A method and device for extracting key information of radar simulation countermeasure system

Abstract

The present invention discloses a method and device for extracting key information of a radar simulation confrontation system, which relates to the field of radar simulation confrontation technology, including S1, determining the size N of a text block from which key information needs to be extracted, and encoding it; S2, caching the real-time output information of the console, using a hash table to store cache entries, S3, extracting the corresponding key information in a polling manner; S4, further processing logic for the key information. The method and device for extracting key information of a radar simulation confrontation system is applicable to a key information extraction method of P display output information, and clarifies the method structure for extracting key information, as well as the corresponding neural network construction method. Through the extraction method, key information can be extracted from a large number of original output information in a streaming processing manner, and prompts the operator, thereby improving the operator's key information processing capability.

Description

Technical Field

[0001] The present invention relates to the field of radar simulation confrontation technology, and in particular to a method and device for extracting key information of a radar simulation confrontation system. Background Art

[0002] Radar P display, referred to as PPI, is one of the most commonly used display methods of radar. It takes the radar antenna as the origin and the scanning radius as the circumference to display the radar echo signal on the plane coordinate system. The radar P display has the following characteristics: First, it is centered on the antenna and the scanning radius as the circumference: the coordinate system of the P display is centered on the radar antenna and the scanning radius as the circumference; the radar echo signal is drawn on the P display at an angle proportional to the antenna azimuth and a distance proportional to the radar distance. Second, the echo intensity is represented by light and dark: the strength of the echo intensity on the P display is represented by light and dark. The stronger the echo intensity, the brighter the display; the weaker the echo intensity, the darker the display. Third, it can display a variety of information: in addition to the basic information of the target, the P display can also display other auxiliary information, such as scanning lines, azimuth indications, range circles, antenna elevation angles, target tracking information, etc.

[0003] The radar P display provides intuitive and visual information to the radar operator, and plays an important role in the detection, identification and tracking of radar targets. Its operation mainly includes adjusting scanning parameters, adjusting display parameters, target identification and target tracking.

[0004] However, in actual operation, when the number of targets and the speed of movement do not change dramatically, the radar images are very similar and the differences between adjacent frames are small. At the same time, key information needs to be generated by integrating data from multiple radars. Accurate key information extraction faces the following difficulties:

[0005] 1. Information extraction and integration: Since the radar situation summary needs to accurately and comprehensively reflect the target information, motion trajectory, speed, direction and other key elements detected by the radar system, the radar system often generates a large amount of raw data, but the existing technology cannot effectively extract useful information from this data and integrate it concisely and clearly;

[0006] 2. P-display images usually contain a large amount of information, and the information is updated very quickly. The operator needs to concentrate for a long time to observe and interpret this information. Staring at the P-display image for a long time will cause fatigue to the operator's visual system, leading to symptoms such as decreased vision, dry eyes, and eye pain. In addition, the P-display image contains a large amount of information and is updated very quickly. The operator needs to process a large amount of information in a short period of time, which will cause brain fatigue, decreased attention, and reduced judgment.

[0007] Therefore, it is urgent to improve this shortcoming. The present invention studies and improves the existing structure and shortcomings, and provides a method and device for extracting key information of a radar simulation countermeasure system. Summary of the invention

[0008] The purpose of the present invention is to provide a method and device for extracting key information of a radar simulation countermeasure system to solve the problems raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] In a first aspect, a method for extracting key information of a radar simulation countermeasure system is provided, comprising the following steps:

[0011] S1, determine the size N of the text block from which key information needs to be extracted, where N is an adjustable parameter, and encode the text of length N according to a predefined vocabulary;

[0012] S2. Cache the real-time output information of the console and use a hash table to store cache entries. When a new data request arrives, use the stream identifier to find the corresponding entry in the hash table. If the entry is found, return the cached data. Otherwise, retrieve the data from the source and store it in the cache before returning it.

[0013] S3, taking fixed-length information in a polling manner, inputting it into a key information extraction network, and extracting the corresponding key information;

[0014] The key information extraction network adopts an encoder-decoder structure. The original text information is encoded into tokens according to the vocabulary and then input into the encoder (using a two-layer bidirectional LSTM) one by one to generate a series of encoder hidden states. , at each time step t, the Decoder decoder (single-layer unidirectional LSTM) receives the token of the previous word and has a decoder state , that is, the i-th word of the input, the encoder generates a hidden state sequence At each time step t, the decoder receives the output of the encoder and generates a decoding state ,attention The calculation formula is as follows:

[0015]

[0016] in, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder status, is a vector, Represents each component, is the activation function, defined as:

[0017]

[0018] A selection coefficient P is introduced to control the selection of the vocabulary, and the value range of P is [0,1]. Because the initial vocabulary is fixed, if the output contains new words, it needs to be combined with the global context. is the global context:

[0019]

[0020] Get the selection coefficient P:

[0021]

[0022] in, are all trainable parameters in the network. The T operation represents matrix transposition. is the decoder state at time step t, is the input of the decoder, Sigmoid is a commonly used activation function, defined as:

[0023]

[0024] P is used to determine, is to select the word w in the vocabulary, the word w in the vocabulary is recorded as , or copy from the input sequence (because it is a new word, it does not exist in the vocabulary, intuitively understood as a word that the system does not know), use the following formula to determine the distribution of the word:

[0025]

[0026] If w is an out-of-vocabulary word (OOV), then is 0. If w is not a word in the original text, then The value is 0 (because the word does not exist in the original text, the attention is 0); you can add additional vocabulary expansion logic, regularly record the oov words, and choose to add the oov words to the predefined vocabulary for subsequent use; another available option is that if the output information is stable and there is no oov output word, simplify the logic of processing oov, and the value of P is directly 0;

[0027] In addition, choose to introduce a vector , to suppress the generation of repeated content; intuitively speaking, it is to suppress the excessive attention to certain words, resulting in too much key information surrounding certain words, thus generating repeated content:

[0028]

[0029] The sum of the attention of all previous decoder steps is 0, because in the first step, no words in the original text are covered, and the attention of all words is 0. If you choose to use it, the formula is updated to:

[0030]

[0031] in, and is a trainable parameter, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder state, t is the time step, i is the i-th word, and the subscript i in the formula represents the i-th component. and have i, and and They are all vectors. A coverage loss function is introduced to suppress repeated words and avoid outputting repeated content. The loss function is defined as follows:

[0032]

[0033] The above loss function will be less than 1, because the sum of all attention distributions will be 1. Introducing this part into the global, the loss function of the entire network is defined as:

[0034]

[0035] in, It is an adjustable parameter to modify the final result;

[0036] S4. Perform further processing logic for key information.

[0037] Furthermore, in step S2, for the real-time information caching result of the console, there are several data streams with completely different semantics, the semantics of the data streams include but are not limited to target positioning, target tracking, and post-annotation processing, and the data streams can be distributed and cached for separate processing.

[0038] Furthermore, in step S2, each entry in the hash table is indexed by a flow identifier, the value of the entry is the cached data of the flow, and the key of the hash table is the unique identifier of the information flow, using the ID of the information source, and the value is the information flow data itself.

[0039] Furthermore, in step S2, the cache is updated regularly to ensure the timeliness of the cached data, and the cache update strategy adopts a FIFO (first in, first out) strategy to eliminate data in the order in which they are inserted. If information processing is too slow and exceeds a predefined threshold, cache update measurement is triggered.

[0040] Furthermore, in step S2, the cache optimization operation includes gzip compression of the cache data to reduce storage space and transmission consumption. The specific cache processing scheme is adjusted according to the actual application scenario. When selecting the hash function and adjusting the load factor, the search efficiency and space utilization must be weighed.

[0041] Furthermore, in step S4, further processing logic specifically includes:

[0042] Verification and cleaning: Check whether the extracted key information is valid, reasonable or in accordance with predefined rules, and remove duplicate, irrelevant or incorrectly formatted information;

[0043] Classification and labeling: Classify the extracted key information according to the set logic or rules, and label the key information to facilitate subsequent search, query or processing;

[0044] Relation and merging: For related key information, establish the connection between the key information and merge the related key information into a complete information or structure;

[0045] Storage and indexing: Save the extracted and processed key information to a database, file or other storage medium, and create an index to improve query speed, especially when the data volume is large;

[0046] Analysis and visualization: Conduct statistical analysis of key information, such as frequency, trend, etc., and present the analysis results in charts, images or other visual forms;

[0047] Trigger subsequent operations: trigger specific business logic or operations based on the extracted key information, such as sending notifications, executing commands, etc.

[0048] Feedback and optimization: Use processing results or user feedback to optimize the key information extraction network or processing logic, and adjust network parameters, improve algorithms, or update processing logic based on feedback;

[0049] Security and privacy protection: Ensure the security of key information during the processing, prevent data leakage or tampering, and comply with privacy protection regulations to ensure that personal data is not abused.

[0050] In the second aspect, a device for extracting key information of a radar simulation confrontation system is provided, which is applied to the key information extraction method of a radar simulation confrontation system as mentioned above. The device includes: a memory, a processor, and computer program instructions stored on the memory and executable on the processor. When the processor executes the computer program instructions, the key information extraction method of the radar simulation confrontation system as mentioned above is implemented.

[0051] In a third aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, the method for extracting key information of a radar simulation countermeasure system as described above is used.

[0052] The present invention provides a method and device for extracting key information of a radar simulation countermeasure system, which has the following beneficial effects:

[0053] The present invention proposes a key information extraction method suitable for P display image output information, and clarifies the method structure for extracting key information and the corresponding neural network construction method. Through the extraction method, key information can be extracted from a large amount of original output information in a streaming processing manner, and prompts are given to the operator, thereby improving the operator's key information processing ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A processing flow chart of a method for extracting key information of a radar simulation countermeasure system according to the present invention;

[0055] Figure 2 The present invention is a schematic diagram of the main structure of a key information extraction network of a method for extracting key information of a radar simulation countermeasure system. DETAILED DESCRIPTION

[0056] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0057] like Figure 1-Figure 2 As shown, a method for extracting key information of a radar simulation countermeasure system comprises the following steps:

[0058] S1, determine the size N of the text block from which key information needs to be extracted, where N is an adjustable parameter, and encode the text of length N according to a predefined vocabulary;

[0059] S2. Cache the real-time output information of the console. For the real-time information cache results of the console, there are several data streams with completely different semantics. The semantics of the data stream include but are not limited to target positioning, target tracking, and post-annotation processing, and the data streams can be distributed and cached and processed separately. A hash table is used to store cache entries. Each entry in the hash table is indexed by a stream identifier. The value of the entry is the cache data of the stream, and the key of the hash table is the unique identifier of the information stream. The ID of the information source is used, and the value is the information stream data itself. When a new data request arrives, the stream identifier is used to find the corresponding entry in the hash table. If the entry is found, If the cache data is not received, the cache data is returned. Otherwise, the data is retrieved from the source and stored in the cache before being returned. The cache is updated regularly to ensure the timeliness of the cache data. The cache update strategy adopts the FIFO (first-in-first-out) strategy to eliminate the data in the order in which it is inserted. If the information processing is too slow and exceeds the predefined threshold, the cache update measurement is triggered. In addition, the cache optimization operation includes gzip compression of the cache data to reduce storage space and transmission consumption. The specific cache processing scheme is adjusted according to the actual application scenario. When selecting the hash function and adjusting the load factor, the search efficiency and space utilization must be weighed.

[0060] S3, taking fixed-length information in a polling manner, inputting it into a key information extraction network, and extracting the corresponding key information;

[0061] The key information extraction network adopts an encoder-decoder structure. The original text information is encoded into tokens according to the vocabulary and then input into the encoder (using a two-layer bidirectional LSTM) one by one to generate a series of encoder hidden states. , at each time step t, the Decoder decoder (single-layer unidirectional LSTM) receives the token of the previous word and has a decoder state , that is, the i-th word of the input, the encoder generates a hidden state sequence At each time step t, the decoder receives the output of the encoder and generates a decoding state ,attention The calculation formula is as follows:

[0062]

[0063] in, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder status, is a vector, Represents each component, is the activation function, defined as:

[0064]

[0065] A selection coefficient P is introduced to control the selection of the vocabulary, and the value range of P is [0,1]. Because the initial vocabulary is fixed, if the output contains new words, it needs to be combined with the global context. is the global context:

[0066]

[0067] Get the selection coefficient P:

[0068]

[0069] in, are all trainable parameters in the network. The T operation represents matrix transposition. is the decoder state at time step t, is the input of the decoder, Sigmoid is a commonly used activation function, defined as:

[0070]

[0071] P is used to determine, is to select the word w in the vocabulary, the word w in the vocabulary is recorded as , or copy from the input sequence (because it is a new word, it does not exist in the vocabulary, intuitively understood as a word that the system does not know), use the following formula to determine the distribution of the word:

[0072]

[0073] If w is an out-of-vocabulary word (OOV), then is 0. If w is not a word in the original text, then The value is 0 (because the word does not exist in the original text, the attention is 0); you can add additional vocabulary expansion logic, regularly record the oov words, and choose to add the oov words to the predefined vocabulary for subsequent use; another available option is that if the output information is stable and there is no oov output word, simplify the logic of processing oov, and the value of P is directly 0;

[0074] In addition, choose to introduce a vector , to suppress the generation of repeated content; intuitively speaking, it is to suppress the excessive attention to certain words, resulting in too much key information surrounding certain words, thus generating repeated content:

[0075]

[0076] The sum of the attention of all previous decoder steps is 0, because in the first step, no words in the original text are covered, and the attention of all words is 0. If you choose to use it, the formula is updated to:

[0077]

[0078] in, and is a trainable parameter, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder state, t is the time step, i is the i-th word, and the subscript i in the formula represents the i-th component. and have i, and and They are all vectors. A coverage loss function is introduced to suppress repeated words and avoid outputting repeated content. The loss function is defined as follows:

[0079]

[0080] The above loss function will be less than 1, because the sum of all attention distributions will be 1. Introducing this part into the global, the loss function of the entire network is defined as:

[0081]

[0082] in, It is an adjustable parameter to modify the final result;

[0083] S4. Further processing logic is performed on key information. The further processing logic specifically includes:

[0084] Verification and cleaning: Check whether the extracted key information is valid, reasonable or in accordance with predefined rules, and remove duplicate, irrelevant or incorrectly formatted information;

[0085] Classification and labeling: Classify the extracted key information according to the set logic or rules, and label the key information to facilitate subsequent search, query or processing;

[0086] Relation and merging: For related key information, establish the connection between the key information and merge the related key information into a complete information or structure;

[0087] Storage and indexing: Save the extracted and processed key information to a database, file or other storage medium, and create an index to improve query speed, especially when the data volume is large;

[0088] Analysis and visualization: Conduct statistical analysis of key information, such as frequency, trend, etc., and present the analysis results in charts, images or other visual forms;

[0089] Trigger subsequent operations: trigger specific business logic or operations based on the extracted key information, such as sending notifications, executing commands, etc.

[0090] Feedback and optimization: Use processing results or user feedback to optimize the key information extraction network or processing logic, and adjust network parameters, improve algorithms, or update processing logic based on feedback;

[0091] Security and privacy protection: Ensure the security of key information during the processing, prevent data leakage or tampering, and comply with privacy protection regulations to ensure that personal data is not abused.

[0092] A device for extracting key information of a radar simulation confrontation system is applied to the above-mentioned method for extracting key information of a radar simulation confrontation system. The device comprises: a memory, a processor, and computer program instructions stored in the memory and executable on the processor. When the processor executes the computer program instructions, the above-mentioned method for extracting key information of a radar simulation confrontation system is implemented.

[0093] A computer-readable storage medium stores computer-executable instructions, which are used to implement the above-mentioned method for extracting key information of a radar simulation countermeasure system when executed by a processor.

[0094] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. A method for extracting key information of a radar simulation countermeasure system, characterized in that: The following steps are involved: S1, determine the size N of the text block from which key information needs to be extracted, and encode the text of length N according to a predefined vocabulary; S2. Cache the real-time output information of the console and use a hash table to store cache entries. When a new data request arrives, use the stream identifier to find the corresponding entry in the hash table. If the entry is found, return the cached data. Otherwise, retrieve the data from the source and store it in the cache before returning it. S3, taking fixed-length information in a polling manner, inputting it into a key information extraction network, and extracting the corresponding key information; The key information extraction network adopts an encoder-decoder structure. The original text information is encoded into tokens according to the vocabulary and then input into the encoder one by one to generate a series of encoder hidden states. , at each time step t, the Decoder receives the token of the previous word and has a decoder state , that is, the i-th word of the input, the encoder generates a hidden state sequence At each time step t, the decoder receives the output of the encoder and generates a decoding state ,attention The calculation formula is as follows: in, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder status, is a vector, Represents each component, is the activation function, defined as: A selection coefficient P is introduced to control the selection of the vocabulary, and the value range of P is [0,1]. Because the initial vocabulary is fixed, if the output contains new words, it needs to be combined with the global context. is the global context: Get the selection coefficient P: in, are all trainable parameters in the network. The T operation represents matrix transposition. is the decoder state at time step t, is the input of the decoder, Sigmoid is a commonly used activation function, defined as: P is used to determine, is to select the word w in the vocabulary, the word w in the vocabulary is recorded as , or copy from the input sequence, use the following formula to determine the word distribution: If w is a word outside the vocabulary, then is 0. If w is not a word in the original text, then The value is 0; you can add additional vocabulary expansion logic, regularly record oov words, and choose to add oov words to the predefined vocabulary for subsequent use; another available option is that if the output information is stable and there is no oov output word, simplify the logic of processing oov, and the value of P is directly 0; In addition, choose to introduce a vector , to suppress the generation of repeated content; intuitively speaking, it is to suppress the excessive attention to certain words, resulting in too much key information surrounding certain words, thus generating repeated content: is the sum of attentions from all previous decoder steps, The value of is 0, because in the first step, no words in the original text are covered, and the attention of all words is 0. If you choose to use it, the formula is updated to: in, and is a trainable parameter, These are the parameters that need to be trained in the network. is the encoder hidden state, is the decoder state, t is the time step, i is the i-th word, and the subscript i in the formula represents the i-th component. and have i, and and They are all vectors. A coverage loss function is introduced to suppress repeated words and avoid outputting repeated content. The loss function is defined as follows: The above loss function will be less than 1, because the sum of all attention distributions will be 1. Introducing this part into the global, the loss function of the entire network is defined as: in, It is an adjustable parameter to modify the final result; S4. Perform further processing logic for key information.

2. The method for extracting key information of a radar simulation countermeasure system according to claim 1, characterized in that: In step S2, for the real-time information caching result of the console, there are several data streams with completely different semantics, the semantics of the data streams include but are not limited to target positioning, target tracking, and post-annotation processing, and the data streams can be distributed and cached and processed separately.

3. The method for extracting key information of a radar simulation countermeasure system according to claim 1, characterized in that: In step S2, each entry in the hash table is indexed by a flow identifier, the value of the entry is the cached data of the flow, and the key of the hash table is the unique identifier of the information flow, using the ID of the information source, and the value is the information flow data itself.

4. The method for extracting key information of a radar simulation countermeasure system according to claim 1, characterized in that: In step S2, the cache is updated regularly to ensure the timeliness of the cached data, and the cache update strategy adopts the FIFO strategy to eliminate data in the order in which they are inserted. If information processing is too slow and exceeds a predefined threshold, cache update measurement is triggered.

5. The method for extracting key information of a radar simulation countermeasure system according to claim 1, characterized in that: In step S2, the cache optimization operation includes gzip compression of the cache data to reduce storage space and transmission consumption. The specific cache processing scheme is adjusted according to the actual application scenario. When selecting the hash function and adjusting the load factor, the search efficiency and space utilization must be weighed.

6. The method for extracting key information of a radar simulation countermeasure system according to claim 1, characterized in that: In step S4, further processing logic specifically includes: Verification and cleaning: Check whether the extracted key information is valid, reasonable or in accordance with predefined rules, and remove duplicate, irrelevant or incorrectly formatted information; Classification and labeling: Classify the extracted key information according to the set logic or rules, and label the key information; Relation and merging: For related key information, establish the connection between the key information and merge the related key information into a complete information or structure; Storage and indexing: Save the extracted and processed key information to a database, file or other storage medium, and create an index to increase query speed; Analysis and visualization: Conduct statistical analysis of key information and present the analysis results in charts, images or other visual forms; Trigger subsequent operations: trigger specific business logic or operations based on the extracted key information; Feedback and optimization: Use processing results or user feedback to optimize the key information extraction network or processing logic, and adjust network parameters, improve algorithms, or update processing logic based on feedback; Security and privacy protection: Ensure the security of key information during the processing, prevent data leakage or tampering, and comply with privacy protection regulations to ensure that personal data is not abused.

7. A device for extracting key information of a radar simulation countermeasure system, characterized in that: The device includes: a memory, a processor, and computer program instructions stored in the memory and executable on the processor. When the processor executes the computer program instructions, the method for extracting key information of a radar simulation countermeasure system as described in any one of claims 1 to 6 above is implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method for extracting key information of a radar simulation countermeasure system as described in any one of claims 1 to 6 above.

Images