Multi-channel radar target detection method based on signal-to-noise ratio weighted fusion

Abstract

This invention proposes a multi-channel radar target detection method based on signal-to-noise ratio (SNR) weighted fusion. The implementation steps are as follows: initializing a multi-channel radar target detection system; calculating local test statistics for each radar channel and performing a monotonic transformation; calculating the SNR weighted value for each local test statistic for each radar channel; calculating the global threshold of the multi-channel radar target detection system through a signal fusion center; and obtaining the target detection result. In this invention, the signal fusion center performs weighted fusion of the local test statistics of the multi-channel radar using the SNR weighted value of each radar channel to obtain the global test statistics. When the local test statistics have the same distribution, the global test statistics contain both local test statistics information and SNR information. When the local test statistics have different distributions, the global test statistics contain both local test statistics and SNR information, reducing information loss and effectively improving the detection probability.

Description

Technical Field

[0001] This invention belongs to the field of signal processing technology and relates to a multi-channel radar target detection method based on signal-to-noise ratio weighted fusion. Background Technology

[0002] A multi-channel radar system consists of a fusion center and multiple radar channels. The multiple radar channels receive echo signals from the same monitoring area. Each radar channel preprocesses the received signal to form a local test statistic, which is then transmitted to the fusion center via wireless or wired communication. The fusion center fuses all the local test statistics to determine whether the target exists within the monitoring area.

[0003] In multi-channel radar systems, various factors such as radar channel parameters, the number of range cells receiving echo signals, echo signal distribution, different background noise levels, and different detection algorithms can lead to different distributions of the local test statistics calculated for each radar channel. If the distributions of the local test statistics are the same, these test statistics are said to be isomorphic; otherwise, they are heterogeneous. Due to these differences between radar channels, the test statistics of practical multi-channel radars are usually heterogeneous.

[0004] In multi-channel radar systems, the optimal fusion design criterion is to directly apply the Niemann-Pearson likelihood ratio fusion criterion to the echo signals of local radar channels. While this method is optimal, it requires knowledge of specific parameters such as amplitude, phase, and noise covariance of the received echo signals from each station. These parameters are usually unknown. Therefore, local radar channels typically preprocess the echo signals to form local test statistics, and then apply the likelihood ratio fusion criterion to these local test statistics to obtain the global test statistics. When some parameters in the received echo signals are unknown, common constant false alarm rate (CFAR) detection algorithms can be used in local radar channels to preprocess the echo observations, such as the generalized likelihood ratio detection algorithm, the cell-average CFAR detection algorithm, and the adaptive matched filtering algorithm. If the amplitude and noise power in the echo signals are known, a square-law detection algorithm can be used in local radar channels to preprocess the echo signals. However, the distribution function of the global test statistics obtained by likelihood ratio fusion based on local test statistics is generally difficult to express with a specific closed-form expression, making it difficult to accurately obtain the global threshold. In addition, for scenarios where multiple parameters of the echo signal are unknown, a generalized likelihood ratio fusion criterion can be used. This fusion criterion performs maximum likelihood estimation on the unknown parameters of all echo signals, and then substitutes the estimated values ​​of the unknown parameters into the likelihood ratio of the echo signal to obtain a global test statistic. When the echo signals received by the radar channels are statistically independent, this statistic is expressed as a weighted sum of the test statistics of the generalized likelihood ratio detection algorithm for the local radar channels. Furthermore, there is a decision fusion criterion, commonly including the AND and OR fusion criteria. During decision fusion, the local radar channel determines whether a target exists in the monitored area based on its received echo signals. The local decision signals (0 and 1) are transmitted to the fusion center. After receiving the local decision signals, the fusion center uses the AND and OR fusion criteria to make a fusion decision. Because the fusion center only receives binary decision signals from the radar channels, there is significant information loss, resulting in a generally low detection probability for multi-channel radar target detection systems using decision fusion detection algorithms.

[0005] When the distribution of local test statistics is the same, the global test statistics of existing multi-channel radar target detection methods are obtained by weighted summation of local test statistics, where the weights do not include signal-to-noise ratio (SNR) information. Theoretical analysis shows that the higher the SNR of the echo signal received by a local radar channel, the higher the probability of that local radar channel detecting the target. Therefore, the corresponding local test statistics have a higher weight in the fusion process and play a greater role in the final decision.

[0006] When the distribution of local test statistics is different, the global test statistics of existing multi-channel radar target detection methods are obtained by receiving binary decision information transmitted from local channels to the fusion center and performing decision fusion calculation. Not only does it not contain signal-to-noise ratio information, but the amount of information contained in the global test statistics is also small due to the binary decision information received by the fusion center. This results in a low detection probability after the multi-channel radar target detection system adopts the decision fusion detection algorithm. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and propose a multi-channel radar target detection method based on signal-to-noise ratio weighted fusion to solve the technical problem of low detection probability in the prior art.

[0008] The technical approach of this invention is as follows: First, a multi-channel radar target detection system is established. Then, the local test statistic for each radar channel is calculated. A monotonic function is selected, and a monotonic transformation is performed on the local test statistic to obtain a new local test statistic. The log-likelihood ratio function is then calculated. The Taylor expansion point of the log-likelihood ratio function is then calculated. Using the first derivative of the log-likelihood ratio, the signal-to-noise ratio weighting value of the monotonically transformed local test statistic for each radar channel is calculated. Based on the weighting value and the monotonically transformed local test statistic, the global test statistic is calculated. The global test statistic is then used to determine whether a target exists in the target detection region Ω at the fusion center. The specific implementation steps are as follows:

[0009] (1) Initialize the multi-channel radar target detection system:

[0010] Initializing the multi-channel radar target detection system includes the signal fusion center C. e And N radar channels R = {r n |n=1,2,…,N}, target detection region Ω, where N≥2, r n This represents the nth radar channel;

[0011] (2) Calculate the local test statistic for each radar channel and perform a monotonic transformation on it:

[0012] Each radar channel r n Based on the received echo signal, its own local test statistic z is calculated using its own detection algorithm. n And through the selected monotonic function g n (x) with respect to the local test statistic z n Perform a monotonic transformation to obtain r for each radar channel. n The corresponding local test statistic q after monotonic transformation n ;

[0013] (3) Calculate the signal-to-noise ratio weighted value of each local test statistic for each radar channel:

[0014] Each radar channel r n Based on the assumption of target existence, the estimated signal-to-noise ratio of the echo signal is λ. n time q n probability density function Calculate the log-likelihood ratio function l n (q|λ n Taylor expansion point of ) And the log-likelihood ratio function l is used n (q|λ n ), through λ n and Calculate the local test statistic q n Signal-to-noise ratio weighted value w n We obtain the signal-to-noise ratio weighted set W = {w_n} for N test statistics Q. n |n=1,2,…,N}, where l n (q|λ n The expression λ represents the estimated signal-to-noise ratio of the echo signal under the assumption that the target exists. n Time radar channel r n The test statistic q n The log-likelihood ratio function;

[0015] (4) The signal fusion center calculates the global threshold of the multi-channel radar target detection system:

[0016] Signal Fusion Center C e Through each radar channel r n Signal-to-noise ratio weighted value w n The global test statistic Y of the multi-channel radar target detection system is obtained by weighted fusion of N local test statistics; at the same time, the distribution function F0(y) of the global test statistic Y of the multi-channel radar target detection system under the assumption that the target does not exist in the target detection area Ω is calculated; and the global threshold g of the multi-channel radar target detection system is calculated through F0(y).

[0017] (5) Obtain the target detection results:

[0018] Signal Fusion Center C e Determine whether Y≥g holds true. If yes, then there is a target in the target detection region Ω; otherwise, there is no target in Ω.

[0019] Compared with the prior art, the present invention has the following advantages:

[0020] In this invention, each radar channel undergoes a monotonic transformation of its local test statistics using a monotonic function that satisfies certain conditions. The signal fusion center then weights and fuses the local test statistics of the multi-channel radar using the signal-to-noise ratio (SNR) weighted value of each radar channel to obtain the global test statistics of the multi-channel radar target detection system. When the local test statistics have the same distribution, the global test statistics contain both local test statistics information and SNR information. When the local test statistics have different distributions, the global test statistics contain both local test statistics and SNR information. This avoids the significant information loss inherent in existing technologies and effectively improves the detection probability. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the implementation of the present invention;

[0022] Figure 2 This is a simulation comparison chart of the detection probabilities of the present invention and existing technologies in different scenarios. Detailed Implementation

[0023] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Example 1: The number of radar channels is set to 4. The generalized likelihood ratio detection algorithm is used for the local test statistics of the radar channels. The total signal-to-noise ratio of the received signals of the radar channels is set to be between 0 and 20 dB, and the signal-to-noise ratio is 10:6:2:1.

[0025] Reference Figure 1 The present invention includes the following steps:

[0026] Step 1) Initialize the multi-channel radar target detection system:

[0027] Initializing the multi-channel radar target detection system includes the signal fusion center C. e And N radar channels R = {r n |n=1,2,…,N}, target detection region Ω, where N≥2, r n This represents the nth radar channel;

[0028] In this Example 1, N = 4.

[0029] Step 2) Calculate the local test statistic for each radar channel and perform a monotonic transformation on it:

[0030] Each radar channel r n Based on the received echo signal, its own local test statistic z is calculated using its own detection algorithm. n And through the selected monotonic function g n (x) with respect to the local test statistic z nPerform a monotonic transformation to obtain r for each radar channel. n The corresponding local test statistic q after monotonic transformation n ;

[0031] Each radar channel r n The chosen monotonic function g n (x), the selection criteria are:

[0032] (2a) According to each radar channel r n The local test statistic z n Calculate the random variable T0, and calculate the distribution function of the random variable T0 under the assumption that the objective does not exist.

[0033]

[0034]

[0035] Among them, a n Let a represent a random real number. n >0, This represents the probability density function of T0 under the assumption that the target does not exist;

[0036] (2b) Determine the distribution function Can it be expressed by a closed-form expression? If so, for each radar channel r n The chosen monotonic function is g. n If (x) = x, then choose a distribution function that satisfies x. The monotonic function g that can be expressed in closed form n (x):

[0037]

[0038]

[0039] q n =g n (z n )

[0040] Where, q n Indicates the use of satisfying The monotonic function of the closed-form expression with respect to the local test statistic z n The local test statistic obtained by performing a monotonic transformation, T′0, represents the result of passing through q. n The calculated random variable, Let T′0 represent the probability density function under the assumption that the target does not exist.

[0041] By selecting the monotonic function g n (x) with respect to the local test statistic z nPerform a monotonic transformation to obtain the local test statistic q after the monotonic transformation. n The transformation formula is:

[0042] q n =g n (z n ).

[0043] Each radar channel receives single or multiple pulse echo signals from the same monitoring area. The received echo signals may contain only signals from the detection unit, or signals from both the detection unit and reference units. There may be only one detection unit, containing both clutter and target echo signals, or only clutter echo signals. There may be multiple reference units, containing only clutter echo signals. The number of reference units can vary between different radar channels. Each radar can employ different signal processing algorithms for the received echo signals, resulting in different configurations of the test statistics.

[0044] Each radar channel r n The local test statistic z n Generalized likelihood ratio detection algorithm, square law detection algorithm, or cell average constant false alarm rate detection algorithm can be used, but are not limited to these three detection algorithms.

[0045] In Example 1, all radar channels are tested using the test statistic of the generalized likelihood ratio detection algorithm. The calculation method of the test statistic of the generalized likelihood ratio detection algorithm is described below.

[0046] Through the nth radar channel r n The received echo signal contains the echo signal from one detection unit and K. n The echo signals of each reference cell, wherein the received echo signal vector x of the detection cell. n (0), and the nth radar channel r n The received echo signal vector x of the j-th reference element n (j), j = 1, 2, ..., K n Calculate r n Detection statistics z n The detection statistics Z of the radar channel set R are obtained, Z = {z} n The detection statistic z is obtained using the generalized likelihood ratio detection algorithm for |n=1,2,3,4}. n The calculation formula is:

[0047]

[0048] in, This indicates that in n radar channels r n The observed target steering vector, L nRepresents the nth radar channel r n The number of snapshots, K n Represents the nth radar channel r n The number of reference units, S n The calculation formula is as follows:

[0049]

[0050] in,(·) H This indicates that the vector inside the parentheses is conjugate transposed.

[0051] In Example 1, the number of reference units for the four radar channels is set to be different, namely 32, 29, 26, and 23, and the number of snapshots for all four radar channels is set to 16. Under the assumption that no target exists within the target detection area Ω, z... n The distribution of follows a Pareto distribution, and its probability density function is expressed as:

[0052]

[0053] Due to the random variable z n The distribution function of a random variable obtained by weighted summation is difficult to obtain using a closed-form expression; therefore, a monotonic function g needs to be chosen. n (x)=(K n +1)log(x) with respect to z n Perform a monotonic transformation, and the test statistic after the monotonic transformation is expressed as q. n =(K n +1)log(z n Under the assumption that the target does not exist within the target detection region Ω, z n After monotonic transformation, q is obtained. n The probability density function is expressed as:

[0054]

[0055] It can be seen that q n The distribution of follows an exponential distribution. In this case, the distribution function of the random variable T′0 obtained by weighted summation of the exponentially distributed random variables is... Although it is not a closed expression, it can be expressed mathematically, therefore, given the value of t'... The value can be solved exactly, and similarly, given... The value of t' can also be accurately calculated after the value of t is obtained.

[0056] Step 3) Calculate the signal-to-noise ratio weighted value for each local test statistic for each radar channel:

[0057] Each radar channel r n Based on the assumption of target existence, the estimated signal-to-noise ratio of the echo signal is λ.n time q n probability density function Calculate the log-likelihood ratio function l n (q|λ n Taylor expansion point of ) And the log-likelihood ratio function l is used n (q|λ n ), through λ n and Calculate the local test statistic q n Signal-to-noise ratio weighted value w n We obtain the signal-to-noise ratio weighted set W = {w_n} for N test statistics Q. n |n=1,2,…,N}, where l n (q|λ n The expression λ represents the estimated signal-to-noise ratio of the echo signal under the assumption that the target exists. n Time radar channel r n The test statistic q n The log-likelihood ratio function;

[0058] Log-likelihood ratio function l n (q|λ n Taylor expansion point of ) and the local test statistic q n Signal-to-noise ratio weighted value w n The calculation formulas are as follows:

[0059]

[0060]

[0061]

[0062] in, The signal-to-noise ratio estimate in the echo signal under the assumption of target existence is λ. n The nth radar channel r n The corresponding monotonic transformation of the detection statistic q n probability density function; l n (q|λ n ) represents q n The log-likelihood ratio function, This represents the first derivative of the function [·] with respect to the independent variable [*]. This indicates that the (·) function is in the independent variable The value at time, This indicates the nth radar channel r under the assumption that the target does not exist. n Detection statistic q nThe probability density function is log(·), which represents the logarithm to the base natural number (·).

[0063] In Example 1, the log-likelihood ratio function l n (q|λ n The calculation formula is:

[0064]

[0065] Where Φ represents the confluence hypergeometry function, Where Γ(·) represents the gamma function. and The expression is

[0066]

[0067]

[0068] Among them, w n The local test statistic q represents n The weighted value, Taylor expansion point The specific values ​​can be obtained through numerical calculation using the computer software MATLAB:

[0069]

[0070] Step 4) The signal fusion center calculates the global threshold of the multi-channel radar target detection system:

[0071] Signal Fusion Center C e Through each radar channel r n Signal-to-noise ratio weighted value w n The global test statistic Y of the multi-channel radar target detection system is obtained by weighted fusion of N local test statistics; at the same time, the distribution function F0(y) of the global test statistic Y of the multi-channel radar target detection system under the assumption that the target does not exist in the target detection area Ω is calculated; and the global threshold g of the multi-channel radar target detection system is calculated through F0(y).

[0072] Signal Fusion Center C e Through each radar channel r n weighted value w n The weighted fusion of N local test statistics is calculated using the following formula:

[0073]

[0074] The distribution function F0(y) is calculated using the following formula:

[0075]

[0076]

[0077] Among them, f Y (y|H0) represents the probability density function of the global test statistic Y of the multi-channel radar target detection system under the assumption that the target does not exist. This represents the convolution operation. This indicates the nth radar channel r under the assumption that the target does not exist in Ω. n The corresponding local test statistic q after monotonic transformation n The probability density function.

[0078] The global threshold g is calculated using the following formula:

[0079]

[0080] Among them, P fa This represents the global false alarm probability of a multi-channel radar target detection system. It represents the inverse function of the distribution function F0(y).

[0081] In Example 1, under the assumption that the target does not exist, the global test statistic Y of the multi-channel radar target detection system follows a weighted exponential distribution, and the threshold is accurately calculated using the distribution function of Y. In their 2011 paper, "Signal fusion-based target detection algorithm for spatial diversity radar," Zhou SH and Liu HW mentioned that the distribution function of a random variable following a weighted exponential distribution can be expressed by a complex mathematical expression after complex derivation. Although this mathematical expression is not a closed-form expression, it allows for the accurate solution of the global threshold. Accurate solution of the global threshold is one of the conditions for the successful implementation of the detector.

[0082] Step 5) Obtain the target detection results:

[0083] Signal Fusion Center C e Determine whether Y≥g holds true. If yes, then there is a target in the target detection region Ω; otherwise, there is no target in Ω.

[0084] Example 2 adjusts the detection algorithm used in the radar channels. Radar channels r1, r2, and r3 employ the generalized likelihood ratio detection algorithm, with corresponding reference cell numbers K1, K2, and K3 of 32, 29, and 26, respectively, and snapshot numbers L1, L2, and L3 of 16 each. Radar channel r4 uses the square-law detection algorithm. The total signal-to-noise ratio (SNR) of the received signals in the radar channels is set between 0 and 25 dB, with an SNR ratio of 10:6:2:1, while other parameters remain unchanged.

[0085] In Example 2, radar channel r4 uses a square-law detection algorithm, and the square-law detector is a single-pulse signal detector. Therefore, in radar channel r4, the echo signal contains only one echo pulse signal x4(0) from the detection unit, rather than a vector form. In the square-law detector, the background noise power and signal-to-noise ratio are known. n Detection statistics z n The calculation formula is:

[0086]

[0087] Furthermore, since the distribution of the detection statistic z4 of r4 within the target detection region Ω under the assumption of a target follows an exponential distribution, a monotonic function is chosen:

[0088]

[0089] For r n Detection statistics z n The test statistic q is obtained by performing a monotonic transformation. n , making q n Weighted summation yields a random variable that follows a weighted exponential distribution. Then, the chosen monotonic function g is used... n (x) with respect to the local test statistic z n Perform a monotonic transformation to obtain the local test statistic q after the monotonic transformation. n The transformation formula is:

[0090]

[0091] In Example 2, the local test statistic q for each radar channel n The formula for calculating the signal-to-noise ratio weighting is as follows:

[0092] First, the local test statistic q n The log-likelihood ratio function l n (q|λ n The calculation formula is:

[0093]

[0094] The nth radar channel r n The local test statistic q n The signal-to-noise ratio weighted value is calculated using the following formula:

[0095]

[0096] Among them, Taylor expansion point The calculation formula is as follows:

[0097]

[0098] Wherein, probability density function For the specific expressions of n = 1, 2, 3, see Example 1. The probability density function distribution of the local test statistic q4 of the 4th radar channel under the assumptions of target presence and absence is expressed as:

[0099]

[0100] Where μ4 represents the power of the noise signal received by the fourth radar channel.

[0101] The global threshold calculation method and the method for making a decision on the target detection area in Example 2 are the same as those in Example 1.

[0102] The technical effects of the present invention will be further explained below with reference to simulation experiments:

[0103] 1. Simulation conditions and content:

[0104] Simulation conditions: In a multi-channel radar system, four radar channels and one signal fusion center are set up. The false alarm probability of the fusion center and each radar channel is set to 10. -4 .

[0105] In Example 1, all four radar channels use the generalized likelihood ratio detection algorithm. Each radar channel has 16 snapshots, and the number of reference cells are 32, 29, 26, and 23, respectively. The total signal-to-noise ratio of the received echo signals from the four radar channels ranges from 0dB to 20dB, and their signal-to-noise ratio ratio is 10:6:2:1.

[0106] In Example 2, radar channels 1, 2, and 3 all use the generalized likelihood ratio detection algorithm, with 16 snapshots and 32, 29, and 26 reference cells respectively. Radar channel 4 uses the square law detection algorithm, with a background noise power of 1. The total signal-to-noise ratio of the received echo signals from the four radar channels ranges from 0dB to 25dB, and their signal-to-noise ratio ratios are 7:5:3:1.

[0107] The simulation process used the following hardware and software environment: Hardware: Intel Core i7-4790 CPU, 3.60GHz, 8GB main memory; Software: Windows 10 Professional, MATLAB simulation software.

[0108] Simulation experiments were conducted on the detection probabilities of Embodiment 1 of the present invention and the equally weighted and modified weighted multi-channel radar target detection algorithm when the distribution forms of the local test statistics are the same, and on Embodiment 2 of the present invention and the decision fusion detection algorithm when the distribution forms of the local test statistics are different. The results are as follows: Figure 2(a) and Figure 2 As shown in (b).

[0109] 2. Simulation Result Analysis:

[0110] Reference Figure 2 (a) The horizontal axis represents the sum of the signal-to-noise ratios of the echo signals received by all radar channels, in dB, ranging from 0 dB to 20 dB. The vertical axis represents the detection probability value of the detection algorithm, ranging from 0 to 1.

[0111] Reference Figure 2 (b) The horizontal axis represents the sum of the signal-to-noise ratios of the echo signals received by all radar channels, in dB, ranging from 0 dB to 25 dB. The vertical axis represents the detection probability value of the detection algorithm, ranging from 0 to 1.

[0112] from Figure 2 As can be seen in (a), since the present invention introduces signal-to-noise ratio information into the weights, the detection probability curves obtained by the present invention (SW) are all higher than the detection probability curves of the equal weighted multi-channel generalized likelihood ratio detection algorithm (SGLRT) and the modified weighted multi-channel generalized likelihood ratio detection algorithm (MGLRT) in the prior art, which can improve the detection probability by up to about 20%.

[0113] from Figure 2 As can be seen in (b), the detection method (SW) based on signal-to-noise ratio weighted fusion adopted in this invention has a higher detection probability than the existing decision fusion detection algorithm. In the decision fusion algorithm, the OR fusion criterion detection algorithm has a higher detection probability than the AND fusion criterion detection algorithm. Compared with the OR fusion criterion detection algorithm, this invention can improve the detection probability by up to about 25%.

[0114] In summary, this invention can not only improve the detection probability of traditional multi-channel radar target detection systems based on the generalized likelihood ratio detection algorithm in scenarios where the distribution of local test statistics is the same, but also improve the detection probability of multi-channel radar target detection systems in scenarios where the distribution of local test statistics is different.

Claims

1. A multi-channel radar target detection method based on signal-to-noise ratio weighted fusion, characterized in that, Includes the following steps: (1) Initialize the multi-channel radar target detection system: Initializing the multi-channel radar target detection system includes the signal fusion center C. e And N radar channels R = {r n |n=1,2,…,N}, target detection region Ω, where N≥2, r n This represents the nth radar channel; (2) Calculate the local test statistic for each radar channel and perform a monotonic transformation on it: Each radar channel r n Based on the received echo signal, its own local test statistic z is calculated using its own detection algorithm. n And through the selected monotonic function g n (x) with respect to the local test statistic z n Perform a monotonic transformation to obtain r for each radar channel. n The corresponding local test statistic q after monotonic transformation n ; (3) Calculate the signal-to-noise ratio weighted value of each local test statistic for each radar channel: Each radar channel r n Based on the assumption of target existence, the estimated signal-to-noise ratio of the echo signal is λ. n time q n The probability density function, Calculate the log-likelihood ratio function l n (q|λ n Taylor expansion point And the log-likelihood ratio function l is used n (q|λ n ), through λ n and Calculate the local test statistic q n Signal-to-noise ratio weighted value w n We obtain the signal-to-noise ratio weighted set W = {w_n} for N test statistics Q. n |n=1,2,…,N}, where l n (q|λ n The expression λ represents the estimated signal-to-noise ratio of the echo signal under the assumption that the target exists. n radar channel r n The test statistic q n The log-likelihood ratio function; (4) The signal fusion center calculates the global threshold of the multi-channel radar target detection system: Signal Fusion Center C e Through each radar channel r n Signal-to-noise ratio weighted value w n The global test statistic Y of the multi-channel radar target detection system is obtained by weighted fusion of N local test statistics; at the same time, the distribution function F0(y) of the global test statistic Y of the multi-channel radar target detection system under the assumption that the target does not exist in the target detection area Ω is calculated; and the global threshold g of the multi-channel radar target detection system is calculated through F0(y). (5) Obtain the target detection results: Signal Fusion Center C e Determine whether Y≥g holds true. If yes, then there is a target in the target detection region Ω; otherwise, there is no target in Ω.

2. The method according to claim 1, characterized in that, Each radar channel r described in step (2) n The chosen monotonic function g n (x), the selection criteria are: (2a) According to each radar channel r n The local test statistic z n Calculate the random variable T0, and calculate the distribution function of the random variable T0 under the assumption that the objective does not exist. Among them, a n Let a represent a random real number. n >0, f T0 (t|H0) represents the probability density function of T0 under the assumption that the target does not exist; (2b) Determine the distribution function Can it be expressed by a closed-form expression? If so, for each radar channel r n The chosen monotonic function is g. n If (x) = x, then choose a distribution function that satisfies x. The monotonic function g that can be expressed in closed form n (x): q n =g n (z n ) Where, q n Indicates the use of satisfying The monotonic function of the closed-form expression with respect to the local test statistic z n The local test statistic obtained by performing a monotonic transformation, T′0, represents the result of passing through q. n The calculated random variable, Let T′0 represent the probability density function under the assumption that the target does not exist.

3. The method according to claim 1, characterized in that, The step (2) described above involves selecting a monotonic function g. n (x) with respect to the local test statistic z n Perform a monotonic transformation to obtain the local test statistic q after the monotonic transformation. n The transformation formula is: q n =g n (z n )。 4. The method according to claim 1, characterized in that, The log-likelihood ratio function l mentioned in step (3) n (q|λ n Taylor expansion point and the local test statistic q n Signal-to-noise ratio weighted value w n The calculation formulas are as follows: in, The signal-to-noise ratio estimate in the echo signal under the assumption of target existence is λ. n The nth radar channel r n The corresponding monotonic transformation of the detection statistic q n probability density function; l n (q|λ n ) represents q n The log-likelihood ratio function, This represents the first derivative of the function [·] with respect to the independent variable [*]. This indicates that the (·) function is in the independent variable The value at time, This indicates the nth radar channel r under the assumption that the target does not exist. n Detection statistic q n The probability density function is log(·), which represents the logarithm to the base natural number (·).

5. The method according to claim 1, characterized in that, The signal fusion center C mentioned in step (4) e Through each radar channel r n Signal-to-noise ratio weighted value w n The weighted fusion of N local test statistics is calculated using the following formula:

6. The method according to claim 1, characterized in that, The distribution function F0(y) mentioned in step (4) is calculated using the following formula: Among them, f Y (y|H0) represents the probability density function of the global test statistic Y of a multi-channel radar target detection system under the assumption that the target does not exist. This represents the convolution operation. This indicates the nth radar channel r under the assumption that the target does not exist in Ω. n The corresponding local test statistic q after monotonic transformation n The probability density function.

7. The method according to claim 1, characterized in that, The global threshold g mentioned in step (4) is calculated using the following formula: Among them, P fa This represents the global false alarm probability of a multi-channel radar target detection system. It represents the inverse function of the distribution function F0(y).

Images