CN103076600B - Radar target recognition method based on multi-azimuth pulse-E technology - Google Patents

Abstract

The invention discloses a radar target recognition method based on multi-azimuth E pulse technology. Assuming that there are M targets in the radar target library, use actual measurement or simulation to obtain the echo data of multiple azimuth angles of each target, and establish a radar target echo database. The echo data in the library are used as training samples; use actual measurement or simulation to obtain The first step is to use the echo data of any azimuth angle of each target in the radar target library as a test sample; use the multi-azimuth E pulse training algorithm and training samples to construct multiple E pulses for each target in the radar target library; use the test The sample tests the recognition effect of the multi-directional E-pulse technology. The scheme of the invention significantly improves the recognition probability and anti-noise ability of the E-pulse technology for radar targets.

Description

Radar Target Recognition Method Based on Multi-Azimuth E Pulse Technology

technical field

The invention belongs to the technical field of radar target recognition, in particular to a radar target recognition method based on multi-directional E-pulse technology.

Background technique

Since the 1990s, the pole characteristics of radar targets have been widely used in short-pulse radar target recognition. The pole is the complex natural resonant frequency of the target, which is only determined by the characteristics of the target itself, such as shape, size, material, etc., and the target attitude and Radar polarization is irrelevant, so the use of pole features for target recognition can overcome the shortcomings of features such as scattering centers that vary with target attitude. The initial target recognition method based on pole features is to extract the poles of known targets in advance and build a database. After receiving the time-domain echo of the radar target to be identified, extract the poles from the echo and compare them with the poles in the database. thereby identifying the target. Due to the low signal-to-noise ratio of the received radar echo signal, it is difficult to accurately extract the pole from it in actual operation, so this method is not practical. In the early 1990s, some scholars proposed the E pulse technology, which allows the poles of radar targets to be extracted in advance in an environment with a large signal-to-noise ratio, and then these poles are used to construct E pulses to construct each target After receiving the echo of the target to be identified, the E pulse of each known target is used to convolve with the time after the echo. If the convolution result is 0, the target to be identified is identified as the corresponding known target. Obviously, this method does not need to extract the target pole from the echo signal received in real time, which reduces the requirement on the signal-to-noise ratio of the echo signal.

However, it should be pointed out that the traditional E-pulse technology usually uses the echo of a single azimuth of the target to extract the pole of the target, and then only constructs one E-pulse for each target, and considers that the constructed E-pulse is consistent with the echo of any azimuth of the corresponding target The convolution is all 0, so as to achieve the purpose of identifying the target. The theoretical basis for this understanding is that the target pole has nothing to do with the orientation of the incident wave, but it ignores that the residue corresponding to the pole is related to the orientation. The residue represents the contribution of the corresponding pole to the back time of the echo. At some angles, some poles The corresponding residue may be very small, which means that the corresponding pole cannot be well excited. Using a pole extraction algorithm such as the matrix beam method, only the pole with a relatively large residue can be extracted, and the poles with a large residue in different orientations are not necessarily the same , which leads to the fact that the poles extracted from different orientations are not necessarily the same. In this case, if only the pole extracted from a certain angle is used to construct the E-pulse, the E-pulse may have no effect on echoes from other azimuths, which will eventually lead to a low recognition rate when the traditional E-pulse technology is actually used.

Contents of the invention

The purpose of the present invention is to provide a radar target recognition method based on multi-azimuth E-pulse technology. The method constructs multiple E-pulse for different angle areas of each target, and covers the different-angle echoes of the entire target, which can significantly Improving the recognition probability and anti-noise ability of E-pulse technology for radar targets can provide important reference materials for radar target recognition methods.

The technical scheme that realizes the object of the present invention is:

The first step, assuming that there are M targets in the radar target library, using actual measurement or simulation to obtain the echo data of multiple azimuth angles of each target, and establishing the radar target echo database, the echo data in the library are all used as training samples;

In the second step, use the actual measurement or simulation to obtain the echo data of any azimuth angle of each target in the radar target library in the first step, as a test sample;

The third step is to use the multi-directional E pulse training algorithm and the training samples in the first step to train multiple E pulses for each target in the radar target library;

The fourth step is to use the test samples in the second step to test the recognition effect of the multi-directional E-pulse technology. Even if the test sample in the second step is convolved with multiple E pulses of each target, the target corresponding to the minimum value among all convolution results is the recognition result.

The echo data in the first step is based on the echo data obtained by the short-pulse radar system. The rule for selecting multiple azimuth angles is to select an angle every 5 to 10 degrees within the azimuth variation range of the target.

Assuming that in the first step, the azimuth angle range of the multiple echo data of each target is φ ⁱ _min to φ ⁱ _max , and the angular interval is △φ _i , the specific steps of the third step multi-azimuth E pulse training algorithm are as follows:

The first step is to initialize i=1, j=1, where i represents the target number, j represents the E pulse number, let n _i =0, and n _i represents the number of E pulses constructed by the i-th target, for the second to For each target in the Mth target, 5~8 angles are evenly selected from the angle range of its azimuth angle, and the echoes of the corresponding angles are obtained from the radar target echo database, and then the E pulse technology is used to obtain the corresponding angle echoes according to each angle. Constructs an E-pulse for the echoes of , thus constructing multiple temporary E-pulses for each of the 2nd to Mth targets;

In the second step, the entire echo angle area (φ ⁱ _min , φ ⁱ _max ) of the i-th target is marked as an invalid area, indicating that no E pulse is currently valid for the echo in this area;

The third step is to record the central angle of the invalid angle area of the i-th target as φ _j , use the echo of this angle to construct an E pulse through E pulse technology, and record it as E ⁱ _j ;

The fourth step is to calculate the convolution of an echo in the invalid area of the i-th target and the E pulses of all targets. If the E pulse with the smallest convolution result is the E pulse of the i-th target, it indicates that the i-th target is correctly identified, the angle corresponding to the echo is included in the effective area, indicating that the E pulse of the i-th target is valid for the echo of this angle; when all the echoes in the invalid area of the i-th target are executed In one-step operation, store E ⁱ _j , n _i =n _i +1;

The fifth step, if all the angles of the i-th target have been included in the effective area, then record n _i as the total number of E pulses of the i-th target, and go to the sixth step; otherwise, j=j+1, go to third step;

The sixth step, i=i+1; if i<M, clear all temporary E pulses of the i-th target, and set j=1, go to the second step; if i>M, end the execution.

Compared with the prior art, the present invention has the following remarkable advantages: (1) The correct recognition rate is improved. Multiple E-pulses are constructed for different angle areas of each target, so that the constructed E-pulses can cover the echoes of all directions of the target, and the identification probability is significantly improved; (2) The ability to resist noise is enhanced. After adding noise in the echo, the radar target recognition method based on multi-azimuth E-pulse technology has a stronger recognition ability than the traditional method. the

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Description of drawings

Figure 1 is a model diagram of the F22, F35, and VFY218 radar targets, (a) F22 model (b) F35 model (c) VFY218 model.

Fig. 2 is the recognition rate of the multi-directional E-pulse technology proposed by the present invention under different signal-to-noise ratios.

Figure 3 shows the recognition rate of the traditional E-pulse technology under different signal-to-noise ratios, (a) the recognition rate of the E-pulse constructed using the 10-degree echo (b) the recognition rate of the E-pulse constructed using the 90-degree echo (c) Discrimination rate of E-pulses constructed using 170-degree echoes.

Detailed ways

The present invention is a radar target recognition method based on multi-azimuth E-pulse technology. Firstly, the echo data of multiple azimuth angles of each target are obtained through actual measurement or simulation as training data, and then the echo data of each target at any azimuth is obtained through actual measurement or simulation. The data is used as test data, and multiple E pulses are trained for each target in the radar target library through the multi-directional E pulse training algorithm, and finally the test data is recognized. This method can improve the correct recognition rate and has strong anti-noise ability.

In the present invention, based on the radar target identification method of multi-azimuth E pulse technology, the steps are as follows:

The first step, assuming that there are M targets in the radar target library, directly use the short-pulse radar to obtain the echoes of multiple azimuth angles of the target, or establish a geometric model of the target, and use the simulation algorithm in computational electromagnetics to obtain multiple azimuth angles of the target These echoes are used to form the radar target echo database, and the echo data in the database are used as training samples. Note that the general rule for selecting multiple azimuth angles is to select an angle every 5~10 degrees within the azimuth range of the target. For example, for an axisymmetric target such as an airplane, assuming that the circle around the airplane in the horizontal direction is 360 degrees, the nose corresponds to 0 and 360 degrees, and the tail corresponds to 180 degrees, then we only need to select the half of 0~180 degrees to calculate the angles of multiple angles Echo, the angular interval is generally selected as 10 degrees;

The second step is to use actual measurement or simulation to obtain the echo data of any azimuth angle of each target in the first step radar target library. At this time, the azimuth angle can be selected arbitrarily, and the obtained echo is used as a test sample;

In the third step, multiple E-pulses are trained for each target in the radar target library using the multi-azimuth E-pulse training algorithm and the training samples from the first step. Assuming that in the first step, the angle range of the echo data of multiple azimuth angles of each target is φ ⁱ _min to φ ⁱ _max , and the angular interval is △φ _i , the specific steps of the multi-azimuth E-pulse training algorithm are as follows:

1. Initialize i=1, j=1, where i represents the target number, j represents the E pulse number, let n _i =0, n _i represents the number of E pulses constructed by the i-th target, for the 2nd to the Mth For each of the three targets, 5 to 8 angles are uniformly selected from the angle range of its azimuth angle, and the echoes of the corresponding angles are obtained from the radar target echo database, and then the echoes of each angle are obtained by using the E pulse technology. Wave constructs one E-pulse, thereby constructing multiple temporary E-pulses for each of the 2nd to Mth targets. For aircraft targets, echoes at five angles of 0°, 45°, 90°, 135°, and 180° can be selected within the range of 0° to 180°, and an E pulse is constructed for each angle of echo using E pulse technology. Thus, five E pulses are obtained;

2. Mark the entire echo angle area (φ ⁱ _min , φ ⁱ _max ) of the i-th target as an invalid area, indicating that there is currently no E pulse valid for the echo in this area;

3. Record the center angle of the invalid angle area of the i-th target as φ _j , use the echo of this angle to construct an E pulse through the E pulse technology, and record it as E ⁱ _j ;

4. Calculate the convolution of an echo in the invalid area of the i-th target with the E pulses of all targets. If the E pulse with the smallest convolution result is the E pulse of the i-th target, it indicates that the i-th target is correct Identification, the angle corresponding to the echo is included in the effective area, indicating that the E pulse of the i-th target is valid for the echo of the angle. When this step is performed on all the echoes in the invalid area of the i-th target, store E ⁱ _j , n _i =n _i +1;

5. If all angles of the i-th target have been included in the effective area, record ni as the total number of E pulses of the i-th target, and go to step 6. Otherwise, j=j+1, go to step three;

Sixth, i=i+1. If i<M, clear all temporary E pulses of the i-th target, set j=1, and go to step 2; if i>M, end the execution.

The E-pulse technology mentioned in the present invention "constructs an E-pulse technology using the E-pulse technology for the echo of a certain angle" is a mature technology, and it is introduced in documents [1], [2], [3] The specific implementation steps are introduced in detail in the second section of document [1], the third section of document [2], and the second to fifth sections of document [3]. The documents are as follows: [1] Xiao Shunping, Guo Guirong, Zhuang Zhaowen, "Target Recognition Method Based on Waveform Synthesis", Electronic Countermeasures, No. 2, 1993. [2]Rothwell, E.; Nyquist, D.; Kun-Mu Chen; Drachman, B.; , "Radar target discrimination using the extinction-pulse technique," Antennas and Propagation, IEEE Transactions on , vol.33, no. 9, pp. 929- 937, Sep 1985[3]Rothwell, E.; Kun-Mu Chen; Nyquist, D.; Weimin Sun; , "Frequency domain E-pulse synthesis and target discrimination," Antennas and Propagation, IEEE Transactions on, vol.35, no.4, pp. 426-434, Apr 1987.

The fourth step is to use the test samples in the second step to test the recognition effect of the multi-directional E-pulse technology. In the third step, multiple E pulses are constructed for each target in the radar target library. The method of testing the recognition effect is to calculate the convolution value of a test sample and multiple E pulses of each target, and all convolution results The target corresponding to the minimum value in is the recognition result. This process is carried out for all test samples, and the correct recognition rate can be calculated by counting the number of correct recognitions.

In order to verify the effectiveness of the method of the present invention, combined with different objectives, the traditional E-pulse technology and the multi-directional E-pulse technology proposed by the present invention are used to carry out a simulation experiment comparison. First, use the commercial software ANSYS to model the three radar targets of F22, F35, and VFY218, as shown in Figure 1. Then use the method of moments to calculate the single-station scattering field of the target from 3MHz to 600MHz, and the frequency sweep interval is 3MHz, then add a Gaussian window to the frequency sweep data and perform inverse Fourier transform to obtain the target in a certain direction with a modulated Gaussian pulse as The time-domain echo of the excitation, by which the VV polarization time-domain echo of each target is calculated, the elevation angle is 5°, the azimuth angle varies from 0° to 180°, and the interval is 5°, 0° corresponds to the nose Cone direction, obviously, each target can get 37 echoes from different angles. In the following experiments, we use odd echoes as training data and even echoes as testing data.

Fig. 2 has shown the recognition rate and the average recognition rate of the present invention to F22, F35, VFY218 three kinds of targets under different signal-to-noise ratio environments. Figure 3 shows the recognition rate and average recognition rate of the traditional E-pulse technology for F22, F35, and VFY218 three targets in different SNR environments. The recognition rate and average recognition rate of the test, Figure 3(b) is the recognition rate and average recognition rate of the three targets using the 90-degree echo to construct the E pulse test, and Figure 3(c) is the three targets using the 170-degree echo Recognition rate and average recognition rate of wave structure E pulse test. It can be found by observation that when the signal-to-noise ratio is 20db, the average recognition rate of the present invention can reach 95%, and the average recognition rate of the E pulse that only utilizes 10 degree or 90 degree or 170 degree echo structure is 47%, respectively. 60%, 41%, which proves that the present invention can significantly improve the recognition rate and anti-noise ability.

Claims (3)

1. a radar target recognition method based on multi-directional E-pulse technology, is characterized in that the steps are as follows: The first step, assuming that there are M targets in the radar target library, using actual measurement or simulation to obtain the echo data of multiple azimuth angles for each target, establish a radar target echo database, and the echo data in the library are used as training samples; The second step is to use the actual measurement or simulation to obtain the echo data of any azimuth angle of each target in the radar target library in the first step, as a test sample; In the third step, multiple E pulses are constructed for each target in the radar target library by using the multi-directional E pulse training algorithm and the training samples in the first step; The fourth step is to use the test sample in the second step to test the recognition effect of the multi-directional E pulse technology; Assuming that in the first step, the azimuth angle range of the multiple echo data of each target is φ ⁱ _min to φ ⁱ _max , and the angular interval is △φ _i , the specific steps of the third step multi-azimuth E pulse training algorithm are as follows: 3.1, initialize i=1, j=1, where i represents the target number, j represents the E pulse number, let n _i =0, n _i represents the number of E pulses constructed by the i-th target, for the 2nd to the Mth For each of the three targets, 5 to 8 angles are uniformly selected from the angle range of its azimuth angle, and the echoes of the corresponding angles are obtained from the radar target echo database, and then the echoes of each angle are obtained by using the E pulse technology. wave constructs an E-pulse, thereby constructing multiple temporary E-pulses for each of the 2nd to Mth targets; 3.2. Mark the entire echo angle area (φ ⁱ _min , φ ⁱ _max ) of the i-th target as an invalid area, indicating that there is currently no E pulse valid for the echo in this area; 3.3. Record the center angle of the invalid angle area of the i-th target as φ _j , use the echo of this angle to construct an E pulse through E pulse technology, and record it as E ⁱ _j ; 3.4. Calculate the convolution of an echo in the invalid area of the i-th target with the E pulses of all targets. If the E pulse with the smallest convolution result is the E pulse of the i-th target, it indicates that the i-th target is correct Identification, the angle corresponding to the echo is included in the effective area, indicating that the E pulse of the i-th target is valid for the echo of this angle; when this step is performed for all echoes in the invalid area of the i-th target When , store E ⁱ _j , n _i =n _i +1; 3.5, if all angles of the i-th target have been included in the effective area, then record n _i as the total number of E pulses of the i-th target, go to 3.6; otherwise, j=j+1, go to 3.3; 3.6, i=i+1; if i<M, clear all temporary E pulses of the i-th target, and set j=1, go to 3.2; if i>M, end the execution. 2. the radar target recognition method based on multi-directional E-pulse technology according to claim 1, is characterized in that: the echo data in the first step is the echo data that obtains based on the short-pulse radar system, multiple azimuth angles The rule of angle selection is to select an angle every 5~10 degrees within the azimuth angle range of the target. 3. the radar target recognition method based on multi-directional E-pulse technology according to claim 1, is characterized in that: make the test sample in the second step and a plurality of E-pulse convolutions of each target in the 4th step, all The target corresponding to the minimum value in the convolution result is the recognition result.