[0065] A comprehensive performance evaluation of the deredundant data is conducted from three core dimensions: information integrity, data purity, and feature discriminability. The information integrity evaluation focuses on whether the data fully covers the key feature information required for target detection and checks for any missing or omitted key features. The data purity evaluation focuses on verifying whether there are any remaining redundant components or interference information in the data that have not been removed, ensuring that the proportion of effective information in the data meets the requirements. The feature discriminability evaluation analyzes the degree to which target-related features in the data can be distinguished from other irrelevant features, and judges whether the features have clear and identifiable unique attributes. The evaluation results of these three dimensions are systematically integrated to form a quantitative description that can comprehensively reflect the overall performance of the deredundant data, resulting in a comprehensive evaluation characterization of the low signal-to-noise ratio echo data.

[0066] The preset evaluation benchmarks are based on the actual needs of target detection in low signal-to-noise ratio echo data and industry standards. They cover the minimum compliance standards for information integrity, the qualified threshold for data purity, and the minimum requirements for feature recognition. The specific values ​​of each indicator in the comprehensive evaluation representation are compared with the corresponding preset evaluation benchmarks one by one to determine whether each indicator meets or exceeds the benchmark requirements. All evaluation indicators that meet the benchmark conditions and their corresponding feature information are selected. These qualified feature information together constitute an effective representation of low signal-to-noise ratio echo data.

[0067] Based on the decision-making logic and priority ranking rules of target detection, the various feature information in the effective representation is hierarchically divided, and the priority order of core key features, important auxiliary features and general reference features is clarified. First, the basic framework of target detection is built on the basis of core key features. Then, important auxiliary features are integrated into the framework to supplement detailed information. Finally, the detection results are improved by combining general reference features. At the same time, logical conflicts that may exist between different feature information are eliminated in the integration process to ensure that all effective representations can form a unified, coherent and accurate detection result that reflects the actual situation of the target, and obtain target detection data with low signal-to-noise ratio echo data.

[0068] The beneficial effects are that by conducting multi-dimensional performance evaluation of the deredundant data, accurately comparing it with preset benchmark values, and integrating it into the system at the decision level, the final target detection data can fully retain effective information and eliminate invalid interference, while also ensuring the accuracy and reliability of the detection data. This allows the target detection results to truly reflect the target situation in the low signal-to-noise ratio echo data, providing high-quality detection data support for subsequent related application scenarios, and further improving the overall effect and practical value of redundancy-removed target detection in low signal-to-noise ratio echo data.

[0069] like Figure 2 The diagram shown is a functional block diagram of a low signal-to-noise ratio echo data redundancy removal target detection system provided in an embodiment of the present invention.

[0070] The low signal-to-noise ratio (SNR) echo data redundancy removal target detection system 100 described in this invention can be installed in an electronic device. Depending on the functions implemented, the low SNR echo data redundancy removal target detection system 100 may include a data processing module 101, a waveform generation module 102, a dimensionality reduction mapping module 103, a criterion generation module 104, a mutual information optimization module 105, and a comprehensive judgment module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0071] In this embodiment, the functions of each module / unit are as follows: The data processing module 101 is used to acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. The waveform generation module 102 is used to perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and to perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform data of the low signal-to-noise ratio echo data. The dimensionality reduction mapping module 103 is used to perform modal deconstruction on the standardized echo data according to the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimensionality reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. The criterion generation module 104 is used to perform evolutionary analysis on the low-dimensional embedded data based on the distribution trajectory and cohesion characteristics in the low-dimensional embedded data, obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. The mutual information optimization module 105 is used to perform mutual information optimization on the low-dimensional embedded data based on the dynamic screening criteria to obtain redundant data of the low signal-to-noise ratio echo data. The comprehensive analysis module 106 is used to perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

[0072] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0073] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0075] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0076] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for redundancy removal and target detection using low signal-to-noise ratio echo data, characterized in that, The method includes: S1. Acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. S2. Perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform of the low signal-to-noise ratio echo data. S3. Based on the target waveform, perform modal decomposition on the standardized echo data to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimensionality reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. S4. Based on the distribution trajectory and aggregation characteristics in the low-dimensional embedded data, perform evolutionary analysis on the low-dimensional embedded data to obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. S5. Based on the dynamic screening criteria, perform mutual information optimization on the low-dimensional embedded data to obtain the deredundant data of the low signal-to-noise ratio echo data. S6. Perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

2. The method for redundancy removal and target detection of low signal-to-noise ratio echo data as described in claim 1, characterized in that, The process of acquiring low signal-to-noise ratio (SNR) echo data and performing structured parsing on the low SNR echo data to obtain standardized echo data includes: Acquire low signal-to-noise ratio echo data; The low signal-to-noise ratio echo data is sparsely decoupled to obtain the primitive waveform sequence of the low signal-to-noise ratio echo data; Residual extraction is performed on the primitive waveform sequence to obtain the background noise of the low signal-to-noise ratio echo data; Based on the background noise, noise cancellation is performed on the low signal-to-noise ratio echo data to obtain clean waveform data of the low signal-to-noise ratio echo data. The clean waveform data is time-domain registered to obtain the standardized echo data of the low signal-to-noise ratio echo data.

3. The method for redundancy removal and target detection of low signal-to-noise ratio echo data as described in claim 1, characterized in that, The time-frequency analysis of the standardized echo data is performed to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and the time-frequency difference characteristics are filtered in multiple dimensions to obtain the target waveform of the low signal-to-noise ratio echo data, including: The time-frequency characteristic distribution of the low signal-to-noise ratio echo data is obtained by performing joint time-frequency domain characterization on the standardized echo data; The time-frequency difference characteristics of the low signal-to-noise ratio echo data are obtained by extracting the difference components from the time-frequency feature distribution. The time-frequency difference features are orthogonally projected to obtain the projection components of the low signal-to-noise ratio echo data. The projection components are subjected to feature extraction to obtain the core feature components of the low signal-to-noise ratio echo data; The core feature components are reconstructed to obtain the target waveform of the low signal-to-noise ratio echo data.

4. The method for redundancy removal target detection from low signal-to-noise ratio echo data as described in claim 1, characterized in that, The step of performing modal decomposition on the standardized echo data based on the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and performing dimensionality reduction mapping on the decomposition coefficients to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data, includes: The standardized echo data is interpreted in layers to obtain the structured representation components of the echo data; By correlating and coupling the structured representation components, the co-expression coefficients of the low signal-to-noise ratio echo data are obtained; Based on the collaborative expression coefficients, the structured representation components are reconstructed exponentially to obtain the decomposition coefficients of the low signal-to-noise ratio echo data; A topological graph of the decomposition coefficients is constructed using the decomposition coefficients as nodes and the correlation strength between the decomposition coefficients as edges. Embedding analysis is performed on the topological map to obtain low-dimensional embedded data of the low signal-to-noise ratio echo data.

5. The method for redundancy removal target detection from low signal-to-noise ratio echo data as described in claim 4, characterized in that, The embedding analysis of the topological map yields low-dimensional embedded data of the low signal-to-noise ratio echo data, including: Calculate the adaptive weights of the feature vectors in the topological graph, wherein the formula for calculating the adaptive weights is as follows: ； in, For the first Adaptive weights for each feature vector. For the first The eigenvalues ​​of each eigenvector The total number of the feature vectors. This is an adjustment factor used to control the degree of nonlinearity in the weight distribution; Based on the adaptive weights, the feature vectors are weighted and fused to obtain the weighted feature space of the low signal-to-noise ratio echo data; The decomposition coefficients are mapped to the weighted feature space to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data.

6. The method for redundancy removal and target detection of low signal-to-noise ratio echo data as described in claim 1, characterized in that, The process involves performing evolutionary analysis on the low-dimensional embedded data based on its distribution trajectory and aggregation characteristics to obtain the mode consistency law of the low signal-to-noise ratio echo data. This mode consistency law is then deconstructed in reverse to derive the dynamic screening criteria for the low signal-to-noise ratio echo data, including: The distribution trajectory of the low-dimensional embedded data is tracked to obtain the trajectory evolution sequence of the low signal-to-noise ratio echo data; Agglomeration characteristics are identified in the trajectory evolution sequence to obtain the evolution characteristics of the low signal-to-noise ratio echo data; Based on the evolutionary characteristics, a mode evaluation is performed on the low signal-to-noise ratio echo data to obtain the evolutionary mode of the low signal-to-noise ratio echo data. The evolutionary patterns are jointly verified to obtain the pattern consistency rules of the low signal-to-noise ratio echo data; Based on the pattern consistency rule, the low signal-to-noise ratio echo data is extracted in reverse according to the rules to obtain the dynamic screening criteria for the low signal-to-noise ratio echo data.

7. The method for redundancy removal target detection from low signal-to-noise ratio echo data as described in claim 6, characterized in that, Based on the pattern consistency rule, the low signal-to-noise ratio (SNR) echo data is subjected to rule-based reverse extraction to obtain dynamic screening criteria for the low SNR echo data, including: By systematically constructing the pattern consistency rules, a rule framework for the low signal-to-noise ratio echo data is obtained; The core constraints of the rule framework are extracted to obtain the key constraints of the low signal-to-noise ratio echo data. The key constraints are translated into the filtering logic for the low signal-to-noise ratio echo data. The filtering logic is criterion-based and integrated to obtain the dynamic filtering criteria for the low signal-to-noise ratio echo data.

8. The method for redundancy removal and target detection of low signal-to-noise ratio echo data as described in claim 1, characterized in that, The step of optimizing the mutual information of the low-dimensional embedded data based on the dynamic screening criteria to obtain the deredundant data of the low signal-to-noise ratio echo data includes: Based on the dynamic screening criteria, the low-dimensional embedded data is targeted to obtain candidate features of the low signal-to-noise ratio echo data. The candidate features are subjected to a correlation measurement to obtain a correlation measurement representation of the candidate features; Based on the aforementioned correlation metric, redundancy is removed from the low signal-to-noise ratio echo data to obtain the de-redundant data of the low signal-to-noise ratio echo data.

9. The method for redundancy removal target detection from low signal-to-noise ratio echo data as described in claim 1, characterized in that, The process of comprehensively analyzing the deredundant data to obtain the final detection data of the low signal-to-noise ratio echo data includes: The performance of the deredundant data is evaluated to obtain a comprehensive evaluation characterization of the low signal-to-noise ratio echo data; The comprehensive evaluation characterization is compared with the preset evaluation benchmark value by magnitude to obtain an effective characterization of the low signal-to-noise ratio echo data. The effective characterizations are integrated at the decision level to obtain the target detection data from the low signal-to-noise ratio echo data.

10. A target detection system for deduplicating low signal-to-noise ratio echo data, characterized in that, The system for implementing the low signal-to-noise ratio echo data redundancy removal target detection method according to claim 1 includes: The data processing module is used to acquire low signal-to-noise ratio echo data and perform structured parsing on the low signal-to-noise ratio echo data to obtain standardized echo data of the low signal-to-noise ratio echo data. The waveform generation module is used to perform time-frequency analysis on the standardized echo data to obtain the time-frequency difference characteristics of the low signal-to-noise ratio echo data, and to perform multi-dimensional filtering on the time-frequency difference characteristics to obtain the target waveform data of the low signal-to-noise ratio echo data. The dimension reduction mapping module is used to perform modal deconstruction on the standardized echo data according to the target waveform to obtain the decomposition coefficients of the low signal-to-noise ratio echo data, and perform dimension reduction mapping on the decomposition coefficients to obtain the low-dimensional embedded data of the low signal-to-noise ratio echo data. The criterion generation module is used to perform evolutionary analysis on the low-dimensional embedded data based on the distribution trajectory and cohesion characteristics in the low-dimensional embedded data, obtain the pattern consistency law of the low signal-to-noise ratio echo data, and perform reverse deconstruction on the pattern consistency law to obtain the dynamic screening criteria of the low signal-to-noise ratio echo data. The mutual information optimization module is used to perform mutual information optimization on the low-dimensional embedded data based on the dynamic screening criteria to obtain redundant data of the low signal-to-noise ratio echo data. The comprehensive analysis module is used to perform comprehensive analysis on the deredundancy data to obtain the target detection data of the low signal-to-noise ratio echo data.

Images