Training method of complex field layered radar anti-interference pattern joint parameter decision network

Abstract

The invention provides a training method of a complex field layered radar anti-interference pattern joint parameter decision network. The method comprises the following steps: constructing a complex field interference feature extraction network; constructing a layered anti-interference decision model HCD2QNet comprising an outer-layer anti-interference style decision network and an inner-layer anti-interference parameter decision network; cascading the complex field interference feature extraction network and the layered anti-interference decision model HCD2QNet to obtain a complex field layered radar anti-interference pattern joint parameter decision network; the method comprises the following steps: receiving an active interference signal transmitted by a jammer, obtaining a radar echo with a plurality of interference signals according to the received signal, and training a complex domain layered radar anti-interference pattern joint parameter decision network according to the radar echo; the obtained complex field layered radar anti-interference pattern joint parameter decision network realizes perception decision integration from original radar echo signals to anti-interference measures, and enhances the flexibility and efficiency of radar anti-interference decision confrontation.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a training method for a complex domain layered radar anti-interference pattern joint parameter decision network. Background Art

[0002] In modern electronic warfare, the interference environment faced by radar systems is complex and changeable. Manual extraction of interference features is difficult to meet the radar's needs for accurate and real-time perception of the battlefield environment, affecting the subsequent radar anti-interference decision-making and confrontation effects; traditional radar anti-interference decision-making methods rely on expert experience to manually design confrontation strategies, with insufficient autonomous decision-making capabilities, difficult to guarantee confrontation performance, and low real-time performance, and are unable to cope with the current flexible and changeable intelligent interference scenarios; existing radar anti-interference decisions are concentrated at the anti-interference style level (or only decide on a single parameter but ignore the anti-interference style), ignoring the impact of anti-interference parameters on the interference effect, destroying the correlation between the anti-interference style and the anti-interference parameters, resulting in poor application of anti-interference strategies in confrontation environments.

[0003] However, the prior art has the following disadvantages:

[0004] 1. Most of the current radar anti-interference decision methods manually extract feature information such as interference bandwidth, power and angle to construct the environment state, and then use Q learning or deep Q network and other methods to obtain the Q value of different anti-interference styles, and decide the final anti-interference style according to the maximum Q value. The shortcomings of this type of method are that the type of manually extracted features depends on expert experience, and the quality of the features is difficult to guarantee, which reduces the accuracy of the anti-interference decision results; the manually extracted feature types are fixed and cannot cope with the current flexible and changeable intelligent interference scenarios, which reduces the radar's autonomous decision-making ability; the manually selected feature types are limited and cannot adapt to complex interference signals, and the perception of the interference environment is low, making it difficult to fully characterize the complex interference confrontation environment; the above input feature deficiencies directly limit the accuracy of the anti-interference decision algorithm.

[0005] 2. Existing radar anti-interference decision methods mainly focus on using intelligent models such as deep Q-networks or decision trees to select anti-interference styles, such as reference [1] Zhao Jiachen, Zhang Jindong, Li Ziyu. Radar intelligent decision generation algorithm based on deep reinforcement learning [J]. Modern Radar, 2022, 44(12): 25-33. Manually extract feature information such as the number, power and type of interference sources, blind source separation anti-interference style in, or optimize a single anti-interference parameter under a fixed anti-interference style, such as reference [2] Y. Fang, S. Wei, L. Zhang, Z. Wu and J. Wu, "Online Emission Policy Selection for Radar Antijamming Using Bandit-Optimized Policy Search," in IEEE Transactions on Aerospace and Electronic Systems, vol. 60, no. 3, pp. 3132-3147, June 2024, doi: 10.1109 / TAES.2024.3358793. Ss. The shortcoming of this type of method is that it only independently decides on the anti-interference style or anti-interference parameters, which destroys the coupling between the two and weakens the effectiveness and flexibility of the radar's autonomous anti-interference decision-making, thereby reducing the radar's interference countermeasure performance and limiting the interference scenarios in which the anti-interference decision-making method can be applied.

[0006] 3. The current radar anti-interference decision-making technology decomposes the interference environment perception (i.e., interference environment feature extraction) and anti-interference decision-making into two independent parts, without considering the correlation between interference feature extraction and anti-interference decision-making, resulting in low efficiency of radar anti-interference decision-making and difficulty in effectively combating time-varying interference environment. Summary of the invention

[0007] In order to solve the above problems existing in the prior art, the present invention provides a training method for a complex domain layered radar anti-interference pattern joint parameter decision network, which specifically includes:

[0008] In a first aspect, the present invention provides a training method for a complex domain layered radar anti-interference pattern joint parameter decision network, comprising:

[0009] S1, constructing a complex domain interference feature extraction network, the complex domain interference feature extraction network includes a complex interference echo real part network branch and a complex interference echo imaginary part network branch, the complex interference echo real part network branch is used to extract echo real part features, and the complex interference echo imaginary part network branch is used to extract echo imaginary part features;

[0010] S2, constructing a hierarchical anti-interference decision model HCD2QNet, which includes an outer anti-interference style decision network and an inner anti-interference parameter decision network. The outer anti-interference style decision network is used to determine the best anti-interference style according to the interference environment state vector output by the complex domain interference feature extraction network, and the inner anti-interference parameter decision network is used to determine the best anti-interference parameters according to the interference environment state vector and the best anti-interference style output by the complex domain interference feature extraction network;

[0011] S3, cascade the complex domain interference feature extraction network and the hierarchical anti-interference decision model HCD2QNet to obtain the complex domain hierarchical radar anti-interference style joint parameter decision network, wherein the output of the complex domain interference feature extraction network is used as the input of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet, and the output of the complex domain interference feature extraction network and the output of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet are used together as the input of the inner anti-interference parameter decision network;

[0012] S4, receives the active jammer signal transmitted by the jammer, obtains the radar echo with the complex jammer signal according to the received signal, and trains the complex domain hierarchical radar anti-jamming style joint parameter decision network according to the radar echo with the complex jammer signal.

[0013] In a second aspect, the present invention further provides a complex domain layered radar anti-interference pattern joint parameter decision method, comprising:

[0014] Receive the active jammer signal and get the radar echo S with complex jammer signal t ′;

[0015] The radar echo S with complex interference signal t ′, input a complex domain layered radar anti-interference pattern joint parameter decision network to obtain an optimal anti-interference pattern and an optimal anti-interference parameter, wherein the complex domain layered radar anti-interference pattern joint parameter decision network is trained by any of the complex domain layered radar anti-interference pattern joint parameter decision network training methods provided in the first aspect;

[0016] Implement the best anti-interference style and best anti-interference parameters to complete the confrontation.

[0017] In a third aspect, the present invention further provides an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0018] Memory, used to store computer programs;

[0019] The processor is used to implement any method provided in the first aspect or the second aspect when executing the program stored in the memory.

[0020] Beneficial effects of the present invention:

[0021] The present invention provides a training method for a complex domain layered radar anti-interference style joint parameter decision network, which comprises a complex domain interference feature extraction network, the complex domain interference feature extraction network comprising a complex interference echo real part network branch and a complex interference echo imaginary part network branch, the complex interference echo real part network branch is used to extract echo real part features, and the complex interference echo imaginary part network branch is used to extract echo imaginary part features; a layered anti-interference decision model HCD2QNet is constructed, the layered anti-interference decision model HCD2QNet comprises an outer anti-interference style decision network and an inner anti-interference parameter decision network, the outer anti-interference style decision network is used to determine an optimal anti-interference style according to an interference environment state vector output by the complex domain interference feature extraction network, and the inner anti-interference parameter decision network is used to determine an optimal anti-interference parameter according to an interference environment state vector output by the complex domain interference feature extraction network and the optimal anti-interference style; the complex domain interference feature extraction network and the layered anti-interference decision model HCD2QNet are cascaded to obtain a complex domain A layered radar anti-interference style joint parameter decision network, wherein the output of the complex domain interference feature extraction network is used as the input of the outer anti-interference style decision network of the layered anti-interference decision model HCD2QNet, and the output of the complex domain interference feature extraction network and the output of the outer anti-interference style decision network of the layered anti-interference decision model HCD2QNet are used together as the input of the inner anti-interference parameter decision network; an active interference signal emitted by a jammer is received, and a radar echo with a complex interference signal is obtained according to the received signal, and a complex domain layered radar anti-interference style joint parameter decision network is trained according to the radar echo with the complex interference signal. The obtained complex domain layered radar anti-interference style joint parameter decision network avoids the separation of interference feature extraction and anti-interference decision steps, retains the correlation between the two, realizes the perception decision integration from the original radar echo signal to the anti-interference measure, enhances the flexibility and efficiency of radar anti-interference decision confrontation, and has the advantage of being widely used in complex time-varying intelligent interference scenarios.

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flowchart of a training method for a complex domain layered radar anti-interference pattern joint parameter decision network provided by the present invention;

[0024] Figure 2 A schematic diagram of the architecture of a complex domain layered radar anti-interference pattern joint parameter decision network provided by the present invention;

[0025] Figure 3 A schematic diagram of a simulation result provided by the present invention. DETAILED DESCRIPTION

[0026] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0027] In view of the deficiencies of the above-mentioned prior art, the present invention proposes a complex domain layered radar anti-interference style joint parameter decision method: constructing a hierarchical model HCD2QNet, deciding the anti-interference style through the outer model, and selecting the anti-interference parameters through the inner model, so as to realize the joint parameter decision of the best anti-interference style; designing a complex domain feature extraction network to fully perceive the complex interference environment characteristics; by cascading the complex domain feature extraction network and the hierarchical model HCD2QNet, the perception and decision integration from the original radar echo signal to the anti-interference measure is realized, so as to avoid the problems of manual extraction of interference signal features, low decision efficiency, low accuracy, poor autonomy and insufficient flexibility in the anti-interference decision process, and can effectively combat a variety of complex radar interference scenarios.

[0028] Figure 1 A flowchart of a training method for a complex domain layered radar anti-interference pattern joint parameter decision network provided by the present invention is shown in FIG. Figure 1 As shown, the method includes:

[0029] S1. Construct a complex domain interference feature extraction network.

[0030] The complex domain interference feature extraction network includes a complex interference echo real part network branch and a complex interference echo imaginary part network branch.

[0031] The complex interference echo real part network branch is used to extract the echo real part features.

[0032] The complex interference echo imaginary part network branch is used to extract the echo imaginary part features.

[0033] like Figure 2 As shown, optionally, the real part of the complex interference echo includes: a first convolutional pooling layer, a second convolutional pooling layer and a first global maximum pooling layer connected in sequence.

[0034] Optionally, the imaginary part of the complex interference echo includes: a third convolutional pooling layer, a fourth convolutional pooling layer, and a second global maximum pooling layer connected in sequence, and feature splicing of the first global maximum pooling layer in the real part of the complex interference echo and the second global maximum pooling layer in the imaginary part of the complex interference echo.

[0035] Optionally, the detailed parameters of each layer of the complex domain interference feature extraction network are shown in Table 1:

[0036] Table 1

[0037]

[0038] The complex domain interference feature extraction network designed by the present invention is used to perceive the interference environment, which overcomes the limitations of the existing anti-interference decision algorithm in manually extracting features and the problem of being unable to adapt to complex interference signals, achieves a full characterization of the interference environment, and improves the accuracy of subsequent radar anti-interference decisions.

[0039] S2. Construct a hierarchical anti-interference decision model HCD2QNet.

[0040] The hierarchical anti-interference decision model HCD2QNet includes an outer anti-interference style decision network and an inner anti-interference parameter decision network.

[0041] The outer anti-interference style decision network is used to determine the best anti-interference style according to the interference environment state vector output by the complex domain interference feature extraction network.

[0042] The inner anti-interference parameter decision network is used to determine the optimal anti-interference parameters according to the interference environment state vector and the optimal anti-interference style output by the complex domain interference feature extraction network.

[0043] like Figure 2 As shown, optionally, the outer anti-interference style decision network includes: a first fully connected layer, a second fully connected layer, a third fully connected layer, a fourth fully connected layer and a first element addition module, wherein the output of the first fully connected layer is the input of the second fully connected layer; the output of the second fully connected layer is the input of the third fully connected layer and the fourth fully connected layer respectively; the output of the third fully connected layer and the output of the fourth fully connected layer are element-added in the first element addition module.

[0044] like Figure 2 As shown, optionally, the inner layer anti-interference parameter decision network includes: a 5th fully connected layer, a 6th fully connected layer, a 7th fully connected layer, an 8th fully connected layer and a second element addition module; wherein the output of the first element addition module and the output of the 5th fully connected layer are the inputs of the 6th fully connected layer; the output of the 6th fully connected layer is the input of the 7th fully connected layer and the 8th fully connected layer respectively; the output of the 7th fully connected layer and the output of the 8th fully connected layer are subjected to element addition operation in the second element addition module.

[0045] Optionally, the detailed parameters of each layer of the hierarchical anti-interference decision model HCD2QNet are shown in Table 2:

[0046] Table 2

[0047]

[0048] The hierarchical anti-interference decision model HCD2QNet constructed by the present invention, in which the outer model decides the anti-interference style, and the inner model synchronously selects the corresponding anti-interference parameters, solves the problem that the existing anti-interference decision method cannot simultaneously decide the anti-interference style and anti-interference parameters, realizes the joint matching decision of the best anti-interference style and anti-interference parameters, and improves the accuracy and confrontation performance of radar confrontation decision.

[0049] S3. Cascade the complex domain interference feature extraction network and the hierarchical anti-interference decision model HCD2QNet to obtain the complex domain hierarchical radar anti-interference style joint parameter decision network.

[0050] Among them, the output of the complex domain interference feature extraction network is used as the input of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet, and the output of the complex domain interference feature extraction network and the output of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet are used together as the input of the inner anti-interference parameter decision network.

[0051] S4. Receive the active interference signal transmitted by the jammer, obtain the radar echo with the complex interference signal according to the received signal, and train the complex domain hierarchical radar anti-interference style joint parameter decision network according to the radar echo with the complex interference signal.

[0052] In a possible implementation, step S4 includes the following steps S41-S43:

[0053] S41, setting the anti-interference reward function of the outer anti-interference style decision network and the anti-interference reward function of the inner anti-interference parameter decision network, the corresponding expressions are:

[0054]

[0055] Among them, R o represents the anti-interference reward function of the outer anti-interference style decision network, J t+1 and J t They represent the interference signals at time t+1 and the current time t, respectively, JSR t+1 and JSR t They represent the interference-to-signal ratio of the interference signal at time t+1 and the current time t, respectively.

[0056] R I =10R ISR (J t ,a t ),

[0057] represents the inner anti-interference parameter decision network anti-interference reward function, R ISR (J t ,at ) represents the interference signal J to the radar t Execute anti-interference style The change rate of interference energy suppression ratio after ISR (J t ,a t ) is calculated as:

[0058]

[0059] Represents the anti-interference action, the anti-interference style executed at the current time t and anti-interference parameters Together they constitute and They respectively represent the maximum amplitude of the residual interference signal and the maximum amplitude of the target signal in the radar echo after the anti-interference action is performed.

[0060] S42. Construct a training loss function.

[0061] Optionally, construct a training loss function, including:

[0062] Construct the training loss function of the outer anti-interference style decision network, and the corresponding expression is:

[0063]

[0064] Among them, L outer represents the training loss function of the outer anti-interference style decision network, represents the anti-interference reward of the outer anti-interference style decision network, and They represent the outputs of the outer anti-interference style decision network at time t and time t+1 respectively, θ represents the parameters of the outer anti-interference style decision network, and γ represents the first discount factor, with a value range of [0,1];

[0065] Construct the training loss function of the inner anti-interference parameter decision network:

[0066]

[0067] Among them, L outer represents the training loss function of the inner anti-interference parameter decision network, represents the inner anti-interference parameter decision network anti-interference reward, and They represent the outputs of the inner anti-interference parameter decision network at time t and time t+1 respectively, σ represents the parameters of the inner anti-interference parameter decision network, and λ represents the second discount factor, with a value range of [0,1].

[0068] S43. Set the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, the training parameters including the number of iterations M, the network learning rate setting, the batch size, and the update frequency interval H of the outer anti-interference style decision network in the layered anti-interference decision model HCD2QNet.

[0069] Exemplarily, the number of iterations M is set to 10000, the network learning rate is set to 0.0002, the batch size is 64, and the update frequency interval H of the outer anti-interference style decision network in the hierarchical anti-interference decision model HCD2QNet is set to 10.

[0070] S44. Based on the anti-interference reward function of the outer anti-interference style decision network, the anti-interference reward function of the inner anti-interference parameter decision network, the training loss function and the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, the active interference signal transmitted by the jammer is received, and according to the received signal, a radar echo with the complex interference signal is obtained, and according to the radar echo with the complex interference signal, the complex domain layered radar anti-interference style joint parameter decision network is trained.

[0071] Optionally, based on the anti-interference reward function of the outer anti-interference style decision network, the anti-interference reward function of the inner anti-interference parameter decision network, the training loss function and the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, an active interference signal transmitted by a jammer is received, and a radar echo with a complex interference signal is obtained according to the received signal, and the complex domain layered radar anti-interference style joint parameter decision network is trained according to the radar echo with the complex interference signal, including:

[0072] Iterate and execute the following steps S81-S84, and execute step S85 once every H times after executing S83, until the set number of iterations M is reached, where H is the update frequency interval of the outer anti-interference style decision network in the set hierarchical anti-interference decision model HCD2QNet;

[0073] S81, receiving the active jamming signal emitted by the jammer at the current time t, and obtaining the radar echo S with the complex jamming signal t , and the radar echo S with complex interference signal t Input the complex domain interference feature extraction network to extract the interference environment state vector V t , and the interference environment state vector V t The outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet is input, and the anti-interference style with the maximum value output by the outer anti-interference style decision network is determined as the optimal anti-interference style this time.

[0074] S82, the interference environment state vector V tThe inner anti-interference parameter decision network in the hierarchical anti-interference decision model HCD2QNet is input jointly with the best anti-interference style this time, and the anti-interference parameter in the maximum value output by the inner anti-interference parameter decision network is determined as the anti-interference parameter index this time, and the best anti-interference parameter is determined according to the best anti-interference style and the anti-interference parameter index this time, as well as the correspondence between the anti-interference style, the anti-interference parameter index and the anti-interference parameter.

[0075] Optionally, the correspondence between the anti-interference style, the anti-interference parameter index and the anti-interference parameter is shown in Table 3:

[0076] Table 3

[0077]

[0078]

[0079] S83, re-receive the active jamming signal transmitted by the jammer, and obtain a new radar echo S with a complex jamming signal according to the received new signal t+1 .

[0080] S84, based on the best anti-interference pattern, the best anti-interference parameters, and the radar echo S with complex interference signals t and the radar echo S with complex interference signal t+1 , determine the network loss value of the training loss function of the inner anti-interference parameter decision network, obtain the first gradient according to the network loss value of the training loss function of the inner anti-interference parameter decision network and the gradient descent algorithm, and back-propagate the first gradient to the inner anti-interference parameter decision network in the hierarchical anti-interference decision model HCD2QNet through the Adam optimization algorithm to update the network parameters.

[0081] S85, radar echo S with complex interference signals according to the best anti-interference pattern this time t and the radar echo S with complex interference signal t+1 , determine the network loss value of the training loss function of the outer anti-interference style decision network, obtain the second gradient according to the network loss value of the training loss function of the outer anti-interference style decision network and the gradient descent algorithm, and back-propagate the second gradient to the outer anti-interference style decision network and the complex domain interference feature extraction network of the hierarchical anti-interference decision model HCD2QNet through the Adam optimization algorithm to update the network parameters.

[0082] The present invention provides a training method for a complex domain layered radar anti-interference style joint parameter decision network, which comprises a complex domain interference feature extraction network, the complex domain interference feature extraction network comprising a complex interference echo real part network branch and a complex interference echo imaginary part network branch, the complex interference echo real part network branch is used to extract echo real part features, and the complex interference echo imaginary part network branch is used to extract echo imaginary part features; a layered anti-interference decision model HCD2QNet is constructed, the layered anti-interference decision model HCD2QNet comprises an outer anti-interference style decision network and an inner anti-interference parameter decision network, the outer anti-interference style decision network is used to determine an optimal anti-interference style according to an interference environment state vector output by the complex domain interference feature extraction network, and the inner anti-interference parameter decision network is used to determine an optimal anti-interference parameter according to an interference environment state vector output by the complex domain interference feature extraction network and the optimal anti-interference style; the complex domain interference feature extraction network and the layered anti-interference decision model HCD2QNet are cascaded to obtain a complex domain A layered radar anti-interference style joint parameter decision network, wherein the output of the complex domain interference feature extraction network is used as the input of the outer anti-interference style decision network of the layered anti-interference decision model HCD2QNet, and the output of the complex domain interference feature extraction network and the output of the outer anti-interference style decision network of the layered anti-interference decision model HCD2QNet are used together as the input of the inner anti-interference parameter decision network; an active interference signal emitted by a jammer is received, and a radar echo with a complex interference signal is obtained according to the received signal, and a complex domain layered radar anti-interference style joint parameter decision network is trained according to the radar echo with the complex interference signal. The obtained complex domain layered radar anti-interference style joint parameter decision network avoids the separation of interference feature extraction and anti-interference decision steps, retains the correlation between the two, realizes the perception decision integration from the original radar echo signal to the anti-interference measure, enhances the flexibility and efficiency of radar anti-interference decision confrontation, and has the advantage of being widely used in complex time-varying intelligent interference scenarios.

[0083] The present invention also provides a flow chart of a complex domain layered radar anti-interference pattern joint parameter decision method, the method comprising the following steps 1-3:

[0084] 1. Receive the active jamming signal emitted by the jammer and obtain the radar echo S with the complex jamming signal t ′.

[0085] 2. The radar echo S with complex interference signal t ′, input the complex domain layered radar anti-interference style joint parameter decision network to obtain the optimal anti-interference style and the optimal anti-interference parameters, the complex domain layered radar anti-interference style joint parameter decision network is trained by any of the complex domain layered radar anti-interference style joint parameter decision networks provided in the above embodiments.

[0086] 3. Implement the best anti-interference style and best anti-interference parameters to complete the confrontation.

[0087] The effect of the present invention can be further illustrated by the following simulation experiment.

[0088] 1. Simulation conditions

[0089] The hardware platform of the simulation experiment of the present invention is: Intel(R) Core(TM) i7-10700 CPU, 2.90GHz, 64G memory, and NVIDIA GeForce RTX 3090 GPU.

[0090] The software platform for the simulation experiment of the present invention is: Pycharm2021.

[0091] 2. Simulation results and analysis

[0092] Under the above simulation conditions, the present invention is used to conduct 5000 countermeasures against radar echoes interfered by linear frequency shift, and the results of HCD2QNet's decision on the best anti-interference pattern and anti-interference parameters are as follows: Figure 3 And as shown in Table 4. Figure 3 Figures (a) and (b) are the profit curves of the outer anti-interference decision network and the inner anti-interference parameters, respectively. Table 4 shows the anti-interference styles and specific parameters decided by 5000 confrontations.

[0093] Table 4

[0094]

[0095] from Figure 3 It can be seen that after 1000 confrontations, HCD2QNet can quickly achieve high-return decisions, that is, select anti-interference styles and parameters with high reward values. The results in Table 4 show that HCD2QNet can accurately decide the best anti-interference style and anti-interference parameters.

[0096] In summary, the complex domain layered radar anti-interference style joint parameter decision method provided by the present invention realizes the perception decision integration from the original radar echo signal to the anti-interference measure. The outer layer interference style decision network in the layered anti-interference decision model HCD2QNet selects the interference style, and the inner layer anti-interference parameter decision network selects the anti-interference parameter, so as to realize the joint decision of the best anti-interference style and parameters, thereby improving the accuracy, efficiency and flexibility of radar anti-interference decision.

[0097] The present invention can be applied to various radar platforms such as synthetic aperture and frequency agility, and can make efficient and accurate anti-interference pattern and anti-interference parameter decisions for various types of interference signals, thereby improving the flexibility and intelligence of radar interference countermeasures. Currently, it has been tested in airborne measured synthetic aperture radar interference echo data. The present invention can automatically decide the best anti-interference pattern and parameter combination according to the interference type and interference parameters, guide the radar to complete the anti-interference processing process, and ensure the radar imaging and detection requirements after anti-interference, which proves the effectiveness and practicality of the method of the present invention.

[0098] The present invention also provides a structure of an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus.

[0099] Memory, used to store computer programs;

[0100] The processor is used to implement the steps provided in the above method embodiment when executing the program stored in the memory.

[0101] The communication interface is used for communication between the above electronic device and other devices.

[0102] As for the electronic device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the specific contents and beneficial effects and other related matters can be referred to the partial description of the method embodiment.

[0103] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0104] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A training method for a complex domain layered radar anti-interference pattern joint parameter decision network, characterized in that: include: S1, constructing a complex domain interference feature extraction network, wherein the complex domain interference feature extraction network includes a complex interference echo real part network branch and a complex interference echo imaginary part network branch, wherein the complex interference echo real part network branch is used to extract echo real part features, and the complex interference echo imaginary part network branch is used to extract echo imaginary part features; S2, constructing a hierarchical anti-interference decision model HCD2QNet, wherein the hierarchical anti-interference decision model HCD2QNet includes an outer anti-interference style decision network and an inner anti-interference parameter decision network, wherein the outer anti-interference style decision network is used to determine the optimal anti-interference style according to the interference environment state vector output by the complex domain interference feature extraction network, and the inner anti-interference parameter decision network is used to determine the optimal anti-interference parameters according to the interference environment state vector output by the complex domain interference feature extraction network and the optimal anti-interference style; S3, cascading the complex domain interference feature extraction network and the hierarchical anti-interference decision model HCD2QNet to obtain a complex domain hierarchical radar anti-interference style joint parameter decision network, wherein the output of the complex domain interference feature extraction network is used as the input of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet, and the output of the complex domain interference feature extraction network and the output of the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet are used together as the input of the inner anti-interference parameter decision network; S4, receiving the active jamming signal transmitted by the jammer, obtaining a radar echo with the complex jamming signal according to the received signal, and training the complex domain layered radar anti-jamming style joint parameter decision network according to the radar echo with the complex jamming signal.

2. The training method according to claim 1, characterized in that: The real part of the complex interference echo includes: The first convolutional pooling layer, the second convolutional pooling layer and the first global maximum pooling layer are connected in sequence.

3. The training method according to claim 2, characterized in that: The imaginary part of the complex interference echo includes: The third convolutional pooling layer, the fourth convolutional pooling layer, and the second global maximum pooling layer are connected in sequence, and the features of the first global maximum pooling layer in the real part of the complex interference echo and the second global maximum pooling layer in the imaginary part of the complex interference echo are spliced.

4. The training method according to claim 3, characterized in that: The outer anti-interference style decision network includes: The first fully connected layer, the second fully connected layer, the third fully connected layer, the fourth fully connected layer and the first element addition module; Among them, the output of the first fully connected layer is the input of the second fully connected layer; the output of the second fully connected layer is the input of the third fully connected layer and the fourth fully connected layer respectively; the output of the third fully connected layer and the output of the fourth fully connected layer are element-added in the first element addition module.

5. The training method according to claim 4, characterized in that: The inner layer anti-interference parameter decision network includes: The 5th fully connected layer, the 6th fully connected layer, the 7th fully connected layer, the 8th fully connected layer and the second element addition module; Among them, the output of the first element addition module and the output of the 5th fully connected layer are the input of the 6th fully connected layer; the output of the 6th fully connected layer is the input of the 7th fully connected layer and the 8th fully connected layer respectively; the output of the 7th fully connected layer and the output of the 8th fully connected layer are subjected to element addition operation in the second element addition module.

6. The method according to claim 5, characterized in that The S4 comprises: The anti-interference reward function of the outer anti-interference style decision network and the anti-interference reward function of the inner anti-interference parameter decision network are set, and the corresponding expressions are: Among them, R o represents the anti-interference reward function of the outer anti-interference style decision network, J t+1 and J t They represent the interference signals at time t+1 and the current time t, respectively, JSR t+1 and JSR t They represent the interference-to-signal ratio of the interference signal at time t+1 and the current time t, respectively. R I =10R ISR (J t ,a t ), represents the inner anti-interference parameter decision network anti-interference reward function, R ISR (J t ,a t ) represents the interference signal J to the radar t Execute anti-interference style The change rate of interference energy suppression ratio after ISR (J t ,a t ) is calculated as: Represents the anti-interference action, the anti-interference pattern a executed at the current time t Ot and anti-interference parameters Together they constitute and They respectively represent the maximum amplitude of the residual interference signal and the maximum amplitude of the target signal in the radar echo after the anti-interference action is performed; Construct training loss function; Setting the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, wherein the training parameters include the number of iterations M, the network learning rate setting, the batch size, and the update frequency interval H of the outer anti-interference style decision network in the layered anti-interference decision model HCD2QNet; Based on the anti-interference reward function of the outer anti-interference style decision network, the anti-interference reward function of the inner anti-interference parameter decision network, the training loss function and the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, an active interference signal transmitted by a jammer is received, and a radar echo with a complex interference signal is obtained according to the received signal; and the complex domain layered radar anti-interference style joint parameter decision network is trained according to the radar echo with the complex interference signal.

7. The method according to claim 6, characterized in that The constructing of the training loss function includes: The training loss function of the outer anti-interference style decision network is constructed, and the corresponding expression is: Among them, L outer represents the training loss function of the outer anti-interference style decision network, represents the anti-interference reward of the outer anti-interference style decision network, and They represent the outputs of the outer anti-interference style decision network at time t and time t+1 respectively, θ represents the parameters of the outer anti-interference style decision network, and γ represents the first discount factor, with a value range of [0,1]; Construct the training loss function of the inner layer anti-interference parameter decision network: Among them, L outer represents the training loss function of the inner anti-interference parameter decision network, represents the inner anti-interference parameter decision network anti-interference reward, and They represent the outputs of the inner anti-interference parameter decision network at time t and time t+1 respectively, σ represents the parameters of the inner anti-interference parameter decision network, and λ represents the second discount factor, with a value range of [0,1].

8. The method according to claim 7, characterized in that The method comprises: receiving an active interference signal transmitted by a jammer based on the anti-interference reward function of the outer anti-interference style decision network, the anti-interference reward function of the inner anti-interference parameter decision network, the training loss function and the training parameters of the complex domain layered radar anti-interference style joint parameter decision network, obtaining a radar echo with a complex interference signal according to the received signal, and training the complex domain layered radar anti-interference style joint parameter decision network according to the radar echo with the complex interference signal, including: Iterate and execute the following steps S81-S84, and execute step S85 once every H times after executing S83, until the set number of iterations M is reached, where H is the update frequency interval of the outer anti-interference style decision network in the hierarchical anti-interference decision model HCD2QNet; S81, receiving the active jamming signal emitted by the jammer at the current time t, and obtaining the radar echo S with the complex jamming signal t , and the radar echo S with complex interference signal t Input the complex domain interference feature extraction network to extract the interference environment state vector V t , and the interference environment state vector V t Input the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet, and determine the anti-interference style in the maximum value output by the outer anti-interference style decision network as the best anti-interference style for this time; S82, the interference environment state vector V t The inner anti-interference parameter decision network in the hierarchical anti-interference decision model HCD2QNet is inputted jointly with the best anti-interference pattern this time, and the anti-interference parameter in the maximum value output by the inner anti-interference parameter decision network is determined as the anti-interference parameter index this time, and the best anti-interference parameter is determined according to the best anti-interference pattern this time and the anti-interference parameter index this time, as well as the correspondence between the anti-interference pattern, the anti-interference parameter index and the anti-interference parameter; S83, re-receive the active jamming signal transmitted by the jammer, and obtain a new radar echo S with a complex jamming signal according to the received new signal t+1 ; S84, according to the best anti-interference pattern, the best anti-interference parameters, and the radar echo S with complex interference signals t and the radar echo S with complex interference signal t+1 , determine the network loss value of the training loss function of the inner anti-interference parameter decision network, obtain a first gradient according to the network loss value of the training loss function of the inner anti-interference parameter decision network and the gradient descent algorithm, back-propagate the first gradient to the inner anti-interference parameter decision network in the hierarchical anti-interference decision model HCD2QNet through the Adam optimization algorithm, and update the network parameters; S85, according to the best anti-interference pattern of this time, the radar echo S with complex interference signals t and the radar echo S with complex interference signal t+1 , determine the network loss value of the training loss function of the outer anti-interference style decision network, obtain a second gradient according to the network loss value of the training loss function of the outer anti-interference style decision network and the gradient descent algorithm, and back-propagate the second gradient to the outer anti-interference style decision network of the hierarchical anti-interference decision model HCD2QNet and the complex domain interference feature extraction network through the Adam optimization algorithm to update the network parameters.

9. A complex domain layered radar anti-interference pattern joint parameter decision method, characterized in that: include: Receive the active jammer signal and get the radar echo S with complex jammer signal t ′; The radar echo S with complex interference signal t ′, input a complex domain layered radar anti-interference pattern joint parameter decision network to obtain an optimal anti-interference pattern and an optimal anti-interference parameter, wherein the complex domain layered radar anti-interference pattern joint parameter decision network is trained by the training method of the complex domain layered radar anti-interference pattern joint parameter decision network as described in any one of claims 1-8; The optimal anti-interference style and the optimal anti-interference parameters are implemented to complete the confrontation.

10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing any of the methods described in claims 1-9 when executing a program stored in a memory.

Images