A fractal detection method for continuous wave radar in rain clutter environment

Abstract

The present invention discloses a fractal detection method for continuous wave radar in rain clutter environment, which can solve the problem of effectively detecting weak target echoes in the main lobe area of ​​rain clutter. The implementation steps are: de-frequency modulating and two-dimensional FFT processing of antenna received signals to obtain complex envelope signals containing cloud and rain reflections and weak moving target echoes; normalization and constant false alarm detection to obtain large energy peak detection points; construct fractal processing data of detection points; on the basis of solving fractal vectors of different dimensions, obtain the mean and variance of fractal features of different dimensions; and detect target echoes according to fractal features. The present invention can be applied to continuous wave radars to effectively eliminate false alarms caused by rain clutter when the radar antenna received signal contains a large amount of rain clutter, and at the same time, utilize constant false alarm and fractal two-step detection methods to have the ability to detect weak target echoes in the main lobe area of ​​rain clutter.

Description

Technical Field

[0001] The invention belongs to the field of radar technology and relates to radar target detection. Specifically, it is a continuous wave radar fractal detection method in a rain clutter environment. The method is mainly used in complex meteorological detection environments such as clouds and rain to achieve effective detection of weak target echoes in the main lobe area of ​​rain clutter. Background Art

[0002] With the development of unmanned aerial technology, continuous wave intelligence reconnaissance radar is required to effectively detect micro-targets such as low-altitude drones in complex terrains such as cities, suburbs, woodlands, and mountains, as well as in complex weather conditions such as strong winds and heavy rains. The traditional target detection method is mainly a constant false alarm detection method, including CA_CFAR, GO_CFAR, OS_CFAR, etc., which is essentially an energy detection method that uses the ratio of the target echo energy to the energy of the noise background to achieve target detection. This type of target detection method can achieve better detection of targets under a noise background. However, if the signal received by the radar contains rain clutter data, since rain clutter is different from ground stationary clutter and has Doppler characteristics similar to those of moving target echoes, it is difficult to use traditional clutter cancellation methods such as MTI to cancel it, resulting in a high false alarm rate during the detection process, affecting the radar detection performance and tracking effect. At the same time, more importantly, since rain clutter has a certain spectral width, the moving target echo will have a greater probability of falling into the main lobe area of ​​rain clutter, resulting in traditional detection methods that not only produce false alarms, but also produce large missed alarms. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the traditional energy-based constant false alarm detection method, and propose a continuous wave radar fractal detection method in a rain clutter environment, which can be applied to complex meteorological detection environments such as clouds and rain, and realize effective detection of weak target echoes in the main lobe area of ​​rain clutter.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A continuous wave radar fractal detection method in a rain clutter environment comprises the following steps:

[0006] Step 1: De-frequency modulate and perform two-dimensional FFT processing on the signal received by the continuous wave radar to obtain the complex envelope signal S containing the clutter reflected by clouds and rain and the echo of weak moving targets. c (n,i), n = 1, 2, ... N, i = 1, 2, ... I, N represents the total number of range units, I represents the total number of Doppler units;

[0007] Step 2: S c (n,i) is normalized along the Doppler dimension according to the following formula to obtain the normalized echo point signal Secho (n,i):

[0008]

[0009] in,

[0010] Step 3: For each normalized echo point signal S echo (n,i) performs constant false alarm detection, when S echo When the amplitude of (n,i) is less than the detection threshold η, S is directly determined. echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0, the echo point detection is finished; otherwise, go to step 4 to continue execution;

[0011] Step 4: Construct the echo point signal S echo The fractal processing matrix S of (n,i) x (k,m):S x (k,m)=S echo (nR ra +k-1,iR do +m-1),k=1,2,…,2R ra +1,m=1,2,…,2R do +1

[0012] Among them, R ra and R do They are the range unit range and Doppler unit range of fractal processing respectively;

[0013] Step 5: Solve the fractal processing matrix S x The maximum value S max , judge S max Is it equal to S echo (n,i), if it is not equal, it is directly judged as a non-target detection point, and the detection result S is set det (n, i) = 0, the echo point detection is completed, otherwise go to step 6 to continue execution;

[0014] Step 6: Using fractals to process the matrix S x (k,m) solve the echo point signal S echo (n,i) fractal vector X in p dimension p , p takes values ​​of 2 and 3 respectively;

[0015] Step 7: Solve for the p-dimensional fractal vector X p The mean and variance

[0016]

[0017]

[0018] Where B represents the p-dimensional fractal vector X p Length;

[0019] Step 8: Solve for the ratio R of the fractal means of the 2nd and 3rd dimensions a and the maximum variance V M :

[0020]

[0021]

[0022] Step 9: Determine R a Is it greater than the threshold? and V M Is it less than the threshold? If all are satisfied, then S is determined echo (n,i) is the target echo detection point, and the detection result S is set det (n,i)=1, otherwise S is determined echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0.

[0023] Furthermore, in step 6, X p The specific solution steps are as follows:

[0024] Step 6-1, set the echo point signal S echo The fractal vector of (n,i) in dimension p is X p The empty set, that is, X p =[];

[0025] Step 6-2, initialize distance point s=1;

[0026] Step 6-3, update the fractal vector X p as follows:

[0027] X p =[X p S z S y ]

[0028] Among them, S z and S y Represent the fractal processing matrix S x (k,m) is the data of the right and left half of Doppler. The specific expression is as follows:

[0029] S z =[S x (Rra +1+s,R do +1+ps)S x (R ra +1-s,R do +1+ps)]

[0030] S y =[S x (R ra +1+s,R do +1-p+s)S x (R ra +1-s,R do +1-p+s)]

[0031] Step 6-4, update the distance starting point s=s+1, if s is less than p, go to step 6-3 to continue execution, otherwise the operation ends.

[0032] The present invention has the following advantages:

[0033] 1. The present invention can effectively eliminate false alarms caused by rain clutter. The present invention adds a first-level fractal detection method on the basis of the traditional constant false alarm method. Through the differentiated fractal characteristics of rain clutter and target echo, the peaks caused by rain clutter can be effectively identified and eliminated, so it has a lower false alarm rate than the traditional method.

[0034] 2. The present invention has a better target detection probability. Since the present invention adds a first-level fractal detection method on the basis of the traditional constant false alarm method, a detection threshold can be set in the constant false alarm detection process. Then, the false alarm introduced by the low threshold is eliminated by using the fractal detection technology based on the feature of the latter level, so that the present invention can achieve detection of weaker target echoes under the condition of a lower false alarm rate.

[0035] 3. The present invention can be applied to continuous wave radars to effectively eliminate false alarms caused by rain clutter when the radar antenna receives signals containing a large amount of rain clutter. At the same time, the constant false alarm and fractal two-step detection method is used to detect weak target echoes in the main lobe area of ​​rain clutter. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the continuous wave radar fractal detection method in rain clutter environment of the present invention;

[0037] Figure 2 It is the result of two-dimensional FFT processing of the echo signal containing rain clutter received by the antenna;

[0038] Figure 3 This is the result of performing constant false alarm detection on the data after two-dimensional FFT processing;

[0039] Figure 4 It is the result obtained by using the fractal detection method used in this patent on the data after two-dimensional FFT processing;

[0040] Figure 5 It is the result of tracking a pedestrian target located in the main lobe area of ​​rain clutter using the patented fractal detection method. DETAILED DESCRIPTION

[0041] Reference Figure 1 , a fractal detection method for continuous wave radar in rain clutter environment, the implementation steps are as follows:

[0042] Step 1: After de-modulation and two-dimensional FFT processing of the signal received by the continuous wave radar, the complex envelope signal S containing the clutter reflected by clouds and rain and the echo of weak moving targets is obtained. c (n,i), n = 1, 2, ... N, i = 1, 2, ... I, N represents the total number of range units, I represents the total number of Doppler units;

[0043] Step 2: S c (n,i) is normalized along the Doppler dimension according to the following formula to obtain the normalized signal S echo (n,i):

[0044]

[0045] in

[0046] Step 3: For each normalized echo point signal S echo (n,i) performs constant false alarm detection, when S echo If the amplitude of (n,i) is less than the detection threshold η, it is directly judged as a non-target detection point and the detection result S is set. det (n,i)=0, the echo point detection is finished; otherwise, when S echo If the amplitude of (n,i) is greater than or equal to the detection threshold η, go to step 4 to continue execution.

[0047] Step 4: Constructing the Echo Dot S echo (n,i) fractal processing matrix S x , its expression is: S x (k,m)=S echo (nR ra +k-1,iR do +m-1),k=1,2,…,2R ra +1,m=1,2,…,2R do +1

[0048] R ra and Rdo They are the distance unit range and Doppler unit range of fractal processing respectively.

[0049] Step 5: Solve the fractal processing matrix S x The maximum value S max , judge S max Is it equal to S echo (n,i), if not equal, directly determine S echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0, the echo point detection is finished, otherwise go to step 6 to continue execution;

[0050] Step 6: Use fractals to process data S x (k,m) solve for echo point S echo (n,i) fractal vector X in p dimension p , p takes values ​​of 2 and 3 respectively, X p The specific solution steps are as follows:

[0051] 6-1) First set the echo point S echo The fractal vector of (n,i) in dimension p is X p The empty set, that is, X p =[];

[0052] 6-2) Initialize distance point s=1;

[0053] 6-3) Update the fractal vector X p The formula is as follows:

[0054] X p =[X p S z S y ]

[0055] Where S z and S y Represent fractal processing data S x (k,m) is the data of the right and left half of Doppler. The specific expression is as follows

[0056] S z =[S x (R ra +1+s,R do +1+ps)S x (R ra +1-s,R do +1+ps)]

[0057] S y =[S x (R ra +1+s,Rdo +1-p+s)S x (R ra +1-s,R do +1-p+s)]

[0058] 6-4) Update the distance starting point s=s+1. If s is less than p, go to step 6-3) to continue execution, otherwise the operation ends.

[0059] Step 7: Solve for the p-dimensional fractal vector X p The mean and variance As shown below:

[0060]

[0061]

[0062] Where B represents the p-dimensional fractal vector X p Length.

[0063] Step 8: Solve for the ratio R of the fractal means of the 2nd and 3rd dimensions a and the maximum variance V M , as follows:

[0064]

[0065]

[0066] Step 9: Determine R a Is it greater than the set threshold? and V M Is it less than the set threshold? If all are satisfied, then S is determined echo (n,i) is the target echo detection point, and the detection result S is set det (n,i)=1, otherwise S is determined echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0.

[0067] 1) Test verification conditions:

[0068] Next, the performance of the present invention is experimentally analyzed. A continuous wave radar is used for data collection, and the data collection environment is a heavy rain environment in the suburbs of the city, that is, the collected echo data contains not only static clutter in the field but also moving target clutter caused by cloud and rain echoes. The target types collected during this test process are slow targets such as pedestrians. Since pedestrians move slowly, the Doppler caused by pedestrian movement just falls in the main lobe area of ​​rain clutter, so rain clutter overlaps with pedestrian target echoes, resulting in greater difficulty in target detection.

[0069] 2) Experimental results

[0070] Figure 2 This is the result of directly performing two-dimensional FFT processing on the echo signal containing rain clutter received by the antenna. It can be seen from the figure that on the range Doppler diagram of the echo signal, there are not only clutter near the zero frequency, but also strong peaks caused by rain clutter in the non-zero Doppler region. These peaks will cause more false alarms to the traditional detection method during the detection process.

[0071] Figure 3 The data after two-dimensional FFT processing is directly performed on the echo signal containing rain clutter received by the antenna. The processing results of the data are processed using the traditional OS_CFAR detection algorithm. It can be seen that a large number of false alarms caused by rain clutter appear in the rain clutter area, resulting in the inability to achieve effective target detection, indicating that traditional methods are difficult to achieve signal processing in rain clutter environments.

[0072] Figure 4 The data after two-dimensional FFT processing is directly performed on the echo signal containing rain clutter received by the antenna, and the result is obtained using the fractal detection method used in this patent invention. It can be seen from the figure that the fractal detection method of the present invention can not only effectively eliminate the false alarm caused by rain clutter, but also realize weak target echo detection in the main lobe area of ​​​​rain clutter, so it has a better target detection effect.

[0073] Figure 5 After using the fractal detection method of the present invention, pedestrian target detection and tracking effects are performed. It can be seen that it can realize continuous detection and tracking of pedestrian targets, indicating that the detection method proposed in the present invention can realize effective detection in a rain clutter environment, thereby achieving accurate detection and tracking of weak moving targets.

[0074] In summary, the present invention utilizes the difference in fractal characteristics between rain clutter and moving target echoes, and adds a first-level fractal detection process on the basis of the traditional constant false alarm detection method, thereby eliminating the false alarm caused by rain clutter, and realizing effective detection of weak target echoes in the main lobe area of ​​rain clutter. Therefore, compared with the traditional constant false alarm detection method based only on energy, the present invention has a more efficient detection capability.

Claims

1. A continuous wave radar fractal detection method in a rain clutter environment, characterized in that: The steps include: Step 1: De-frequency modulate and perform two-dimensional FFT processing on the signal received by the continuous wave radar to obtain the complex envelope signal S containing the clutter reflected by clouds and rain and the echo of weak moving targets. c (n,i), n = 1, 2, ... N, i = 1, 2, ... I, N represents the total number of range units, I represents the total number of Doppler units; Step 2: S c (n,i) is normalized along the Doppler dimension according to the following formula to obtain the normalized echo point signal S echo (n,i): in, Step 3: For each normalized echo point signal S echo (n,i) performs constant false alarm detection, when S echo When the amplitude of (n,i) is less than the detection threshold η, S is directly determined. echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0, the echo point detection is finished; otherwise, go to step 4 to continue execution; Step 4: Construct the echo point signal S echo The fractal processing matrix S of (n,i) x (k,m): S x (k,m)=S echo (n-R ra +k-1,i-R do +m-1),k=1,2,…,2R ra +1,m=1,2,…,2R do +1 Among them, R ra and R do They are the range unit range and Doppler unit range of fractal processing respectively; Step 5: Solve the fractal processing matrix S x The maximum value S max , judge S max Is it equal to S echo (n,i), if it is not equal, it is directly judged as a non-target detection point, and the detection result S is set det (n, i) = 0, the echo point detection is completed, otherwise go to step 6 to continue execution; Step 6: Using fractals to process the matrix S x (k,m) solve the echo point signal S echo (n,i) fractal vector X in p dimension p , p takes values ​​of 2 and 3 respectively; Step 7: Solve for the p-dimensional fractal vector X p The mean and variance Where B represents the p-dimensional fractal vector X p Length; Step 8: Solve for the ratio R of the fractal means of the 2nd and 3rd dimensions a and the maximum variance V M : Step 9: Determine R a Is it greater than the threshold? and V M Is it less than the threshold? If all are satisfied, then S is determined echo (n,i) is the target echo detection point, and the detection result S is set det (n,i)=1, otherwise S is determined echo (n,i) is a non-target detection point, and the detection result S is set det (n,i)=0.

2. The continuous wave radar fractal detection method in a rain clutter environment according to claim 1, characterized in that: In step 6, X p The specific solution steps are as follows: Step 6-1, set the echo point signal S echo The fractal vector of (n,i) in dimension p is X p The empty set, that is, X p =[]; Step 6-2, initialize distance point s=1; Step 6-3, update the fractal vector X p as follows: X p =[X p S z S y ] Among them, S z and S y Represent the fractal processing matrix S x (k,m) is the data of the right and left half of Doppler. The specific expression is as follows: S z =[S x (R ra +1+s,R do +1+p-s)S x (R ra +1-s,R do +1+p-s)] S y =[S x (R ra +1+s,R do +1-p+s)S x (R ra +1-s,R do +1-p+s)] Step 6-4, update the distance starting point s=s+1, if s is less than p, go to step 6-3 to continue execution, otherwise the operation ends.

Images