CN117407319B - Target generation method and device for radar system software test and electronic equipment - Google Patents
Target generation method and device for radar system software test and electronic equipment Download PDFInfo
- Publication number
- CN117407319B CN117407319B CN202311723807.3A CN202311723807A CN117407319B CN 117407319 B CN117407319 B CN 117407319B CN 202311723807 A CN202311723807 A CN 202311723807A CN 117407319 B CN117407319 B CN 117407319B
- Authority
- CN
- China
- Prior art keywords
- target
- time step
- motion
- jth
- simulation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000004088 simulation Methods 0.000 claims abstract description 100
- 238000013522 software testing Methods 0.000 claims abstract description 45
- 238000009826 distribution Methods 0.000 claims abstract description 31
- 238000005295 random walk Methods 0.000 claims description 36
- 238000012549 training Methods 0.000 claims description 27
- 101100369993 Mus musculus Tnfsf10 gene Proteins 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 239000000203 mixture Substances 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 4
- 230000004044 response Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 abstract description 71
- 230000006870 function Effects 0.000 description 11
- 230000008569 process Effects 0.000 description 10
- 238000001514 detection method Methods 0.000 description 7
- 230000001133 acceleration Effects 0.000 description 5
- 230000010354 integration Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000013179 statistical model Methods 0.000 description 3
- 206010063385 Intellectualisation Diseases 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003252 repetitive effect Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000013515 script Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3672—Test management
- G06F11/3684—Test management for test design, e.g. generating new test cases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3672—Test management
- G06F11/3688—Test management for test execution, e.g. scheduling of test suites
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The disclosure provides a target generation method and device for radar system software testing, and electronic equipment, wherein the method comprises the following steps: acquiring attribute types of targets to be generated, wherein the attribute types comprise a target type, a position, a speed type and a motion mode; generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder; generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information; and generating a motion trail of the simulation target in the motion mode based on the initial position and the position information. According to the scheme, the generated simulation target is closer to the attribute distribution of the real target through generating the attributes such as the speed and the position of the simulation target, and the generated simulation target has the real motion characteristic in the test scene through generating the motion trail of the simulation target.
Description
Technical Field
The disclosure relates to the technical field of radar testing, and in particular relates to a target generation method and device for radar system software testing and electronic equipment.
Background
Radar systems are an important component in modern military, and are used to perform a variety of functions such as detecting, tracking, and identifying targets. The software in the radar system can realize the intellectualization and automation of the radar system, and improve the efficiency and accuracy of the radar system. Many complex algorithms and models exist in software in a radar system, and if the software algorithms are unstable, the detection, tracking and identification capabilities of the radar system are reduced, so that the performance and accuracy of the radar system are affected. Therefore, radar system software testing is becoming increasingly important to ensure that the software quality of the radar system meets prescribed standards and requirements.
At present, in the radar system software testing process, the problem of insufficient testing data exists, mainly because the testing data of the radar system are often required to be obtained in actual use, and the testing data in actual use are often difficult to obtain.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a target generation method, apparatus, and electronic device for radar system software testing that are capable of generating realistic and diversified targets for testing and evaluating radar software.
In a first aspect, an embodiment of the present disclosure provides a target generation method for radar system software testing, the method including:
acquiring attribute types of targets to be generated, wherein the attribute types comprise a target type, a position, a speed type and a motion mode;
generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder;
generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information;
and generating a motion trail of the simulation target in the motion mode based on the initial position and the position information.
In a second aspect, embodiments of the present disclosure provide a target generating apparatus for radar system software testing, the apparatus comprising:
the target track generation module is used for acquiring attribute types of targets to be generated, wherein the attribute types comprise a target type, a position, a speed type and a motion mode; generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder;
The motion trail generation module is used for generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information; and generating a motion trail of the simulation target in the motion mode based on the initial position and the position information.
In a third aspect, an embodiment of the present disclosure provides an electronic device, including:
a processor; and
a memory in which a program is stored,
wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the steps of the target generation method for radar system software testing according to the first aspect.
In a fourth aspect, embodiments of the present disclosure provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the target generation method for radar system software testing according to the first aspect.
According to the target generation method, the target generation device and the electronic equipment for the radar system software test, the attribute type of the target to be generated is obtained, the attribute type comprises the target type, the position, the speed type and the motion mode, the condition-variable automatic encoder is used for generating the simulation target meeting the target type, the initial position and the speed information of the simulation target, further, the position information corresponding to different time steps in the motion mode is generated based on the initial position and the speed information, and the motion track of the simulation target in the motion mode is generated based on the initial position and the position information. By adopting the scheme, the generated simulation targets are more close to the attribute distribution of the real targets by generating the attributes such as the speed, the position and the like of the simulation targets, and the generated simulation targets have more real motion characteristics in a test scene by generating the motion trail of the simulation targets, so that accurate, vivid and diversified target data are provided for the test of radar software, the performance of the software is better comprehensively evaluated and verified, and the method is beneficial to improving the design and algorithm of a radar system and improving the reliability and accuracy of the radar system in practical application.
Drawings
FIG. 1 is a flow chart of a target generation method for radar system software testing according to an exemplary embodiment of the present disclosure;
FIG. 2 is a flow diagram of a target generation method for radar system software testing according to another exemplary embodiment of the present disclosure;
FIG. 3 is a flow chart of a target generation method for radar system software testing according to yet another exemplary embodiment of the present disclosure;
fig. 4 is a schematic structural view of a target generating apparatus for radar system software testing according to an exemplary embodiment of the present disclosure.
Detailed Description
In order to make the objects, technical solutions and advantages of the present disclosure more apparent, the present disclosure will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present disclosure.
Radar systems are an important component in modern military, and are used to perform a variety of functions such as detecting, tracking, and identifying targets. While software in radar systems plays a very important role, they are responsible for various functions in radar systems, such as system control, signal processing, target tracking, etc. The software in the radar system can realize the intellectualization and automation of the radar system, and improve the efficiency and accuracy of the radar system. Meanwhile, the software can also realize the networking of the radar system and realize the joint combat of a plurality of radar systems. With the continuous development of radar technology, the software proportion of radar systems is gradually increased, and the software proportion is an indispensable part in radar systems. According to statistics, the proportion of software in the radar system at present reaches more than 50%, and the proportion of software is increased along with the continuous expansion of the functions of the radar system.
Many complex algorithms and models exist in software in a radar system, and if the software algorithms are unstable, the detection, tracking and identification capabilities of the radar system are reduced, so that the performance and accuracy of the radar system are affected. If the software has loopholes and potential safety hazards, the radar system is easy to attack and destroy, so that the execution and the result of military operations are affected. Therefore, the radar system software test is more and more important, and the software quality of the radar system is ensured to reach the specified standard and requirement.
The software test in the radar field has remarkable development in the aspects of automation, simulation, environment simulation, security test and application of agile and DevOps methods. These trends help to improve the quality, reliability and safety of radar systems and meet changing demands. Testing of radar system software typically requires simulation of different environmental and target conditions. The development of simulation and emulation technology enables testers to better simulate various scenes and conditions and carry out comprehensive tests on the radar system. Through simulation and emulation, indexes such as performance, detection range, target recognition capability and the like of the radar system can be effectively verified. The current radar system software testing method mainly comprises the following steps:
1) Automated testing: in order to improve the test efficiency and accuracy, the automatic test is widely applied to the radar system software test. By writing test scripts and using an automatic test tool, a large number of test cases can be automatically executed, and detailed test reports are generated, so that time and labor cost can be saved, and the coverage rate and consistency of the test can be improved.
2) Multilevel testing strategy: the radar system software test adopts a multi-level test strategy, including unit test, integrated test, system test, acceptance test and the like. The unit test is mainly aimed at testing each module of the software, so that the normal function of each module is ensured. The integration test tests interactions and collaboration between different modules. The system test is to comprehensively test the whole radar system, including the aspects of function, performance, stability and the like. The acceptance test is the last gateway to ensure that the software meets the user requirements and specification requirements.
3) High reliability and safety requirements: testing of radar system software requires a high degree of reliability and safety. Radar systems have very high demands on accuracy and stability of the system in military and aerospace applications. Accordingly, test engineers are required to design and execute various test cases to ensure that software can function properly under various complex environments and emergency conditions.
4) The data-driven test method comprises the following steps: in the radar system software test, a data-driven test method is widely adopted. By using various real and simulated data, a test engineer can verify the performance and accuracy of the software under different data inputs. This approach can help discover potential problems and vulnerabilities and optimize and improve software.
5) Continuous integration and continuous testing: to ensure the quality and stability of the software, radar system software testing employs a method of continuous integration and continuous testing. The continuous integration means that the codes of the developer are continuously integrated into a backbone code base, and automatic construction and testing are performed. The continuous test is to perform continuous test and verification by an automatic test tool on the basis of continuous integration, and discover and repair problems in time.
Currently, some bottlenecks and problems are also encountered in the radar system software testing process, which mainly comprise the following aspects:
1. insufficient test data
Test data is the basis of the test, but in radar system software testing, the test data is often difficult to obtain. This is mainly because the test data of the radar system often needs to be acquired in actual use, which is often difficult to acquire.
2. Unstable test environment
The stability of the test environment has a great influence on the accuracy of the test results. In the radar system software test, the test environment is often complex, including multiple aspects of radar system hardware, software, network, etc., which makes the stability of the test environment difficult to ensure.
3. Defective test tool
Test tools are an important means of testing, but in radar system software testing, test tools are often difficult to obtain. This is mainly because the test tools of radar systems often require specialized equipment and software support, which tend to be relatively expensive.
4. Generating object types and insufficient scenes
The targets generated by the random generation algorithm may lack realism and fidelity and may not fully simulate complex scenes. The physical model algorithm is complex to implement, requiring more computing resources and an accurate physical model. Statistical model algorithms require accurate statistical models and adequate data support, and may present challenges for specific target types.
In order to better improve the software quality of the radar system in China and cover more target types in the test process, the invention provides a target generation method for the radar system software test, the scheme can be used for simulating targets in the environment where the radar system is located, and the detection, tracking and positioning capabilities of the software under different target conditions can be evaluated by generating targets with different types, sizes, speeds and movement modes; various scenarios and situations can be created to test the functionality and performance of radar software under different target conditions. The scheme of the present disclosure is mainly applicable to the following aspects:
1) Target detection test: targets of different sizes and shapes are generated and placed in the working area of the radar, then the target detection algorithm of the radar software is tested, and whether the software can accurately detect the targets and give the correct position and characteristic information is observed.
2) Target tracking test: by using the scheme, moving targets comprising targets with different speeds and moving modes can be generated, the targets are introduced into a radar monitoring area, a target tracking algorithm of software is tested, and the tracking performance of the software is evaluated, wherein the tracking performance comprises continuous tracking, motion prediction and track estimation capability of the targets.
3) Multi-objective treatment test: by using the scheme, a plurality of targets can be generated, and the processing capacity of the software for the plurality of targets is tested, including detecting, tracking and identifying the plurality of targets at the same time.
4) Target characteristic test: targets with different characteristics, such as different radar reflection sections, different speed ranges, different acceleration changes and the like, are generated, and the performance and the accuracy of the targets in a complex scene are evaluated through the processing capacity of test software for the characteristics of the different targets.
3) Target movement pattern test: according to the scheme, different target motion modes such as linear motion, circular motion and the like are simulated, and the adaptability and the accuracy of the test software to different motion modes are evaluated according to the detection and tracking capabilities of the test software to the targets in the different motion modes.
The method and the device for generating the target for testing the radar system software and the electronic equipment are explained in detail below with reference to the accompanying drawings.
Fig. 1 is a flowchart of a target generating method for radar system software testing according to an exemplary embodiment of the present disclosure, where the method may be performed by a target generating apparatus for radar system software testing provided by the embodiments of the present disclosure, and the target generating apparatus for radar system software testing may be implemented in software and/or hardware and may be integrated in an electronic device, where the electronic device may be a computer or other device.
As shown in fig. 1, the target generation method for radar system software testing includes the following steps:
step 101, obtaining attribute types of an object to be generated, wherein the attribute types comprise an object type, a position, a speed type and a motion mode.
For example, a user interface may be provided in the electronic device for a user to select a type of attribute of the object to be generated, including, but not limited to, a type of object, a location of the object, a type of speed of the object, a movement pattern, a size of the object, a radar cross section, and so forth. Wherein the target type may be, for example, pedestrians, vehicles, spheres, dots, etc.; the speed type of the target may include speed, acceleration, angular velocity, etc.; the motion pattern may be, for example, uniform linear motion, uniform circular motion, uniform acceleration motion, and the like.
In the embodiment of the disclosure, the attribute type of the target to be generated can be set by the user, so that the electronic equipment can acquire the attribute type of the target to be generated.
Step 102, generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder.
Wherein, according to the different speed types set by the user, the generated speed information of the simulation target is also different. For example, if the speed type set by the user is acceleration, the generated speed information of the simulation target is acceleration information; the speed type set by the user is speed, and the generated speed information of the simulation target is speed information; and if the speed type set by the user is angular speed, the generated speed information of the simulation target is angular speed information.
In the embodiment of the disclosure, after the electronic device acquires the attribute type of the object to be generated, a condition-variable automatic encoder (Conditional Variational Auto Encoder, CVAE) may be utilized to generate the simulation object satisfying the acquired object type and the initial position and speed information of the simulation object.
CVAE is a generation model based on an automatic encoder (Auto encoder). Unlike the conventional automatic encoder, the conditional variance automatic encoder models the data generation process through a conditional probability distribution, thereby enabling more efficient generation of data having specific conditions.
The model structure of the CVAE is similar to that of a conventional variational automatic encoder (Variational Auto Encoder, VAE), but the CVAE models the data generation process by a conditional probability distribution. Specifically, CThe VAE hypothesis data is generated from an unnoticed continuous random variable z, which obeys an a priori distribution P θ (z), and the data generation process may be performed by a conditional probability distribution P θ (X|z) modeling. Where X represents the observed data and θ represents the model parameters. In the model structure of CVAE, there is one encoder and one decoder. Wherein the encoder portion attempts to learn q θ (z|x), corresponding to the hidden representation x of the learning data or x encoded into the hidden (probability encoder) representation; the decoder section attempts to learn P θ (x|z) decoding hidden representation input space.
In training the CVAE model, two loss functions need to be minimized: reconstruction loss and KL divergence loss. Wherein the reconstruction loss is the expected negative logarithmic likelihood of the observed data, namely:
reconstruction loss = -logq θ (z|x)+logP θ (X|z)。
KL divergence loss is q θ KL divergence between (z|x) and p (z), namely:
KL divergence loss = d_ { KL } (q θ (z|x)||p(z))。
The objective function of the entire training process can be expressed as:
objective function = reconstruction loss + KL divergence loss.
By minimizing the objective function described above, the CVAE model can be learned to a more accurate conditional probability distribution and a more efficient data generation process when training the CVAE. In the target generation process of the scheme, the CVAE is applied to learn the distribution mode of the target attribute and generate the vivid target attribute.
In this embodiment, with the trained CVAE, a simulation target satisfying the target type may be generated, and initial position and speed information of the simulation target may be generated.
For example, the user sets the type of the object to be generated as a pedestrian, and also sets the position of the object to be generated, the speed as an angular speed, and the motion mode as uniform circular motion. The CVAE can be used for generating a virtual pedestrian as a simulation target, and generating the initial position of the pedestrian and the angular velocity information of the pedestrian doing uniform circular motion.
It should be noted that, the present disclosure is explained by using CVAE to generate a simulation target and related attribute information of the simulation target, but not limited to the present disclosure, and in practical application, attribute information of the simulation target may be generated by using a statistical model, a physical model, or a data driving method, so as to ensure that the generated attribute can truly reflect the characteristics of the target.
And step 103, generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information.
In the embodiment of the present disclosure, after the initial position and speed information of the generated simulation target are determined, position information corresponding to different time steps in the corresponding motion mode may be generated based on the initial position and speed information.
As an example, when the motion pattern is uniform linear motion, the position information corresponding to the t-th time step may be determined by the following formula, assuming that the position information corresponding to the t-th time step is denoted as (x, y)
x=x 0 +Vx*t；
y=y 0 +Vy*t；
Wherein, (x) 0 ,y 0 ) For the initial position, vx is the velocity of the generated simulation target in the x direction, vy is the velocity of the generated simulation target in the y direction.
As another example, when the motion pattern is uniform circular motion, the position information corresponding to the t-th time step may be determined by the following formula, assuming that the position information corresponding to the t-th time step is denoted as (x, y)
x=x 0 +R*cos(ωt)；
y=y 0 +R*sin(ωt)；
Wherein, (x) 0 ,y 0 ) For the initial position, ω is the angular velocity of the generated simulation target, and R is the radius of the circular motion of the simulation target.
And 104, generating a motion track of the simulation target in the motion mode based on the initial position and the position information.
In the embodiment of the disclosure, after the position information of the simulation target corresponding to different time steps is determined, the motion trail of the simulation target in the corresponding motion mode can be generated based on the initial position and the position information corresponding to different time steps.
As an example, assuming that the simulated environment is a non-interference environment, the initial position and the position information of the subsequent different time steps may be directly determined as the motion trajectory of the simulation target. Assuming that the position information of each of the N time steps is determined to be X1, X2,..and XN, the motion trajectory of the simulation target may be expressed as x= { X1, X2,., XN }.
As another example, a disturbance may be added to the position information corresponding to each time step, so as to obtain a new position corresponding to each time step, and the new position corresponding to each time step is determined as the motion trail of the simulation target. For example, assume that the disturbance ε n Obeys a normal distribution N (0, sigma) n ) Wherein sigma n Is disturbance epsilon n Assuming that the position information of each of the N time steps is determined to be X1, X2,..xn, where X1 is the initial position and XN is the end position, the new position to which the disturbance is added can be expressed as: xi' =xi+epsilon_i (i=1, 2,3,., N), where epsilon_i is a normal-compliant distribution N (0, σ) randomly generated for the ith time step n ) The motion trajectory of the simulated target may be expressed as x= { X1', X2',..x, XN ' }.
As another example, a new position corresponding to the current time step may be determined from the position information of the previous time step and the random disturbance of the current time step, and the new position corresponding to each time step may be further determined as the motion trajectory of the simulation target. For example, assume that the disturbance ε n Obeys a normal distribution N (0, sigma) n ) Wherein sigma n Is disturbance epsilon n Assuming that the position information of each of the N time steps is determined to be X1, X2,..xn, where X1 is the initial position and XN is the end position, the new position to which the disturbance is added can be expressed as: xi' =x (i-1) +epsilon_i (i=2, 3., N), where epsilon_i is a normal-compliant distribution N (0, σ) randomly generated for the ith time step n ) Disturbance of (2)The motion trajectory of the variable, simulation target, may be expressed as x= { X1, X2',..and XN' }.
As yet another example, the motion profile of the simulated target may be determined based on the initial position and the random disturbance. For example, assume that the disturbance ε n Obeys a normal distribution N (0, sigma) n ) Wherein sigma n Is disturbance epsilon n If the initial position is x_1, then the position information for the different time steps can be expressed as x_i=x_ { i-1} +ε_i (i=2, 3,.., N), ε_i being a normal-compliant distribution N (0, σ) randomly generated for the i-th time step n ) X_i (i=2, 3,., N) are intermediate positions that are generated from a previous position plus a random disturbance, and the motion trajectory of the simulation target can be expressed as x= { x_1, x_2,., x_n }.
In the disclosed embodiments, since the disturbance is a random variable, the location of the different time steps generated by adding the random disturbance is also a series of random variables, so the motion profile generated each time may be different. At the same time, the disturbance epsilon can be adjusted according to the needs n The standard deviation of the simulation target is changed to change the pattern of the motion trail of the simulation target, so that the pattern is more in line with the behavior characteristics of the real target, the authenticity of the simulation target is improved, and the simulation target has real motion characteristics in a test scene.
According to the target generation method for radar system software testing, the attribute type of the target to be generated is obtained, the attribute type comprises the target type, the position, the speed and the motion mode, the condition variation automatic encoder is utilized to generate the simulation target meeting the target type, the initial position and the speed information of the simulation target, further, the position information corresponding to different time steps in the motion mode is generated based on the initial position and the speed information, and the motion track of the simulation target in the motion mode is generated based on the initial position and the position information. By adopting the scheme, the generated simulation targets are more close to the attribute distribution of the real targets by generating the attributes such as the speed, the position and the like of the simulation targets, and the generated simulation targets have more real motion characteristics in a test scene by generating the motion trail of the simulation targets, so that accurate, vivid and diversified target data are provided for the test of radar software, the performance of the software is better comprehensively evaluated and verified, and the method is beneficial to improving the design and algorithm of a radar system and improving the reliability and accuracy of the radar system in practical application.
In an alternative embodiment of the present disclosure, a random walk model may be pre-trained, the network parameters of which include the normal distribution N (0, σ) to which the disturbance variable is subjected n ) Standard deviation sigma of n And the standard deviation lambda of the normal distribution N (0, lambda) obeyed by the wind variable, and the pattern of the generated motion trail of the simulation target is changed by adding disturbance and wind factors, so that the accuracy and the reality of the motion trail are ensured. Specifically, as shown in fig. 2, the training process of the random walk model includes the following steps:
step 201, a training sample is obtained, wherein the training sample comprises simulation positions and real target tracks corresponding to N time steps respectively, the number of the time steps of the real target track is N, and N is a positive integer.
In the embodiment of the disclosure, a plurality of training samples may be obtained for training a random walk model, where each training sample includes N simulation positions corresponding to N time steps and a real target track, and the number of time steps of the real target track is also N.
Step 202, inputting the simulation positions corresponding to the N time steps into a random walk model to obtain a virtual target track, wherein the network parameters of the random walk model comprise a normal distribution N (0, sigma) obeyed by disturbance variables n ) Standard deviation sigma of n And the standard deviation lambda of the normal distribution N (0, lambda) to which the wind variable is subjected.
In embodiments of the present disclosure, the simulated positions corresponding to the N time steps, respectively, are input into a random walk model, which may be based on current network parameters (i.e., the standard deviation σ of the disturbance n And standard deviation lambda of wind force) to generate a virtual target track.
For example, the virtual target trajectory may be expressed as:
X_i=X_{i-1}+ε_i+f_i。
where X_i is the simulated position of the ith time step, X_ { i-1} is the simulated position of the previous time step, ε_i is the disturbance of the ith time step, and f_i is the wind force of the ith time step.
Step 203 calculates an error between the virtual target trajectory and the real target trajectory.
In the embodiment of the disclosure, in order to optimize the pattern of the target track, parameters of the random walk model, such as disturbance epsilon, can be adjusted n Standard deviation sigma corresponding to wind power n And lambda. In particular, the optimal σ can be solved by minimizing the error between the virtual target trajectory and the real target trajectory n And lambda. Assuming that the real target trajectories are y_1, y_2,..y_n, the error between the virtual target trajectory and the real target trajectory can be expressed as:
E=sum((X_i-Y_i))/N；
wherein X_i is the simulation position of the ith time step in the virtual target track, and Y_i is the simulation position of the ith time step in the real target track.
Step 204, updating the standard deviation sigma based on the error and the learning rate corresponding to the current training round number by adopting a preset optimization algorithm n And the standard deviation lambda is used for obtaining the trained random walk model until the error between the virtual target track output by the random walk model and the corresponding real target track reaches a preset error threshold value.
The preset optimization algorithm may be an optimization algorithm commonly used at present, such as a gradient descent algorithm.
In the embodiment of the disclosure, after determining the error between the virtual target track and the real target track, a preset optimization algorithm can be adopted, and the standard deviation sigma is updated based on the learning rate corresponding to the error and the current training wheel number n And standard deviation lambda to solve for the optimal sigma n And lambda until the error between the virtual target track output by the random walk model and the corresponding real target track reaches a preset error threshold value, and obtaining the optimal sigma at the moment n And lambda, obtaining a trained random walk model.
Wherein the false isLet the disturbance standard deviation of the current optimum be sigma n The current optimal standard deviation of wind power is lambda (i.e. the current network parameter of the random walk model), and the updated standard deviation sigma n And standard deviation λ is expressed as:
σ n =σ n -learning_rate*(partial_derivative_E_with_respect_to_σ n )；
λ=λ-learning_rate*(partial_derivative_E_with_respect_to_λ)；
wherein E represents the error between the virtual target track and the real target track, learning_rate represents the learning rate corresponding to the current training wheel number, and partial_differential_E_with_recovery_to_sigma n Representing the error E with respect to the current standard deviation sigma n Partial_differential_e_with_recovery_to_λ represents the partial derivative of error E with respect to the current standard deviation λ.
In the embodiment of the disclosure, the standard deviation sigma is as described above n And repeatedly iterating the updating mode of lambda, and continuously optimizing the parameters of the random walk model until the error E reaches a preset error threshold. Finally, using the optimized model parameters, a virtual target track more conforming to the real target track can be generated.
According to the target generation method for radar system software testing, the network parameters of the random walk model are optimized, the optimal disturbance standard deviation and wind power standard deviation are solved, the trained random walk model is obtained, and a foundation is laid for generating a more accurate and vivid target motion trail.
In an optional embodiment of the present disclosure, when generating a motion trajectory of a simulation target in a motion mode, initial position and position information may be input into a trained random walk model, and a virtual target trajectory output by the trained random walk model may be obtained as a motion trajectory of the simulation target in the motion mode. Wherein, the motion trail is expressed as:
X_i=X_{i-1}+ε_i+f_i；
Wherein x_i represents the position of the ith time step, x_ { i-1} represents the position of the (i-1) th time step, ε_i represents the disturbance variable of the ith time step, f_i represents the wind variable of the ith time step, x_1 is the initial position, x_n is the end position, i=2, 3,...
In the embodiment of the disclosure, the motion rule of the simulation target is generated by using the trained random walk model, so that the generated motion trail is more real.
In some application scenarios, there are multiple targets, so generating multiple simulated targets and simulating cooperative motion between them has great help in testing the processing capability of radar system software for multiple targets. The target generation method for radar system software testing can generate a plurality of targets. Specifically, as shown in fig. 3, the target generation method for radar system software testing of the present disclosure may further include the steps of:
step 301, obtaining a target number K of targets to be generated and a step number M of time steps.
For example, the number of targets to be generated (denoted as K) and the number of time steps (denoted as M) included in the motion trajectory of each target may be set by the user through the user interface, and after the user completes the setting, the electronic device may obtain the number of targets K and the number of time steps M.
Step 302, obtaining a mean, a covariance of a jth time step when the gaussian mixture model generates the jth target and a probability of occurrence of the jth target at the jth time step, wherein j=1, 2, 3.
Among these, the gaussian mixture model (Gaussian Mixture Model, GMM) is a model for describing probability distributions, which can combine multiple gaussian distributions to generate a more complex data distribution.
As an example, when each target is generated, the mean and covariance corresponding to each time step may be randomly generated, and then the probability of the generated target occurring at that time step may be determined based on the mean and covariance of the same time step. For example, for the jth time step when the jth object is generated, the jth object generated by the Gaussian mixture model may be calculated at the jth time using a Softmax functionThe probability of occurrence of a step (also referred to as a weight), a specific calculation formula can be expressed as: w_j (t) =exp (-0.5 x_t-mu_j (t)) 2 The method comprises the steps of (a) and (b), wherein x_t represents an observed value of a t-th time step, mu_j (t) represents a mean value of the t-th time step when a j-th target is generated, sigma_j (t) represents covariance of the t-th time step when the j-th target is generated, and Z_t is a normalization constant.
As another example, when each target is generated, only the mean and variance corresponding to the first time step of the first target may be randomly generated, and then the mean and variance are updated once after each time step, so as to obtain the mean and variance corresponding to the next time step.
Specifically, the initial mean and initial covariance of the gaussian mixture model may be randomly generated, and then the probability of occurrence of the jth target at the t time step may be determined according to the following formula:
w_j(t)=exp(-0.5*(x_t-mu_j(t)) 2 /sigma_j(t))/Z_t；
wherein x_t represents an observed value of a t-th time step, mu_j (t) represents an average value of the t-th time step when the j-th target is generated, sigma_j (t) represents a covariance of the t-th time step when the j-th target is generated, Z_t is a normalization constant, an average value mu_1 (1) of the 1-th time step of the 1-th target is the initial average value, and a covariance sigma_1 (1) of the 1-th time step of the 1-th target is the initial covariance;
the mean and covariance of the (t+1) th time step in generating the jth target is determined by the following formula:
。
where x_j (T) represents the observed value of the T-th time step, mu_j (t+1) represents the mean value of the t+1th time step when the j-th target is generated, w_j (T) represents the probability that the j-th target appears at the T-th time step, and sigma_j (t+1) represents the covariance of the t+1th time step when the j-th target is generated.
That is, in this example, the initial average value generated randomly is taken as the average value of the 1 st time step when the 1 st object is generated, the initial covariance generated randomly is taken as the covariance of the 1 st time step when the 1 st object is generated, and then the probability w_1 (1) that the 1 st object appears at the 1 st time step is determined according to the above formula for calculating the probability of occurrence. Then, according to the formula for determining the mean value and covariance of the next time step, determining the mean value and covariance of the 2 nd time step to obtain mu_1 (2) and sigma_1 (2), determining the probability w_1 (2) of the 1 st target occurring in the 2 nd time step according to mu_1 (2) and sigma_1 (2), and so on to obtain the mean value, covariance and probability of each time step when each target is generated. It can be understood that, each time step corresponding to the average value, the covariance and the occurrence probability of M time steps are generated, a motion track of a target can be generated, and the average value and the covariance of the 1 st time step of the latter target are updated after the occurrence probability corresponding to the M time steps of the former target is determined.
And 303, generating a motion track corresponding to the jth target based on the mean value and covariance of the jth time step when the jth target is generated and the probability of the jth target occurring in the jth time step.
In the embodiment of the disclosure, after determining the mean value and covariance of the jth time step when the jth target is generated and the probability that the jth target appears in the jth time step, the motion trail corresponding to the jth target can be generated based on the data.
The position x_j (t) of the t-th time step in the motion track corresponding to the j-th target is expressed as follows:
;
where w_j (T) represents the probability that the jth target appears at the jth time step, mu_j (T) represents the mean of the jth time step when the jth target is generated, sigma_j (T) represents the covariance of the jth time step when the jth target is generated, randn (1, d) represents a vector randomly obtained from a normal distribution, d represents the dimension of the jth target, and d has a value of 2 for the target on a two-dimensional plane, for example.
The object determined in the above manner has certain characteristics (such as speed and position) of a real object.
According to the target generation method for radar system software testing, a plurality of targets can be generated through the Gaussian mixture model, so that a more challenging and real test scene is provided, the radar system software in a multi-target scene is tested, and the complexity and the authenticity of the test are increased.
In order to implement the above embodiment, the present disclosure also provides a target generating apparatus for radar system software testing.
Fig. 4 is a schematic structural diagram of a target generating apparatus for radar system software testing, which may be implemented in software and/or hardware and may be integrated in an electronic device, according to an exemplary embodiment of the present disclosure. As shown in fig. 4, the target generating apparatus 40 for radar system software testing includes: a target track generation module 401 and a motion track generation module 402, wherein:
the target track generation module 401 is configured to obtain an attribute type of a target to be generated, where the attribute type includes a target type, a position, a speed type, and a motion mode; generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder;
a motion trail generation module 402, configured to generate position information corresponding to different time steps in the motion mode based on the initial position and the speed information; and generating a motion trail of the simulation target in the motion mode based on the initial position and the position information.
In one embodiment, the target generating device 40 for radar system software testing further comprises a first acquisition module and a training module, wherein,
the first acquisition module is used for acquiring a training sample, wherein the training sample comprises simulation positions and real target tracks corresponding to N time steps respectively, the number of the time steps of the real target track is N, and N is a positive integer;
training module for:
inputting the simulation positions corresponding to the N time steps into a random walk model to obtain a virtual target track, wherein network parameters of the random walk model comprise normal distribution N (0, sigma) obeyed by disturbance variables n ) Standard deviation sigma of n And the standard deviation λ of the normal distribution N (0, λ) to which the wind variable is subjected;
calculating an error between the virtual target track and the real target track;
updating the standard deviation sigma based on the error and the learning rate corresponding to the current training round number by adopting a preset optimization algorithm n The standard deviation lambda is used for obtaining a trained random walk model until the error between the virtual target track output by the random walk model and the corresponding real target track reaches a preset error threshold;
wherein the updated standard deviation sigma n And standard deviation λ is expressed as:
σ n =σ n -learning_rate*(partial_derivative_E_with_respect_to_σ n )；
λ=λ-learning_rate*(partial_derivative_E_with_respect_to_λ)；
wherein E represents the error between the virtual target track and the real target track, learning_rate represents the learning rate corresponding to the current training wheel number, and partial_differential_E_with_recovery_to_sigma n Representing the error E with respect to the current standard deviation sigma n Partial_differential_e_with_recovery_to_λ represents the partial derivative of error E with respect to the current standard deviation λ.
In one embodiment, the motion trajectory generation module 402 is further configured to:
inputting the initial position and the position information into the trained random walk model, and acquiring a virtual target track output by the trained random walk model as a motion track of the simulation target in the motion mode;
wherein, the motion trail is expressed as:
X_i=X_{i-1}+ε_i+f_i；
wherein x_i represents the position of the ith time step, x_ { i-1} represents the position of the (i-1) th time step, ε_i represents the disturbance variable of the ith time step, f_i represents the wind variable of the ith time step, x_1 is the initial position, x_n is the end position, i=2, 3,...
In one embodiment, the motion trajectory generation module 402 is further configured to:
and in response to the motion mode being uniform linear motion, determining position information (x, y) corresponding to a t-th time step through the following formula:
x=x 0 +Vx*t；
y=y 0 +Vy*t；
Wherein, (x) 0 ,y 0 ) For the initial position, vx is the velocity of the simulation target in the x direction, and Vy is the velocity of the simulation target in the y direction.
In one embodiment, the motion trajectory generation module 402 is further configured to:
and in response to the motion mode being uniform circular motion, determining position information (x, y) corresponding to a t-th time step through the following formula:
x=x 0 +R*cos(ωt)；
y=y 0 +R*sin(ωt)；
wherein, (x) 0 ,y 0 ) And for the initial position, omega is the angular speed of the simulation target, and R is the radius of the circular motion of the simulation target.
In one embodiment, the target generating device 40 for radar system software testing further comprises:
the second acquisition module is used for acquiring the target number K of the targets to be generated and the step number M of the time steps;
a third acquisition module configured to acquire a mean, a covariance, and a probability of occurrence of a jth target at the jth time step when the gaussian mixture model generates the jth target, where j=1, 2,3,..k, t=1, 2,3,..m;
the generation module is used for generating a motion track corresponding to the jth target based on the mean value and covariance of the jth time step when the jth target is generated and the probability of the jth target in the jth time step;
The position x_j (t) of the t-th time step in the motion track corresponding to the j-th target is expressed as follows:
;
wherein w_j (T) represents the probability that the jth target appears at the jth time step, mu_j (T) represents the mean of the jth time step when the jth target is generated, sigma_j (T) represents the covariance of the jth time step when the jth target is generated, randn (1, d) represents a vector randomly obtained from a standard normal distribution, and d represents the dimension of the jth target.
In one embodiment, the third acquisition module is further configured to:
randomly generating an initial mean and an initial covariance of the Gaussian mixture model;
determining the probability of occurrence of said jth target at said t time step by the following formula:
w_j(t)=exp(-0.5*(x_t-mu_j(t)) 2 /sigma_j(t))/Z_t；
wherein x_t represents an observed value of a t-th time step, mu_j (t) represents an average value of the t-th time step when a j-th target is generated, sigma_j (t) represents a covariance of the t-th time step when the j-th target is generated, Z_t is a normalization constant, an average value mu_1 (1) of a 1-th time step of a 1-th target is the initial average value, and a covariance sigma_1 (1) of the 1-th time step of the 1-th target is the initial covariance;
the mean and covariance of the (t+1) th time step in generating the jth target is determined by the following formula:
。
The target generation device for radar system software testing provided by the embodiment of the disclosure can execute any target generation method for radar system software testing applicable to electronic equipment, and has the corresponding functional modules and beneficial effects of the execution method. Details of the embodiments of the apparatus of the present disclosure that are not described in detail may refer to descriptions of any of the embodiments of the method of the present disclosure.
The embodiment of the disclosure also provides an electronic device, including: a processor; and a memory storing a program, wherein the program comprises instructions that when executed by the processor cause the processor to perform the steps of the target generation method for radar system software testing according to any of the foregoing embodiments, and are not described in detail herein for the sake of avoiding repetition of the description.
The embodiments of the present disclosure further provide a computer readable storage medium, on which a computer program is stored, where the computer program when executed by a processor implements the steps of the target generation method for radar system software testing according to any of the foregoing embodiments, and in order to avoid repetitive description, a description is omitted herein.
The embodiments of the present disclosure further provide a computer program product, including a computer program, where the computer program, when executed by a processor, implements the steps of the target generation method for radar system software testing according to any of the foregoing embodiments, and will not be described herein in detail for the sake of avoiding repetitive description.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided by the present disclosure may include at least one of non-volatile and volatile memory, among others. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, or the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static random access memory (Static Random Access Memory, SRAM), dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples represent only a few embodiments of the present disclosure, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that variations and modifications can be made by those skilled in the art without departing from the spirit of the disclosure, which are within the scope of the disclosure. Accordingly, the scope of protection of the present disclosure should be determined by the following claims.
Claims (8)
1. A target generation method for radar system software testing, comprising:
acquiring attribute types of targets to be generated, wherein the attribute types comprise a target type, a position, a speed type and a motion mode;
generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder;
generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information;
generating a motion trail of the simulation target in the motion mode based on the initial position and the position information;
wherein the method further comprises:
Obtaining a training sample, wherein the training sample comprises simulation positions and real target tracks corresponding to N time steps respectively, the number of the time steps of the real target track is N, and N is a positive integer;
inputting the simulation positions corresponding to the N time steps into a random walk model to obtain a virtual target track, wherein network parameters of the random walk model comprise normal distribution N (0, sigma) obeyed by disturbance variables n ) Standard deviation sigma of n And the standard deviation λ of the normal distribution N (0, λ) to which the wind variable is subjected;
calculating an error between the virtual target track and the real target track;
updating the standard deviation sigma based on the error and the learning rate corresponding to the current training round number by adopting a preset optimization algorithm n The standard deviation lambda is used for obtaining a trained random walk model until the error between the virtual target track output by the random walk model and the corresponding real target track reaches a preset error threshold;
wherein the updated standard deviation sigma n And standard deviation λ is expressed as:
σ n =σ n -learning_rate*(partial_derivative_E_with_respect_to_σ n )；
λ=λ-learning_rate*(partial_derivative_E_with_respect_to_λ)；
wherein E represents the error between the virtual target track and the real target track, learning_rate represents the learning rate corresponding to the current training wheel number, and partial_differential_E_with_recovery_to_sigma n Representing the error E with respect to the current standard deviation sigma n Partial_differential_e_with_recovery_to_λ represents the partial derivative of error E with respect to the current standard deviation λ;
the generating a motion trail of the simulation target in the motion mode based on the initial position and the position information comprises the following steps:
inputting the initial position and the position information into the trained random walk model, and acquiring a virtual target track output by the trained random walk model as a motion track of the simulation target in the motion mode;
wherein, the motion trail is expressed as:
X_i=X_{i-1}+ε_i+f_i；
wherein x_i represents the position of the ith time step, x_ { i-1} represents the position of the (i-1) th time step, ε_i represents the disturbance variable of the ith time step, f_i represents the wind variable of the ith time step, x_1 is the initial position, x_n is the end position, i=2, 3,...
2. The target generation method for radar system software testing according to claim 1, wherein the generating position information corresponding to different time steps in the movement mode based on the initial position and the speed information includes:
And in response to the motion mode being uniform linear motion, determining position information (x, y) corresponding to a t-th time step through the following formula:
x=x 0 +Vx*t；
y=y 0 +Vy*t；
wherein, (x) 0 ,y 0 ) For the initial position, vx is the velocity of the simulation target in the x direction, and Vy is the velocity of the simulation target in the y direction.
3. The target generation method for radar system software testing according to claim 1, wherein the generating position information corresponding to different time steps in the movement mode based on the initial position and the speed information includes:
and in response to the motion mode being uniform circular motion, determining position information (x, y) corresponding to a t-th time step through the following formula:
x=x 0 +R*cos(ωt)；
y=y 0 +R*sin(ωt)；
wherein, (x) 0 ,y 0 ) And for the initial position, omega is the angular speed of the simulation target, and R is the radius of the circular motion of the simulation target.
4. The target generation method for radar system software testing of claim 1, further comprising:
acquiring the target number K of targets to be generated and the step number M of time steps;
acquiring a mean value, a covariance of a jth time step when a gaussian mixture model generates the jth target and a probability of the jth target occurring at the jth time step, wherein j=1, 2,3,..k, t=1, 2,3,..m;
Generating a motion track corresponding to a jth target based on the mean value and covariance of the jth time step when the jth target is generated and the probability of the jth target occurring in the jth time step;
the position x_j (t) of the t-th time step in the motion track corresponding to the j-th target is expressed as follows:
;
wherein w_j (T) represents the probability that the jth target appears at the jth time step, mu_j (T) represents the mean of the jth time step when the jth target is generated, sigma_j (T) represents the covariance of the jth time step when the jth target is generated, randn (1, d) represents a vector randomly obtained from a standard normal distribution, and d represents the dimension of the jth target.
5. The method for generating targets for radar system software testing of claim 4, wherein said obtaining a mean, covariance, and probability of occurrence of a jth target at said jth time step when said gaussian mixture model generates said jth target comprises:
randomly generating an initial mean and an initial covariance of the Gaussian mixture model;
determining the probability of occurrence of said jth target at said t time step by the following formula:
w_j(t)=exp(-0.5*(x_t-mu_j(t))2/sigma_j(t))/Z_t；
Wherein x_t represents an observed value of a t-th time step, mu_j (t) represents an average value of the t-th time step when a j-th target is generated, sigma_j (t) represents a covariance of the t-th time step when the j-th target is generated, Z_t is a normalization constant, an average value mu_1 (1) of a 1-th time step of a 1-th target is the initial average value, and a covariance sigma_1 (1) of the 1-th time step of the 1-th target is the initial covariance;
the mean and covariance of the (t+1) th time step in generating the jth target is determined by the following formula:
。
6. a target generating apparatus for radar system software testing, comprising:
the target track generation module is used for acquiring attribute types of targets to be generated, wherein the attribute types comprise a target type, a position, a speed type and a motion mode; generating a simulation target meeting the target type and initial position and speed information of the simulation target by using a condition variation automatic encoder;
the motion trail generation module is used for generating position information corresponding to different time steps in the motion mode based on the initial position and the speed information; generating a motion track of the simulation target in the motion mode based on the initial position and the position information;
The first acquisition module is used for acquiring a training sample, wherein the training sample comprises simulation positions and real target tracks corresponding to N time steps respectively, the number of the time steps of the real target track is N, and N is a positive integer;
training module for:
inputting the simulation positions corresponding to the N time steps into a random walk model to obtain a virtual target track, wherein network parameters of the random walk model comprise normal distribution N (0, sigma) obeyed by disturbance variables n ) Standard deviation sigma of n And the standard deviation λ of the normal distribution N (0, λ) to which the wind variable is subjected;
calculating an error between the virtual target track and the real target track;
updating the standard deviation sigma based on the error and the learning rate corresponding to the current training round number by adopting a preset optimization algorithm n The standard deviation lambda is used for obtaining a trained random walk model until the error between the virtual target track output by the random walk model and the corresponding real target track reaches a preset error threshold;
wherein the updated standard deviation sigma n And standard deviation λ is expressed as:
σ n =σ n -learning_rate*(partial_derivative_E_with_respect_to_σ n )；
λ=λ-learning_rate*(partial_derivative_E_with_respect_to_λ)；
wherein E represents the error between the virtual target track and the real target track, learning_rate represents the learning rate corresponding to the current training wheel number, and partial_differential_E_with_recovery_to_sigma n Representing the error E with respect to the current standard deviation sigma n Partial_differential_e_with_recovery_to_λ represents the partial derivative of error E with respect to the current standard deviation λ;
the motion trail generation module is further used for:
inputting the initial position and the position information into the trained random walk model, and acquiring a virtual target track output by the trained random walk model as a motion track of the simulation target in the motion mode;
wherein, the motion trail is expressed as:
X_i=X_{i-1}+ε_i+f_i；
wherein x_i represents the position of the ith time step, x_ { i-1} represents the position of the (i-1) th time step, ε_i represents the disturbance variable of the ith time step, f_i represents the wind variable of the ith time step, x_1 is the initial position, x_n is the end position, i=2, 3,...
7. An electronic device, comprising:
a processor; and
a memory in which a program is stored,
wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the steps of the target generation method for radar system software testing according to any one of claims 1-5.
8. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the object generation method for radar system software testing of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311723807.3A CN117407319B (en) | 2023-12-15 | 2023-12-15 | Target generation method and device for radar system software test and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311723807.3A CN117407319B (en) | 2023-12-15 | 2023-12-15 | Target generation method and device for radar system software test and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117407319A CN117407319A (en) | 2024-01-16 |
| CN117407319B true CN117407319B (en) | 2024-04-02 |
Family
ID=89496575
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311723807.3A Active CN117407319B (en) | 2023-12-15 | 2023-12-15 | Target generation method and device for radar system software test and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117407319B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107003398A (en) * | 2015-10-22 | 2017-08-01 | 尤尼克塞克有限公司 | For the method for testing of the virtual radar signature signal of use of automobile safety radar system |
| CN111694287A (en) * | 2020-05-14 | 2020-09-22 | 北京百度网讯科技有限公司 | Obstacle simulation method and device in unmanned simulation scene |
| CN113673749A (en) * | 2021-07-28 | 2021-11-19 | 广东省科学院智能制造研究所 | Vehicle and pedestrian map track prediction method and device |
| CN114332418A (en) * | 2021-12-20 | 2022-04-12 | 中国电子科技集团公司第五十二研究所 | Target track simulation method and device |
| CN115438569A (en) * | 2022-07-29 | 2022-12-06 | 深圳元戎启行科技有限公司 | Automatic driving scene generation method and training method and device of related model |
| CN117031419A (en) * | 2023-07-10 | 2023-11-10 | 中国电子科技集团公司第十四研究所 | Radar target simulation device of air mobile platform |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11816901B2 (en) * | 2020-03-04 | 2023-11-14 | Nec Corporation | Multi-agent trajectory prediction |
| CN112904897B (en) * | 2021-01-26 | 2022-02-15 | 华中科技大学 | Track optimization method and equipment for underwater accurate launching of ocean sensor |
-
2023
- 2023-12-15 CN CN202311723807.3A patent/CN117407319B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107003398A (en) * | 2015-10-22 | 2017-08-01 | 尤尼克塞克有限公司 | For the method for testing of the virtual radar signature signal of use of automobile safety radar system |
| CN111694287A (en) * | 2020-05-14 | 2020-09-22 | 北京百度网讯科技有限公司 | Obstacle simulation method and device in unmanned simulation scene |
| CN113673749A (en) * | 2021-07-28 | 2021-11-19 | 广东省科学院智能制造研究所 | Vehicle and pedestrian map track prediction method and device |
| CN114332418A (en) * | 2021-12-20 | 2022-04-12 | 中国电子科技集团公司第五十二研究所 | Target track simulation method and device |
| CN115438569A (en) * | 2022-07-29 | 2022-12-06 | 深圳元戎启行科技有限公司 | Automatic driving scene generation method and training method and device of related model |
| CN117031419A (en) * | 2023-07-10 | 2023-11-10 | 中国电子科技集团公司第十四研究所 | Radar target simulation device of air mobile platform |
Non-Patent Citations (4)
| Title |
|---|
| 一种基于目标运动模型的雷达点迹模拟方法;林盛;刘军伟;徐伟;;舰船电子对抗;20180625(第03期);51-54+69 * |
| 区间聚类结合区域划分法实现WSN随机游走模型跟踪;孙武;李小刚;祁荣宾;钟伟民;;华东理工大学学报(自然科学版);20151230(第06期);81-85 * |
| 基于运动方程的目标航迹模拟方法;王兴申;田康生;刘婕;;现代电子技术;20060315(第05期);22-24+27 * |
| 基于高斯混合-变分自编码器的轨迹预测算法;张显炀;朱晓宇;林浩申;刘刚;安喜彬;;计算机工程;20200715;46(第07期);56-63 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117407319A (en) | 2024-01-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tuncali et al. | Simulation-based adversarial test generation for autonomous vehicles with machine learning components | |
| JP7233807B2 (en) | Computer-implemented method, computer system, and computer program for simulating uncertainty in artificial neural networks | |
| KR20190031318A (en) | Domain Separation Neural Networks | |
| CN109902018B (en) | Method for acquiring test case of intelligent driving system | |
| CN106774312B (en) | Method and device for determining moving track of mobile robot | |
| CN109977571B (en) | Simulation calculation method and device based on data and model mixing | |
| CN105183993A (en) | Integrated simulation platform and method for electromagnetic rail gun | |
| CN115735107A (en) | Computer-implemented method for providing a test procedure for a traffic scenario to be tested | |
| CN111932451A (en) | Method and device for evaluating repositioning effect, electronic equipment and storage medium | |
| CN112147973A (en) | Inspection system, method and apparatus for selecting a true test and test system | |
| CN117407319B (en) | Target generation method and device for radar system software test and electronic equipment | |
| CN111895990B (en) | Method for detecting and repositioning mobile robot based on multi-index binding frame | |
| Innes et al. | Automated testing with temporal logic specifications for robotic controllers using adaptive experiment design | |
| Orihara et al. | Approximation of time-consuming simulation based on generative adversarial network | |
| US11514268B2 (en) | Method for the safe training of a dynamic model | |
| JP2020024633A (en) | Training data evaluation device, training data evaluation method, and program | |
| Nair et al. | Maximum likelihood uncertainty estimation: Robustness to outliers | |
| US20230221726A1 (en) | Method for the qualification of a control with the aid of a closed-loop simulation process | |
| Lauttia | Adaptive Monte Carlo Localization in ROS | |
| CN108509684B (en) | Steering engine and dynamic load simulator matching design method | |
| CN112560881A (en) | Object identification method and device and data processing method | |
| CN116168062B (en) | 3D target tracking method and device | |
| Ellis et al. | A mapping of assurance techniques for learning enabled autonomous systems to the systems engineering lifecycle | |
| Sun et al. | Medium-Fidelity Evaluation and Modeling for Perception Systems of Intelligent and Connected Vehicles | |
| Prokhorov | Toward next generation of autonomous systems with AI |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |