A low sidelobe transmitting and receiving beam forming method based on cylindrical array

Abstract

The application relates to a low-sidelobe transmitting and receiving beam forming method based on a cylindrical array, which is used for solving the problem of high sidelobe level of the existing cylindrical phased array radar. The application utilizes the cylindrical array to decouple the cylindrical array into a uniform linear array in the elevation direction and a uniform circular array or a circular arc array in the azimuth direction at a certain beam pointing direction, and performs low-sidelobe processing on the elevation direction and the azimuth direction respectively during transmitting and receiving beam forming, so that the low-sidelobe effect of the whole array is realized, wherein the transmitting beam adopts only phase weighting, and the receiving beam adopts amplitude and phase weighting.

Description

TECHNICAL FIELD

[0001] The application relates to a low-sidelobe beam forming method based on a cylindrical phased array radar, and simultaneously comprises forming of a transmitting and receiving low-sidelobe beam, and belongs to the field of phased array radar beam forming. BACKGROUND

[0002] The conformal array not only has good aerodynamic characteristics, can save carrier space and realize wide-angle scanning, but also can reduce the RCS (radar scattering cross section) and the interaction between the antenna and the radome. Therefore, the conformal array has broad development and application prospect in the field of phased array radars.

[0003] The cylindrical array, as a typical conformal array, combines the advantages of circular arrays and linear arrays. Therefore, it can inherit the distortionless pattern of the circular array in the azimuth direction while making up for the shortcomings of the circular array in the elevation direction, that is, the too wide beam in the elevation direction and the loss of target resolution in the elevation direction.

[0004] However, the cylindrical array, while combining the advantages of circular arrays and linear arrays, also inherits the shortcomings of circular arrays, mainly in the form of high sidelobe level in the azimuth direction. The high sidelobe of the transmitting beam is easy to cause the beam to irradiate the ground, greatly increasing the intensity of the ground clutter; when the sidelobe of the receiving beam is too high, the target echo entering from the sidelobe is detected, causing false alarm. However, most of the methods for reducing sidelobes involve density weighting or amplitude weighting. For the transmitting beam, although density weighting can effectively reduce the sidelobe, the complexity of the array layout and the feed network will be greatly increased, and although amplitude weighting is simple and easy to implement, it will cause difficulties in system physical implementation, such as:

[0005] 1. The phased array radar often needs to perform beam scanning quickly, and frequent and rapid control of the excitation current size not only has a very high cost, but also is difficult to implement in engineering;

[0006] 2. Amplitude weighting will cause the excitation current to decrease, and the transmitting power cannot reach the full bias, reducing the detection distance and the detection ability of weak targets of the radar.

[0007] Therefore, it is crucial to design a low-sidelobe transmitting and receiving beam forming scheme suitable for the cylindrical phased array radar. SUMMARY

[0008] The application provides a transmitting and receiving low sidelobe beam forming method based on a cylindrical array to solve the problem of high sidelobe level of the transmitting and receiving beams of the existing cylindrical phased array radar.

[0009] Further, in the receiving beam forming process, the weight of each array element is searched by using an improved virtual interference method when the azimuth dimension is processed, the improved virtual interference method refers to that K virtual interferences are uniformly set outside the 3dB beam width of the receiving beam azimuth pattern, the interference intensity of the K virtual interferences is iterated constantly to update the autocorrelation matrix of the interference receiving signal, the autocorrelation matrix of the non-ideal receiving signal is recalculated through the updated interference receiving signal autocorrelation matrix, and then the weight W t corresponding to each iteration is calculated according to the ADBF principle, and the pattern is obtained, so that the pattern is constantly close to the ideal low sidelobe pattern, when the cost function meets the requirements, the expected weight W t is obtained, and the corresponding pattern is the final pattern meeting the requirements.

[0010] Further, the calculation process of the interference intensity Γ t,k of the kth virtual interference in the tth iteration is as follows: first, the sidelobe levels of the ideal pattern Pattern0 and the pattern obtained in the (t-1)th iteration are calculated at the position of the kth virtual interference, and the difference is obtained, so that the sidelobe level difference D t,k at the position of the kth virtual interference in the tth iteration is obtained, k=1…K; then, the interference intensity Γ t,k of the kth virtual interference obtained in the last iteration is corrected by using D t-1,k , and the non-negative correction value is selected as the interference intensity Γ t,k of the kth virtual interference in the tth iteration.

[0011] Further, the weight W tThe product of the steering vector of the desired beam pointing and the inverse matrix of the autocorrelation matrix of the non-ideal received signal, the autocorrelation matrix of the non-ideal received signal comprising the autocorrelation matrix of the interference received signal and the autocorrelation matrix of the noise received signal, both of which are L rows and L columns square matrices, wherein L is the number of elements used in the azimuth beam forming, wherein the noise is constant.

[0012] Further, the cost function after the tth update of the weight W t The cost function after the tth update of the weight W t The difference between the highest sidelobe SLL1 of the obtained directional diagram and the expected overall sidelobe level SLL0.

[0013] Further, in the receiving beam forming process, the Dolph-Chebyshev synthesis method is used for low sidelobe processing in the elevation dimension; in the transmitting beam forming process, the particle swarm algorithm is used to search the phase weight of each element when low sidelobe processing is performed on the elevation dimension and the azimuth dimension respectively.

[0014] Beneficial effects

[0015] Compared with the prior art, the present application has the following advantages:

[0016] 1. The present application can greatly simplify the process of cylindrical array beam forming by performing low sidelobe processing on the elevation dimension and the azimuth dimension respectively in the transmitting and receiving beam forming processes.

[0017] 2. The transmitting beam uses the phase-only weighting method, which can realize low sidelobe transmitting beam without losing transmitting gain compared with the conventional amplitude and phase weighting method, thereby ensuring the power of the radar.

[0018] 3. The receiving beam uses the improved virtual interference method in the azimuth dimension, which can realize low sidelobe beam forming and effectively solve the problem of high sidelobe of the uniform circular / arc array directional diagram by uniformly setting a large number of virtual interferences outside the 3dB beam width and iteratively optimizing the strength of the virtual interferences to constantly approach the ideal low sidelobe directional diagram, and the sidelobe level can generally reach about -30dB in practical application. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The cylindrical phased array radar array element layout schematic diagram described in the present application;

[0020] Figure 2 The unweighted directional diagram of the 16-element linear array in the transmitting beam elevation dimension;

[0021] Figure 3 The low sidelobe directional diagram obtained by phase-only weighting of the 16-element linear array in the transmitting beam elevation dimension;

[0022] Figure 4 - the unweighted pattern of activating 22 adjacent elements in the azimuth dimension of the 66-element circular ring array;

[0023] Figure 5 - the low side lobe pattern obtained by phase-only weighting of activating 22 adjacent elements in the azimuth dimension of the 66-element circular ring array;

[0024] Figure 6 - the low side lobe pattern obtained by the virtual interference method of activating 22 adjacent elements in the azimuth dimension of the 66-element circular ring array;

[0025] Figure 7 - the low side lobe pattern obtained by the Dolph-Chebyshev synthesis method of the 16-element linear array in the elevation dimension;

[0026] Figure 8 - a schematic diagram of the method of the present application. DETAILED DESCRIPTION

[0027] The beam forming method of the present application will be explained in detail below with reference to the accompanying drawings.

[0028] The cylindrical array of the present application has MxN elements, all of which are uniformly distributed on the surface of the cylindrical array, as shown in the drawing. Figure 1 The entire cylindrical array has M circular rings from top to bottom, each of which corresponds to an N-element uniform circular array, and the adjacent circular rings are uniformly spaced. If viewed from the curved surface of the cylindrical array, N M-element uniform linear arrays are uniformly arranged on the curved surface of the cylindrical array.

[0029] The above beam forming process of the cylindrical array follows the following steps:

[0030] S01: The cylindrical array array of the present application is arranged as shown in the drawing. Figure 1 The array is numbered from bottom to top M uniform circular ring arrays, which are 1, 2, 3, …, M circular rings. As described above, the M uniform circular arrays are identical, so for a single uniform circular array, the N elements inside the circular array are numbered, which are 1, 2, 3, …, N elements. Similarly, the cylindrical array can be numbered according to N M-element uniform linear arrays, which are 1, 2, 3, …, N linear arrays. The uniform linear array is numbered inside, which is 1, 2, 3, …, M elements.

[0031] S02: When the cylindrical array transmits beam, for the convenience of solving the weight of each array element, the beam is first decoupled into two dimensions of azimuth and elevation, the azimuth beam is formed by the circular array, and the elevation beam is formed by the linear array. Considering the blocking effect of the cylindrical array, that is, when forming a beam in a certain direction, the elements on the back not only cannot help the beam, but also will raise the beam sidelobe. Assuming that the number of array elements participating in beam forming in the azimuth dimension is P and the number of array elements participating in beam forming in the elevation dimension is M, the azimuth beam is formed by a uniform circular arc array composed of P elements on the circular array, and the elevation beam is formed by all M elements in the elevation dimension.

[0032] S03: Cylindrical array transmit beam forming. In order to ensure that the gain of the transmit beam does not decrease, the transmit beam forming discards the common amplitude weighting and amplitude-phase weighting, and adopts the phase-only weighting method. This weighting method can reduce the beam sidelobe while keeping the beam width and array antenna gain basically unchanged. In order to obtain the phase values of the array elements, the present application first decouples the cylindrical array into an M-element uniform linear array in the elevation dimension and an N-element uniform circular array in the azimuth dimension, and then uses the phase weight search method based on the particle swarm algorithm to obtain the phase weight of each array element. The specific steps are as follows:

[0033] S03-1: First, set the expected sidelobe level DSLL. For the elevation dimension, the array is a uniform linear array, for the azimuth dimension, the array is a uniform circular array, and for multiple beams, the array is a circular arc array. The specific expected sidelobe level DSLL is related to the array type and the number of array elements. The more the number of array elements, the lower the DSLL. For example, the DSLL of a 16-element uniform linear array is about -15.4 dB, which is 2 dB lower than the actual sidelobe level ASLL without weighting, and the DSLL of a 66-element uniform circular array is -12 dB, which is 1.6 dB lower than the ASLL without weighting.

[0034] S03-2: Search for the phase weight to be solved by using the particle swarm algorithm. Taking the equivalent uniform linear array in the elevation dimension as an example:

[0035] a) Construct L particles, each particle has M dimensions, that is, each particle contains the phase weight of M array elements, and one dimension represents the phase weight of one array element. If the array beam pointing direction is the normal direction, then according to the symmetry of the phase, each particle has M / 2 dimensions;

[0036] b) Initialize the random value of the M dimensions of each particle, that is, the initial phase weight, and pay attention to ensure that the phase weight is between -π and π;

[0037] c) Construct the fitness function. The fitness function used in the present application is constructed as F = DSLL - ASLL (it is worth noting that the sidelobe level here is the commonly used logarithmic value);

[0038] d) Set the parameters of the particle swarm algorithm, such as learning rate c1, c2, which can be set to 1-2, weighting factor w, which can be set to 0.1-0.9, and velocity, which can be set to -1-1;

[0039] e) Search using the particle swarm algorithm, calculate the current fitness function value at each iteration, and pay attention to whether the current particle velocity and position are out of bounds. The lower bound of the velocity is set to -1, and the upper bound is set to 1. The position boundary is -π-π. If it is out of bounds, it is limited to the boundary value. When the fitness function meets the requirements (such as F < 0.1 dB), exit the loop to get the final phase weight. If the number of iterations set by the algorithm is reached and the loop is still not exited, the expected side lobe level (ASLL) should be reduced and then searched again;

[0040] The velocity represents the speed of change of the particle position in the particle swarm algorithm, and the position is the phase weight.

[0041] S03-3: Use the phase weight that meets the conditions in 2 to perform beam synthesis to obtain the directivity pattern.

[0042] Figure 3 With Figure 5 The low-sidelobe transmit directivity pattern obtained by the above method is compared with the unweighted directivity pattern in 1 and 2, and it can be seen that the sidelobes of the transmit beam are effectively suppressed. Figure 2 and Figure 4

[0043] ​S04: Cylindrical array receiving beamforming. Since more flexible DBF digital beamforming technology can be adopted for reception, amplitude weighting or amplitude and phase weighting is used for the receiving beam to achieve a lower sidelobe level. For the uniform linear array in the elevation dimension, the Dolph-Chebyshev method is adopted in this invention for pattern synthesis, which can minimize the main lobe broadening while achieving the specified sidelobe level. For the azimuth dimension, since the azimuth dimension is a circular arc array with L elements evenly distributed when forming the beam, common low-sidelobe methods for linear arrays and planar arrays cannot be used. Assume that the array has a total of Q lobes, including 1 main lobe and Q - 1 sidelobes. According to the principle of ADBF adaptive beamforming, when there are R different-direction interference signals (R < Q - 1) incident on the pattern in the spatial domain, ADBF can generate adaptive zeros in R directions, and the depth of the zeros is positively correlated with the intensity of the interference in that direction. This invention is inspired by ADBF adaptive beamforming. That is, if more than Q - 1 virtual interference signals are set in the pattern, ADBF will not be able to form adaptive zeros in the directions of each interference, but will reduce the overall interference received by the entire array to the lowest level. When the number of virtual interference signals is much larger than the number of array elements, the overall low-sidelobe effect of the array pattern can be obtained. By iteratively adjusting the intensity of the virtual interference signals, the desired sidelobe level can be achieved. The improved virtual interference method proposed in this invention is specifically as follows: By constructing a cost function and uniformly setting a large number of virtual interference signals in the region outside the 3dB beamwidth of the receiving beam azimuth dimension pattern, the "large number" means greater than 4(Q - 1), and continuously iterating the intensity of the virtual interference signals to update the autocorrelation matrix of the interference received signals, and calculating the corresponding weight W t , making the cost function gradually approach the value of 0, so as to obtain the desired weight W t , and finally obtaining a pattern that meets the requirements. Among them, the cost function after the t-th update of the weight W t is equal to the difference between the highest sidelobe SLL1 of the pattern obtained under the weight W t and the desired overall sidelobe level SLL0.

[0044] Now assume that the echo signal is an S-band narrowband signal with a frequency of 2.9 - 3.1 GHz, the radius of the uniform circular array is 0.55 meters, and the total number of array elements is N = 66. Using L = 22 adjacent array elements to form a receiving beam to illustrate the specific steps:

[0045] S04-1: First, determine the desired beam pointing θ0 and 3dB beamwidth BW0, then preset a desired overall sidelobe level SLL0 (such as SLL0 = -30 dB), and set an ideal pattern according to θ0, BW0, and SLL0, denoted as Pattern0;

[0046] S04-2: Place K initial virtual interference signals γ in the region outside the 3dB beamwidth SLL0k k = 1…K, where the size of K is related to the spatial sampling interval d1 and the number of array elements, and is generally taken as... Where T represents the total number of iterations, the initial virtual interference intensity can generally be set to 0, and no interference is placed in the area within the 3dB beamwidth BW0. It should be noted here that as subsequent iterations and optimizations continue, while the sidelobe level gradually decreases, the main lobe width expands accordingly. Therefore, it is necessary to redetermine the 3dB beamwidth after each iteration and set the interference intensity within the 3dB beamwidth to zero.

[0047] S04-3: Based on the ADBF principle, the weight W0 without virtual interference is obtained, as shown in the following formula:

[0048]

[0049] W0 represents the initial weights at t=0. W0 is then used to form the initial beam pattern, denoted as Pattern1, where S0 is the steering vector pointing towards the desired beam. This represents the inversion of the autocorrelation matrix U0 of the initial non-ideal received signal, U t U is the autocorrelation matrix of the non-ideal received signal. t Including the autocorrelation matrix of the received signal with interference and the autocorrelation matrix of the received signal with noise, U t Both autocorrelation matrices are square matrices with L rows and L columns, where L is the number of array elements used in azimuth beamforming. The noise remains unchanged, and the autocorrelation matrix of the initial interference received signal can be set to an all-zero matrix.

[0050] S04-4: Iterate over the interference intensity of each virtual interference.

[0051] Wherein, the interference intensity Γ of the k-th (k=1…K) virtual interference in the t-th (t=1…T) iteration is... t , k The update process is as follows: At the k-th virtual interference location, the sidelobe levels of the ideal pattern Pattern0 and the pattern obtained in the (t-1)-th iteration are calculated and subtracted to obtain the sidelobe level difference D. t,k In the first iteration of t, the sidelobe levels of the ideal pattern Pattern0 and the initial pattern Pattern1 are calculated and subtracted.

[0052] Then, using the sidelobe level difference D t,k The interference intensity Γ of the k-th virtual interference in the previous iteration t-1,k A correction is made. Since the corrected value may be negative, a non-negative correction value is selected as the updated interference intensity Γ of the k-th virtual interference. t,k The specific calculation formula is as follows:

[0053] σ t，k t-1，k +u*D t，k

[0054] Γ t，k t，k

[0055] wherein, for Γ t-1,k , when t = 1, i.e. the first iteration, the initial value is set to 0, and u is a constant, generally between 1 and 10, and u can be adjusted according to actual conditions to make the iteration result converge faster;

[0056] Then, the autocorrelation matrix of the interference receiving signal is updated by continuously iterating the interference intensity of the K virtual interferences, and the autocorrelation matrix of the non-ideal receiving signal is recalculated through the updated autocorrelation matrix of the interference receiving signal, which is a known common knowledge and is as follows:

[0057] The latest interference intensity Γ t,k of the updated kth virtual interference is used to recalculate the autocorrelation matrix R t,k of the interference receiving signal calculated at the kth virtual interference in the tth iteration, and the autocorrelation matrix R t,K of the interference receiving signal calculated at the Kth virtual interference in the tth iteration is obtained through iterative calculation, wherein the calculation formula of R t,k is as follows:

[0058]

[0059] In the above formula, V k represents the corresponding vector of the array manifold matrix at the kth virtual interference position, i.e. the kth column of the array manifold matrix, which is L rows and 1 column, and R t,k is L rows and L columns;

[0060] The autocorrelation matrix U t,K of the non-ideal receiving signal in the current tth iteration is obtained through R t , and the specific formula is as follows:

[0061] U t = R t,K + R n

[0062] In the above formula, R n is the autocorrelation matrix of the noise receiving signal, which is L rows and L columns.

[0063] After obtaining U t , the weight W t corresponding to the tth iteration is calculated through the formula in S04-3; then the weight W​​​t Calculate the current pattern.

[0064] S04-5: Construct the cost function F, and update the weight W t Calculate the cost function F, when the cost function reaches the requirement (for example, F≤0.2dB) in a certain loop, then jump out of the loop, at this time the weight obtained by iteration is the solution; if the number of iteration loops reaches the set number and still not jump out, it means that the level of sidelobe is too low or the number of iteration is too small, increase the number of iteration or relax the requirement of sidelobe level until the cost function reaches the requirement, at this time the weight W t The formed pattern meets the requirement. The cost function F is the difference between the highest sidelobe SLL1 of the pattern obtained by the weight W and the expected overall sidelobe level SLL0, the closer F is to 0, the closer the current pattern is to the expected pattern, the weight W t after the tth update is equal to the cost function F t The highest sidelobe SLL1 of the pattern obtained by the weight W

[0065] Figure 6 Compared with Figure 7 The low sidelobe pattern of the azimuth and elevation receiving beam obtained by the above method, and the low sidelobe pattern of the azimuth and elevation receiving beam obtained by the prior art Figure 4 and Figure 2 Compared with the prior art, the receiving pattern and the transmitting pattern are the same without weighting, and the sidelobe is greatly reduced.

Claims

1. A method for low sidelobe transmit / receive beamforming based on a cylindrical array, characterized in that: When a cylindrical array is pointed to a certain beam, it is decoupled into a uniform linear array for elevation and a uniform circular array or arc array for azimuth. When the transmit and receive beams are formed, low sidelobe processing is performed on the elevation and azimuth dimensions respectively, and the low sidelobe effect of the whole array is finally achieved. Among them, the transmit beam uses phase-only weighting and the receive beam uses amplitude-phase weighting. During the receiving beamforming process, when performing low sidelobe processing in the azimuth dimension, an improved virtual interference method is used to search for the weights of each array element. This improved virtual interference method involves constructing a cost function and uniformly distributing it in the region outside the 3dB beamwidth of the receiving beam's azimuth pattern. K A virtual interference, constantly affecting K The interference intensity of each virtual interference is iteratively used to update the autocorrelation matrix of the received interference signal. The autocorrelation matrix of the non-ideal received signal is then recalculated using the updated autocorrelation matrix. Finally, the weights corresponding to each iteration are calculated based on the ADBF principle. And the radiation pattern is continuously approximated to the ideal low sidelobe radiation pattern. When the cost function meets the requirements, the desired weights are obtained. At this point, the corresponding radiation pattern is the final radiation pattern that meets the requirements.

2. The method for low sidelobe transmit / receive beamforming based on a cylindrical array according to claim 1, characterized in that: No. The second iteration The interference strength of each virtual interference The calculation process is as follows: First, in the first... k At each of the virtual interference locations, calculate the ideal radiation pattern Pattern0 and the... t The sidelobe levels of the radiation pattern obtained in the -1 iteration are subtracted to obtain the current level. t The second iteration k Sidelobe level difference at each virtual interference location , Then, using Correct the first iteration obtained k The interference strength of each virtual interference Choose a non-negative correction value as the first... The second iteration k The interference strength of each virtual interference .

3. A method for low sidelobe transmit / receive beamforming based on a cylindrical array according to claim 1 or 2, characterized in that: The weights corresponding to each iteration It is equal to the product of the steering vector of the desired beam direction and the inverse of the autocorrelation matrix of the non-ideal received signal. The autocorrelation matrix of the non-ideal received signal includes the autocorrelation matrix of the interference received signal and the autocorrelation matrix of the noise received signal. Both autocorrelation matrices are square matrices of L rows and L columns, where L is the number of array elements used in azimuth beamforming, and the noise remains unchanged.

4. The method for low sidelobe transmit / receive beamforming based on a cylindrical array according to claim 3, characterized in that: The weight is updated for the tth time. The subsequent cost function is the weight. The highest sidelobe of the obtained radiation pattern With the expected overall sidelobe level The difference.

5. The method for low sidelobe transmit / receive beamforming based on a cylindrical array according to claim 1, characterized in that: During the receiving beamforming process, the Dolph-Chebyshev synthesis method is used to process low sidelobes in the elevation dimension; during the transmitting beamforming process, when processing low sidelobes in the elevation and azimuth dimensions respectively, the particle swarm optimization algorithm is used to search for the phase weight of each array element.

Images