阅读说明：本技术 一种毫米波雷达非直视多目标定位方法 (Non-direct-view multi-target positioning method for millimeter wave radar ) 是由 崔国龙 郑晶月 郭世盛 吴佩仑 魏雅琦 贾超 孔令讲 杨晓波 于 2021-08-03 设计创作，主要内容包括：本发明公开一种毫米波雷达非直视多目标定位方法,应用于城市环境拐角后的非直视目标定位技术领域,针对现有技术缺少精准定位非直视多目标的算法方案的问题,本发明利用毫米波雷达对隐藏在拐角后的多个非直视目标进行准确定位；根据电磁波的传播现象,分析了多径信号主要传播路径为双程一次反射路径；本发明通过联合所有通道相位信息应用非参数化的最小方差无失真响应算法估计镜像目标到达角；结合场景几何先验信息,通过坐标转换、镜像映射等处理完成了非直视多目标准确定位。(The invention discloses a millimeter wave radar non-direct-view multi-target positioning method, which is applied to the technical field of non-direct-view target positioning behind corners of urban environments and aims at solving the problem that the prior art lacks an algorithm scheme for accurately positioning non-direct-view multi-targets; analyzing that the main propagation path of the multipath signal is a two-way primary reflection path according to the propagation phenomenon of the electromagnetic wave; estimating the arrival angle of a mirror image target by combining all channel phase information and applying an unparameterized minimum variance distortionless response algorithm; and combining scene geometric prior information, and completing non-direct-view multi-target accurate positioning through coordinate conversion, mirror image mapping and other processing.)

1. A millimeter wave radar non-direct-view multi-target positioning method is characterized by comprising the following steps of: the target is located at a non-direct-view position behind a corner, the method comprising the steps of:

s1, establishing an electromagnetic propagation model based on the primary reflection path, and calculating the distance of the mirror image target;

s2, obtaining the azimuth angle of the mirror image target by adopting an MVDR angle estimation method for all channel signals at the mirror image target distance unit;

and S3, calculating the position coordinates of the non-direct-view target by combining the scene geometric prior information.

2. The off-line-of-sight multi-target positioning method for the millimeter wave radar as recited in claim 1, wherein the step S1 specifically comprises the following substeps:

s11, eliminating static clutter in the environment by adopting a moving target indication method, and reserving echoes of the moving target;

s12, obtaining a target range profile by adopting pulse compression processing;

s13, the range profile is subjected to incoherent superposition, and the range profile after accumulation is detected by a one-dimensional unit average constant false alarm detection method to obtain the mirror image target distance R.

3. The off-line-of-sight multi-target positioning method for the millimeter wave radar as claimed in claim 2, wherein the step S32 is specifically as follows: and carrying out incoherent superposition on the range images, calculating a detection threshold corresponding to each range unit in the superposed signals, if the value of the range unit is greater than or equal to the calculated detection threshold, indicating that a target exists in the range unit, thereby obtaining the mirror image target distance, otherwise, indicating that no target exists in the range unit.

4. The off-line-of-sight multi-object positioning method for the millimeter wave radar as recited in claim 3, wherein the calculation formula of the detection threshold of the specific distance unit is as follows:

wherein, P_fRepresentsProbability of false alarm, N_rRepresenting the unit u to be detected_m,nNumber of reference cells in (h, j).

5. The off-line-of-sight multi-target positioning method for the millimeter wave radar as claimed in claim 1, wherein the step S2 is specifically as follows: inputting the range profile processed in step S1 into an MVDR beam forming spatial filter, calculating the average power of the output signal, and then performing spectral peak search on the average power to find out the angle corresponding to the peak point, i.e., the azimuth angle of the current frame mirror image target.

6. The off-line-of-sight multi-target positioning method for the millimeter wave radar as claimed in claim 1, wherein the step S3 is specifically as follows: and resolving the position coordinate of the mirror image target by combining the mirror image target distance and the mirror image target azimuth angle obtained in the target distance detection algorithm, then carrying out coordinate system conversion on the position coordinate of the mirror image target, and obtaining the position coordinate of the non-direct-view target by utilizing the building layout information according to the mirror image mapping relation between the mirror image target and the real target.

Technical Field

The invention belongs to the technical field of target positioning, and particularly relates to a target positioning technology in an urban corner scene.

Background

The non-direct-view target detection and positioning under the urban environment can be widely applied to the fields of urban street fighting, anti-terrorism and stability, target rescue and the like. The traditional radar usually detects a target under direct-view (Line of Sight, LOS), but a Non-direct-view (NLOS) target under an urban environment cannot directly reach the target in a detection blind area due to shielding of buildings, so that a direct-view detection mode using electromagnetic waves fails, and only a Non-direct-view multipath detection mode can be adopted. The non-direct-view multi-path detection mainly utilizes diffraction and reflection paths of electromagnetic waves in a complex environment to detect and position a non-direct-view target in a building, and is one of research hotspots and difficulties in the radar field at the present stage.

Many research institutes at home and abroad develop the non-direct-view target detection and positioning of corner scenes. In 2019, French aviation laboratory scholars applied a handheld wide-beam Radar and proposed a non-direct-view target detection and positioning method based on a matching subspace filtering method (K.Thai, O.Rabaste, J.Bosse, et al.Detection-localization Algorithms in the Around and the corner Radar project [ J ]. IEEE Transactions on Aerospace and Electronic Systems,2019,55(6):2658 and 2673). The actual measurement experiment proves that the method can realize the positioning of the non-direct-view target after turning, but the method has large calculation amount when calculating the matching grid in a detection area, and a large amount of false targets introduced by multipath echo strong ambiguity are remained in the positioning result. In 2020, a university of national defense science and technology millimeter wave Radar is placed at two different positions to respectively collect primary reflection path echo data (H.Du, C.Fan, Z.Chen, C.Cao, and X.Huang, NLOS Target Localization with an L-Band UWB Radar sight Grid [ J ], progressive In electromagnetic Research M,2020, pp.45-56.) of a Target, and the non-direct-view single Target positioning after turning is realized by using an elliptical cross positioning method, but the method needs to move a detection Radar to two stations, is inconvenient to operate and easily causes the problem that detection results of different stations are not matched, so that the multi-Target is difficult to be accurately positioned. In 2020, the university of electronic technology learns to apply the proposed positioning algorithm based on multi-channel phase comparison (s.guo, q.zhao, g.cui, s.li, l.kong and x.yang, beam Corner target positioning Using Small Aperture meter Wave Radar in NLOS under Environment [ J ], IEEE Journal of Selected Targets in Applied Earth observation and Remote Sensing,2020, pp.460-470.) to the primary reflection path echo data of the target, so as to realize the positioning of the non-direct-view target after the Corner, but limited by the direction-finding accuracy of the multi-channel phase comparison method, and the positioning result has the problem of point spread.

Disclosure of Invention

In order to solve the technical problem, the invention provides a millimeter wave radar non-direct-view multi-target positioning method, which adopts an MVDR angle measurement method in combination with geometric prior information to accurately calculate the position of a non-direct-view target hidden at a corner.

The technical scheme adopted by the invention is as follows: a millimeter wave radar non-direct-view multi-target positioning method comprises the following steps based on application scenes: the target is located at a non-direct-view position behind a corner, the method comprising the steps of:

s1, establishing an electromagnetic propagation model based on the primary reflection path, and calculating the distance of the mirror image target;

s2, obtaining the azimuth angle of the mirror image target by adopting an MVDR angle estimation method for all channel signals at the mirror image target distance unit;

and S3, calculating the position coordinates of the non-direct-view target by combining the scene geometric prior information.

Step S1 specifically includes the following substeps:

s11, eliminating static clutter in the environment by adopting a moving target indication method, and reserving echoes of the moving target;

s12, obtaining a target range profile by adopting pulse compression processing;

s13, the range profile is subjected to incoherent superposition, and the range profile after accumulation is detected by a one-dimensional unit average constant false alarm detection method to obtain the mirror image target distance R.

Step S2 specifically includes: inputting the range profile processed in step S1 into an MVDR beam forming spatial filter, calculating the average power of the output signal, and then performing spectral peak search on the average power to find out the angle corresponding to the peak point, i.e., the azimuth angle of the current frame mirror image target.

Step S3 specifically includes: and resolving the position coordinate of the mirror image target by combining the mirror image target distance and the mirror image target azimuth angle obtained in the target distance detection algorithm, then carrying out coordinate system conversion on the position coordinate of the mirror image target, and obtaining the position coordinate of the non-direct-view target by utilizing the building layout information according to the mirror image mapping relation between the mirror image target and the real target.

The invention has the beneficial effects that: the invention utilizes the millimeter wave radar to position the non-direct-view multiple targets behind the corner; analyzing that the main propagation path of the electromagnetic wave of the corner scene is a primary reflection path according to the propagation phenomenon of the electromagnetic wave; in addition, the invention effectively utilizes the MVDR angle measurement method to realize the accurate positioning of the non-direct-view multiple targets. The actual measurement result shows that the method can obtain accurate positioning results of a plurality of targets in a non-direct-view detection scene after the corner is dealt with.

Drawings

FIG. 1 is an electromagnetic wave propagation model of a corner scene of a building

FIG. 2 is a schematic diagram of coordinate transformation;

fig. 3 is a scene diagram of the actual measurement test.

FIG. 4 is a three-jog target range profile.

FIG. 5 shows the results of MVDR goniometry positioning.

Detailed Description

In order to facilitate the understanding of the technical contents of the present invention by those skilled in the art, the present invention will be further explained with reference to the accompanying drawings.

The invention discloses a non-direct-view multi-target positioning method of a millimeter wave radar, which is based on a detection scene as shown in figure 1, wherein the scene comprises a corner C, a wall 1 and a wall 2. Adopt MIMO radar system to survey the target that is hidden, the radar contains 2 transmitting antenna, 4 receiving antenna, and the radar antenna interval is half wavelength distance, and the radar is placed in one side at the corner, uses the antenna array center as original point (0,0), establishes rectangular coordinate system with the original point, and wherein, wall body 1,2 are parallel with coordinate system y axle, and 2 distance y axle lengths on the wall body are 4.5 m. Target Q is in the NLOS region between wall 1 and wall 2 passageway, sets for to survey non-direct-view target Q and only has a two-way primary reflection route, and the electromagnetic wave propagation process can be described as the electromagnetic wave is launched from the radar and is returned to the radar along former way behind the reflection of wall 2 to target position department, and the electromagnetic wave propagation route is: o → W → Q → W → O. The processing flow of the method comprises the following steps:

step 1: non-line-of-sight target distance measurement

In order to eliminate the influence of background static clutter in radar echoes, a Moving Target Indicator (MTI) method is adopted to eliminate the static clutter in the environment and retain the echoes of the dynamic Target. Let z be the target echo of the electromagnetic wave transmitted by the mth antenna and received by the nth antenna_m,n(t) for two adjacent echo signals starting from the ith period, e.g. z_m,n(t, i) and z_m,n(t, i-1), which can be expressed as:

z'_m,n(t,i)＝z_m,n(t,i)-z_m,n(t,i-1)

wherein, z'_m,n(t, i) represents the echo signal of the i-th cycle after the MTI.

Then to the post-MTI echo signal z'_m,n(t, i) obtaining a target range profile using pulse compression. Therefore, the range image data of the ith data cycle transmitted by the mth antenna and received by the nth antenna is defined as:

x_m,n(i)＝[x_m,n(i，1),…,x_m,n(i,j),…,x_m,n(i,N_c)]

where x is_m,n(i, j) represents the amplitude of the jth range bin of the ith cycle, N_cRepresenting the number of range cells.

To improve the detection performance of the target, the range profile x is subjected to_m,nWith N_TNon-coherent for one frame per periodSuperposition, assuming range profile x_m,nTotal N_AMultiple cycles, then N is total after accumulation of multiple cycle range profile_F＝N_A/N_TFrame distance image. The incoherent superposition is usually N_TThe absolute value of the periodic range image is then accumulated for processing, and the non-coherent stacking process can be represented as:

wherein u is_m,n(h) Represents the signal after the incoherent superposition of the h-th frame, h is 1,2, …, N_FIn the above formula, | · | represents an absolute value.

In this example N_TThe value is 128.

After the distance images are subjected to incoherent superposition, a one-dimensional Cell Averaging-Constant False Alarm Rate (CA-CFAR) method is adopted to carry out the incoherent superposition on the accumulated distance images u_m,n(h) And detecting to obtain the distance unit index and the distance value of the target. For the signal u after accumulation_m,n(h) The jth distance unit of (1), the detection threshold TH_jCan be expressed as:

wherein P is_fRepresenting false alarm probability, N_rRepresenting the unit u to be detected_m,nNumber of reference cells in (h, j).

After the detection threshold is calculated, the value of the distance unit where the target is located can be obtained according to a self-adaptive judgment criterion, wherein the judgment criterion is as follows:

in the formula H₁Indicates the presence of the target hypothesis, H₀Indicating no target hypothesis, executing the above formula judgment criterion for all range cells, and if the target hypothesis is present, retaining the range cell indexAnd obtaining the target distance R detected by the h frame.

Step 2: non-direct-view target Minimum Variance Distortionless Response (MVDR) angle measurement

For a uniform line model, the direction vector a (θ) of the antenna received signal can be defined as:

a(θ)＝[1 e^-jφ … e^-j(K-1)φ]^T

φ＝2πdsinθ/λ

where φ is the phase difference between adjacent antennas.

A minimum variance distortion free response (MVDR) beamforming algorithm is then employed to estimate the target signal azimuth. The output of the hollow-domain filter in the MVDR beam forming is as follows:

y(j)＝w^Hx(j)

wherein, w^HIs the weight vector of the spatial filter after conjugate rotation, and has w ═ w₁ w₂ … w_K]^TT denotes transposition, H denotes conjugate transposition, K denotes the number of array elements, and the signal x (j) is the input signal of the spatial filter and is also the jth range bin signal where the target is located.

The average power P (θ) of the output signal is:

P(θ)＝E{|y(j)|²}＝E{w^Hx(j)x^H(j)w}

＝w^HRw

wherein R ═ E { x (j) x^H(j) Is the spatial autocorrelation matrix.

In order to minimize the average output power P (θ) of the MVDR spatial filter, signals and noise in other directions are suppressed as much as possible. Constructing a conditional extremum problem as follows:

the average output power can be solved by applying the Lagrange multiplier method as follows:

in the range of [ - π, π]Internally varying theta in a (theta) to obtain P_MVDRAnd (theta) changing the curve, then performing spectral peak search on the P (theta), and finding out theta corresponding to the peak point, namely the azimuth angle of the current frame target.

And step 3: non-direct-view target mirror mapping location acquisition

By analyzing the electromagnetic propagation path of the corner scene, the obtained object distance and azimuth angle are both from the mirror image object Q ', and the position coordinates (x ', y ') of the mirror image object can be calculated as:

for the convenience of analysis, we define the coordinates of the target in a new coordinate system, and as can be seen from fig. 2, the coordinates of the obtained mirror image target are in a coordinate system established by the radar array viewing angle, i.e. an X ' OY ' coordinate system, so that the coordinates (X ', y ') of the X ' OY ' coordinate system in which the mirror image target Q ' is located can be converted into a new coordinate system XOY to obtain new coordinates (X ', y ')/X₀,y₀) The coordinate conversion formula is as follows:

wherein phi is an included angle between the radar array direction and the horizontal direction, and needs to be obtained through experimental measurement in actual positioning. After obtaining the new coordinates of the mirror image target, the coordinates (x, y) of the real target can be obtained by combining the mirror image symmetry principle and the geometric information, and the calculation formula is as follows:

where L is the lateral distance of the wall 2 from the center of the array, which is a priori known information.

The following provides a specific embodiment of the present invention based on a measurement test.

For multiple targets behind the corner, the measurement scene is shown in fig. 3, the radar height is 1.2m, if the radar is taken as the center origin, the coordinate of the corner C is (0.28,0.49) m, the transverse distance between the wall 2 and the radar is 2.55m, and the included angle phi between the radar and the horizontal direction is 46.6 degrees. Target 1 was located at the (-3.2,3.9) m position, target 2 was located at the (-1.9,2.57) m position, and target 3 was located at the (-2.6,5.58) m position. The processing steps according to the invention:

step 1: target distance measurement

First, the distance images of three targets are obtained through the steps of MTI and pulse compression, as shown in FIG. 4. The propagation distances of the target 1, the target 2 and the target 3 are theoretically calculated to be 6.98m, 7.55m and 8.49m, which are matched with target distance tracks at 7.05m, 7.55m and 8.62m in the range image, and the target corresponding to each distance track is marked in 4. After CFAR detection, a target distance value can be obtained.

Step 2: calculating the target arrival angle by adopting an MVDR angle measurement method

And step 3: calculating target position in combination with building layout information

The target distance and the arrival angle can be calculated according to the step 1 and the step 2, then the target position is obtained through the steps of coordinate conversion, mirror symmetry and the like, the prior information can be referred for removing the false points outside the building, and the obtained positioning result is shown in fig. 5.

And 4, step 4: calculating target positioning error

In order to measure the positioning effect of the MVDR angle measurement positioning method, the positioning errors of three non-direct-view targets adopting the MVDR positioning method are calculated as follows:

here, theFor each frame, the estimated value of the target position, (x, y) is the true value of the target position.

TABLE 1 mean positioning error of three micro-motion targets

Positioning error (m)	Object 1	Object 2	Target 3
				MVDR method	0.154	0.090	0.167

The average positioning error of 3 targets of the MVDR positioning method is calculated according to the positioning error formula, as shown in table 1. It can be seen that the MVDR positioning method has good positioning effect, and the average positioning errors of 3 targets are less than 0.17 m.

The non-direct-view multi-target positioning method applicable to the rear of the corner can accurately position the rear multi-target of the corner, and the accuracy and the effectiveness of the method are verified.

It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.