Radar operating state recognition method, device, equipment, medium and program product

Abstract

The present disclosure provides a method, apparatus, device, medium and program product for identifying the operating state of a radar, which can be applied to the field of radar technology. The method for identifying the operating state of a radar includes: extracting features from the timing information of the arrival times of a pulse sequence to obtain feature information of the pulse sequence, where the feature information includes pulse timing sequence modulation features and sequence metric features; based on the feature information, using an associated embedding method to obtain associated features of the pulse sequence, where the associated features represent features obtained by associating the pulse timing sequence modulation features, sequence metric features, pulse width features and radio frequency features of multiple pulses in the pulse sequence with each other; performing semantic recognition on the associated features to determine the operating state of the radar.

Description

Technical Field

[0001] The present disclosure relates to the field of radar technology, and in particular, to a method, device, equipment, medium, and program product for identifying the working state of a radar. Background Art

[0002] A radar is an electronic device that uses radio waves to measure the distance and direction to a specific target, as well as the size and moving speed of the specific target. With the development of science and technology, most radars are used to search for or track targets when attacking missiles or guiding missiles. Therefore, a radar is an important and indispensable electronic device in modern warfare.

[0003] In the process of implementing the concept of the present disclosure, the inventors found that in the related art, when there are incomplete observation conditions such as pulse loss, false pulses, and measurement errors in the obtained radar pulse sequence, resulting in incomplete data, the accuracy of radar working state identification is relatively low. Summary of the Invention

[0004] In view of the above problems, the present disclosure provides a method, device, equipment, medium, and program product for identifying the working state of a radar.

[0005] According to a first aspect of the present disclosure, there is provided a method for identifying the working state of a radar, including: extracting features from the timing information of the arrival times of a pulse sequence to obtain feature information of the pulse sequence, where the feature information includes pulse timing sequence modulation features and sequence metric features;

[0006] Based on the above feature information, using an associated embedding method to obtain associated features of the pulse sequence, where the associated features represent features obtained by associating the pulse timing sequence modulation features, sequence metric features, pulse width features, and radio frequency features of multiple pulses in the pulse sequence with each other;

[0007] Performing semantic recognition on the above associated features to determine the working state of the radar.

[0008] According to an embodiment of the present disclosure, the timing information of the arrival times includes N arrival times, and the extracting features from the timing information of the arrival times of the pulse sequence to obtain feature information of the pulse sequence includes:

[0009] Based on the least squares method, fitting N points corresponding to the N arrival times in the timing information of the arrival times to obtain a fitting curve corresponding to the timing information of the arrival times;

[0010] For the nth arrival time, determining the minimum distance between the point corresponding to the nth arrival time and the fitting curve as the target distance, where N is a positive integer greater than 0, and n is a positive integer less than or equal to N;

[0011] Perform baseline removal on the N target distances corresponding to N arrival times to obtain the characteristic information of the above pulse sequence.

[0012] According to an embodiment of the present disclosure, based on the above characteristic information, an association embedding method is used to obtain the association features of the above pulse sequence, including:

[0013] Perform normalization processing on the pulse width feature of the above pulse sequence to obtain a normalized pulse width feature;

[0014] Perform normalization processing on the radio frequency feature of the above pulse sequence to obtain a normalized radio frequency feature;

[0015] Based on the above characteristic information, the above normalized pulse width feature, and the above normalized radio frequency feature, an association embedding method is used to obtain the association features of the above pulse sequence.

[0016] According to an embodiment of the present disclosure, based on the above characteristic information, the above normalized pulse width feature, and the above normalized radio frequency feature, an association embedding method is used to obtain the association features of the above pulse sequence, including:

[0017] Perform dilated convolution processing on the above characteristic information with the parameter information being the first preset value to obtain a first association feature;

[0018] Perform dilated convolution processing on the above normalized pulse width feature with the parameter information being the second preset value to obtain a second association feature;

[0019] Perform dilated convolution processing on the above normalized radio frequency feature with the parameter information being the third preset value to obtain a third association feature;

[0020] Concatenate the above first association feature, the above second association feature, and the above third association feature to obtain the association features of the above pulse sequence.

[0021] According to an embodiment of the present disclosure, the above semantic recognition is performed on the above association features to determine the above radar working state, including:

[0022] Perform pre-convolution processing on the above association features to obtain target association features;

[0023] Input the above target association features into a two-stage residual shrinkage network to obtain the semantic features of the above pulse sequence;

[0024] Input the above semantic features into a bidirectional gated recurrent unit to determine the above radar working state.

[0025] According to an embodiment of the present disclosure, the above inputting the above target association features into a two-stage residual shrinkage network to obtain the semantic features of the above pulse sequence, including:

[0026] Input the above target correlation features into the residual block structure to obtain second target correlation features;

[0027] Process the above second target correlation features to obtain semantic features of the above pulse sequence.

[0028] According to an embodiment of the present disclosure, the above inputting the above semantic features into a bidirectional gated recurrent unit to determine the above radar operating state includes:

[0029] Input the above semantic features into a bidirectional gated recurrent unit to obtain a plurality of feature vectors;

[0030] Generate an operating state label for each pulse in the above pulse sequence according to a target feature vector, where the above target feature vector is the last feature vector among the above plurality of feature vectors;

[0031] According to the operating state label of each pulse above, obtain the probability of the operating state corresponding to each pulse in the above pulse sequence;

[0032] Determine the above radar operating state according to the probability of the operating state corresponding to each pulse in the above pulse sequence.

[0033] A second aspect of the present disclosure provides a radar operating state recognition device, including: a first obtaining module, a second obtaining module, and a determining module.

[0034] The first obtaining module is configured to extract features from the arrival time sequence information of the pulse sequence to obtain the feature information of the above pulse sequence, where the above feature information includes pulse time sequence modulation features and sequence metric features;

[0035] The second obtaining module is configured to, based on the above feature information, adopt an associated embedding method to obtain the associated features of the above pulse sequence, where the above associated features represent features obtained by associating the pulse time sequence modulation features of a plurality of pulses in the above pulse sequence, the above sequence metric features, pulse width features, and radio frequency features;

[0036] The determining module is configured to perform semantic recognition on the above associated features to determine the operating state of the above radar.

[0037] A third aspect of the present disclosure provides an electronic device, including: one or more processors; a memory for storing one or more computer programs, where the above one or more processors execute the above one or more computer programs to implement the steps of the above method.

[0038] The fourth aspect of the present disclosure further provides a computer-readable storage medium, on which a computer program or instructions are stored, and when the computer program or instructions are executed by a processor, the steps of the above method are implemented.

[0039] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, and when the computer program or instructions are executed by a processor, the steps of the above method are implemented.

[0040] According to the radar working state recognition method, device, electronic device, medium and program product provided by the present disclosure, by extracting features from the arrival time sequence information of the pulse sequence, feature information including pulse time sequence modulation features and sequence metric features can be obtained. The associated embedding method is used to perform associated embedding on the feature information to obtain the associated features of the pulse sequence. By performing semantic recognition on the associated features, the working state of the radar can be determined. The associated embedding of the feature information can improve the representation ability of the feature information of the pulse sequence, and further improve the representation ability of the radar working state corresponding to the pulse sequence. And the semantic recognition of the associated features improves the accuracy of radar working state recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Through the following description of the embodiments of the present disclosure with reference to the drawings, the above content and other objects, features and advantages of the present disclosure will become clearer. In the drawings:

[0042] Figure 1 Schematically shows an application scenario diagram of the radar working state recognition method according to an embodiment of the present disclosure;

[0043] Figure 2 Schematically shows a flowchart of the radar working state recognition method according to an embodiment of the present disclosure;

[0044] Figure 3 Schematically shows a flowchart of the radar working state recognition method according to another embodiment of the present disclosure;

[0045] Figure 4 Schematically shows a schematic diagram of a hybrid neural network model according to an embodiment of the present disclosure;

[0046] Figure 5 Schematically shows a flowchart of the radar working state recognition method according to another embodiment of the present disclosure;

[0047] Figure 6 Schematically shows a structural block diagram of the radar working state recognition device according to an embodiment of the present disclosure; and

[0048] Figure 7 Schematically shows a block diagram of an electronic device suitable for implementing the radar working state recognition method according to an embodiment of the present disclosure. Detailed implementation manners

[0049] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the following detailed description, for the sake of explanation, numerous specific details are set forth in order to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments can also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present disclosure.

[0050] The terms used herein are merely for describing specific embodiments and are not intended to limit the present disclosure. The terms "including", "comprising" and the like used herein indicate the presence of the described features, steps, operations and / or components, but do not exclude the presence or addition of one or more other features, steps, operations or components.

[0051] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0052] In the case of using expressions such as "at least one of A, B, and C, etc.", generally, it should be interpreted according to the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but not be limited to a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0053] In the technical solution of the present disclosure, the user information involved (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties. Moreover, the processing of relevant data, such as collection, storage, use, processing, transmission, provision, disclosure, and application, complies with relevant laws, regulations, and standards, takes necessary confidentiality measures, does not violate public order and good customs, and provides corresponding operation entrances for users to choose to authorize or refuse.

[0054] In the process of implementing the present disclosure, it is found that in the related art, the working state of a radar is mainly identified by using traditional recognition methods that rely on feature parameters and template matching. However, affected by detection means and detection systems, when there are factors such as pulse loss, false pulses, and measurement noise interference in the obtained radar pulse sequence, the data has the characteristics of incompleteness and mutilation, which will lead to a low accuracy in identifying the working state of the radar.

[0055] In the related art, the recognition method that relies on template matching extracts the feature parameters of the entire pulse sequence and matches and recognizes them with the accumulated radar feature parameter templates, so as to realize the recognition of the radar working state. Machine learning methods similar to template matching can include working state recognition methods such as support vector machines and decision trees. Some features can be manually extracted and appropriate classifiers can be designed to realize the recognition of the radar working state. However, for a multi-functional radar with flexible beam pointing and complex and variable modulation methods, the accumulated fixed feature parameter templates usually cannot completely represent all the parameter modes of the radar, and at the same time, they cannot adapt to the situations such as pulse loss, false pulses, and noise interference caused by the complex and variable electromagnetic environment.

[0056] The related art also includes a radar working state recognition method based on deep learning, including a model from pulse sequence to label (Seq2one) and a model from pulse sequence to label sequence (Seq2seq). The model from pulse sequence to label globally recognizes the entire pulse sequence, extracts the features of the entire pulse sequence, and recognizes it as a label category. The model from pulse sequence to label sequence predicts the working state of each pulse to obtain a serialized label type. However, for a multi-functional radar with flexible beam pointing and flexible and variable parameters, the recognition accuracy of the working state of the model from pulse sequence to label and the model from pulse sequence to label sequence is relatively low. In addition, in the related art, the first-order difference features that can only represent the domain relationship, that is, the pulse repetition interval (PRI), are usually used as the main feature parameter for recognition. In the case of serious false pulses and missing pulses, the working state of the radar cannot be effectively recognized.

[0057] In view of this, the embodiments of the present disclosure provide a method for recognizing the working state of a radar, including: extracting the feature of the arrival time sequence information of the pulse sequence to obtain the feature information of the pulse sequence, where the feature information includes the pulse time sequence modulation feature and the sequence metric feature; based on the feature information, using the associated embedding method to obtain the associated feature of the pulse sequence, where the associated feature represents the feature obtained by associating the pulse time sequence modulation feature, the sequence metric feature, the pulse width feature, and the radio frequency feature; performing semantic recognition on the associated feature to determine the working state of the radar.

[0058] Figure 1 Schematically shows an application scenario diagram of the radar working state recognition method according to an embodiment of the present disclosure.

[0059] As Figure 1 shown, the application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0060] Users can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0061] The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, and desktop computers, etc.

[0062] The server 105 may be a server providing various services, such as a background management server (only as an example) that supports websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and process data such as received user requests, etc., and feedback the processing results (such as web pages, information, or data obtained or generated according to user requests, etc.) to the terminal devices.

[0063] It should be noted that the radar working state recognition method provided by the embodiments of the present disclosure can generally be executed by the server 105. Correspondingly, the radar working state recognition device provided by the embodiments of the present disclosure can generally be set in the server 105. The radar working state recognition method provided by the embodiments of the present disclosure can also be executed by a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105. Correspondingly, the radar working state recognition device provided by the embodiments of the present disclosure can also be set in a server or a server cluster different from the server 105 and capable of communicating with the terminal devices 101, 102, 103 and / or the server 105.

[0064] It should be understood that Figure 1 the number of terminal devices, networks, and servers in [[ ]] is merely illustrative. According to implementation requirements, there can be any number of terminal devices, networks, and servers.

[0065] Based on the scenario described below Figure 1 through [[ ]], Figures 2 to 6 the radar working state recognition method of the disclosed embodiments will be described in detail.

[0066] Figure 2 Schematically shows a flowchart of the radar working state recognition method according to an embodiment of the present disclosure.

[0067] As Figure 2 shown, the radar working state recognition method 200 of this embodiment includes operations S210 to S230.

[0068] In operation S210, the feature extraction of the arrival time sequence information of the pulse sequence is performed to obtain the feature information of the pulse sequence.

[0069] In operation S220, based on the feature information, the associated embedding method is used to obtain the associated features of the pulse sequence.

[0070] In operation S230, semantic recognition is performed on the associated features to determine the working state of the radar.

[0071] According to an embodiment of the present disclosure, the radar may include a pulse radar, and the pulse radar can continuously emit pulses, so the pulse radar emits a pulse sequence. The radar may also include a multi-functional radar, and the multi-functional radar can represent a relatively important electromagnetic radiation source target. The multi-functional radar has the characteristics of agility and variability and can perform multiple tasks simultaneously. The receiver can receive the pulse sequence emitted by the radar under complete conditions or receive the pulse sequence under incomplete conditions. For example, the incomplete conditions may be situations such as measurement errors of the receiver and false pulses included in the pulse sequence.

[0072] According to an embodiment of the present disclosure, the arrival time sequence information can represent the time of arrival (TOA) of the pulse sequence. The arrival time of the pulse sequence can represent the moment when the receiver receives the pulse sequence emitted by the radar.

[0073] According to an embodiment of the present disclosure, the feature information may include pulse time sequence modulation features and sequence metric features. The pulse time sequence modulation features can represent the features of modulating the pulse sequence. For example, the modulation method of the pulse sequence can be a fixed modulation method or a jittered modulation method and other modulation methods. The sequence metric features are obtained by performing feature extraction on the pulse sequence and can be data features included in the pulse sequence.

[0074] According to an embodiment of the present disclosure, feature extraction can be performed on the arrival time sequence information of a pulse sequence to obtain the feature information of the pulse sequence, and the feature information can be a distance vector. The process of performing feature extraction on the arrival time sequence information of a pulse sequence to obtain the feature information of the pulse sequence can also be referred to as metric coding of the arrival time sequence information.

[0075] According to an embodiment of the present disclosure, the associated feature can represent a feature obtained by associating the pulse timing sequence modulation feature, the sequence metric feature, the pulse width feature, and the radio frequency feature of multiple pulses in the pulse sequence with each other. For example, the associated feature can be a feature obtained by associating the pulse timing sequence modulation feature, the sequence metric feature, the pulse width feature, and the radio frequency feature of the 5 pulses before and after the 10th pulse in the pulse sequence to the 10th pulse. The association of the pulse timing sequence modulation feature, the sequence metric feature, the pulse width feature, and the radio frequency feature can be an association between timings. Based on the feature information, by using the associated embedding method, the associated feature of the pulse sequence can be obtained.

[0076] According to an embodiment of the present disclosure, semantic recognition can represent the process of understanding according to the meaning and context of the language. The working state of the radar can represent the working mode in which the radar is in, and the working state of the radar can include a search state, a tracking state, a scanning state, etc. By performing semantic recognition on the associated feature, a speech recognition result can be obtained, and according to the semantic recognition result, the working state of the radar can be determined.

[0077] According to an embodiment of the present disclosure, by performing feature extraction on the arrival time sequence information of the pulse sequence, feature information including the pulse timing sequence modulation feature and the sequence metric feature can be obtained. By using the associated embedding method to perform associated embedding on the feature information, the associated feature of the pulse sequence can be obtained. By performing semantic recognition on the associated feature, the working state of the radar can be determined. The associated embedding of the feature information can improve the representation ability of the feature information of the pulse sequence, and further improve the representation ability of the radar working state corresponding to the pulse sequence. And the semantic recognition of the associated feature improves the accuracy of the radar working state recognition.

[0078] According to an embodiment of the present disclosure, performing feature extraction on the arrival time sequence information of the pulse sequence to obtain the feature information of the pulse sequence includes: based on the least squares method, fitting N points corresponding to N arrival times in the arrival time sequence information to obtain a fitting curve corresponding to the arrival time sequence information; for the nth arrival time, determining the minimum distance between the point corresponding to the nth arrival time and the fitting curve as the target distance, where N is a positive integer greater than 0, and n is a positive integer less than or equal to N; performing baseline removal processing on the N target distances corresponding to the N arrival times to obtain the feature information of the pulse sequence.

[0079] According to an embodiment of the present disclosure, the arrival time sequence information may include N arrival times. The values of the N arrival times can be placed in a rectangular coordinate system. The abscissa of the rectangular coordinate system can represent the sequence number of the arrival time, and the ordinate of the rectangular coordinate system can represent the value of the arrival time. In the rectangular coordinate system, each arrival time can correspond to a point. For example, the point corresponding to the i-th arrival time is (i, TOA i ).

[0080] According to an embodiment of the present disclosure, based on the least squares method, N points corresponding to the N arrival times in the arrival time sequence information can be fitted to obtain a fitting curve corresponding to the arrival time sequence information.

[0081] According to an embodiment of the present disclosure, the fitting curve can be Equation (1):

[0082] y = k * x + b (1)

[0083] where k represents the slope of the fitting curve, b represents the intercept of the fitting curve, x represents the independent variable of the fitting curve, and y represents the dependent variable of the fitting curve.

[0084] According to an embodiment of the present disclosure, the vector synthesized by the intercept and slope of the fitting curve can be transposed to obtain a first vector. The first vector C can be denoted as Equation (2):

[0085] C = [k, b] T (2)

[0086] According to an embodiment of the present disclosure, the formula of the least squares method can be used to solve the first vector C, as follows Equation (3):

[0087] C = (A T A) -1 A T D (3)

[0088] where A represents a second vector, D represents a third vector, D = [TOA 1 TOA 2 TOA 3 ... TOA N T .

[0089] According to an embodiment of the present disclosure, based on the least squares method, the first vector C can be obtained, so that the slope and intercept of the fitting curve can be obtained, and thus the fitting curve corresponding to the arrival time sequence information can be obtained.

[0090] ​According to an embodiment of the present disclosure, in the case of obtaining a fitting curve, for the nth arrival time, the minimum distance between the point corresponding to the nth arrival time and the fitting curve can be calculated, and the minimum distance between the point corresponding to the nth arrival time and the fitting curve can be determined as the target distance, the target distance distance n as shown in the following formula (4):

[0091]

[0092] where abs() represents taking the absolute value.

[0093] According to an embodiment of the present disclosure, there can be multiple target distances, and the number of target distances can correspond to the number of arrival times. In the case where the arrival time sequence information includes N arrival times, the target distance can include N.

[0094] According to an embodiment of the present disclosure, baseline removal processing can fix the target distance at the same scale. By performing baseline removal processing on the N target distances corresponding to the N arrival times, the characteristic information of the pulse sequence can be obtained. The characteristic information Distance is as shown in the following formula (5):

[0095]

[0096] where Distance = {distance 1 , distance 2 , ……, distance N}

[0097] According to an embodiment of the present disclosure, formula (5) can represent that the difference between the characteristic information and the mean value of the N target distances is assigned to the characteristic information.

[0098] According to an embodiment of the present disclosure, based on the least squares method, a fitting curve corresponding to the arrival time sequence information can be obtained. The minimum distances between the N points corresponding to the N arrival times in the arrival time sequence information and the fitting curve can be determined as the target distances, and baseline removal processing is performed on the target distances to obtain the characteristic information of the pulse sequence. Through baseline removal processing, the target distances can be fixed at the same scale, thereby improving the regularity of the arrival time values in the time sequence.

[0099] According to an embodiment of the present disclosure, based on the characteristic information, using the associated embedding method, the associated characteristics of the pulse sequence are obtained, including: normalizing the pulse width characteristic of the pulse sequence to obtain the normalized pulse width characteristic. Normalizing the radio frequency characteristic of the pulse sequence to obtain the normalized radio frequency characteristic. Based on the characteristic information, the normalized pulse width characteristic, and the normalized radio frequency characteristic, using the associated embedding method, the associated characteristics of the pulse sequence are obtained.

[0100] According to an embodiment of the present disclosure, the normalization process may represent a process of converting dimensional data into dimensionless data through transformation. By performing normalization processing on the pulse width (PW) feature of the pulse sequence, a normalized pulse width feature can be obtained. By performing normalization processing on the radio frequency (RF) feature of the pulse sequence, a normalized RF feature can be obtained.

[0101] According to an embodiment of the present disclosure, the associated embedding may represent a process of embedding the feature information of adjacent pulses of the pulse to be feature-embedded in the pulse sequence, the normalized pulse width feature, and the normalized RF feature into the pulse to be feature-embedded according to weights. Among them, the weights may be parameter information corresponding to performing dilated convolution processing on the feature information, the normalized pulse width feature, and the normalized RF feature respectively. For example, a target pulse to be feature-embedded can be determined from the pulse sequence. For the target pulse in the pulse sequence, E pulses before the target pulse and M pulses after the target pulse can be determined, where both E and M are positive integers greater than 1. E and M may be equal or unequal. Based on this, the normalized pulse width feature of the target pulse, the normalized pulse width features of the E pulses respectively, and the normalized pulse width features of the M pulses respectively can be processed according to weights to obtain the processed embedded normalized pulse width feature. The processed embedded normalized pulse width feature can be embedded into the target pulse to complete the associated embedding operation for the target pulse. Thus, the association between the normalized pulse width features of multiple pulses in the pulse sequence can be achieved. The feature information and the normalized RF features between multiple pulses can also be associated by a similar method as described above, which is not elaborated herein. For example, the feature information, the normalized pulse width feature, and the normalized RF feature of 5 pulses before and after the 20th pulse can be embedded into the 20th pulse according to different weights. Based on the feature information, the normalized pulse width feature, and the normalized RF feature, by using the associated embedding method, the associated feature of the pulse sequence can be obtained.

[0102] According to an embodiment of the present disclosure, by using the associated embedding method, the associated feature of the pulse sequence can be obtained, improving the association relationship between the feature information of the pulse sequence, thereby improving the representation ability of the feature information of the pulse sequence.

[0103] According to an embodiment of the present disclosure, based on the feature information, the normalized pulse width feature, and the normalized radio frequency feature, an associated embedding method is adopted to obtain the associated feature of the pulse sequence, including: performing dilated convolution processing on the feature information with the parameter information being the first preset value to obtain the first associated feature; performing dilated convolution processing on the normalized pulse width feature with the parameter information being the second preset value to obtain the second associated feature; performing dilated convolution processing on the normalized radio frequency feature with the parameter information being the third preset value to obtain the third associated feature; and splicing the first associated feature, the second associated feature, and the third associated feature to obtain the associated feature of the pulse sequence.

[0104] According to an embodiment of the present disclosure, the first associated feature can represent the feature obtained by associatively embedding the feature information of adjacent pulses according to the parameter information. The second associated feature can represent the feature obtained by associatively embedding the normalized pulse width feature of adjacent pulses according to the parameter information. The third associated feature can represent the feature obtained by associatively embedding the normalized radio frequency feature of adjacent pulses according to the parameter information. Since the first preset value, the second preset value, and the third preset value can be the same or different, the number of adjacent features in the first associated feature, the number of adjacent features in the second associated feature, and the number of adjacent features in the third associated feature can be the same or different.

[0105] According to an embodiment of the present disclosure, the parameter information can represent the convolution kernel of the dilated convolution processing. The first preset value can be set according to requirements. The second preset value can be set according to requirements. The third preset value can be set according to requirements. The first preset value, the second preset value, and the third preset value can be the same or different. Dilated convolution processing can be performed on the feature information with the parameter information being the first preset value to obtain the first associated feature. Dilated convolution processing can be performed on the normalized pulse width feature with the parameter information being the second preset value. Dilated convolution processing can be performed on the normalized radio frequency feature with the parameter information being the third preset value to obtain the third associated feature. For example, when the first preset value, the second preset value, and the third preset value are the same, dilated convolution processing is respectively performed on the feature information, the normalized pulse width feature, and the normalized radio frequency feature, and the first associated feature corresponding to the feature information, the first associated feature corresponding to the normalized pulse width feature, and the third associated feature corresponding to the normalized radio frequency feature can be obtained.

[0106] According to an embodiment of the present disclosure, by splicing the first associated feature, the second associated feature, and the third associated feature, the associated feature of the pulse sequence can be obtained. The first associated feature, the second associated feature, and the third associated feature can be spliced according to the channel dimension. For example, when the size of the first associated feature is (1, 128), the size of the second associated feature is (1, 128), and the size of the third associated feature is (1, 128), the size of the spliced associated feature can be (3, 128).

[0107] According to an embodiment of the present disclosure, based on the associated embedding method, the associated feature Data of the pulse sequence is obtained as shown in the following formula (6):

[0108] Data = concat(FE(TOA pre ), FE(PW pre ), FE(RF pre )) (6)

[0109] Wherein, FE() represents associated embedding, TOA pr e represents the first associated feature, PW pre represents the second associated feature, RF pre represents the third associated feature, and concat() represents splicing according to the channel dimension.

[0110] According to an embodiment of the present disclosure, by respectively performing dilated convolution processing on the feature information, the normalized pulse width, and the normalized radio frequency, the first associated feature, the second associated feature, and the third associated feature can be obtained, and the first associated feature, the second associated feature, and the third associated feature are spliced. Through the dilated convolution processing, the information range is increased, and at the same time, the size of the output feature map can be ensured to remain unchanged, thereby improving the representation ability of the feature information of the pulse sequence.

[0111] Figure 3 Schematically shows a flowchart of a radar working state recognition method according to another embodiment of the present disclosure.

[0112] As Figure 3 shown, the radar working state recognition method 300 of this embodiment includes operations S310 to S370.

[0113] In operation S310, feature extraction is performed on the timing information of the arrival time of the pulse sequence to obtain the feature information of the pulse sequence.

[0114] In operation S320, normalization processing is performed on the pulse width feature of the pulse sequence to obtain the normalized pulse width feature.

[0115] In operation S330, normalization processing is performed on the radio frequency feature of the pulse sequence to obtain the normalized radio frequency feature.

[0116] In operation S340, perform dilated convolution processing on the feature information with the parameter information being the first preset value to obtain the first associated feature.

[0117] In operation S350, perform dilated convolution processing on the normalized pulse width feature with the parameter information being the second preset value to obtain the second associated feature.

[0118] In operation S360, perform dilated convolution processing on the normalized radio frequency feature with the parameter information being the third preset value to obtain the third associated feature.

[0119] In operation S370, splice the first associated feature, the second associated feature, and the third associated feature to obtain the associated feature of the pulse sequence.

[0120] Operations S310 to S370 can refer to the descriptions of other embodiments of the present disclosure and will not be elaborated herein.

[0121] According to an embodiment of the present disclosure, perform semantic recognition on the associated feature to determine the radar working state, including: perform pre-convolution processing on the associated feature to obtain the target associated feature. Input the target associated feature into a two-stage residual shrinkage network to obtain the semantic feature of the pulse sequence. Input the semantic feature into a bidirectional gated recurrent unit to determine the radar working state.

[0122] According to an embodiment of the present disclosure, performing pre-convolution processing on the associated feature can represent adding a convolutional layer to the associated feature, thereby increasing the dimension of the associated feature. For example, the pre-convolution of the associated feature can be adding a one-dimensional convolutional layer to the associated feature.

[0123] According to an embodiment of the present disclosure, the Residual Shrinkage Network (RSNet) is an improved convolutional neural network. By introducing an identity path, it can reduce the difficulty of training a hybrid neural network model.

[0124] According to an embodiment of the present disclosure, the two-stage residual shrinkage network is a cascade of two residual shrinkage networks, and the output of the first residual shrinkage network is the input of the second residual shrinkage network. Performing pre-convolution on the associated feature can obtain the target associated feature. Inputting the target associated feature into the two-stage residual shrinkage network can obtain the semantic feature of the pulse sequence. Each residual shrinkage network in the two-stage residual shrinkage network includes a residual block structure, a batch normalization module, and a spatial attention module.

[0125] According to an embodiment of the present disclosure, a bidirectional gated recurrent unit (GRU) can represent the introduction of a gate control mechanism in a recurrent neural network, thereby achieving multi-dimensional input and single output. It can also iteratively input pulse description words (PDWs) of signal pulses separately from the forward and backward directions. The pulse description words can include features such as the radio frequency of the pulse, the pulse width, and the arrival time. By inputting semantic features into the bidirectional gated recurrent unit, the probability that each pulse belongs to each working state can be obtained. Based on the probability that each pulse belongs to each working state, the working state of the radar can be determined.

[0126] According to an embodiment of the present disclosure, by inputting target-associated features into a two-level residual shrinkage network to extract semantic features of the pulse sequence, the problems of gradient disappearance and gradient explosion can be improved. Moreover, in the spatial attention module, the attention weights of different regions can be adaptively learned, which can promote the processing of the pulse sequence. By inputting the semantic features into the bidirectional gated recurrent unit, based on the probability that each pulse belongs to each working state, the working state of the radar can be determined, improving the accuracy of radar working state recognition.

[0127] According to an embodiment of the present disclosure, inputting target-associated features into a two-level residual shrinkage network to obtain semantic features of the pulse sequence includes: inputting target-associated features into a residual block structure to obtain second target-associated features. Processing the second target-associated features to obtain semantic features of the pulse sequence.

[0128] According to an embodiment of the present disclosure, inputting target-associated features into the residual block structure of the first residual shrinkage network can obtain target-associated features processed by the residual block structure of the first residual shrinkage network. The target-associated features processed by the residual block structure of the first residual shrinkage network can be skip-connected with the target-associated features, thereby obtaining third target-associated features.

[0129] According to an embodiment of the present disclosure, the third target-associated features can be input into the batch normalization module of the first residual shrinkage network for batch normalization processing to obtain first normalized associated features. Inputting the first normalized associated features into the spatial attention module of the second residual shrinkage network for processing to obtain the first semantic features of the pulse sequence.

[0130] According to an embodiment of the present disclosure, the first semantic features of the pulse sequence output by the first residual shrinkage network are input into the residual block structure of the second residual shrinkage network for processing to obtain target-associated features processed by the residual block structure of the second residual shrinkage network. The target-associated features processed by the residual block structure of the second residual shrinkage network can be skip-connected with the first semantic features of the pulse sequence, thereby obtaining fourth target-associated features.

[0131] According to an embodiment of the present disclosure, the fourth target associated feature is input into the batch normalization module of the second residual shrinkage network for batch normalization processing to obtain the second normalized associated feature. The second normalized associated feature is input into the spatial attention module of the second residual shrinkage network for processing to obtain the second semantic feature of the pulse sequence.

[0132] According to an embodiment of the present disclosure, the semantic feature includes any one of the first semantic feature and the second semantic feature. The spatial attention module of the first residual shrinkage network can output the first semantic feature, and the spatial attention module of the second residual shrinkage network can output the second semantic feature. The second target associated feature includes any one of the third target associated feature and the fourth target associated feature. The residual block structure of the first residual shrinkage network can output the third target associated feature, and the residual block structure of the second residual shrinkage network can output the fourth target associated feature.

[0133] According to an embodiment of the present disclosure, by inputting the target associated feature into the residual block structure of the residual shrinkage network, the skip connection of the residual block structure can improve the problems of gradient disappearance and gradient explosion, and a spatial attention module is introduced into the residual shrinkage network, thereby improving the processing efficiency of the pulse sequence.

[0134] According to an embodiment of the present disclosure, inputting the semantic feature into the bidirectional gated recurrent unit to determine the radar working state includes: inputting the semantic feature into the bidirectional gated recurrent unit to obtain a plurality of feature vectors. According to the target feature vector, generate the working state label of each pulse in the pulse sequence. According to the working state label of each pulse, obtain the probability of the working state corresponding to each pulse in the pulse sequence. According to the probability of the working state corresponding to each pulse of the pulse sequence, determine the radar working state.

[0135] According to an embodiment of the present disclosure, inputting the semantic feature into the bidirectional gated recurrent unit can obtain a plurality of feature vectors, and the target feature vector can be the last feature vector among the plurality of feature vectors.

[0136] According to an embodiment of the present disclosure, the target feature vector is output to the pulse-level label generation layer to generate the working state label of each pulse, and the pulse-level label generation layer can be composed of a pulse-level fully connected layer and a softmax function.

[0137] According to an embodiment of the present disclosure, the softmax function is as follows in formula (7):

[0138] O t = W * faltten LB (y n ) + Q

[0139]

[0140] Among them, faltten represents the flattening function, faltten LB (y n ) ∈ R L×B×H , y n = ∈ R L×B·H , L represents the length of the pulse sequence, B represents the number of segments included in the pulse sequence, H represents the number of hidden layers, output represents the output of the softmax function, O t represents the output of the fully connected layer, O i represents the output of the i-th node of the fully connected layer, O j represents the output of the j-th node of the fully connected layer, W represents the first weight parameter of the fully connected layer, Q represents the second weight parameter of the fully connected layer, and K represents the number of nodes in the fully connected layer.

[0141] According to an embodiment of the present disclosure, according to the target feature vector, the working state label of each pulse in the pulse sequence can be generated at the pulse level label, and the working state labels of each pulse in the pulse sequence can form a pulse sequence label. According to the working state label of each pulse in the pulse sequence, the probability of the working state corresponding to each pulse in the pulse sequence can be obtained. According to the probability of the working state corresponding to each pulse of the pulse sequence, the radar working state can be determined. For example, if the probabilities of the working state corresponding to the 10th pulse are 0.5, 0.2, 0.2, and 0.1 respectively, then the working state corresponding to the probability of 0.5 can be determined as the working state corresponding to the 10th pulse.

[0142] According to an embodiment of the present disclosure, by inputting the semantic feature into the bidirectional gated recurrent unit, multiple feature vectors can be obtained, and the last feature vector can be the target vector. The target vector is output to the pulse level label generation layer to generate the working state label of each pulse in the pulse sequence, so that the probability of the working state corresponding to each pulse in the pulse sequence can be obtained, and then the radar working state can be determined, improving the accuracy of radar working state recognition.

[0143] According to an embodiment of the present disclosure, the hybrid neural network model can include semantic feature extraction and working state label recognition. Semantic feature extraction can be obtained by pre-convolving the associated features to obtain the target associated features and inputting the target associated features into the two-stage residual shrinkage network. The recognition of the working state label can be obtained by inputting the semantic feature into the bidirectional gated recurrent unit to obtain the target feature vector and according to the target feature vector. According to the working state label, the probability of the working state corresponding to each pulse in the pulse sequence can be obtained. According to the probability of the working state corresponding to each pulse of the pulse sequence, the radar working state can be determined.

[0144] According to an embodiment of the present disclosure, the pulse sequence may include a plurality of pulse sequences, and each pulse sequence may correspond to a working state sequence. In the case where an error exists in the working state recognition in the working state sequence, a penalty may be imposed according to a loss function. The loss function E on N working state sequence training samples is as follows in formula (8):

[0145]

[0146] Wherein, represents the loss of the i-th working state sequence, L i represents the number of pulses in the i-th working state sequence, y j represents the j-th label in the pulse sequence label, represents the index of the output probability value corresponding to the t-th pulse in the working state sequence.

[0147] Figure 4 Schematically shows a schematic diagram of a hybrid neural network model according to an embodiment of the present disclosure.

[0148] As Figure 4 shown, the hybrid neural network model 400 of this embodiment includes a pre-convolution module 410, a two-stage residual shrinkage network 420, a bidirectional gated recurrent unit 430, and a pulse-level label generation layer 440.

[0149] According to an embodiment of the present disclosure, the associated features may be input into the pre-convolution module 410 for pre-convolution processing to obtain target associated features. The two-stage residual shrinkage network includes a first residual shrinkage network and a second residual shrinkage network. Each residual shrinkage network includes a residual block structure 421, a batch normalization module 422, and a spatial attention module 423. The bidirectional gated recurrent unit 430 may output a plurality of feature vectors. In the case where the number of feature vectors is N, the N feature vectors may be denoted as y 1 , y 2 , ……, y N , and the target feature vector may be input into the pulse-level label generation layer 440, where the target feature vector is the last feature vector y N among the N feature vectors. Wherein, the symbol between the residual block structure 421 and the batch normalization module 422 indicates that the target associated features output by the pre-convolution module 410 may be skip-connected with the target associated features processed by the residual block structure 421. The features obtained by skip-connecting the target associated features output by the pre-convolution module 410 and the target associated features processed by the residual block structure 421 may be input into the batch normalization module 422.

[0150] Figure 5A flowchart of a radar operating state recognition method according to another embodiment of the present disclosure is schematically shown.

[0151] As Figure 5 shown, the radar operating state recognition method 500 of this embodiment includes operations S510 to S550.

[0152] In operation S510, feature extraction of the arrival time.

[0153] In operation S520, normalization processing of the pulse width.

[0154] In operation S530, normalization processing of the radio frequency.

[0155] In operation S540, the first correlation feature 510, the second correlation feature 520, and the third correlation feature 530 are correlated and embedded to obtain a correlation feature.

[0156] In operation S550, the hybrid neural network model processes the correlation feature to obtain the operating state of the radar.

[0157] According to an embodiment of the present disclosure, by extracting the features of the arrival time, the first correlation feature 510 can be obtained. The normalization processing of the pulse width can obtain the second correlation feature 520. The normalization processing of the radio frequency can obtain the third correlation feature 530. The first correlation feature 510, the second correlation feature 520, and the third correlation feature 530 are correlated and embedded to obtain a correlation feature. The correlation feature is input into the hybrid neural network model, and the hybrid neural network model processes the correlation feature to obtain the operating state of the radar.

[0158] Based on the above radar operating state recognition method, the present disclosure also provides a radar operating state recognition device. The following will be combined with Figure 6 to describe this device in detail.

[0159] Figure 6 A structural block diagram of a radar operating state recognition device according to an embodiment of the present disclosure is schematically shown.

[0160] As Figure 6 shown, the radar operating state recognition device 600 of this embodiment includes a first obtaining module 610, a second obtaining module 620, and a determining module 630.

[0161] The first obtaining module 610 is configured to extract the features of the arrival time sequence information of the pulse sequence to obtain the feature information of the pulse sequence, where the feature information includes the pulse time sequence modulation feature and the sequence metric feature. In one embodiment, the first obtaining module 610 may be configured to perform the operation S210 described above, which will not be elaborated here.

[0162] A second obtaining module 620, configured to obtain the correlation features of the pulse sequence by using a correlation embedding method based on the feature information, where the correlation features represent the features obtained by correlating the pulse timing sequence modulation features, sequence metric features, pulse width features, and radio frequency features of multiple pulses in the pulse sequence with each other. In one embodiment, the second obtaining module 620 may be configured to perform the operation S220 described above, which will not be elaborated herein.

[0163] A determining module 630, configured to perform semantic recognition on the correlation features to determine the working state of the radar. In one embodiment, the determining module 630 may be configured to perform the operation S230 described above, which will not be elaborated herein.

[0164] According to an embodiment of the present disclosure, the arrival time sequence information includes N arrival times. The first obtaining module 610 includes:

[0165] A first obtaining sub-module, configured to fit N points corresponding to the N arrival times in the arrival time sequence information based on the least squares method to obtain a fitting curve corresponding to the arrival time sequence information;

[0166] A second obtaining sub-module, configured to, for the nth arrival time, determine the minimum distance between the point corresponding to the nth arrival time and the fitting curve as the target distance, where N is a positive integer greater than 0, and n is a positive integer less than or equal to N;

[0167] A third obtaining sub-module, configured to perform baseline removal processing on the N target distances corresponding to the N arrival times to obtain the feature information of the pulse sequence.

[0168] According to an embodiment of the present disclosure, the second obtaining module 620 includes:

[0169] A fourth obtaining sub-module, configured to perform normalization processing on the pulse width feature of the pulse sequence to obtain a normalized pulse width feature;

[0170] A fifth obtaining sub-module, configured to perform normalization processing on the radio frequency feature of the pulse sequence to obtain a normalized radio frequency feature;

[0171] A sixth obtaining sub-module, configured to obtain the correlation features of the pulse sequence by using a correlation embedding method based on the feature information, the normalized pulse width feature, and the normalized radio frequency feature.

[0172] According to an embodiment of the present disclosure, the sixth obtaining sub-module includes:

[0173] A first obtaining unit, configured to perform dilated convolution processing on the feature information with a parameter information of a first preset value to obtain a first correlation feature;

[0174] A second obtaining unit, configured to perform dilated convolution processing with parameter information being a second preset value on the normalized pulse width feature to obtain a second correlation feature;

[0175] A third obtaining unit, configured to perform dilated convolution processing with parameter information being a third preset value on the normalized radio frequency feature to obtain a third correlation feature;

[0176] A fourth obtaining unit, configured to splice the first correlation feature, the second correlation feature, and the third correlation feature to obtain a correlation feature of the pulse sequence.

[0177] According to an embodiment of the present disclosure, the determining module 630 includes:

[0178] A first determining sub-module, configured to perform pre-convolution processing on the correlation feature to obtain a target correlation feature;

[0179] A second determining sub-module, configured to input the target correlation feature into a two-stage residual shrinkage network to obtain a semantic feature of the pulse sequence;

[0180] A third determining sub-module, configured to input the semantic feature into a bidirectional gated recurrent unit to determine the radar working state.

[0181] According to an embodiment of the present disclosure, the second determining sub-module includes:

[0182] A first determining unit, configured to input the target correlation feature into a residual block structure to obtain a second target correlation feature;

[0183] A second determining unit, configured to process the target correlation feature to obtain a semantic feature of the pulse sequence.

[0184] According to an embodiment of the present disclosure, the third determining sub-module includes:

[0185] A third determining unit, configured to input the semantic feature into a bidirectional gated recurrent unit to obtain a plurality of feature vectors;

[0186] A fourth determining unit, configured to generate a working state label for each pulse in the pulse sequence according to a target feature vector, where the target feature vector is the last feature vector among the plurality of feature vectors;

[0187] A fifth determining unit, configured to obtain a probability of the working state corresponding to each pulse in the pulse sequence according to the working state label of each pulse;

[0188] A sixth determining unit, configured to determine the radar working state according to the probability of the working state corresponding to each pulse in the pulse sequence.

[0189] According to an embodiment of the present disclosure, any of the first obtaining module 610, the second obtaining module 620, and the determining module 630 may be combined and implemented in one module, or any one of them may be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first obtaining module 610, the second obtaining module 620, and the determining module 630 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on substrate, a system on package, an application specific integrated circuit (ASIC), or may be implemented by any other reasonable means such as hardware or firmware for integrating or packaging circuits, or may be implemented in any one of the three implementation manners of software, hardware, and firmware, or in an appropriate combination of any several of them. Alternatively, at least one of the first obtaining module 610, the second obtaining module 620, and the determining module 630 may be at least partially implemented as a computer program module, and when the computer program module is run, corresponding functions may be executed.

[0190] Figure 7 A block diagram of an electronic device suitable for implementing the radar operating state recognition method according to an embodiment of the present disclosure is schematically shown.

[0191] As Figure 7 shown, the electronic device 700 according to an embodiment of the present disclosure includes a processor 701, which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage section 708 into a random access memory (RAM) 703. The processor 701 may include, for example, a general microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), etc. The processor 701 may also include on-board memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0192] In the RAM 703, various programs and data required for the operation of the electronic device 700 are stored. The processor 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. The processor 701 performs various operations of the method flow according to the embodiments of the present disclosure by executing the programs in the ROM 702 and / or the RAM 703. It should be noted that the programs may also be stored in one or more memories other than the ROM 702 and the RAM 703. The processor 701 may also perform various operations of the method flow according to the embodiments of the present disclosure by executing the programs stored in the one or more memories.

[0193] According to an embodiment of the present disclosure, the electronic device 700 may further include an input / output (I / O) interface 705, and the input / output (I / O) interface 705 is also connected to the bus 704. The electronic device 700 may further include one or more of the following components connected to the I / O interface 705: an input portion 706 including a keyboard, a mouse, etc.; an output portion 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage portion 708 including a hard disk, etc.; and a communication portion 709 including a network interface card such as a LAN card, a modem, etc. The communication portion 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 710 as needed so that a computer program read therefrom is installed into the storage portion 708 as needed.

[0194] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist separately without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present disclosure is implemented.

[0195] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program may be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include one or more memories other than the above-described ROM 702 and / or RAM 703 and / or ROM 702 and RAM 703.

[0196] An embodiment of the present disclosure also includes a computer program product, which includes a computer program that contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to enable the computer system to implement the radar working state recognition method provided by the embodiment of the present disclosure.

[0197] When the computer program is executed by the processor 701, it executes the above functions defined in the system / apparatus of the embodiment of the present disclosure. According to an embodiment of the present disclosure, the above-described systems, apparatuses, modules, units, etc. may be implemented by computer program modules.

[0198] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and be downloaded and installed through the communication part 709, and / or be installed from the removable medium 711. The program code included in the computer program may be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0199] In such an embodiment, the computer program may be downloaded and installed from the network through the communication part 709, and / or be installed from the removable medium 711. When the computer program is executed by the processor 701, it executes the above functions defined in the system of the embodiment of the present disclosure. According to an embodiment of the present disclosure, the above-described systems, devices, apparatuses, modules, units, etc. may be implemented by computer program modules.

[0200] According to embodiments of the present disclosure, program code for executing the computer programs provided by the embodiments of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, such as Java, C++, Python, the "C" language, or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (for example, by using an Internet service provider to connect through the Internet).

[0201] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the above-mentioned module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks can occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0202] Those skilled in the art can understand that the features recited in the various embodiments and / or claims of the present disclosure can be combined or / and combined in various ways, even if such combinations or combinations are not explicitly recited in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features recited in the various embodiments and / or claims of the present disclosure can be combined and / or combined in various ways. All such combinations and / or combinations fall within the scope of the present disclosure.

[0203] The embodiments of the present disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments have been described separately above, this does not mean that the measures in each embodiment cannot be used advantageously in combination. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art can make various substitutions and modifications, and these substitutions and modifications should fall within the scope of the present disclosure.

Claims

1. A radar working state identification method, comprising: Extracting the characteristics of the arrival time sequence information of the pulse sequence to obtain characteristic information of the pulse sequence, wherein the characteristic information includes pulse timing sequence modulation characteristics and sequence measurement characteristics; Based on the feature information, a correlation embedding method is used to obtain a correlation feature of the pulse sequence, wherein the correlation feature represents a feature obtained by correlating pulse timing sequence modulation features, sequence metric features, pulse width features and radio frequency features of multiple pulses in the pulse sequence; Performing semantic recognition on the associated features to determine the working state of the radar; The arrival time sequence information includes N arrival times, and the feature extraction of the arrival time sequence information of the pulse sequence to obtain the feature information of the pulse sequence includes: Based on the least squares method, N points corresponding to the N arrival times in the arrival time sequence information are fitted to obtain a fitting curve corresponding to the arrival time sequence information; For the nth arrival time, the minimum distance between the point corresponding to the nth arrival time and the fitting curve is determined as the target distance, where N is a positive integer greater than 0, and n is a positive integer less than or equal to N; Baseline removal is performed on the N target distances corresponding to the N arrival times to obtain characteristic information of the pulse sequence.

2. The method according to claim 1, wherein: The method of obtaining the correlation feature of the pulse sequence based on the feature information by adopting the correlation embedding method includes: Normalizing the pulse width characteristics of the pulse sequence to obtain a normalized pulse width characteristic; Normalizing the radio frequency characteristics of the pulse sequence to obtain normalized radio frequency characteristics; Based on the feature information, the normalized pulse width feature and the normalized radio frequency feature, an association embedding method is adopted to obtain the association feature of the pulse sequence.

3. The method according to claim 2, wherein: The step of obtaining the correlation feature of the pulse sequence by using a correlation embedding method based on the feature information, the normalized pulse width feature and the normalized radio frequency feature includes: Performing a dilated convolution process on the feature information with parameter information being a first preset value to obtain a first associated feature; Performing a dilated convolution process on the normalized pulse width feature with parameter information being a second preset value to obtain a second correlation feature; Performing a dilated convolution process on the normalized radio frequency feature, wherein parameter information is a third preset value, to obtain a third correlation feature; The first correlation feature, the second correlation feature and the third correlation feature are concatenated to obtain the correlation feature of the pulse sequence.

4. The method according to claim 1, wherein: The performing semantic recognition on the associated features to determine the radar working state includes: Performing pre-convolution processing on the associated features to obtain target associated features; Inputting the target association features into a two-stage residual shrinkage network to obtain semantic features of the pulse sequence; The semantic features are input into a bidirectional gate cycle control unit to determine the radar working state.

5. The method according to claim 4, wherein: The step of inputting the target association features into a two-stage residual shrinkage network to obtain the semantic features of the pulse sequence includes: Inputting the target-related feature into a residual block structure to obtain a second target-related feature; The second target association feature is processed to obtain a semantic feature of the pulse sequence.

6. The method according to claim 5, wherein: The step of inputting the semantic feature into a bidirectional gate loop control unit to determine the radar working state includes: Inputting the semantic features into a bidirectional gate recurrent control unit to obtain a plurality of feature vectors; Generate a working state label of each pulse in the pulse sequence according to a target feature vector, wherein the target feature vector is the last feature vector among the multiple feature vectors; Obtaining the probability of the working state corresponding to each pulse in the pulse sequence according to the working state label of each pulse; The radar working state is determined according to the probability of the working state corresponding to each pulse of the pulse sequence.

7. A radar working status identification device, wherein: The device comprises: The first obtaining module is used to extract the characteristics of the arrival time sequence information of the pulse sequence to obtain the characteristic information of the pulse sequence, wherein the characteristic information includes the pulse timing sequence modulation characteristics and the sequence measurement characteristics; A second obtaining module is used to obtain the correlation features of the pulse sequence by adopting the correlation embedding method based on the feature information, wherein the correlation features represent the features obtained by associating the pulse timing sequence modulation features, sequence metric features, pulse width features and radio frequency features of the multiple pulses in the pulse sequence with each other; A determination module, used for performing semantic recognition on the associated features to determine the working state of the radar; Wherein, the arrival time sequence information includes N arrival times; First get the module, including: A first obtaining submodule is used to fit N points corresponding to the N arrival times in the arrival time sequence information based on the least squares method to obtain a fitting curve corresponding to the arrival time sequence information; The second obtaining submodule is used to determine, for the nth arrival time, the minimum distance between the point corresponding to the nth arrival time and the fitting curve as the target distance, wherein N is a positive integer greater than 0, and n is a positive integer less than or equal to N; The third obtaining submodule is used to perform baseline removal processing on the N target distances corresponding to the N arrival times to obtain characteristic information of the pulse sequence.

8. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images