A knowledge graph-based phased array radar threat level estimation method

Abstract

The application relates to a knowledge graph-based phased array radar threat level estimation method and belongs to the field of threat level estimation. The method comprises the following steps: determining threat level factors affecting the phased array radar and quantifying each threat level factor; analyzing the influence of different threat level factors on the threat level of different types of phased array radars, and establishing a dynamic Bayesian network; traversing the parameter changes of the phased array radar in different working scenes, inputting the parameter changes into the dynamic Bayesian network, outputting corresponding threat level estimation probability values, and obtaining massive experimental data; performing knowledge extraction processing on the experimental data, constructing entity relationships, and establishing a knowledge graph; establishing a query system to output relevant phased array radar threat level query results; comparing the threat levels of the output results to obtain threat degree sorting results of each target. The application can effectively enhance the knowledge correlation and reasoning ability of the knowledge graph, and realizes accurate target sorting of multiple phased array radars.

Description

Technical Field

[0001] This invention belongs to the field of threat level estimation, specifically involving a knowledge graph-based method for estimating the threat level of phased array radar, which is applicable to the ranking of threat levels of phased array radar in multiple operating scenarios. Background Technology

[0002] Phased array radar possesses flexible airspace scanning capabilities and wavefront agility, allowing it to autonomously adopt different operating modes based on varying operational environments and mission requirements. Its significant performance advantages have led to its widespread application in 21st-century local conflicts. Accurate threat level estimation can effectively determine the priority sequence for multi-target phased array radar strikes, enabling a global grasp of battlefield information and providing technical and intelligence support for subsequent operational strategies and policy formulation. However, for the highly flexible and complex phased array radar system, traditional threat level estimation methods suffer from poor reliability and are unable to adapt to the changing electromagnetic environment and emerging weaponry.

[0003] Currently, the main threat level estimation methods are divided into multi-attribute decision-making methods and machine learning methods. Multi-attribute decision-making methods utilize existing prior knowledge and decision information to rank a number of targets or select the best option from a number of alternatives. However, because they require manual assignment of weights to each attribute, they are highly subjective and prone to errors. Machine learning methods transform the ranking and decision-making problem into a non-linear prediction problem. They output and rank targets by directly estimating their threat values. Compared to multi-attribute decision-making methods, machine learning networks can better correlate threat information with temporal information, accumulate feature information, and possess a certain degree of reasoning ability. However, machine learning methods also rely on expert-assigned conditional probabilities and transition probabilities, and a single model cannot adjust threat estimation for multiple scenarios. Knowledge graphs possess powerful knowledge association and relearning capabilities. Through semantic relationships between entities, they offer advantages such as fast location speed and accurate search, providing strategic decision support functions. Threat estimation, as a reference method for reasoning and decision-making, can be largely met by knowledge graphs. By constructing a knowledge graph system, rapid association and analysis of target phased array radar characteristics can be achieved.

[0004] Currently, traditional threat level estimation methods rely on prior knowledge and lack the ability to adaptively adjust parameters to suit different battlefield environments. Furthermore, due to the lack of publicly available datasets and textual data in the field of phased array radar, a method is needed to fill this knowledge gap. Summary of the Invention

[0005] The technical problem to be solved by this invention is:

[0006] To address the problems of traditional threat level estimation methods, this invention provides a knowledge graph-based threat level estimation method. This method solves the issues of insufficient publicly available datasets in the phased array radar field, the inability to establish knowledge graphs, and the difficulty in accurate threat level estimation across multiple scenarios.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0008] A knowledge graph-based method for estimating the threat level of phased array radar, characterized by comprising:

[0009] Identify the threat level factors affecting phased array radar, and scientifically quantify and analyze each threat level factor by establishing membership degrees;

[0010] The impact of different threat level factors on the threat level of different phased array radar models is analyzed, and the corresponding conditional probability tables are given. A dynamic Bayesian network is established based on the conditional probability tables.

[0011] By iterating through the parameter changes of the phased array radar under different working scenarios, inputting them into a dynamic Bayesian network, and outputting the corresponding threat level estimation probability value, a massive amount of experimental data is obtained.

[0012] The experimental data is processed by knowledge extraction, entity relationships are constructed, and a knowledge graph is established.

[0013] Establish a query system: Calculate cosine similarity based on the input phased array radar information and phased array radar information in the knowledge graph; output relevant phased array radar threat level query results based on the cosine similarity.

[0014] By comparing the threat levels of the output results, the threat ranking of each target in the target work scenario is obtained.

[0015] A further technical solution of the present invention: the scientific quantitative analysis of each threat level factor by establishing membership degrees includes:

[0016] The carrier threat membership degree μ1(x) is represented as:

[0017]

[0018] The repetition frequency membership degree μ2(x) is expressed as:

[0019]

[0020] The pulse width threat membership degree μ3(x) is represented as:

[0021]

[0022] Where x represents the threat level factor.

[0023] A further technical solution of the present invention: the establishment of a dynamic Bayesian network based on a conditional probability table specifically includes:

[0024] A dynamic Bayesian network can be defined as <B0, B1, B2>, where B0 refers to the Bayesian network model at the initial time T0, B2 refers to the state of the Bayesian network at the next time step, and B1 refers to the transition model between the two, which is used to represent the change in the state of the Bayesian network between two adjacent time points. The time point interval can be set according to the application background requirements. Let the prior probability of the initial time corresponding to B0 be P0, then in the transition model B1, the transition probability from X[t] to X[t+1] is P(X[t+1]|X[t]), which finally yields the time step B2, and so on.

[0025] A further technical solution of the present invention: the calculation of cosine similarity based on the input phased array radar information and the phased array radar information in the knowledge graph specifically involves:

[0026] Cosine_similarity=(A·B) / (||A||×||B||)

[0027] Wherein, vector A and vector B are the phased array radar information input and the phased array radar information in the knowledge graph, respectively, (A·B) represents the dot product of vector A and vector B, ||A|| represents the magnitude of vector A, and ||B|| represents the magnitude of vector B.

[0028] A further technical solution of the present invention: the step of outputting the relevant phased array radar threat level query result based on cosine similarity specifically includes:

[0029] If the cosine similarity is lower than the set threshold, the threat level of the most relevant query result is used as the final threat level estimate; if it is higher than the set threshold, the threat levels of the 5 most relevant targets are averaged and used as the final threat level output.

[0030] A computer system is characterized by comprising: one or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method described above.

[0031] A computer-readable storage medium is characterized by storing computer-executable instructions, which, when executed, are used to implement the above-described method.

[0032] A computer program product is characterized by including computer-executable instructions, which, when executed, are used to implement the above-described method.

[0033] The beneficial effects of this invention are as follows:

[0034] This invention provides a threat level estimation method based on knowledge graphs. By comprehensively analyzing the factors that affect threat level assessment, it independently builds experimental scenarios to obtain massive amounts of experimental data, establishes a threat factor knowledge base and a threat estimation reasoning knowledge base, and constructs a phased array radar knowledge graph.

[0035] Compared to traditional methods, this invention can address the threat level estimation needs in different scenarios. By continuously enriching the accumulation of prior knowledge and improving experimental data, it can effectively enhance the knowledge association and reasoning capabilities of the knowledge graph, thereby achieving accurate target ranking for multi-target phased array radar. Attached Figure Description

[0036] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0037] Figure 1 This is a flowchart of a knowledge graph-based method for estimating the threat level of phased array radar.

[0038] Figure 2 Dynamic Bayesian network architecture diagram.

[0039] Figure 3 Schematic diagram of the target simulation working scenario. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0041] This invention provides a knowledge graph-based method for estimating the threat level of phased array radar, such as... Figure 1 As shown, it includes the following steps:

[0042] Step 1: Comprehensively analyze the factors affecting the threat level of phased array radar, specifically divided into pulse signal parameters (represented by pulse descriptors) and the kinematic behavior of the platform.

[0043] Step 2: Scientifically quantify and analyze the factors of each threat level by establishing membership degrees. Based on the impact of parameter magnitude on the threat level, they can be divided into interval type, benefit type, and cost type. Establish membership degree functions for different attributes and give the corresponding high, medium, and low evaluation levels.

[0044] The carrier threat membership degree μ1(x) is represented as:

[0045]

[0046] The repetition frequency membership degree μ2(x) is expressed as:

[0047]

[0048] The pulse width threat membership degree μ3(x) is represented as:

[0049]

[0050] Among them, carrier frequency threat membership is interval type, frequency repetition rate membership is benefit type, pulse width threat membership is cost type, and x is the threat level factor.

[0051] Step 3: Analyze the impact of different threat factors on the threat levels of different phased array radar models, provide the corresponding conditional probability tables, and establish a dynamic Bayesian network.

[0052] A dynamic Bayesian network can be defined as <B0, B1, B2>, where B0 refers to the Bayesian network model at the initial time T0, B2 refers to the state of the Bayesian network at the next time step, and B1 refers to the transition model between the two, which is used to represent the change in the state of the Bayesian network between two adjacent time points. The time point interval can be set according to the application background requirements. Let the prior probability of the initial time corresponding to B0 be P0, then in the transition model B1, the transition probability from X[t] to X[t+1] is P(X[t+1]|X[t]), which finally yields the time step B2, and so on.

[0053] Step 4: Traverse the parameter changes of the phased array radar under different working scenarios, input them into the dynamic Bayesian network, output the corresponding threat level estimation probability value, and obtain massive experimental data.

[0054] Step 5: Extract knowledge from the data, construct entity relationships, and build a knowledge graph.

[0055] Step 6: Establish a query system to calculate cosine similarity between the input phased array radar information and the phased array radar information in the knowledge graph.

[0056] Cosine_similarity=(A·B) / (||A||×||B||) (4)

[0057] Here, vectors A and B respectively contain the attribute values ​​of the phased array radar. (A·B) represents the dot product of vectors A and B, ||A|| represents the magnitude of vector A, and ||B|| represents the magnitude of vector B.

[0058] Output the relevant phased array radar threat level query results. If the cosine similarity of the query results is lower than the set threshold, the threat level of the most relevant query result is used as the final threat level estimate. If it is higher than the set threshold, the threat levels of the 5 most relevant targets are averaged and used as the final threat level output.

[0059] Step 7: Compare the threat level probability values ​​of the output results to obtain the threat ranking results of each target in the target work scenario.

[0060] To enable those skilled in the art to better understand the present invention, the present invention will be described in detail below with reference to specific embodiments.

[0061] Example 1:

[0062] Step 1: Consider the following scenario: There is one measurement station in space, and the phased array radars are physically spaced apart from each other, mounted on platforms such as fighter jets, ground-based radars, and shipborne radars. The main monitoring station is located on the protected target. The target phased array radar models and prior information are shown in Table 1. The electromagnetic environment of a certain phased array radar is simulated, and the obtained phased array radar pulse signal data and the kinematic behavior of the mounted platforms are shown in Table 2.

[0063] Step 2: Based on the membership function of different threat factors, give the corresponding evaluation level.

[0064] Step 3: Analyze the relationship between various threat factors and the final threat level for different phased array radar weapon models, provide the corresponding conditional probability tables, and establish a dynamic Bayesian network. The network architecture is as follows: Figure 2 As shown.

[0065] Step 4: Based on the pulse signal parameters emitted by each phased array radar and the kinematic behavior of the platform, three levels of changes are made and the data are re-input into the dynamic Bayesian network to obtain new threat level estimates. This process is repeated for different operating modes of various phased array radar models to obtain a large amount of experimental data.

[0066] Step 5: Extract knowledge from the data, establish entity relationships, and input them into the knowledge graph. The established entity relationship table for the phased array radar is shown in Table 3.

[0067] Step 6: Establish a query system based on the Cypher language to query the output results of the closest parameters from the knowledge graph;

[0068] Step 7: Output the relevant phased array radar threat level query results. If the cosine similarity of the query results is lower than the set threshold, the threat level of the most relevant query result is used as the final threat level estimate. If it is higher than the set threshold, the threat levels of the five most relevant targets are averaged and used as the final threat level output. The accuracy of the ranking results is evaluated using Normalized Discounted Cumulative Gain (NDCG), Kendall's Tau (TAU in this invention), and Mean Reciprocal Rank (MRR). Table 4 shows a performance comparison between the proposed method and traditional methods.

[0069] Table 1. Information on phased array radar models and weaponry under target operating scenarios.

[0070]

[0071] Table 2. Signal parameters and kinematic behavior of each phased array radar

[0072]

[0073] Table 3. Entity Relationship Table for Phased Array Radars

[0074]

[0075] Table 4 shows the threat level ranking results and index values ​​for phased array radars by different methods.

[0076]

[0077] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in the present invention, and such modifications or substitutions should all be covered within the scope of protection of the present invention.

Claims

1. A knowledge graph-based method for estimating the threat level of phased array radar, characterized in that, include: Identify the threat level factors affecting phased array radar, and scientifically quantify and analyze each threat level factor by establishing membership degrees; including: Carrier frequency threat membership Represented as: Repetition frequency membership Represented as: Pulse width threat membership Represented as: in, Threat level factor; This analysis examines the impact of different threat level factors on the threat levels of different phased array radar models, presents corresponding conditional probability tables, and establishes a dynamic Bayesian network based on these tables. Specifically: Dynamic Bayesian networks can be defined as follows: ,in This refers to the initial moment. Bayesian network model at time, This refers to the state of the Bayesian network at the next time step. This refers to the transition model between the two, used to represent the change in the state of the Bayesian network between two adjacent time points. The time interval can be set according to the application background requirements. The corresponding prior probability at the initial time is Then the transfer model middle, arrive transition probability Finally obtained Time, and so on; By iterating through the parameter changes of the phased array radar under different working scenarios, inputting them into a dynamic Bayesian network, and outputting the corresponding threat level estimation probability value, a massive amount of experimental data is obtained. The experimental data is processed by knowledge extraction, entity relationships are constructed, and a knowledge graph is established. Establish a query system: Calculate cosine similarity based on input phased array radar information and phased array radar information in the knowledge graph; output relevant phased array radar threat level query results based on the cosine similarity; specifically: Where, vector ,vector The input phased array radar information and the phased array radar information from the knowledge graph are respectively. This represents the dot product of vectors A and B. This represents the magnitude of vector A. This represents the magnitude of vector B; By comparing the threat levels of the output results, the threat ranking of each target in the target work scenario is obtained, as follows: If the cosine similarity is lower than the set threshold, the threat level of the most relevant query result is used as the final threat level estimate; if it is higher than the set threshold, the threat levels of the 5 most relevant targets are averaged and used as the final threat level output.

2. A computer system, characterized in that... include: One or more processors, a computer-readable storage medium for storing one or more programs, wherein, when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method of claim 1.

3. A computer-readable storage medium, characterized in that... The device stores computer-executable instructions, which, when executed, are used to implement the method of claim 1.

4. A computer program product, characterized in that... It includes computer-executable instructions, which, when executed, are used to implement the method of claim 1.

Images