A method and system for locating a position of a radiation source

Abstract

Some embodiments of the present application provide a method and system for positioning a radiation source, the method comprising: receiving a plurality of characteristic parameter values sent by at least some devices in a device cluster, wherein each device in the device cluster is configured to receive a signal emitted by a same radiation source, and the characteristic parameter values are obtained by time difference sampling signals, and the time difference sampling signals are obtained by fusing a sampling signal of the signal by any device and sampling signals of the signal by neighboring devices of the any device; constructing a target cost function according to the plurality of characteristic parameter values; and obtaining a target position of the radiation source according to the target cost function. The method provided by the embodiments of the present application can realize efficient and accurate positioning of the position of the radiation source.

Description

Technical Field

[0001] This application relates to the field of passive positioning technology, and more specifically, to a method and system for locating the position of a radiation source. Background Technology

[0002] Because of the advantages of low cost and rapid deployment, swarm devices are widely used in passive positioning fields (e.g., applying drone swarms in search and rescue or industrial inspection).

[0003] Currently, in the field of passive positioning, the location of radiation sources is generally achieved using two-step positioning or direct positioning methods. However, research has found that the two-step positioning method has poor positioning performance; while the direct positioning method has better positioning performance than the two-step method, it has higher computational complexity and cannot achieve distributed positioning of radiation sources. It also requires significant communication resource overhead during the positioning process. Therefore, the direct positioning method cannot be directly applied to device clusters.

[0004] Therefore, how to provide a technical solution for efficiently and accurately locating radiation sources has become an urgent technical problem to be solved. Summary of the Invention

[0005] The purpose of some embodiments of this application is to provide a method and system for locating the position of a radiation source. The technical solutions of the embodiments of this application can effectively reduce computer resource consumption and reduce the communication pressure between devices in the cluster during the positioning calculation process, thereby achieving efficient and accurate positioning of the radiation source.

[0006] In a first aspect, some embodiments of this application provide a method for locating a radiation source, comprising: receiving a plurality of feature parameter values ​​transmitted by at least some devices in a device cluster, wherein each device in the device cluster is used to receive a signal emitted by the same radiation source, the feature parameter values ​​are obtained by time difference sampling signals, the time difference sampling signals are obtained by each device by fusing its own sampling signals of the signal with sampling signals of the signal from adjacent devices; constructing a target cost function based on the plurality of feature parameter values; and obtaining the target location of the radiation source based on the target cost function.

[0007] In some embodiments of this application, each device in the device cluster only shares the sampling signal with its neighboring devices and obtains the feature parameter value through the shared sampling signal. Then, only the feature parameter value is transmitted to the computing node to determine the location of the radiation source. Compared with the sampling signal, the amount of data of the feature parameter value is significantly reduced. Therefore, the embodiments of this application significantly reduce the amount of data transmission between each sampling node device and the computing device. Without reducing the radiation source positioning accuracy (for example, neighboring devices share sampling information to improve positioning accuracy), the amount of data transmitted between devices is also significantly reduced. In other words, some embodiments of this application can obtain the radiation source location with high accuracy with low computer resource overhead.

[0008] In some embodiments, the feature parameter value is obtained by: obtaining the time difference sampling signal based on the sampling signal of the signal from any device and the sampling signal of the signal from the adjacent device; and using a target parameter extraction model deployed in any device to extract features from the time difference sampling signal to obtain a feature parameter value.

[0009] Some embodiments of this application use a target parameter extraction model to extract features from the time difference sampling signals obtained from the sampling signals of any device and adjacent devices, resulting in a feature parameter value. This can accurately obtain the time difference between the signals received by the radiation source between devices, and after being converted into a feature parameter value, it significantly reduces the amount of data in the sampling signal and saves communication overhead.

[0010] In some embodiments, the time difference sampling signal is obtained by preprocessing the original time difference sampling signal, and the preprocessing includes at least one of fitting processing and normalization processing.

[0011] Some embodiments of this application process the initial time difference sampling signal so that the time difference sampling signal can be adapted to the input requirements of the target parameter extraction model, thereby obtaining feature extraction results with high accuracy.

[0012] In some embodiments, the time difference sampling signal is obtained by the following method: any device samples the signal to obtain a sampling signal corresponding to the device; the adjacent device samples the signal to obtain a sampling signal corresponding to the adjacent device, and sends the sampling signal corresponding to the adjacent device to the device; any device obtains the original time difference sampling signal based on its own sampling signal and the sampling signals received from all adjacent devices; the preprocessing is performed on the original time difference sampling signal to obtain the time difference sampling signal.

[0013] Some embodiments of this application achieve information interaction between any device and its neighboring devices by fusing the sampling signals sent by the neighboring devices with its own sampling signals, thereby obtaining an effective time difference sampling signal.

[0014] In some embodiments, the type of the feature parameter value includes: a time difference estimate, a first reference feature value, and a second reference feature value, wherein the time difference estimate is used to characterize the time difference between the reception of the signal by the any device and the adjacent device.

[0015] Some embodiments of this application obtain various types of feature parameter values, providing reliable data support for the subsequent construction of the target cost function, thereby enabling the accurate location of the radiation source.

[0016] In some embodiments, the target cost function is obtained by the following formula:

[0017]

[0018] Wherein, C(u) s Let u be the objective cost function. s Let N be the location of the radiation source to be solved. m This refers to the number of the aforementioned devices. Let be the first reference characteristic value between the k-th device and its adjacent devices. This is the second reference characteristic value between the k-th device and its adjacent devices. This is an estimated time difference between the k-th device and its adjacent devices.

[0019] Some embodiments of this application construct the target cost function of the above formula through multiple feature parameter values, which can reduce the computational complexity and computer resource consumption, while providing a computational basis for obtaining accurate radiation source locations.

[0020] In some embodiments, obtaining the target location of the radiation source based on the target cost function includes: solving the target cost function using a target iteration algorithm to obtain the target location.

[0021] Some embodiments of this application solve the target cost function using a target iterative algorithm, which can quickly obtain the location of the radiation source with high accuracy and improve computational efficiency.

[0022] In some embodiments, the target iteration algorithm includes either Newton's iteration algorithm or gradient descent algorithm.

[0023] Some embodiments of this application can solve the target cost function using various iterative algorithms, which can efficiently and quickly obtain the location of the radiation source and improve computational efficiency.

[0024] Secondly, some embodiments of this application provide a method for locating the position of a radiation source, comprising: each device in a device cluster receiving a signal emitted by the radiation source and sampling the signal to obtain a sampled signal; a fusion node device receiving sampled signals sent from adjacent node devices and obtaining feature parameter values ​​through a time difference sampled signal obtained from the sampled signals; a current computing node device receiving the feature parameter values ​​sent from each fusion node device and obtaining the target position of the radiation source based on a target cost function constructed according to the feature parameter values; wherein the current computing node device and the fusion node device both belong to the device cluster.

[0025] Some embodiments of this application employ a distributed computing method to locate the radiation source. This method is highly flexible and can select any device in the device cluster as the computing node for this task. This method can significantly reduce the overhead of communication and computing resources between devices while maintaining high positioning accuracy.

[0026] Thirdly, some embodiments of this application provide a system for locating the position of a radiation source, comprising: the system including multiple node devices, wherein each node device is configured to: receive a signal emitted by the radiation source and sample the signal to obtain a sampled signal; send the sampled signal to adjacent node devices; obtain a time difference sampled signal based on the received sampled signals from other adjacent node devices, and obtain a feature parameter value based on the time difference sampled signal; construct a target cost function based on the received feature parameter values ​​from other node devices, solve the target cost function, and obtain the target position of the radiation source.

[0027] Fourthly, some embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the method described in any embodiment of the first aspect.

[0028] Fifthly, some embodiments of this application provide an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the method as described in any embodiment of the first aspect.

[0029] Sixthly, some embodiments of this application provide a computer program product comprising a computer program, wherein the computer program, when executed by a processor, can implement the method described in any embodiment of the first aspect. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of some embodiments of this application, the accompanying drawings used in some embodiments of this application will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 System diagrams for locating radiation sources provided for some embodiments of this application;

[0032] Figure 2 One of the flowcharts for a method of locating a radiation source provided in some embodiments of this application;

[0033] Figure 3 A second flowchart illustrating a method for locating a radiation source, provided for some embodiments of this application;

[0034] Figure 4 Structure diagrams of target DNN models provided for some embodiments of this application;

[0035] Figure 5 A block diagram of a device for locating a radiation source is provided for some embodiments of this application;

[0036] Figure 6 A schematic diagram of an electronic device provided for some embodiments of this application. Detailed Implementation

[0037] The technical solutions of some embodiments of this application will now be described with reference to the accompanying drawings.

[0038] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0039] In related technologies, device swarms (e.g., drone swarms) offer the advantage of rapid deployment. Equipping multiple drones in the swarm with wireless communication modules or sensors enables passive localization of radiation sources in civilian or military applications. For example, drone swarms can be used for search and rescue, industrial inspection, and electronic warfare operations. However, due to the large number of drones in the swarm, highly complex algorithms are required for accurate radiation source localization. Generally, passive localization of radiation sources employs two-step or direct localization methods. Two-step localization has poor performance and low accuracy. While traditional direct localization outperforms two-step methods, it is highly complex, cannot achieve distributed localization, and requires significant computer resources to locate the radiation source. As can be seen from the above related technologies, existing radiation source localization methods are highly complex, consume significant computer resources, and incur substantial communication resource costs between swarm devices, making efficient and accurate radiation source localization impossible in resource-constrained drone swarms.

[0040] In view of this, some embodiments of this application provide a method and system for locating a radiation source. This method constructs a target cost function by receiving multiple feature parameter values ​​sent by at least some devices in a device cluster, and obtains the target location of the radiation source based on the target cost function. The method of this application not only obtains the radiation source location with high accuracy but also significantly reduces the overhead of communication and computing resources.

[0041] like Figure 1 As shown, some embodiments of this application provide a system for locating radiation sources, the system's equipment cluster including multiple unmanned aerial vehicles (UAVs) (e.g., Figure 1 The system includes a first UAV 110, a second UAV 120, a third UAV 130, a fourth UAV 140, and a fifth UAV 150, a radiation source 200, and communication paths A, B, C, D, E, F, G, and I between the UAVs. These communication paths enable the exchange of information and data (e.g., sampled signals) between the UAVs. It should be noted that in some embodiments of this application, the computing power of each device (i.e., the UAV) in the device cluster is equivalent. Therefore, any device in the device cluster has the following functions: collecting signals emitted by the received radiation source and obtaining characteristic parameter values; performing the functions of a fusion node device; or performing the functions of a current computing node device. It should be understood that in actual application scenarios, whether a device is a fusion node device or a current computing node device can be set according to the actual communication and computing resources available; this application does not specifically limit this.

[0042] It is understood that in some embodiments of this application, each drone can receive signals emitted by the radiation source 200. Adjacent drones can exchange information. For example, the adjacent devices adjacent to the first drone 110 are the second drone 120 and the fourth drone 140. The first drone 110 can communicate with the second drone 120, and the first drone 110 can also communicate with the fourth drone 140.

[0043] In some embodiments of this application, each device can sample the signal emitted by the received radiation source 200 to obtain a sampled signal. When the first UAV 110 sends the sampled signal to the second UAV 120, the second UAV 120 can fuse its own sampled signal with the sampled signal received from the first UAV 110 to obtain a time difference sampling signal. In this case, the second UAV 120 is a fusion node device. Similarly, when the second UAV 120 sends the sampled signal to the first UAV 110, the first UAV 110 can also fuse its own sampled signal with the sampled signal received from the second UAV 120 to obtain a time difference sampling signal. In this case, the first UAV 110 is a fusion node device. It is easy to understand that there can be multiple fusion node devices.

[0044] In some embodiments of this application, the fusion node device can send the feature parameter values ​​obtained after feature extraction of the time difference sampling signal to the local computing node device (effectively reducing the communication volume between these nodes and the computing node device by sending feature parameter values), and the local computing node device can obtain the target location of the radiation source through its internal calculation logic. Figure 1 It is known that the computing node device in this case is the fifth UAV 150, and each fusion node device can send feature parameter values ​​to the fifth UAV 150. For example, the second UAV 120 can send the extracted feature parameter values ​​to the fifth UAV 150.

[0045] In other embodiments of this application, the devices in the device cluster may be other electronic devices with information processing and transmission functions besides drones, such as sensor devices or microcomputer devices, etc.

[0046] The following is in conjunction with the appendix Figure 2 The present application provides an exemplary embodiment of a method for locating a radiation source, executed by the current computing node device.

[0047] Please see the appendix Figure 2 , Figure 2 A flowchart of a method for locating a radiation source by a computing node device, provided for some embodiments of this application, is shown. The method includes:

[0048] S210, receiving multiple feature parameter values ​​sent by at least some devices in the device cluster, wherein each device in the device cluster is used to receive a signal emitted by the same radiation source, and the feature parameter values ​​are obtained by time difference sampling signals, which are obtained by each device by fusing its own sampling signals of the signal with the sampling signals of the signal from its neighboring devices.

[0049] In some embodiments of this application, a distributed computing method (i.e., each device in the device cluster participates in the computation, and each device is a node device) is used to locate the radiation source. In actual application scenarios, there may be one or more radiation sources. The example of this application can locate the location of one or more radiation sources respectively. Multiple devices in the device cluster can be used as fusion node devices and current computation node devices.

[0050] For example, although any device and its neighboring devices both have the function of sampling signal fusion, to avoid data duplication, any device can act as a fusion node device and send the fused result (i.e., feature parameter value) to the current computing node device. Therefore, at least some devices in the device cluster are fusion node devices, and thus the current computing node device can receive multiple feature parameter values ​​sent by multiple fusion node devices. It should be noted that the current computing node device can be any device in the device cluster. When any device is a fusion node device, any device can fuse its own sampled signal with the sampled signals of its neighboring devices to obtain a time difference sampling signal.

[0051] To achieve the fusion of sampled signals and improve the efficiency of radiation source localization, each node device performs information fusion with its neighboring devices to obtain feature parameter values. It can be understood that since each node device only shares information with its neighboring node devices (i.e., neighboring devices), the data communication volume is significantly reduced while improving the accuracy of the localization results (i.e., the communication volume is reduced compared to each node device sending sampled signals to the current computing node device separately).

[0052] For example, in some embodiments of this application, the time difference sampling signal is obtained by the following method: any device samples the signal to obtain a sampling signal corresponding to the device; the neighboring device samples the signal to obtain a sampling signal corresponding to the device, and sends the sampling signal corresponding to the device to the device; each device obtains the original time difference sampling signal based on its own sampling signal and the sampling signals received from all neighboring devices; the original time difference sampling signal is preprocessed to obtain the time difference sampling signal. The preprocessing includes at least one of fitting processing and normalization processing.

[0053] Since large-scale network communication among device clusters is difficult in real-world scenarios, some embodiments of this application employ a method of communication between adjacent devices, enabling information exchange between cluster devices. For example, each device in the cluster has the function of sampling signals emitted by the received radiation source. Adjacent devices send the sampled signals obtained from the signal sampling to any other device, allowing any device to fuse the sampled signals to obtain an initial time difference sampling signal. To reduce computer resource overhead, any device (i.e., the fusion node device) can also perform fitting processing (e.g., sine fitting) and normalization processing (e.g., sigmoid function normalization) on the initial time difference sampling signal to obtain the time difference sampling signal.

[0054] In some embodiments of this application, the feature parameter value is obtained by the following method: obtaining the time difference sampling signal based on the sampling signal of the signal from any device and the sampling signal of the signal from the adjacent device; and using a target parameter extraction model deployed in any device to extract features from the time difference sampling signal to obtain a feature parameter value.

[0055] For example, in some embodiments of this application, a target parameter extraction model is deployed in the fusion node device. By inputting the time difference sampling signal into the target parameter extraction model, a feature parameter value can be obtained. It should be noted that one or more target parameter extraction models can be deployed in the fusion node device, wherein one target parameter extraction model can extract one feature parameter value. In addition, the target parameter extraction model is obtained by training an initial parameter extraction model, which can be a DNN (Deep Neural Networks) model.

[0056] In some embodiments of this application, the type of the feature parameter value includes: time difference estimate, first reference feature value and second reference feature value, wherein the time difference estimate is used to characterize the time difference between the reception of the signal by any device and the adjacent device.

[0057] For example, in some embodiments of this application, the fusion node device may be equipped with three target parameter extraction models to extract the time difference estimate, the first reference feature value, and the second reference feature value from the time difference sampling signal, respectively.

[0058] S220, construct the target cost function based on the multiple feature parameter values.

[0059] In some embodiments of this application, the target cost function is obtained by the following formula:

[0060]

[0061] Where C(u) is the target cost function, u is the location of the radiation source to be solved, and N m This refers to the number of the aforementioned devices. Let be the first reference characteristic value between the k-th device and its adjacent devices. This is the second reference characteristic value between the k-th device and its adjacent devices. This is the estimated time difference between the k-th device and its adjacent devices. It should be noted that the k-th device can be any device.

[0062] S230, obtain the target location of the radiation source according to the target cost function.

[0063] In some embodiments of this application, S230 may further include: solving the target cost function using a target iteration algorithm to obtain the target position. The target iteration algorithm may be either Newton's iteration algorithm or gradient descent algorithm.

[0064] The following is in conjunction with the appendix Figure 3 The specific implementation process of the method for locating the position of a radiation source provided in some embodiments of this application is illustrated by way of example. It should be noted that some embodiments of this application use... Figure 1 Taking a system for locating radiation sources as an example, we use the fifth UAV 150 as the computing node device and the other UAVs as fusion node devices. The process of locating the radiation source is described in detail below.

[0065] S310, the radiation source emits a signal, and each node device in the equipment cluster receives the signal.

[0066] As an example of this application, Figure 1The drone swarm (as a specific example of a device swarm) contains five drones, each receiving a signal emitted by radiation source 200 from a different location. Each node device is a drone.

[0067] S320, any fusion node device samples the signal to obtain a sampled signal.

[0068] As an example of this application, the i-th UAV is any fusion node device (that is, any node device), and the i-th UAV samples the signal to obtain a sampled signal.

[0069] The sampled signal is obtained using the following formula:

[0070] r i (t)=a i x(t-τ i )+e i (t)

[0071]

[0072] Where, r i (t) represents the sampled signal of the i-th UAV, a i x(t-τ) represents the signal propagation attenuation. i ) represents the signal emitted by the radiation source, τ i Let e ​​be the time delay between the signal emitted by the radiation source and the propagation delay of the signal received by the i-th UAV. i (t) represents additive Gaussian white noise, s i Let be the position of the i-th UAV, c be the speed of signal propagation, and u be the position of the radiation source to be solved.

[0073] As another example of this application, the sampling matrix expression for the sampling signal of the i-th UAV is as follows:

[0074] r i =a i F H Γ i Fx+e i

[0075] in,

[0076]

[0077] In the formula, x is the signal matrix emitted by the radiation source, and e i Let f be the noise matrix corresponding to additive white Gaussian noise. s Let F be the sampling frequency of the i-th UAV, and F be the Fourier transform matrix. H Let Γ be the conjugate transpose of the Fourier transform matrix, N denotes performing an N-point Fourier transform, and Γ be the Γ-coordinate matrix.i Let be the phase offset matrix of the signal.

[0078] S330, any fusion node device receives the sampling signal from the adjacent device, and the fusion node device obtains the original time difference sampling signal based on its own sampling signal and the sampling signal of the adjacent device.

[0079] As an example of this application, any fusion node device is the i-th UAV, and the adjacent device (also referred to as the adjacent node device) is the j-th UAV. First, the j-th UAV sends its sampled signal to the i-th UAV. The i-th UAV then fuses its own sampled signal and the j-th UAV's sampled signal using the following formula to obtain the cross-correlation function (i.e., the original time difference sampled signal):

[0080] R i,j (γ)=r i H F H Γ(γ)Fr j

[0081] Among them, R i,j (γ) is the cross-correlation function between the i-th UAV and the j-th UAV, r i Let F be the sampling matrix corresponding to the sampling signal of the i-th UAV. H Let Γ(γ) be the conjugate transpose of the Fourier transform matrix, Γ(γ) be the phase offset matrix of the signal, γ be the time difference between the time the i-th UAV receives the signal and the time the j-th UAV receives the signal, and r be the time difference between the time the i-th UAV receives the signal and the time the j-th UAV receives the signal. j Let F be the sampling matrix corresponding to the sampling signal of the j-th UAV, and F be the Fourier transform matrix.

[0082] As a specific example of this application, any fusion node device can be Figure 1 The first drone 110 is adjacent to the second drone 120. First, the second drone 120 sends a sampling signal to the first drone 110. The first drone 110 fuses its own sampling signal with the sampling signal of the second drone 120 to obtain the cross-correlation function between the first drone 110 and the second drone 120.

[0083] S340, any fusion node device preprocesses the original time difference sampling signal to obtain the time difference sampling signal.

[0084] As an example of this application, the original time difference sampling signal is first subjected to sine fitting to obtain the first time difference sampling signal, namely:

[0085]

[0086] in, The first time difference sampling signal, The peak point of the waveform corresponding to the first time difference sampled signal. The peak width of the waveform corresponding to the first time difference sampled signal. This represents the peak value of the waveform corresponding to the first time difference sampling signal.

[0087] Secondly, the first time difference sampling signal is normalized using the sigmoid function to obtain the second time difference sampling signal. Since the target parameter extraction model has requirements for the input data, the first time difference sampling signal needs to be normalized using the sigmoid function.

[0088] The S350 uses a target parameter extraction model deployed in any fusion node device to extract features from the time difference sampling signal and obtain a feature parameter value.

[0089] As an example of this application, the time difference sampling signal is input to the target DNN model (as a specific example of a target parameter extraction model) to obtain the feature parameter values ​​output by the target DNN model. Each fusion node device can deploy three target DNN models, each of which can extract one feature parameter value to obtain a time difference estimate, a first reference feature value, and a second reference feature value. In some embodiments of this application, the first reference feature value corresponds to the peak point of the waveform corresponding to the time difference sampling signal, the second reference feature value corresponds to the peak width of the waveform corresponding to the time difference sampling signal, and the time difference estimate corresponds to the peak value of the waveform corresponding to the time difference sampling signal.

[0090] It should be noted that the target DNN model is obtained by training the DNN model using the training dataset.

[0091] S360, the current computing node device receives multiple feature parameter values ​​sent by all fusion node devices, and constructs a target cost function based on the multiple feature parameter values.

[0092] As an example of this application, the fifth UAV 150 receives multiple feature parameter values ​​sent by the first UAV 110, the second UAV 120, the third UAV 130 and the fourth UAV 140 respectively, and constructs a target cost function based on the multiple feature parameter values.

[0093] As a specific example of this application, the objective cost function is obtained by the following formula:

[0094]

[0095] Where C(u) is the target cost function, u is the location of the radiation source to be solved, and N m For the number of fusion devices, Let be the first reference characteristic value between the i-th UAV and its adjacent devices. Let be the second reference characteristic value between the i-th UAV and its adjacent devices. This is the estimated time difference between the i-th UAV and its adjacent devices.

[0096] S370, the computing node device uses a target iteration algorithm to solve the target cost function and obtain the target location of the radiation source.

[0097] As an example of this application, the fifth UAV 150 uses the gradient descent algorithm to solve for the target cost function. The specific solution process is as follows:

[0098] S371, Select any point within the target area as the initial position estimate of the radiation source, which is the h-th position value. At this time, the number of iterations h = 1.

[0099] S372, input the h-th position value into the target cost function;

[0100] S373, based on the objective cost function, obtain the h-th gradient value and the h-th adaptive iteration step size;

[0101] S374, based on the h-th gradient value and the h-th adaptive iteration step size, update the h-th position value to obtain the h+1-th position value;

[0102] S375, if the h-th adaptive iteration step size or iteration number h meets the preset condition, then output the h+1-th position value and use the h+1-th position value as the target position of the radiation source;

[0103] Otherwise, let h = h + 1 and return to S372.

[0104] As a specific example of this application, the h-th gradient value can be obtained in S373 using the following formula:

[0105]

[0106] in, Let h be the gradient value. u is the derivative of the objective cost function. h This is the value at position h.

[0107] The adaptive iteration step size h is obtained using the following formula:

[0108]

[0109] Where, λ h Let u be the adaptive iteration step size for the h-th iteration. hThis is the value at position h.

[0110] The value at position h+1 in S374 can be obtained using the following formula: Among them, u h+1 This is the value at position h+1.

[0111] As a specific example of this application, S375 may specifically include: if the adaptive iteration step size of h is less than a set threshold, then output the updated position value of h+1, or if the iteration number h is equal to a preset value, then output the position value of h+1.

[0112] In addition, in some embodiments of this application, S360 may further include: inputting multiple feature parameters into an equivalent cost function to obtain a target cost function. The equivalent cost function is a distributed computation equivalent cost function derived from a maximum likelihood estimator.

[0113] Taking the example of UAV i and UAV j being adjacent, the initial equivalent cost function derived from the maximum likelihood estimator is as follows:

[0114]

[0115] Where M represents the number of devices in the device cluster. Here, R is the correlation coefficient, is a constant, and R0 is the constant. i,j Let τ be the cross-correlation function between the i-th UAV and the j-th UAV. i,j (u) is the difference between the time when the i-th UAV receives the signal emitted by the radiation source at position u and the time when the j-th UAV receives the signal emitted by the radiation source, a i The propagation attenuation of the signal emitted by the radiation source to the i-th UAV.

[0116] To reduce computational overhead, in some embodiments of this application, the initial equivalent cost function can be simplified, that is, the fixed terms in the initial equivalent cost function can be ignored, resulting in an equivalent cost function that is more convenient for calculating the location of the radiation source, namely:

[0117]

[0118] Where, N m To determine the number of fusion node devices, R is the correlation coefficient. k Let τ be the cross-correlation function between the k-th UAV and its neighboring UAVs. k (u) is the difference between the time when the k-th UAV receives the signal emitted by the radiation source at position u and the time when the adjacent UAV receives the signal emitted by the radiation source.

[0119] To verify the effectiveness of the method for locating radiation sources provided in some embodiments of this application, a comparison table of computational resource consumption was obtained by locating radiation sources using different algorithms, as shown in Table 1. Here, T represents the time spent locating the radiation source. The different algorithms include: CDPD (Centralized Direct Position Determination), DDPD (Distributed Direct Position Determination), and IDDPD (Iterative Distributed Direct Position Determination) (i.e., the method for locating radiation sources provided in some embodiments of this application). Table 1 shows that the computational performance of the method for locating radiation sources provided in some embodiments of this application is superior to existing algorithms. While ensuring the accuracy of radiation source location, the method provided in this application can significantly reduce computational resource consumption within the same runtime.

[0120] Table 1

[0121]

[0122] Please see the appendix Figure 4 Some embodiments of this application also provide a structural diagram of a target DNN model. This target DNN model can be used to extract feature parameter values ​​from time-difference sampling signals, thereby obtaining the target location of the radiation source. Figure 4As can be seen, the target DNN model includes an input layer, which is used to collect input information. For example, the real, imaginary, and moduli of the collected cross-correlation function are organized into a three-dimensional matrix, and the matrix corresponding to the time difference sampling signal after normalization by the sigmoid function is input into the CNN layer. The matrix corresponding to the time difference sampling signal is passed through two one-dimensional CNN layers (CV layer 1, CV layer 2), which are used to extract the waveform features of the time difference sampling signal. The flattening layer (not shown in the figure) connected to the CNN layer flattens the waveform features output by the CNN layer into a one-dimensional vector. The kernel size of the one-dimensional CNN layer is 1, and the number of nodes in the two convolutional layers are 256 and 128, respectively. To prevent overfitting, the target DNN model also includes a dropout layer (not shown in the figure), where the dropout rate is set to 0.02 and the ReLU activation function is used. Then, the one-dimensional vector is input to two fully connected layers (FC Layer 1 and FC Layer 2) connected to the CNN layer. These fully connected layers map the waveform features extracted by the CNN layer to waveform feature parameters (i.e., feature parameter values). The two fully connected layers have 4096 and 2048 nodes respectively. To prevent overfitting, the target DNN model also includes a dropout layer (not shown in the figure) with a dropout rate of 0.02 and the ReLU activation function. Finally, the output layer uses a sigmoid function and a loss function to map the feature parameter values ​​output by the fully connected layer FC Layer 2 to the (0,1) interval, thus completing the feature parameter extraction.

[0123] To ensure the generalization ability of the target DNN model, in some embodiments of this application, a large amount of sample data (e.g., 200,000 samples) is collected to construct a sample dataset during DNN model training, and the sample dataset is divided into a training dataset and a validation dataset. During DNN model training, the Adam optimizer is used, with an initial bias coefficient set to 0.01, a batch size of 64 in each iteration, and an initial learning rate of 1×10⁻⁶. -5 Furthermore, the learning rate is reduced by 90% every five epochs to obtain a well-trained target DNN model.

[0124] In addition, some embodiments of this application also provide a system for locating the position of a radiation source. The system includes multiple node devices, each of which is configured to: receive a signal emitted by the radiation source and sample the signal to obtain a sampled signal; send the sampled signal to adjacent node devices; obtain a time difference sampled signal based on the received sampled signals from other adjacent node devices, and obtain a feature parameter value based on the time difference sampled signal; construct a target cost function based on the received feature parameter values ​​from other node devices, solve the target cost function, and obtain the target position of the radiation source.

[0125] Some embodiments of this application also provide a flowchart of a method for locating the location of a radiation source performed by a system for locating the location of a radiation source. The method includes: each device in a device cluster receiving a signal emitted by the radiation source and sampling the signal to obtain a sampled signal; a fusion node device receiving sampled signals sent from adjacent node devices and obtaining feature parameter values ​​through a time difference sampled signal obtained from the sampled signals; a current computing node device receiving the feature parameter values ​​sent from each fusion node device and obtaining the target location of the radiation source based on a target cost function constructed from the feature parameter values; wherein the current computing node device and the fusion node device both belong to the device cluster.

[0126] Please refer to Figure 5 , Figure 5 The diagram illustrates a block diagram of an apparatus for locating a radiation source according to some embodiments of this application. It should be understood that this apparatus corresponds to the method embodiments described above and is capable of performing the various steps involved in the method embodiments. The specific functions of this apparatus for locating a radiation source can be found in the description above; detailed descriptions are omitted here to avoid repetition.

[0127] Figure 5 The device for locating the position of a radiation source includes at least one software functional module that can be stored in a memory or embedded in the device in the form of software or firmware. The device includes: a receiving module 510, configured at least to receive multiple feature parameter values ​​transmitted by at least some devices in a device cluster, wherein each device in the device cluster is used to receive signals emitted by the same radiation source, and the feature parameter values ​​are obtained through time-difference sampling signals, which are obtained by each device by fusing its own sampling signal of the signal with sampling signals from adjacent devices. A function construction module 520, configured at least to construct a target cost function based on the multiple feature parameter values. A solution module 530, configured at least to obtain the target position of the radiation source based on the target cost function.

[0128] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the aforementioned method, and will not be elaborated further here.

[0129] Some embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can perform the operation of any of the methods corresponding to the methods for locating the location of a radiation source provided in the above embodiments.

[0130] Some embodiments of this application also provide a computer program product, which includes a computer program, wherein when the computer program is executed by a processor, it can implement the operation of any of the methods corresponding to the methods for locating the location of a radiation source provided in the above embodiments.

[0131] like Figure 6 As shown, some embodiments of this application provide an electronic device 600, which includes a memory 610, a processor 620, and a computer program stored in the memory 610 and executable on the processor 620. When the processor 620 reads the program from the memory 610 via a bus 630 and executes the program, it can implement any of the methods included in the above-described method for locating the position of a radiation source.

[0132] Processor 620 can process digital signals and can include various computing architectures. For example, it can be a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements multiple instruction set combinations. In some examples, processor 620 can be a microprocessor.

[0133] The memory 610 can be used to store instructions executed by the processor 620 or data related to the execution of instructions. These instructions and / or data may include code for implementing some or all of the functions of one or more modules described in the embodiments of this application. The processor 620 of this disclosure embodiment can be used to execute the instructions in the memory 610 to implement the methods shown above. The memory 610 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memories well known to those skilled in the art.

[0134] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0136] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for locating a radiation source, characterized in that, include: Receive multiple feature parameter values ​​sent by at least some devices in a device cluster, wherein each device in the device cluster is used to receive a signal emitted by the same radiation source, and the feature parameter values ​​are obtained by time difference sampling signals, which are obtained by any device by fusing its own sampling signals of the signal with the sampling signals of the signal from its neighboring devices. Based on the multiple feature parameter values, construct the target cost function; The target location of the radiation source is obtained based on the target cost function.

2. The method as described in claim 1, characterized in that, The feature parameter values ​​are obtained through the following method: The time difference sampling signal is obtained based on the sampling signal of the signal from any one of the devices and the sampling signal of the signal from the adjacent device; Using the target parameter extraction model deployed in any of the devices, feature extraction is performed on the time difference sampling signal to obtain a feature parameter value.

3. The method as described in claim 2, characterized in that, The time difference sampling signal is obtained by preprocessing the original time difference sampling signal. The preprocessing includes at least one of fitting processing and normalization processing.

4. The method as described in claim 3, characterized in that, The time difference sampling signal is obtained through the following method: The device samples the signal to obtain a sampled signal corresponding to the device. The adjacent device samples the signal to obtain a sampled signal corresponding to the adjacent device, and sends the sampled signal corresponding to the adjacent device to any of the adjacent devices; Each device obtains the original time difference sampling signal based on its own sampling signal and the sampling signals received from all adjacent devices; The original time difference sampling signal is subjected to the preprocessing described above to obtain the time difference sampling signal.

5. The method as described in claim 4, characterized in that, The types of the feature parameter values ​​include: time difference estimate, first reference feature value and second reference feature value, wherein the time difference estimate is used to characterize the time difference between the reception of the signal by any device and the adjacent device.

6. The method as described in claim 5, characterized in that, The target cost function is obtained through the following formula: in, Let the target cost function be... Let be the location of the radiation source to be determined. N m This refers to the number of the aforementioned devices. Let be the first reference characteristic value between the k-th device and its adjacent devices. This is the second reference characteristic value between the k-th device and its adjacent devices. This is an estimated time difference between the k-th device and its adjacent devices.

7. The method according to any one of claims 1-6, characterized in that, Obtaining the target location of the radiation source based on the target cost function includes: The target cost function is solved using a target iterative algorithm to obtain the target position.

8. The method as described in claim 7, characterized in that, The target iterative algorithm includes either Newton's iteration algorithm or gradient descent algorithm.

9. A method for locating a radiation source, characterized in that, The method includes: Each device in the equipment cluster receives the signal emitted by the radiation source and samples the signal to obtain a sampled signal; The fusion node device receives sampling signals sent from neighboring node devices and obtains feature parameter values ​​through the time difference sampling signal obtained from the sampling signals; The computing node device receives the feature parameter values ​​sent by each fusion node device, and obtains the target location of the radiation source based on the target cost function constructed according to the feature parameter values. Both the current computing node device and the fusion node device belong to the device cluster.

10. A system for locating a radiation source, characterized in that, The system includes multiple node devices, each of which is configured as follows: Receive the signal emitted by the radiation source and sample the signal to obtain a sampled signal; Send the sampling signal to adjacent node devices; Based on the received sampling signals from other adjacent node devices, a time difference sampling signal is obtained, and feature parameter values ​​are obtained based on the time difference sampling signal. Based on the characteristic parameter values ​​received from other node devices, a target cost function is constructed, and the target cost function is solved to obtain the target location of the radiation source.

Images