A Radar Jamming Decision Method Based on Knowledge Graph

Abstract

The present invention belongs to the technical field of radar jamming, and specifically relates to a radar jamming decision-making method based on a knowledge graph. The method of the present invention mainly classifies and identifies the intercepted radar signals first, constructs triples for known radars and unknown radars respectively, matches the obtained triples through the knowledge graph, the known radars can be directly matched by name, while the unknown radars are matched by fingerprint features, obtains the corresponding radar knowledge based on the matching results, and evaluates its threat level. Then, a dynamic variable and effective interference is implemented on the radar through an interference strategy table, so as to achieve the purpose of radar jamming.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar jamming, and particularly relates to a radar jamming decision-making method based on a knowledge graph. Background Art

[0002] With the continuous development of the electromagnetic field, modern warfare has gradually become a battle for electromagnetic dominance. As the role of radar becomes increasingly important, jamming measures against radar have also become a very important part of the electronic countermeasure field. Traditional jamming mostly analyzes the radar data received in real time and combines it with a jamming strategy library to make jamming decisions. This data-driven jamming decision-making algorithm has high requirements for data and can only effectively jam when the opponent's radar equipment uses recorded parameters. Moreover, due to the increasing number of radars and jamming equipment invested by both sides in modern electronic warfare, it is easy to face the problem of combinatorial explosion in radar jamming decision-making in actual battlefields. It is very difficult to accurately formulate jamming plans for a large number of targets in a short time using traditional methods. Summary of the Invention

[0003] In view of the above problems, the present invention proposes a radar jamming decision-making method based on a knowledge graph.

[0004] The technical solution of the present invention is as follows:

[0005] A radar jamming decision-making method based on a knowledge graph, characterized by comprising the following steps:

[0006] S1. After classifying and identifying the intercepted radar signal, determine whether the category is known. If it is known, construct a triple structure <XX radar name, fingerprint feature, feature vector R k > (where the radar name is the head entity head, the fingerprint feature is the relationship relation, and the feature vector is the tail entity tail. Therefore, this triple structure forms a fact - the fingerprint feature of XX radar is feature vector R k ); otherwise, if the category is "unknown radar", construct a triple structure <automatically named XX, fingerprint feature, feature vector R unk >, where the automatic naming is based on the set naming rules;

[0007] S2. Use the radar knowledge graph to match the triple structure constructed in S1. The radar entities in the radar knowledge graph include radar entity names, radar fingerprint feature vectors, and other relevant information, such as geographical location, the platform carried, etc. At the same time, the radar entity name is associated with the radar fingerprint feature vector;

[0008] The triple <radar name, fingerprint feature, feature vector R k > with known category is directly matched through the radar name;

[0009] Triple of unknown category <Automatically named, fingerprint feature, feature vector R unk >Match through fingerprint features, specifically by using the obtained fingerprint feature vector R unk Perform correlation analysis with the fingerprint feature vectors R of existing radars in the map to obtain the correlation coefficient ρ:

[0010]

[0011] where n is the number of elements in the fingerprint feature vector, r i is the i-th element in R, r is the mean of the feature vector R, r unki is R unk the i-th element in, is the feature vector R unk mean value of;

[0012] Judge whether the matching is successful by judging whether ρ≥T holds, where T is the set similarity threshold;

[0013] According to the matching result and the relevant information associated with the radar entity in the knowledge graph, such as the platform P carried by the radar, the location A where it has appeared, the historical interference situation J G , the interference signal S it has received, the radars R that have appeared together, etc., and perform knowledge reasoning on the obtained radar information to obtain some supplementary information, then perform threat level assessment based on the obtained knowledge, and then use it for interference decision-making. For the radars with failed matching, the radar fingerprint feature vectors and other relevant information of this radar will be stored in the database to update the knowledge graph, and then use the traditional method to perform threat level assessment through the intercepted radar information and radiation source location information;

[0014] S3. Use the obtained threat level assessment result for interference decision-making, and at the same time use the made interference decision to update the knowledge graph.

[0015] Traditional interference mostly analyzes the real-time received radar data and combines with the interference strategy library for interference decision-making. This data-driven interference decision-making algorithm has high requirements for data and can only perform effective interference when the opponent's radar equipment uses the recorded parameters. The present invention uses the knowledge graph for interference decision-making on the basis of the traditional method, and this method can fully mine the relevant historical information of the radar and update the corresponding knowledge base. Brief Description of the Drawings

[0016] Figure 1 is the flowchart of radar interference decision-making.

[0017] Figure 2 is the interference strategy table.

[0018] Figure 3 It is the state transition diagram after the radar is interfered.

[0019] Figure 4 It is the result of the radar interference decision. Specific implementation manner

[0020] The technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings and simulation examples:

[0021] The present invention is divided into three parts, namely radar individual recognition, knowledge graph analysis and threat level assessment, and interference pattern selection. First, the individual is recognized, and a triple of <radar name, fingerprint feature, feature vector R k > or <automatic naming, fingerprint feature, feature vector R unk > is constructed. Then, knowledge graph analysis is performed. If the individual is recognized as a known category, name matching is performed in the knowledge base; if it does not belong to a known category, the fingerprint features of the radar are extracted for feature matching. If the name matching or feature matching is successful, knowledge reasoning is used to obtain information, and then the threat level assessment is performed; if the feature matching is unsuccessful, the triple is stored in the database and the knowledge graph is updated, and then the threat level assessment is performed using traditional methods. After the threat level assessment, finally, the interference pattern is selected and the knowledge graph is updated.

[0022] As Figure 1 shown, the specific method of the present invention is:

[0023] S1. Radar individual recognition

[0024] In the battlefield environment, the information that can usually be obtained includes the intercepted radar signal and the position information of the radar carrying platform, such as the moving speed v, the operating distance l from us, etc. For the intercepted radar signal, the method of radar emitter individual recognition can be used for classification and recognition analysis, and the known and unknown radar emitter individuals are discussed separately. For the known radar emitter, while obtaining the name of the individual, the fingerprint feature vector R k =[r1, r2,..., r n representing the individual is saved; for the unknown radar of the unknown category, after the individual recognition is completed, it is classified as an unknown category, and it can be named according to the current recognition time and order, such as 2022022202 representing the second unknown radar recognized on February 22, 2022, and its fingerprint feature vector R unk =[r1, r2,..., r n is saved.

[0025] For the known radar emitter, based on the structure of the knowledge graph triple <radar name, fingerprint feature, feature vector R k>Describe the radar features; if the radar does not belong to a known category, it is automatically named to form a triple <Automatic Naming, Fingerprint Feature, Feature Vector R unk >.

[0026] S2. Knowledge Graph Analysis

[0027] In the knowledge graph, the radar-related knowledge domain not only has its entity name, but also associates a radar fingerprint feature vector with a specific radar entity. Therefore, the corresponding radar entity in the graph can be found according to the radar name and its fingerprint feature vector, and the associated platform P, historical interference situation J G , information such as the locations A where it has appeared, etc. can be quickly retrieved. For an unknown radar with only a fingerprint feature vector, its own fingerprint feature vector can be used to perform a correlation analysis with the fingerprint feature vectors R s = [r s1 , r s2 , …, r sn of the existing radars in the graph, and the correlation coefficient ρ between the two can be calculated, as shown in Equation (1).

[0028]

[0029] The value range of the correlation coefficient ρ is [-1, 1]. The larger the absolute value of the coefficient ρ, the higher the similarity between the two. Therefore, a similarity threshold T is set. If ρ ≥ T, the radar matching is successful; otherwise, the radar matching fails. For radars that fail to match using feature vector matching, they need to be stored in the knowledge graph in the form of a triple <Radar Name, Fingerprint Feature, R> to update the knowledge graph.

[0030] S3. Threat Level Assessment

[0031] When the radar emitter is matched based on the knowledge graph, for those with successful matching, an inference process is carried out to quickly retrieve the knowledge related to the current radar, such as the platform P carried by the radar, the interference signals S it has received, the radars R that have appeared together, etc. Based on the information obtained from these inferences, a threat level assessment is further carried out to assist in interference decision-making. For radars that fail to match, the threat level assessment can only be carried out based on the intercepted radar signals and the location information of the emitter.

[0032] According to the characteristics of the target radar and other factors affecting the radar threat level, a quantitative and qualitative evaluation of the threat level is carried out. The radar threat level assessment model is shown in Equation (2).

[0033]

[0034] In the formula, w i ∈ [0, 1] is the threat level of the i-th radar, e ijis the j-th influencing factor of the i-th radar, ω j is the weight corresponding to the current influencing factor. The present invention will select three influencing factors for the radar threat level.

[0035] a. The speed e of the radar-carrying platform i1

[0036] The faster the radiation source platform moves, the lower the probability of successfully intercepting it, and thus the greater the threat to us. The platform speed influencing factor is shown in Equation (3).

[0037]

[0038] In the formula, v i is the speed of the platform carried by the i-th radar, v min and v max are the minimum and maximum speeds of the platform movement respectively.

[0039] b. The operating distance e from the radar i2

[0040] Generally, the closer the radar radiation source is to us, the greater the threat to us. Then the operating distance influencing factor from the radar is shown in Equation (4).

[0041]

[0042] In the formula, l i is the operating distance from the platform carried by the i-th radar to us, l max is the farthest distance from the detected target radar to us.

[0043] c. The radar-carrying platform i3

[0044] The threat levels of the radar to us are different when it is on different combat platforms and performing different tasks. Generally, when the radar is carried on high-threat platforms such as missiles and fighter jets and performing urgent and important tasks, the corresponding influencing factor value is higher; on the contrary, the influencing factor value is lower. In addition, when the radar is carried on the same platform but performing different tasks, the corresponding influencing factor value varies with the tasks. Similarly, if the radar performs the same task but is carried on different platforms, the corresponding influencing factor value will vary with the threat level of the platform. The present invention only gives the value range of this influencing factor here i3 ∈[0,1], and a fixed value can be specifically set according to actual needs.

[0045] Different influencing factors contribute differently to the radar threat level assessment. The present invention sets corresponding weights according to each influencing factor, as shown in Table 1.

[0046] Table 1 Weights of Influence Factors for Radar Threat Levels

[0047]

[0048] Based on the above radar threat level influence factors and combined with the reasoning results of the knowledge graph, the threat levels of other radar radiation sources on the platform where the radar may be carried can be calculated, and further the threat level of the current radar can be evaluated as shown in Equation (5).

[0049] w = max{w i | i = 1, 2, …, n} (5)

[0050] In the formula, n represents the number of all radars on the platform.

[0051] S4. Jamming Mode Selection

[0052] Based on years of experience in electronic countermeasures, an interference library has been established according to the existing data information, which is mainly composed of a radar information library and an interference mode library (interference modes and interference parameters). Among them, the radar library records the working parameter range of the radar in a certain working mode, and the interference mode library records the interferences implemented for a certain radar under different working modes. The specific modes are shown in Table 2.

[0053] Table 2 Example of Interference Library

[0054]

[0055] Therefore, after completing the radar threat level assessment, according to the reference table J of the interference library L and the interference situation J provided in the knowledge graph G after that, the best interference mode J = [J L , J G is launched. At the same time, the interference information launched this time is further supplemented into the knowledge graph to make the knowledge in the radar field more rich and detailed.

[0056] Simulation Example

[0057] Based on the intercepted radar signal information and comprehensive intelligence information from other sources, the present invention hopes to accurately and effectively strike the enemy after the first implementation of interference through the method based on the knowledge graph. However, electronic countermeasures is a dynamic game process. To cope with such a situation, a perfect radar interference decision-making system needs to be established, that is, subsequent countermeasures need to be prepared after the first round of interference. To simulate the dynamic process of radar interference decision-making, the present invention conducts a simulation experiment on radar interference decision-making. First, an interference strategy table is established, as Figure 2 shown.

[0058] After the simulation starts, the interference decision-making for the radar side begins. Information such as the current state S of the radar side, the interference pattern it receives, and the state S' after the radar side is disturbed and transformed is continuously updated, specifically as Figure 3 shown.

[0059] When the state of the radar radiation source target is disturbed to the state with the lowest threat level, it is considered at this time that the interference implemented by our side is effective, and then the interference action will be stopped. The entire confrontation result, including the state transition process of the radar and the corresponding interference implemented by our side, is as Figure 4 shown.

Claims

1. A radar jamming decision-making method based on a knowledge graph, characterized in that Including the following steps: S1. After classifying and identifying the intercepted radar signals, determine whether the signal category is known. If it is known, construct a triple structure <radar name, fingerprint feature, feature vector R k >; otherwise, define it as an unknown radar category and construct a triple structure <automatically named, fingerprint feature, feature vector R unk >, where the automatic naming is based on the set naming rules; S2. Use the radar knowledge graph to match the triple structure constructed in S1. The radar entities in the radar knowledge graph include entity names, and at the same time, the radar entities are associated with radar fingerprint feature vectors; Triples of known categories <radar name, fingerprint feature, feature vector R k >Match directly by radar name; Triple of unknown category <Automatic naming, fingerprint feature, feature vector R unk >Match through fingerprint features, specifically by using the obtained fingerprint feature vector R unk Perform a correlation analysis with the fingerprint feature vector R of the existing radar in the map to obtain the correlation coefficient ρ: where n is the number of elements in the fingerprint feature vector, r i is the i-th element in R, is the mean of the feature vector R, r unki is R unk the i-th element in, is the feature vector R unk the mean of; Judge whether the matching is successful by judging whether ρ≥T holds, where T is a set similarity threshold; Then, based on the radar in the matched knowledge graph, obtain relevant radar information, and obtain corresponding inference knowledge. Based on the obtained knowledge, conduct threat level assessment and use it for interference decision-making. For the radars with failed matching, conduct threat level assessment through intercepted radar information and radiation source location information; S3. Use the obtained threat level assessment results to formulate interference decisions, and at the same time use the made interference decisions to update the knowledge graph.

Images