Laser seeker pseudo-random code identification method and system based on sequence deviation evaluation coefficient

Abstract

The application provides a laser seeker pseudo-random code identification method and system based on sequence deviation evaluation coefficient. The method comprises: obtaining laser echo, infrared thermal and radio frequency data generated when the laser seeker detects a target. Based on the sequence deviation evaluation coefficient, the laser echo data is phase-inverted and adjusted, and then integrated with the infrared thermal and radio frequency data to obtain a target composite data set. According to a preset wavelength separation threshold, each data set is divided into multiple sub-channel data streams according to the corresponding wavelength. Local features related to the detection signal are extracted from all sub-channel data streams by using a convolutional neural network. The local features are matched with a preset pseudo-random code sequence to verify the coincidence degree of the timing mode and the phase characteristics, and a space-time feature set containing the matching result is generated. Based on the feature set, it is determined whether the detection signal contains a target pseudo-random code, and an identification result is obtained. The application improves the identification accuracy and reliability of the laser seeker for the pseudo-random code.

Description

Technical Field

[0001] This application relates to the field of coding and recognition technology, and in particular to a pseudo-random coding and recognition method and system for laser seekers based on sequence deviation evaluation coefficients. Background Technology

[0002] In the practical application of laser semi-active guided weapon systems, the laser seeker is the core sensing component. It receives laser pulses reflected from targets and analyzes the pseudo-random codes carried within them to identify and lock onto specific targets. The pseudo-random code is a specific sequence embedded in the target indication laser pulse, used to distinguish friendly targets from jamming signals and clutter in complex electromagnetic environments. This scenario places stringent requirements on the coding and recognition technology: it must not only maintain stable recognition capabilities in complex environments such as strong electromagnetic interference and multipath reflections, resisting enemy deceptive interference; it must also possess microsecond-level real-time response speeds to match the guidance rhythm during high-speed missile maneuvers; and simultaneously, it must be able to capture subtle deviations in the timing and phase of the coded sequence to ensure high-resolution differentiation of target signals and avoid a decrease in guidance accuracy due to feature misjudgment.

[0003] Currently, a relatively mature technical solution for this requirement is a sequence matching algorithm based on dynamic time warping. This solution uses a pre-set pseudo-random encoded standard timing template to perform time-axis scaling correction on the signal sequence received by the laser seeker, eliminating timing misalignments caused by transmission delays. Then, it calculates the Euclidean distance between the corrected signal sequence and the standard template, and uses a distance threshold to determine whether a match with the target code exists. This solution has some applicability in scenarios with weak interference and has been applied in some low-to-medium altitude guidance scenarios.

[0004] However, this scheme has significant limitations in real-world combat environments: First, it relies solely on laser signal sequences for matching, without incorporating multi-source data such as infrared and radio frequency signals, making it susceptible to noise in environments with strong electromagnetic interference and lacking sufficient anti-interference capabilities. Second, it lacks a specific mechanism for handling phase shifts in the coded sequence; when relative motion between the target and the seeker causes phase deviations, the matching accuracy drops significantly. Third, the computational complexity of time-axis scaling correction increases rapidly with sequence length, resulting in significant response delays when identifying long coded sequences, making it difficult to meet the real-time requirements of high-speed target interception. Fourth, feature extraction relies solely on temporal matching, failing to effectively capture local features of the coded sequence and exhibiting limited ability to distinguish subtle temporal and phase deviations, easily affecting guidance accuracy due to feature misjudgments. These shortcomings restrict its reliable application in complex battlefield environments. Summary of the Invention

[0005] This application provides a method and system for identifying pseudo-random codes of laser seekers based on sequence deviation evaluation coefficients, in order to solve the problems of low accuracy in identifying pseudo-random codes of laser seekers caused by weak noise resistance, sensitivity to phase shift, and poor real-time performance in the prior art.

[0006] In a first aspect, this application provides a pseudo-random coding identification method for laser seekers based on sequence deviation evaluation coefficients, including:

[0007] Acquire laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection;

[0008] Based on the sequence bias evaluation coefficient, the laser echo data is phase-reversed and adjusted. The adjusted laser echo data, the infrared thermal data, and the radio frequency data are then correlated and integrated to obtain the target composite data set.

[0009] Based on a preset wavelength separation threshold, each type of data in the target composite data group is separated into multiple corresponding sub-channel data streams according to the corresponding wavelength;

[0010] A convolutional neural network is used to extract local features from multiple sub-channel data streams corresponding to all data types to obtain local features related to the detection signal.

[0011] The local features are matched and verified with the pseudo-random coding sequence of the preset laser seeker to determine the degree of consistency in terms of timing pattern and phase features, and a spatiotemporal feature set containing the matching results is generated.

[0012] Based on the spatiotemporal feature set, an identification result is determined to indicate whether the signal detected by the laser seeker contains a target pseudo-random code.

[0013] Optionally, the step of using a convolutional neural network to extract local features from multiple sub-channel data streams corresponding to all data types to obtain local features related to the detection signal includes:

[0014] For all types of data, the multiple sub-channel data streams are arranged sequentially according to source type and wavelength range to form an ordered input data string;

[0015] Based on a convolutional neural network, adjacent data points in the input data string are continuously calculated to obtain features of multiple first data segments of different lengths.

[0016] The multiple sub-channel data streams are matched with the features of the first data segment to obtain the features of the second data segment corresponding to different sub-channel data streams under the same type of data;

[0017] The features of the second data segment corresponding to all sub-channel data streams are superimposed to form a comprehensive feature segment. The significantly different parts are selected from the comprehensive feature segment and used as local features.

[0018] Optionally, the step of superimposing the features of the second data segment corresponding to all sub-channel data streams to form a comprehensive feature segment, filtering out the significantly different parts from the comprehensive feature segment, and using the significantly different parts as local features includes:

[0019] Determine the correlation between the features of the second data segment on a preset data structure, and construct a correlation framework based on the correlation;

[0020] Based on the aforementioned association framework, position matching and data overlay are performed on each feature of the second data segment to generate initial overlay features;

[0021] The initial superimposed features are adjusted according to a preset intensity ratio so that the parts of each second data segment feature in the initial superimposed features maintain coordination in expression, thereby obtaining the target superimposed features;

[0022] Based on the target superposition features, the portion whose numerical change exceeds a preset amplitude threshold is selected as the feature portion with significant differences.

[0023] The significantly different feature portions are integrated according to a preset temporal relationship to generate local features related to the detection signal.

[0024] Optionally, the step of performing position matching and data overlay on each feature of the second data segment based on the association framework to generate initial overlay features includes:

[0025] Determine the reference position of each second data segment feature on a preset data structure;

[0026] Multiple second data segment features from the same reference location are merged to form a first combined feature; multiple second data segment features from different reference locations are merged to form a second combined feature.

[0027] Based on the aforementioned association framework, time stamps and wavelength stamps are extracted from different sub-channel data streams under each type of data.

[0028] Based on the time marker and the wavelength marker, the first combined feature and the second combined feature are superimposed to generate an initial superimposed feature.

[0029] Optionally, the step of matching and verifying the local features with a preset pseudo-random encoded sequence of a laser seeker to determine the degree of agreement in terms of both temporal pattern and phase features, and generating a spatiotemporal feature set containing the matching results, includes:

[0030] The time variation information of the local features and the pseudo-random coding sequence of the preset laser seeker are extracted respectively to form the local feature time sequence and the pseudo-random coding time sequence;

[0031] Based on the variation patterns of the local feature time series and the pseudo-random coding time series, mark the time periods in which the local feature time series and the pseudo-random coding time series change in the same way;

[0032] The phase change information of the local features and the pseudo-random coding sequence are extracted respectively to form the local feature phase sequence and the pseudo-random coding phase sequence;

[0033] Based on the variation patterns of the local feature phase sequence and the pseudo-random encoded phase sequence, mark the phase intervals in which the local feature phase sequence and the pseudo-random encoded phase sequence change in the same way;

[0034] By integrating the time period and the phase interval, a spatiotemporal feature set with matching temporal and phase consistency is generated.

[0035] Optionally, the step of marking the time periods in which the changes in the local feature time series sequence and the pseudo-random coding time series are consistent, based on the changing patterns of the local feature time series sequence and the pseudo-random coding time series sequence, includes:

[0036] Based on the preset time scale correspondence, a preset time synchronization benchmark is established;

[0037] Based on the preset time synchronization benchmark, both the local feature time sequence and the pseudo-random encoded time sequence are divided into multiple time windows of equal length.

[0038] Determine the degree to which the upward or downward trend of the local feature time series sequence and the pseudo-random encoded time series sequence are the same within each time window, and quantify the degree of similarity of the trend to obtain the trend matching ratio value.

[0039] Calculate the difference in the magnitude of change between the local feature time series sequence and the pseudo-random encoded time series sequence within each time window;

[0040] The time window in which both the trend matching ratio and the difference in the magnitude of change are within the corresponding preset range is marked as the period of consistent change.

[0041] Optionally, the step of performing phase reversal adjustment on the laser echo data based on the sequence bias evaluation coefficient, and then correlating and integrating the adjusted laser echo data, the infrared thermal data, and the radio frequency data to obtain a target composite data set, includes:

[0042] Based on the sequence bias evaluation coefficient, a set of phase adjustment parameters corresponding to the phase distribution characteristics of the laser echo data is generated;

[0043] Based on the set of phase adjustment parameters, phase reversal adjustment is performed on each data point in the laser echo data to obtain the adjusted laser echo data;

[0044] Based on the adjusted laser echo data, the infrared thermal data, and the radio frequency data, the corresponding data dimension information, data sampling interval information, and data recording time information are extracted respectively, and the correlation between the corresponding data dimension information, data sampling interval information, and data recording time information is determined.

[0045] Based on the aforementioned correlation, an association model is constructed to reflect the correspondence between the adjusted laser echo data, the infrared thermal data, and the radio frequency data.

[0046] Based on the aforementioned correlation model, the adjusted laser echo data, the infrared thermal data, and the radio frequency data are integrated and processed to obtain the target composite data set.

[0047] Secondly, this application provides a pseudo-random coding recognition system for laser seekers based on sequence deviation evaluation coefficients, comprising:

[0048] The acquisition module is used to acquire laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection.

[0049] The adjustment module is used to perform phase reversal adjustment on the laser echo data based on the sequence deviation evaluation coefficient, and to correlate and integrate the adjusted laser echo data, the infrared thermal data and the radio frequency data to obtain the target composite data group.

[0050] The separation module is used to separate each type of data in the target composite data group into multiple corresponding sub-channel data streams according to the corresponding wavelength based on a preset wavelength separation threshold.

[0051] The extraction module is used to extract local features from multiple sub-channel data streams corresponding to all types of data using a convolutional neural network, so as to obtain local features related to the detection signal.

[0052] The generation module is used to match and verify the local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of consistency in terms of temporal pattern and phase features, and generate a spatiotemporal feature set containing the matching results.

[0053] The identification module is used to determine, based on the spatiotemporal feature set, the identification result indicating whether the signal detected by the laser seeker contains a target pseudo-random code.

[0054] Thirdly, this application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked and executed by the processing component to implement a pseudo-random coding recognition method for a laser seeker based on a sequence deviation evaluation coefficient as described in any of the first aspects.

[0055] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements a pseudo-random coding identification method for laser seekers based on sequence deviation evaluation coefficients as described in any of the first aspects.

[0056] This application provides a method for identifying pseudo-random codes in a laser seeker based on a sequence bias evaluation coefficient. The method includes: acquiring laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection; performing phase inversion adjustment on the laser echo data based on the sequence bias evaluation coefficient, and integrating the adjusted laser echo data, infrared thermal data, and radio frequency data to obtain a target composite data group; separating each type of data in the target composite data group into multiple sub-channel data streams according to their corresponding wavelengths based on a preset wavelength separation threshold; extracting local features from the multiple sub-channel data streams corresponding to all data types using a convolutional neural network to obtain local features related to the detection signal; matching and verifying the local features with a preset pseudo-random code sequence of the laser seeker to determine the degree of agreement in terms of temporal pattern and phase characteristics, generating a spatiotemporal feature set containing the matching results; and determining the identification result based on the spatiotemporal feature set to indicate whether the signal detected by the laser seeker contains a target pseudo-random code.

[0057] This application offers the following advantages: By acquiring laser echo data, infrared thermal data, and radio frequency data during the laser seeker's target detection process, it provides comprehensive raw data support for subsequent processing, ensuring the integrity of the data dimensions; by performing phase reversal adjustment on the laser echo data based on the sequence deviation evaluation coefficient, and then integrating the adjusted laser echo data, infrared thermal data, and radio frequency data to obtain a target composite data set, it optimizes the phase accuracy of the laser echo data. Simultaneously, by associating multi-source data, it enhances the complementarity and correlation of the data, laying the foundation for subsequent refined processing; by separating each type of data in the target composite data set into multiple sub-channel data streams according to the corresponding wavelength based on a preset wavelength separation threshold, it achieves refined data classification, allowing data with different wavelength characteristics to be presented independently, improving... The targeted and accurate feature extraction is achieved by using convolutional neural networks to extract local features from multiple sub-channel data streams corresponding to all data types. This yields local features related to the detection signal, enabling efficient capture of key local features in the data through deep learning models, reducing the limitations of manual feature design, and improving the automation and effectiveness of feature extraction. By matching and verifying local features with preset pseudo-random coding sequences, the consistency of temporal patterns and phase features is determined, and a spatiotemporal feature set is generated. This enables matching verification from two dimensions, improving the comprehensiveness and reliability of the matching results. By determining whether the signal detected by the laser seeker contains the target pseudo-random code based on the spatiotemporal feature set, accurate identification of the target pseudo-random code can be achieved, providing a reliable decision-making basis for target detection by the laser seeker.

[0058] Furthermore, when extracting local features using a convolutional neural network, this application first arranges the sub-channel data streams in an ordered manner to form an input data string. The network then obtains the features of the first data segment and matches them to obtain the features of the second data segment. Next, an association framework is constructed to perform position matching, overlay adjustment, and significant difference filtering on the features of the second data segment, ultimately yielding the local features. Through ordered arrangement, precise matching and adjustment, and difference filtering, the targeting and effectiveness of local feature extraction are improved, providing a high-quality feature foundation for subsequent matching verification.

[0059] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1A flowchart illustrating a pseudo-random code recognition method for a laser seeker based on sequence deviation evaluation coefficients, provided in this application embodiment;

[0062] Figure 2 A schematic diagram of a pseudo-random coding recognition system for a laser seeker based on sequence deviation evaluation coefficients is provided in this application embodiment;

[0063] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0064] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0065] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 11, 12, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] To address the low accuracy of pseudo-random code recognition in existing laser seeker technologies, this application provides a method for pseudo-random code recognition based on sequence bias evaluation coefficients. This method employs the following concept: first, it acquires three types of data: laser echo, infrared thermal, and radio frequency data. After adjusting the phase of the laser echo data using the sequence bias evaluation coefficient, the three types of data are integrated, then divided into multiple sub-data streams according to wavelength. Features are extracted using a correlation network, and finally, the data is matched with a preset code to determine the recognition result. This method, by combining three types of data, has stronger anti-interference capabilities; phase adjustment of the laser echo data reduces the impact of phase changes; and feature extraction after data splitting reduces computational complexity, accelerates processing speed, and better captures subtle features, thus effectively solving the problems of existing solutions.

[0068] Figure 1 A flowchart of a pseudo-random coding identification method for laser seekers based on sequence bias evaluation coefficients, provided for embodiments of this application, is shown below. Figure 1 As shown, the method includes:

[0069] S11. Acquire laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection.

[0070] Among them, laser echo data is the data formed by the reflected signal after a laser illuminates a target, reflecting the target's reflection characteristics of the laser; infrared thermal data reflects the infrared heat distribution of the target; and radio frequency data is signal data related to radio waves, used to capture the target's electromagnetic characteristics. Together, these three constitute the raw data describing the target's multi-dimensional characteristics.

[0071] In this embodiment of the application, when the laser seeker detects a target, it simultaneously receives laser reflection signals, target infrared heat information and related radio wave signals, converts these signals into a processable data form and summarizes them, for example, by using different sensors to record the object's reflection, temperature and electromagnetic reaction and then summarizing them into raw information.

[0072] S12. Based on the sequence bias evaluation coefficient, the laser echo data is phase-reversed and adjusted. The adjusted laser echo data, infrared thermal data and radio frequency data are correlated and integrated to obtain the target composite data set.

[0073] Among them, the sequence deviation evaluation coefficient is a parameter that measures the degree to which the phase of the laser echo data deviates from the standard; the phase reversal adjustment is an operation that adjusts the phase of the laser echo data in the opposite direction according to the coefficient to correct the deviation; the target composite data set is a comprehensive data set after the phase-adjusted laser echo data is associated and integrated with infrared thermal data and radio frequency data.

[0074] In this embodiment, the phase deviation of the laser echo data is first determined by the sequence deviation evaluation coefficient, and the phase inversion adjustment is performed accordingly. Then, the adjusted laser echo data is associated and integrated with infrared thermal data and radio frequency data to form a comprehensive data group, similar to correcting recording errors and merging related files.

[0075] S13. Based on the preset wavelength separation threshold, each type of data in the target composite data group is separated into multiple corresponding sub-channel data streams according to the corresponding wavelength.

[0076] The preset wavelength separation threshold is a standard value for dividing different wavelength ranges. Different types of data correspond to different thresholds. In this application embodiment, the size of the threshold is not specifically limited. Sub-channel data streams are subdivided data segments of the same type of data after being divided according to wavelength range. Each segment corresponds to a specific wavelength interval.

[0077] In this embodiment of the application, for each type of data in the target composite data group, according to its corresponding wavelength separation threshold, the part of different wavelength range is split into multiple sub-channel data streams, just like splitting white light into multiple monochromatic beams according to wavelength, which is convenient for individual processing.

[0078] S14. Use a convolutional neural network to extract local features from multiple sub-channel data streams corresponding to all types of data to obtain local features related to the detection signal.

[0079] Convolutional neural networks are computational models that can automatically extract key features from data and capture important information in complex data through multi-layer processing. Local features are key parts of the data that are related to the detected signal and reflect the unique properties of the signal.

[0080] In this embodiment, all sub-channel data streams are input into a convolutional neural network. The model automatically identifies and extracts key features related to the detection signal through layer-by-layer calculation, similar to using intelligent tools to filter out the core features of the target from complex information.

[0081] S15. Match and verify the local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of consistency in both temporal pattern and phase features, and generate a spatiotemporal feature set containing the matching results.

[0082] Among them, the preset pseudo-random coding sequence is a specific coding pattern used to identify the target, similar to an "identity password"; the temporal pattern is the law of data change over time; the phase feature is the positional characteristics of signal fluctuation; and the spatiotemporal feature set is a comprehensive information set that includes the matching results of local features and pseudo-random coding sequences in terms of time and phase.

[0083] In this embodiment of the application, local features are compared with a preset pseudo-random coding sequence to analyze the degree of matching between the two in terms of temporal change patterns and phase fluctuation positions. The results are integrated into a spatiotemporal feature set, just as a similarity report is summarized after comparing the arrangement order and character positions of two passwords.

[0084] S16. Based on the spatiotemporal feature set, determine the identification result used to indicate whether the signal detected by the laser seeker contains the target pseudo-random code.

[0085] The identification result is the conclusion of whether the detection signal contains the target pseudo-random code; the target pseudo-random code is a pre-set specific code used to confirm the identity of the target, and is a key indicator for identifying the target.

[0086] In this embodiment of the application, the matching results of the spatiotemporal feature set are analyzed to determine whether the degree of matching between the local features and the preset pseudo-random coding sequence reaches the target confirmation standard, and then the conclusion is drawn that the detection signal contains the target pseudo-random code, similar to judging whether the identity is consistent based on the password matching degree.

[0087] Based on the above, the following specific example is provided: When a Type A guided weapon detects target B, its laser seeker first collects and stores the laser echo data, infrared thermal data, and radio frequency data of target B; it then performs phase reversal adjustment on the laser echo data based on the sequence bias evaluation coefficient, and integrates the adjusted data with the infrared thermal data and radio frequency data into a target composite data group; according to a preset wavelength separation threshold, the laser echo data in the composite data group is divided into 3 sub-channel data streams, the infrared thermal data into 2, and the radio frequency data into 4; all sub-channel data streams are input into a convolutional neural network to extract local features reflecting the characteristics of the target B detection signal; the local features are compared with a preset pseudo-random coding sequence to generate a spatiotemporal feature set containing timing and phase matching results; by analyzing this feature set, it is finally determined whether the target B detection signal contains a target pseudo-random code.

[0088] By executing S11~S16, this embodiment of the application ensures information integrity through multi-source data collection, improves data accuracy and correlation through phase adjustment and multi-source integration, enhances the targeting of feature extraction by subdividing data by wavelength, efficiently extracts key features with the help of convolutional neural networks, and reduces human limitations, improves the comprehensiveness of results through temporal and phase dual-dimensional matching verification, and finally achieves accurate identification of target pseudo-random codes based on comprehensive analysis, providing a reliable decision basis for target detection of laser seekers, and improving the stability and accuracy of target identification in complex environments.

[0089] In one possible embodiment, S14, a convolutional neural network is used to extract local features from multiple sub-channel data streams corresponding to all data types to obtain local features related to the detection signal, including:

[0090] Step 141: Arrange the multiple sub-channel data streams corresponding to all types of data in order according to source type and wavelength range to form an ordered input data string.

[0091] Among them, the source type refers to the type of original data in the subchannel data stream, such as laser echo data, infrared thermal data, or radio frequency data; the wavelength range refers to the wavelength range corresponding to each subchannel data stream; and the input data string is an ordered data set formed by arranging these subchannel data streams in a certain order.

[0092] In this embodiment, the sub-channel data streams are first categorized by their source type. For example, the sub-channels for laser echo data are first grouped together, followed by the sub-channels for infrared thermal data, and finally the sub-channels for radio frequency (RF) data. Within each source type, the sub-channels are arranged in ascending order according to their corresponding wavelength range, forming an ordered input data string. For instance, if there are 3 sub-channels for laser echo (wavelengths 1, 2, and 3), 2 sub-channels for infrared thermal data (wavelengths 4 and 5), and 4 sub-channels for RF data (wavelengths 6-9), they are arranged as Laser 1, Laser 2, Laser 3, Infrared 4, Infrared 5, RF 6, RF 7, RF 8, and RF 9, forming an ordered data string.

[0093] Step 142: Based on the convolutional neural network, perform continuous calculations on adjacent data points in the input data string to obtain features of multiple first data segments of different lengths.

[0094] Among them, convolutional neural networks are models that can automatically process data and extract features, and can process continuous data like the human eye observes a continuous picture; adjacent data points are data that are consecutive in the input data string; the first data segment feature is a small part of data with different lengths calculated by the model, which reflects the local characteristics of the data.

[0095] In this embodiment, the input data string from step 141 is input into a convolutional neural network. The model slides across the data string using "windows" of different sizes (e.g., covering 3, 5, or 7 data points at a time) to calculate the relationship between adjacent data points within each window, resulting in segments of different lengths. Because the window sizes differ, the segment lengths also differ, and these segments constitute the first data segment features. For example, a small window yields short segments, and a large window yields long segments; each segment reflects the characteristics of a certain part of the data.

[0096] Step 143: Match multiple sub-channel data streams with the features of the first data segment to obtain the features of the second data segment corresponding to different sub-channel data streams under the same data type.

[0097] The second data segment feature is a feature selected from the first data segment features that matches the corresponding subchannel data stream. They belong to the same type of original data but correspond to different subchannels (different wavelengths).

[0098] In this embodiment, the first data segment features are compared one by one with the sub-channel data streams. By judging the similarity of the data change trends (such as upward or downward directions), matching segments are selected, and these segments are the second data segment features. Different sub-channels (different wavelengths) of the same type of original data (such as laser echo) will each correspond to a set of second data segment features.

[0099] Step 144: Overlay the features of the second data segments corresponding to all sub-channel data streams to form a comprehensive feature segment. Select the parts with significant differences from the comprehensive feature segment and use the parts with significant differences as local features.

[0100] Among them, the comprehensive feature segment is the overall feature formed by merging the features of all the second data segments; the significantly different part is the part of the comprehensive feature that changes significantly and is much different from other parts; the local feature is the feature selected from these significantly different parts that can reflect the key characteristics of the detection signal.

[0101] In this embodiment of the application, all features of the second data segments are merged into a comprehensive feature segment. The regions in which the data changes significantly more than other parts (such as the fluctuation of a certain data segment being much larger than other parts) are observed and selected out as local features.

[0102] Based on the above, the following specific example is provided: When the Type A guided weapon processes the data of target B, the three laser echo subchannel data streams, two infrared thermal subchannel data streams, and four radio frequency subchannel data streams obtained from S13 are first classified according to the source order from laser echo, infrared thermal to radio frequency. Within each category, the data streams are arranged from shortest to longest wavelength, forming an ordered input data string. Then, this data string is input into a convolutional neural network. The model uses a sliding window calculation containing 3, 5, and 7 adjacent data points to obtain first data segment features of lengths 3, 5, and 7. Then, these first data segment features are compared with each subchannel data stream to select matching second data segment features (3 groups for laser echo, 2 groups for infrared thermal, and 4 groups for radio frequency). Finally, all second data segment features are merged into a comprehensive feature segment, from which three regions with significant changes are selected as local features reflecting the characteristics of the target B detection signal.

[0103] By executing steps 141 to 144, this embodiment of the application arranges data by source and wavelength to ensure that subsequent processing is carried out in an orderly manner and avoids chaos; it uses a convolutional neural network to extract features of first data segments of different lengths to capture the diverse local characteristics of the data; it uses the matched features of second data segments to accurately correspond to the characteristics of each subchannel data stream, improving targeting; and finally, by merging and filtering significantly different parts, it focuses on local features that can reflect the key characteristics of the detection signal, reduces interference from irrelevant information, and provides a high-quality feature foundation for the subsequent identification of target pseudo-random codes.

[0104] In one possible embodiment, step 144 involves superimposing the features of the second data segments corresponding to all sub-channel data streams to form a comprehensive feature segment, filtering out the significantly different parts from the comprehensive feature segment, and using the significantly different parts as local features, including:

[0105] a1. Determine the correlation between the features of the second data segment on the preset data structure, and construct a correlation framework based on the correlation.

[0106] The preset data structure is a pre-defined data organization form, similar to a fixed table frame, which specifies the arrangement and association of data; the second data segment feature is the feature segment corresponding to different sub-channel data streams obtained in the previous steps; the correlation is the relationship between these feature segments, such as the order of appearance or the mutual influence of numerical changes; the correlation framework is a structure based on the correlation that reflects the relationship between feature segments.

[0107] In this embodiment of the application, the position and relationship of all second data segment features in the preset data structure are examined to determine their correlation, such as which feature segments appear adjacently in the data structure or have similar numerical change patterns. Then, a framework is built based on these correlations to indicate the connection between feature segments. For example, if feature A often appears after feature B appears, the framework reflects the sequential relationship between A and B.

[0108] a2. Based on the association framework, position matching and data overlay are performed on the features of each second data segment to generate initial overlay features.

[0109] Among them, the association framework is the structure constructed in step a1 that reflects the relationship between feature fragments; position matching is to place each second data fragment feature in the corresponding position in the framework; data overlay is to merge the feature data in the corresponding positions; and the initial overlay feature is the preliminary merged feature obtained through position matching and overlay.

[0110] In this embodiment of the application, based on the association framework of step a1, each second data segment feature is placed in the corresponding position in the framework to complete the position matching, and then the feature values ​​of the same or related positions are merged, such as by adding or averaging, to generate the initial superimposed feature. For example, if a certain position in the framework corresponds to the feature segments of laser echo and infrared heat, the values ​​of the two are merged to obtain the initial superimposed feature of that position.

[0111] a3. Adjust the initial superimposed features according to the preset intensity ratio so that the parts of each second data segment feature in the initial superimposed features are coordinated in expression, and obtain the target superimposed features.

[0112] Among them, the initial superposition feature is the preliminary merged feature obtained in step a2; the preset intensity ratio is the ratio of the expression intensity of different feature parts that is set in advance; and the target superposition feature is the superposition feature whose expression is coordinated after intensity adjustment.

[0113] In this embodiment of the application, the expression intensity of each second data segment feature in the initial superimposed feature is examined and adjusted according to a preset intensity ratio, such as reducing the part with a large value and increasing the part with a small value, so that the expression of each part is coordinated and the target superimposed feature is obtained. For example, the preset intensity ratio requires the value to be between 1 and 10. If a part of the initial value is 20, it is reduced to 10, and if a part is 0.5, it is increased to 1.

[0114] a4. Based on the target superposition features, the parts whose numerical change exceeds the preset amplitude threshold are selected as the feature parts with significant differences.

[0115] Among them, the target superimposed feature is the superimposed feature of the coordinated expression of each part obtained in step a3; the numerical change range is the magnitude of the change of the feature value within a certain range; the preset range threshold is a pre-set value to judge whether the change range is significant; the feature part with significant difference is the part whose change range exceeds the threshold.

[0116] In this embodiment of the application, the numerical change range of each part in the target superimposed feature is calculated. For example, a part changes from 10 to 30, which is a change of 20, and another part changes from 5 to 8, which is a change of 3. The change range is then compared with a preset threshold. The part that exceeds the threshold is the feature part with significant difference. For example, if the preset threshold is 10, the part with a change range of 20 is filtered out.

[0117] a5. Integrate the feature parts with significant differences according to the preset time sequence relationship to generate local features related to the detection signal.

[0118] Among them, the feature parts with significant differences are the parts with large changes selected in step a4; the preset time sequence relationship is the time order of the feature parts as specified in advance; and the local features are the key features related to the detection signal obtained after integrating the significantly different parts according to the time sequence.

[0119] In this application embodiment, the temporal order of the feature parts with significant differences is determined, arranged according to a preset temporal relationship, and then integrated into an overall feature, that is, a local feature related to the detection signal. For example, three significantly different parts are arranged and integrated according to their occurrence time to form a local feature containing these three parts.

[0120] This application provides the following specific example: When a Type A guided weapon processes data from target B, it first analyzes the correlation of the second data segment features obtained in step 143 within a preset data structure to construct a correlation framework; based on the framework, it performs position matching and data overlay on each feature to generate initial overlay features; it adjusts the initial overlay features according to a preset intensity ratio to obtain target overlay features with coordinated expression in each part; it calculates the numerical variation amplitude of each part of the target overlay features, and filters out the two parts with variation amplitudes of 8 and 12 that exceed the preset threshold of 5 as significantly different parts; finally, it integrates these parts according to the chronological order of their appearance during the detection process to form local features reflecting the characteristics of the target B detection signal.

[0121] By executing a1~a5, the embodiments of this application clarify the relationship between features by constructing an association framework, integrate relevant information through position matching and data overlay, coordinate feature expression by intensity adjustment, focus key information by screening significantly different parts, and the local features formed by time-series integration can accurately reflect the key information of the detection signal, providing a reliable basis for subsequent identification work and improving the systematicness and effectiveness of feature processing.

[0122] In one possible embodiment, a2, based on the association framework, performs position matching and data overlay on each second data segment feature to generate initial overlay features, including:

[0123] b1. Determine the reference position of each second data segment feature on the preset data structure.

[0124] The preset data structure is a predefined data organization format, similar to a fixed table, which specifies the classification and arrangement of data; the second data segment feature is the feature segment corresponding to different sub-channel data streams obtained in the previous steps; the reference position is the specific position of each second data segment feature in the preset data structure, like the position of a cell in a table.

[0125] In this embodiment of the application, the preset data structure is examined to clarify the classification and range of each position. Then, the attributes of each second data segment feature are compared with the corresponding sub-channel type and wavelength range to determine the specific position of each feature in the data structure. This position is the reference position. For example, if the preset data structure is divided into "sub-channel type + wavelength range", the feature of a certain laser echo sub-channel corresponds to the reference position of "laser echo + short wavelength".

[0126] b2. Merge the features of multiple second data segments at the same reference position to form a first combined feature, and merge the features of multiple second data segments at different reference positions to form a second combined feature.

[0127] The first combined feature is the feature obtained by merging multiple second data segment features at the same reference location; the second combined feature is the feature obtained by merging multiple related second data segment features at different reference locations; the merging process integrates the information of multiple features, such as adding or averaging the values.

[0128] In this embodiment of the application, all second data segment features at the same reference position are identified, and their numerical information is integrated to form a first combined feature. For example, position P1 has 3 feature segments, and the values ​​are added together to obtain the first combined feature. Then, feature segments at different reference positions but related in the association frame are identified, such as the features of positions P1 and P2 always appearing at the same time, and their information is integrated to form a second combined feature.

[0129] b3. Based on the association framework, extract the time stamp and wavelength stamp of different sub-channel data streams under each data of the same type.

[0130] Among them, the association framework is the structure previously constructed that reflects the relationship between feature segments; the time stamp is the time point when each second data segment feature appears during the detection process; the wavelength stamp is the wavelength range corresponding to each feature; and the different sub-channel data streams under the same type of data are sub-data belonging to the same original data type (such as laser echo) but in different wavelength ranges.

[0131] In this embodiment, the time point of each second data segment feature extracted from the associated frame is used as a time marker, and then the wavelength marker is determined according to the wavelength range of the sub-channel data stream corresponding to the feature. For example, if a feature is found to appear in the frame at 5 seconds after detection, corresponding to a wavelength of 3-5μm, the time marker "5 seconds" and the wavelength marker "3-5μm" are extracted.

[0132] b4. Based on time markers and wavelength markers, the first combination feature and the second combination feature are superimposed to generate an initial superimposed feature.

[0133] Among them, the time marker is the time point when the feature appears; the wavelength marker is the wavelength range corresponding to the feature; the first combined feature and the second combined feature are the integrated features obtained in step b2; data overlay is to merge different combined features according to rules; the initial overlay feature is the preliminary integrated feature obtained after overlay.

[0134] In this embodiment, the first combination features and the second combination features are sorted according to the time marker, and features at the same time point are grouped together. Within the same time group, the features are mapped to the corresponding wavelength range according to the wavelength marker, and the feature values ​​in the same range are integrated (such as by adding or averaging) to generate the initial superimposed features. For example, in the features at the 5th second of the time marker, the features with wavelengths of 3-5μm are superimposed and merged.

[0135] This application provides the following specific example: When a Type A guided weapon processes data from target B, it first determines the reference positions (such as P1, P2, P3) of the second data segment features in step 143 according to the subchannel type (laser echo, infrared thermal, radio frequency) and wavelength range; then it merges the features at the same position into the first combined feature (such as adding the values ​​of the two laser echo features of P1 to get F1, and the one infrared thermal feature of P2 to get F2), and merges the features of the associated positions (P1 and P3) into the second combined feature (such as averaging the values ​​to get F3); then it extracts the time stamp (such as F1 being the 6th second, F2 being the 8th second, and F3 being the 6th second) and wavelength stamp (such as F1 and F3 both containing "2-4μm") from the associated frame; finally, it sorts by time, adds F1 (value 10) and F3 (value 8) which are at the 6th second and have a wavelength of "2-4μm", and together with F2, they form the initial superposition feature.

[0136] By executing b1~b4, this embodiment of the application achieves precise feature localization by determining the reference position, integrates scattered features into a combined feature of the system through merging processing, extracts time and wavelength markers to provide a matching basis for feature superposition, and finally the initial superimposed feature formed by superposition not only reflects the continuity of time but also retains the difference in wavelength dimension, thus improving the systematicness and accuracy of feature processing as a whole and laying a reliable foundation for subsequent steps.

[0137] In one possible embodiment, S15, the local features are matched and verified with the preset pseudo-random coding sequence of the laser seeker to determine the degree of agreement in terms of both temporal pattern and phase features, and a spatiotemporal feature set containing the matching results is generated, including:

[0138] Step 151: Extract the time variation information of local features and the preset pseudo-random coding sequence of the laser seeker, respectively, to form a local feature time series sequence and a pseudo-random coding time series sequence.

[0139] Among them, local features are key feature parts extracted from the detection signal, which can reflect the unique characteristics of the signal; the preset pseudo-random coding sequence of the laser seeker is a specific code set in advance to identify the target, similar to the target's "identity password"; time change information is the change of features or codes over time, such as numerical rise and fall; local feature time sequence is a sequence of local feature time change information arranged in chronological order; pseudo-random coding time sequence is a sequence of pseudo-random coding time change information arranged in chronological order.

[0140] In this embodiment, the numerical changes of local features at different time points are observed, these changes are recorded and arranged in chronological order to form a local feature time sequence, such as value 1 at second 1, value 3 at second 2, and value 2 at second 3, forming a corresponding sequence; the preset pseudo-random encoding sequence is processed in the same way, its time changes are recorded and arranged in order to form a pseudo-random encoding time sequence, such as encoding at second 1, value 3 at second 2, and value 2 at second 3, forming a corresponding sequence.

[0141] Step 152: Based on the changing patterns of the local feature time series and the pseudo-random coding time series, mark the time periods in which the local feature time series and the pseudo-random coding time series change in the same way.

[0142] Among them, the local feature time series and the pseudo-random coding time series are the time series formed in step 151; the change pattern is the trend of the value in the series changing with time, such as rising or falling; the time period is a time interval, such as the 2nd to the 4th second.

[0143] In this embodiment of the application, the changing trends of two time series are compared, and the time periods in which the changing directions of the two are consistent are found, such as both rising or both falling at the same time. These time periods are marked. For example, if the value of the local feature time series changes from 3, 5 to 4 in the 2nd to 4th second, and the encoded time series changes from 3, 5 to 4 in the same time, then the 2nd to 4th second is marked as the time period with consistent changes.

[0144] Step 153: Extract the phase change information of the local features and the pseudo-random coding sequence respectively to form the local feature phase sequence and the pseudo-random coding phase sequence.

[0145] Among them, phase change information is the change of phase of a feature or code over time, such as the increase or decrease of phase value; local feature phase sequence is a sequence of phase change information of local features arranged in chronological order; pseudo-random code phase sequence is a sequence of phase change information of pseudo-random codes arranged in chronological order.

[0146] In this embodiment, the phase values ​​of local features at different time points are recorded and arranged in chronological order to form a local feature phase sequence, such as 30° at the 1st second, 60° at the 2nd second, and 45° at the 3rd second, forming a corresponding sequence; the pseudo-random coding sequence is processed in the same way, its phase changes are recorded and arranged in order to form a pseudo-random coding phase sequence, such as coding at 30° at the 1st second, 60° at the 2nd second, and 45° at the 3rd second, forming a corresponding sequence.

[0147] Step 154: Based on the variation patterns of the local feature phase sequence and the pseudo-random coding phase sequence, mark the phase intervals in which the local feature phase sequence and the pseudo-random coding phase sequence change in the same way.

[0148] Among them, the local feature phase sequence and the pseudo-random encoded phase sequence are the phase sequences formed in step 153; the phase interval is the range of phase values, such as 30°-60°; the phase interval with consistent changes is the time period in which the two sequences have the same phase range within the same time period.

[0149] In this embodiment of the application, the phase values ​​at each time point in the two phase sequences are compared, and the time periods with the same phase range are identified and marked as phase intervals with consistent changes. For example, if the local feature phase is 30°-60° in the 2nd-4th second and the encoded phase is also 30°-60° in the same time period, then the time period is marked as a phase interval with consistent changes.

[0150] Step 155: Integrate time periods and phase intervals to generate a spatiotemporal feature set that matches both temporal and phase consistency.

[0151] The spatiotemporal feature set is a collection that integrates time periods and phase intervals, including the matching of time and phase.

[0152] In this embodiment of the application, the time period marked in step 152 and the phase interval marked in step 154 ​​are integrated to record the phase interval situation in each time period and form a spatiotemporal feature set. For example, the 2nd to 4th seconds are a consistent time period, and the phase interval in this time period is 30°-60°. This correspondence is included in the spatiotemporal feature set.

[0153] For example, when a guided weapon of type A processes data from target B, it first extracts the temporal changes of local features (values ​​for seconds 1-5: 1, 3, 2, 4, and 3), forming a local feature time sequence. Simultaneously, it extracts the temporal changes of a preset pseudo-random coding sequence (same time values: 1, 3, 2, 4, and 3), forming a pseudo-random coding time sequence. Comparison reveals that the two sequences exhibit consistent trends in seconds 1-5, and these time periods are marked. Next, it extracts the phase changes of local features (phase for seconds 1-5: 30°, 60°, 45°, 60°, and 30°), forming a local feature phase sequence; it also extracts the phase changes of the coding sequence (same time phase: 30°, 60°, 45°, 60°, and 30°), forming a pseudo-random coding phase sequence. Comparison shows that seconds 1-5 are marked as the phase-consistent interval. Finally, it integrates the time periods (seconds 1-5) and the phase interval (30°-60°) to generate a spatiotemporal feature set containing the time and phase matching cases.

[0154] By executing steps 151 to 155, this embodiment of the application extracts the time and phase change information of local features and pseudo-random coding sequences to form corresponding sequences and marks time periods and phase intervals with consistent changes, ultimately integrating them into a spatiotemporal feature set. This process comprehensively presents the consistency between the two from both time and phase dimensions, avoiding the one-sidedness of single-dimensional judgment, providing a clear and comprehensive basis for subsequent identification of target pseudo-random codes, and improving the reliability of identification.

[0155] In one possible embodiment, step 152, marking the time periods in which the changes in the local feature time series and the pseudo-random coding time series are consistent, based on the changing patterns of the local feature time series and the pseudo-random coding time series, includes:

[0156] c1. Establish a preset time synchronization benchmark based on the preset time scale correspondence.

[0157] The preset time scale correspondence is a pre-defined rule for time marker correspondence, such as 1 second corresponding to 1 scale unit; the time synchronization benchmark is a unified time standard established based on this correspondence, used to unify the time reference of different sequences.

[0158] In this embodiment of the application, a preset time scale correspondence is determined (e.g., 1 scale = 1 second), and a unified time standard is established based on this to ensure that the local feature time series and the pseudo-random encoded time series use the same standard as the time reference.

[0159] c2. Based on the preset time synchronization benchmark, both the local feature time series and the pseudo-random encoded time series are divided into multiple time windows of equal length.

[0160] Among them, the time synchronization benchmark is a unified time standard established by c1; the time window of equal length is a time segment of the same duration (such as 2 seconds / window); the local feature time series sequence and the pseudo-random coding time series sequence are feature and coding change sequences arranged in time.

[0161] In this embodiment of the application, a time window duration (e.g., 2 seconds) is set based on a time synchronization benchmark, and the two sequences are divided into multiple segments of equal duration according to this duration, with each segment corresponding to the same time period content.

[0162] c3. Determine the degree of similarity between the local feature time series and the pseudo-random encoded time series in each time window, and quantify the degree of similarity to obtain the trend matching ratio.

[0163] Among them, the time window is an equal-length segment divided by c2; the degree of trend similarity is the degree of consistency in the direction of numerical rise and fall of the two sequences within the same window; the trend matching ratio is the quantitative result of this degree (e.g., 0.8 indicates that the trends are mostly consistent).

[0164] In this embodiment of the application, the numerical change trends (increasing / decreasing) of the two sequences within each time window are compared, and the proportion of time points with the same trend to the total number of time points is calculated to obtain the trend matching ratio (e.g., if 4 out of 5 time points have the same trend, the ratio is 4 / 5 = 0.8).

[0165] c4. Calculate the difference in the magnitude of change between the local feature time series sequence and the pseudo-random encoded time series sequence within each time window.

[0166] Among them, the magnitude of change is the amount of numerical change of the sequence within the time window (e.g., the magnitude from 1 to 3 is 2); the difference in magnitude of change is the difference in the magnitude of change of the two sequences within the same window.

[0167] In the embodiments of this application, the change amplitude of the two sequences within each time window is calculated (e.g., local feature amplitude = 3-1 = 2, encoding amplitude = 3-1 = 2), and then the difference between the two is calculated (e.g., 2-2 = 0) to obtain the change amplitude difference value.

[0168] c5. The time window in which both the trend matching ratio and the difference in the magnitude of change are within the corresponding preset range is the period of consistent change.

[0169] The preset range is the numerical interval for judging whether the trend and magnitude match (such as ratio 0.7-1, difference 0-1); the period of consistent change is the time window in which the trend and magnitude of the two sequences are both within the preset range.

[0170] In this embodiment of the application, it is checked whether the trend matching ratio and the difference in change magnitude of each time window are both within a preset range, and the windows that meet the conditions are marked as time periods with consistent changes.

[0171] For example, when the A-type guided weapon processes the data of target B, it first establishes a time synchronization benchmark based on the correspondence of "1 scale = 1 second"; then, it divides the two sequences with a total duration of 10 seconds into 5 equal-length windows of 2 seconds each; for each window, it calculates the trend matching ratio (e.g., window 1: the trends at the two time points are the same, ratio = 2 / 2 = 1; window 2: ratio = 1) and the difference in change amplitude (window 1: local feature amplitude = 2, coding amplitude = 2, difference = 0; window 2: difference = 0); finally, according to the preset range (ratio 0.7-1, difference 0-1), it marks the first 3 windows as the period of consistent change.

[0172] By executing c1~c5, this embodiment of the application ensures the consistency of comparison by establishing a unified time benchmark, realizes segmented fine analysis by dividing equal-length windows, and marks consistent time periods by combining quantitative indicators of trend and amplitude, thus comprehensively improving the accuracy of sequence comparison in the time dimension and providing a reliable time matching basis for the identification of target pseudo-random codes.

[0173] In one possible embodiment, S12, based on the sequence bias evaluation coefficient, the laser echo data is phase-reversed and adjusted; the adjusted laser echo data, infrared thermal data, and radio frequency data are correlated and integrated to obtain a target composite data set, including:

[0174] Step 121: Based on the sequence bias evaluation coefficient, generate a set of phase adjustment parameters corresponding to the phase distribution characteristics of the laser echo data.

[0175] Among them, the sequence deviation evaluation coefficient is a parameter that measures the degree to which the phase of the laser echo data deviates from the standard state; the phase distribution characteristics are the distribution of phase values ​​in the laser echo data; and the phase adjustment parameter set is a set of parameters generated by combining the evaluation coefficient and the phase distribution characteristics, which is used to adjust the phase, including the adjustment amplitude and direction for each data point.

[0176] In this embodiment, the phase deviation of the laser echo data is analyzed based on the sequence deviation evaluation coefficient, and the adjustment amplitude and direction of each phase value are determined in combination with the phase distribution characteristics. This information is organized into a phase adjustment parameter set. For example, if a phase value deviates from the standard value by 10° and needs to be adjusted in the opposite direction, the parameter set will record that the phase value needs to be reduced by 10°.

[0177] Step 122: Based on the phase adjustment parameter set, perform phase reversal adjustment on each data point in the laser echo data to obtain the adjusted laser echo data.

[0178] Among them, the phase adjustment parameter set is the adjustment parameter generated in step 121; the data point is the basic unit of laser echo data, each with a corresponding phase value; the phase reversal adjustment is the operation of adjusting the phase of the data point according to the parameters; the adjusted laser echo data is data that is closer to the standard state after phase adjustment.

[0179] In this embodiment, based on the phase adjustment parameter set, each corresponding data point in the laser echo data is found, and its phase value is changed according to the adjustment amplitude and direction in the parameters. For example, the original phase value of 30° is reduced by 10° to 20° according to the parameters to obtain the adjusted laser echo data.

[0180] Step 123: Based on the adjusted laser echo data, infrared thermal data, and radio frequency data, extract their respective data dimension information, data sampling interval information, and data recording time information, and determine the correlation between their respective data dimension information, data sampling interval information, and data recording time information.

[0181] Among them, the three types of data refer to the adjusted laser echo data, infrared thermal data, and radio frequency data; the data dimension information is the number of features contained in the data; the data sampling interval information is the time interval between two acquisitions; the data recording time information is the specific time when the data was recorded; and the correlation is the correspondence between the three types of data in terms of dimension, sampling interval, and recording time.

[0182] In the embodiments of this application, the dimensions, sampling intervals, and recording time information of the three types of data are extracted respectively, and their correspondence is compared and analyzed. For example, it is determined whether they are recorded at the same time, whether the dimensional features are related, and the correlation is summarized.

[0183] Step 124: Based on the correlation, construct a correlation model to reflect the correspondence between the adjusted laser echo data, infrared thermal data, and radio frequency data.

[0184] Among them, the association situation is the correspondence between the three types of data determined in steps 123; the association model is a structure that reflects the correspondence between the three types of data, explaining the correspondence between the features of a certain type of data and the features of other types of data.

[0185] In this embodiment of the application, a structure reflecting the correspondence between the three types of data is constructed based on their correlation. For example, "the laser echo intensity at the same moment corresponds to the infrared thermal temperature and the radio frequency signal intensity" to form an correlation model.

[0186] Step 125: Based on the correlation model, the adjusted laser echo data, infrared thermal data and radio frequency data are integrated and processed to obtain the target composite data set.

[0187] Among them, the association model is the correspondence structure constructed in step 124; the integration process is the operation of merging the three types of data according to the association model; the target composite data group is the integrated data set containing the three types of data information.

[0188] In this embodiment of the application, based on the correspondence of the association model, the corresponding features and time information of the three types of data are merged. For example, the intensity features of the three types of data at the same time are integrated to form a target composite data group.

[0189] For example, when Type A guided weapon processes data from target B, it first generates a phase adjustment parameter set containing adjustment values ​​(e.g., +5°, -10°) for each data point by using the sequence bias evaluation coefficient and the phase distribution characteristics of the laser echo data. Then, it adjusts the phase of each data point in the laser echo data (e.g., from 30° to 20°) according to this parameter set to obtain the adjusted laser echo data. Next, it extracts the dimensions, sampling intervals, and recording time information of the adjusted laser echo data (3D, 0.1-second interval, time 1-10 seconds), infrared thermal data (2D, 0.1-second interval, time 1-10 seconds), and radio frequency data (4D, 0.1-second interval, time 1-10 seconds), and finds that the recording times of the three are one-to-one and the intensity dimensions are related. Based on this, it constructs an association model to clarify the relationship of "corresponding intensity features at the same time". Finally, it merges the features of the corresponding times in the three types of data according to the model to form a target composite data group.

[0190] By executing steps 121 to 125, this embodiment of the application effectively improves data accuracy by generating a precise set of phase adjustment parameters and correcting the phase of laser echo data; by extracting the correlation information of multi-source data and constructing a model, the inherent relationship between data is clearly presented; the target composite data group finally integrated achieves the orderly concentration of multi-source information, preserves data correlation, and provides complete and reliable basic data for subsequent unified analysis and processing, thereby improving the efficiency and reliability of data processing as a whole.

[0191] Figure 2 A schematic diagram of a pseudo-random coding recognition system for a laser seeker based on sequence deviation evaluation coefficients is provided in this application embodiment, as shown below. Figure 2 As shown, the system includes:

[0192] The acquisition module 21 is used to acquire laser echo data, infrared thermal data and radio frequency data generated by the laser seeker during the target detection process.

[0193] The adjustment module 22 is used to perform phase reversal adjustment on the laser echo data based on the sequence deviation evaluation coefficient, and to integrate the adjusted laser echo data, infrared thermal data and radio frequency data to obtain the target composite data group.

[0194] The separation module 23 is used to separate each type of data in the target composite data group into multiple sub-channel data streams according to the corresponding wavelength based on a preset wavelength separation threshold.

[0195] Extraction module 24 is used to extract local features from multiple sub-channel data streams corresponding to all types of data using a convolutional neural network, so as to obtain local features related to the detection signal.

[0196] The generation module 25 is used to match and verify local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of consistency in terms of temporal pattern and phase features, and generate a spatiotemporal feature set containing the matching results.

[0197] The identification module 26 is used to determine the identification result based on the spatiotemporal feature set, which indicates whether the signal detected by the laser seeker contains a target pseudo-random code.

[0198] Figure 2 The laser seeker pseudo-random coding recognition system based on sequence bias evaluation coefficient can perform... Figure 1 The implementation principle and technical effects of the laser seeker pseudo-random code recognition method based on sequence deviation evaluation coefficients described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the laser seeker pseudo-random code recognition system based on sequence deviation evaluation coefficients in the above embodiments have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0199] In one possible design, Figure 2 The laser seeker pseudo-random coding recognition system based on sequence bias evaluation coefficients in the illustrated embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32.

[0200] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0201] The processing component 32 is used to perform the following process: acquiring laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection; performing phase reversal adjustment on the laser echo data based on the sequence deviation evaluation coefficient, and integrating the adjusted laser echo data, infrared thermal data, and radio frequency data to obtain a target composite data group; separating each type of data in the target composite data group into multiple sub-channel data streams according to the corresponding wavelength based on a preset wavelength separation threshold; extracting local features from the multiple sub-channel data streams corresponding to all types of data using a convolutional neural network to obtain local features related to the detection signal; matching and verifying the local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of consistency in both timing pattern and phase features, and generating a spatiotemporal feature set containing the matching results; and determining the identification result based on the spatiotemporal feature set to indicate whether the signal detected by the laser seeker contains the target pseudo-random code.

[0202] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0203] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Random Access Memory (RAM), Static Random-Access Memory (SRAM), Erasable Programmable Read Only Memory (EPROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0204] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0205] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0206] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0207] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0208] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The illustrated embodiment is a pseudo-random coding recognition method for laser seekers based on sequence deviation evaluation coefficients.

[0209] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0210] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units 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. Those skilled in the art can understand and implement this without any creative effort.

[0211] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0212] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A pseudo-random coding recognition method for laser seekers based on sequence bias evaluation coefficients, characterized in that, include: Acquire laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection; Based on the sequence deviation evaluation coefficient, the laser echo data is phase-reversed and adjusted. The adjusted laser echo data, the infrared thermal data, and the radio frequency data are then correlated and integrated to obtain the target composite data set. The sequence deviation evaluation coefficient is a parameter that measures the degree of phase deviation of the laser echo data from the standard. Based on a preset wavelength separation threshold, each type of data in the target composite data group is separated into multiple corresponding sub-channel data streams according to the corresponding wavelength; A convolutional neural network is used to extract local features from multiple sub-channel data streams corresponding to all data types to obtain local features related to the detection signal. The local features are matched and verified with the pseudo-random coding sequence of the preset laser seeker to determine the degree of consistency in terms of timing pattern and phase features, and a spatiotemporal feature set containing the matching results is generated. Based on the spatiotemporal feature set, a recognition result is determined to indicate whether the signal detected by the laser seeker contains a target pseudo-random code; The step of performing phase reversal adjustment on the laser echo data based on the sequence bias evaluation coefficient includes: Based on the sequence bias evaluation coefficient, a set of phase adjustment parameters corresponding to the phase distribution characteristics of the laser echo data is generated; Based on the set of phase adjustment parameters, phase reversal adjustment is performed on each data point in the laser echo data to obtain the adjusted laser echo data; wherein the phase reversal adjustment is a reverse adjustment.

2. The method according to claim 1, characterized in that, The method of using a convolutional neural network to extract local features from multiple sub-channel data streams corresponding to all data types to obtain local features related to the detection signal includes: For all types of data, the multiple sub-channel data streams are arranged sequentially according to source type and wavelength range to form an ordered input data string; Based on a convolutional neural network, adjacent data points in the input data string are continuously calculated to obtain features of multiple first data segments of different lengths. The multiple sub-channel data streams are matched with the features of the first data segment to obtain the features of the second data segment corresponding to different sub-channel data streams under the same type of data; The features of the second data segment corresponding to all sub-channel data streams are superimposed to form a comprehensive feature segment. The significantly different parts are selected from the comprehensive feature segment and used as local features.

3. The method according to claim 2, characterized in that, The process of superimposing the features of the second data segment corresponding to all sub-channel data streams to form a comprehensive feature segment, filtering out the significantly different parts from the comprehensive feature segment, and using the significantly different parts as local features includes: Determine the correlation between the features of the second data segment on a preset data structure, and construct a correlation framework based on the correlation; Based on the aforementioned association framework, position matching and data overlay are performed on each feature of the second data segment to generate initial overlay features; The initial superimposed features are adjusted according to a preset intensity ratio so that the parts of each second data segment feature in the initial superimposed features maintain coordination in expression, thereby obtaining the target superimposed features; Based on the target superposition features, the portion whose numerical change exceeds a preset amplitude threshold is selected as the feature portion with significant differences. The significantly different feature portions are integrated according to a preset temporal relationship to generate local features related to the detection signal.

4. The method according to claim 3, characterized in that, Based on the association framework, the step of performing position matching and data overlay on each feature of the second data segment to generate initial overlay features includes: Determine the reference position of each second data segment feature on a preset data structure; Multiple second data segment features from the same reference location are merged to form a first combined feature, and multiple second data segment features from different reference locations are merged to form a second combined feature. Based on the aforementioned association framework, time stamps and wavelength stamps are extracted from different sub-channel data streams under each type of data. Based on the time marker and the wavelength marker, the first combined feature and the second combined feature are superimposed to generate an initial superimposed feature.

5. The method according to claim 1, characterized in that, The step of matching and verifying the local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of agreement in terms of both temporal pattern and phase features, and generating a spatiotemporal feature set containing the matching results, includes: The time variation information of the local features and the pseudo-random coding sequence of the preset laser seeker are extracted respectively to form the local feature time sequence and the pseudo-random coding time sequence; Based on the variation patterns of the local feature time series and the pseudo-random coding time series, mark the time periods in which the local feature time series and the pseudo-random coding time series change in the same way; The phase change information of the local features and the pseudo-random coding sequence are extracted respectively to form the local feature phase sequence and the pseudo-random coding phase sequence; Based on the variation patterns of the local feature phase sequence and the pseudo-random encoded phase sequence, mark the phase intervals in which the local feature phase sequence and the pseudo-random encoded phase sequence change in the same way; By integrating the time period and the phase interval, a spatiotemporal feature set with matching temporal and phase consistency is generated.

6. The method according to claim 5, characterized in that, The step of marking time periods in which the changes in the local feature time series and the pseudo-random encoded time series are consistent, based on the variation patterns of the local feature time series and the pseudo-random encoded time series, includes: Based on the preset time scale correspondence, a preset time synchronization benchmark is established; Based on the preset time synchronization benchmark, both the local feature time sequence and the pseudo-random encoded time sequence are divided into multiple time windows of equal length. Determine the degree to which the upward or downward trend of the local feature time series sequence and the pseudo-random encoded time series sequence are the same within each time window, and quantify the degree of similarity of the trend to obtain the trend matching ratio value. Calculate the difference in the magnitude of change between the local feature time series sequence and the pseudo-random encoded time series sequence within each time window; The time window in which both the trend matching ratio and the difference in the magnitude of change are within the corresponding preset range is marked as the period of consistent change.

7. The method according to claim 1, characterized in that, The step of correlating and integrating the adjusted laser echo data, the infrared thermal data, and the radio frequency data to obtain a target composite data set includes: Based on the adjusted laser echo data, the infrared thermal data, and the radio frequency data, the corresponding data dimension information, data sampling interval information, and data recording time information are extracted respectively, and the correlation between the corresponding data dimension information, data sampling interval information, and data recording time information is determined. Based on the aforementioned correlation, an association model is constructed to reflect the correspondence between the adjusted laser echo data, the infrared thermal data, and the radio frequency data. Based on the aforementioned correlation model, the adjusted laser echo data, the infrared thermal data, and the radio frequency data are integrated and processed to obtain the target composite data set.

8. A pseudo-random coding recognition system for a laser seeker based on sequence bias evaluation coefficients, characterized in that, include: The acquisition module is used to acquire laser echo data, infrared thermal data, and radio frequency data generated by the laser seeker during target detection. The adjustment module is used to perform phase reversal adjustment on the laser echo data based on the sequence deviation evaluation coefficient, and to correlate and integrate the adjusted laser echo data, the infrared thermal data and the radio frequency data to obtain the target composite data group. The separation module is used to separate each type of data in the target composite data group into multiple corresponding sub-channel data streams according to the corresponding wavelength based on a preset wavelength separation threshold. The extraction module is used to extract local features from multiple sub-channel data streams corresponding to all types of data using a convolutional neural network, so as to obtain local features related to the detection signal. The generation module is used to match and verify the local features with the preset pseudo-random coding sequence of the laser seeker to determine the degree of consistency in terms of temporal pattern and phase features, and generate a spatiotemporal feature set containing the matching results. The identification module is used to determine, based on the spatiotemporal feature set, the identification result indicating whether the signal detected by the laser seeker contains a target pseudo-random code. The step of performing phase reversal adjustment on the laser echo data based on the sequence bias evaluation coefficient includes: Based on the sequence bias evaluation coefficient, a set of phase adjustment parameters corresponding to the phase distribution characteristics of the laser echo data is generated; Based on the set of phase adjustment parameters, phase reversal adjustment is performed on each data point in the laser echo data to obtain the adjusted laser echo data; wherein the phase reversal adjustment is a reverse adjustment.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the pseudo-random coding recognition method for laser seekers based on sequence deviation evaluation coefficients as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The system contains a computer program that, when executed by a computer, implements a pseudo-random coding identification method for laser seekers based on sequence deviation evaluation coefficients as described in any one of claims 1 to 7.

Images