A method for positioning targets in T-shaped mine tunnels based on millimeter-wave radar

Through the target positioning method of T-tunnel in mine based on millimeter wave radar, the electromagnetic propagation path and the echo model are simplified, and the signal attenuation and multipath effect problems of non-sight target positioning of T-tunnel in coal mines are solved, achieving high precision and Low complexity positioning, suitable for underground coal mine operations and disaster rescue.

CN118501835BActive Publication Date: 2024-11-12CHINA UNIV OF MINING & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410568896.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-09
Publication Date
2024-11-12
Estimated Expiration
2044-05-09

AI Technical Summary

Technical Problem

The non-sight target positioning technology of T-shaped tunnels underground in coal mines has signal attenuation and multipath effect problems. Traditional methods are difficult to provide accurate position information, which affects operating efficiency and safety. The application of existing millimeter wave radar technology in L-shaped corner scenarios is insufficient. It is suitable for T-shaped structure tunnels.

Method used

The target positioning method of mine T-tunnel based on millimeter-wave radar is established by simplifying the millimeter-wave electromagnetic propagation path of T-tunnel scenes, and the radar echo model is established, using continuous frequency modulation signals, MTI clutter suppression, pulse compression, CA-CFAR detection and MVDR airspace filtering method solves the false target position, eliminates invalid false targets, and obtains the real target position.

Benefits of technology

It realizes accurate detection and positioning of non-sight targets in T-shaped tunnels underground in coal mines, improves positioning accuracy and robustness, reduces the computational complexity, and is suitable for mine disaster rescue and unmanned driving.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118501835B_ABST
    Figure CN118501835B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for locating a target in a mine T-type tunnel based on a millimeter wave radar. First, a simplified millimeter wave electromagnetic propagation path of a T-type tunnel scene is established and analyzed, and a corresponding FMCW system radar echo model is established; secondly, the radar signal is preprocessed using MTI, pulse compression, and incoherent superposition, and the target distance is extracted using the CA‑CFAR detection method, and the target azimuth is estimated using the MVDR spatial domain filtering method, and the false target point set is solved by combining the distance and azimuth information; finally, the false target point set is matched and eliminated to obtain a valid false target point set, and the valid false target point set is coordinate-transformed using the prior scene information to obtain the real target point set, and the positioning process is completed. The present invention is aimed at the complex environment of the T-type tunnel scene in a coal mine, and can achieve the accuracy and positioning accuracy of a relatively ideal non-line-of-sight target detection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a non-line-of-sight personnel positioning method in a mine, in particular to a target positioning method in a T-shaped tunnel of a mine based on a millimeter-wave radar, and belongs to the technical field of safe production in coal mines. Background Art

[0002] Target positioning technology determines the current position of the target node in the positioning scene through parameter measurement, position solution, result optimization and other steps. It has wide applications in intelligent transportation, industrial production, military confrontation and other fields. However, there are usually many obstacles in the positioning scene. Taking the mine environment of coal mines as an example, there are usually obstacles such as mine cars, tunnel walls, miners, etc. in the mine environment, and there are uncertainties in the position and motion state of obstacles, which makes the positioning environment more complicated. Non-line-of-sight (NLOS) target positioning problems are often faced. As a common underground tunnel structure, the T-type structure tunnel has a particularly prominent non-line-of-sight target positioning problem. Due to the particularity of the T-type structure tunnel, traditional underground positioning technology (such as based on Wi-Fi, Bluetooth, UWB, etc.) often cannot provide accurate location information due to signal attenuation and multipath effects, which not only affects the operation efficiency, but also increases the operation risk, and even leads to the occurrence of mine driving safety accidents. Therefore, the development of a non-line-of-sight target positioning technology suitable for T-type structure tunnels is of great significance to improving the safety and efficiency of underground coal mine operations.

[0003] For the research on non-line-of-sight target positioning in coal mines, early studies mainly focused on the deployment and utilization of wireless sensor networks, attempting to estimate the target position through multi-point measurement and signal strength indication (RSSI). Subsequently, positioning methods based on ultrasonic and geomagnetic induction were also proposed, which improved the accuracy of positioning to a certain extent. However, these technologies still have certain limitations when facing complex underground environments, especially in T-shaped structure tunnels without direct line of sight, and the positioning accuracy and robustness still need to be improved.

[0004] In recent years, millimeter wave radar technology has begun to be applied to non-line-of-sight target positioning due to its unique advantages, such as high resolution, strong penetration ability and sensitivity to small targets. Millimeter wave radar can detect and locate targets in complex underground environments through reflection, refraction and scattering of walls and other obstacles. This feature makes millimeter wave radar an ideal choice for solving the problem of non-line-of-sight target positioning in T-shaped tunnels. However, the existing technology that uses millimeter wave radar to solve the problem of non-line-of-sight target positioning mostly constructs the scene as a typical L-shaped corner scene. There is insufficient discussion on the common T-shaped tunnel scene in coal mines, and the non-line-of-sight target positioning problem applicable to the L-shaped corner scene is not completely applicable to the non-line-of-sight target positioning problem in the T-shaped tunnel scene; in addition, most existing methods rely on the distance information of radar echoes for positioning, and make insufficient use of potential angle information. Summary of the invention

[0005] In view of the problems existing in the above-mentioned prior art, the present invention provides a mine T-tunnel target positioning method based on millimeter-wave radar, which can achieve relatively ideal non-line-of-sight target detection accuracy and positioning precision for the complex environment of the T-tunnel scene in coal mines.

[0006] To achieve the above purpose, the mine T-shaped tunnel target positioning method based on millimeter wave radar specifically includes the following steps:

[0007] Step 1: Simplify the millimeter wave electromagnetic propagation path in the T-shaped tunnel scenario:

[0008] The T-shaped structure lane includes Lane I and Lane II which is perpendicular to Lane I and connected with Lane I. Lane I and Lane II form a T-shaped structure. Lane II includes Wall 3 and Wall 1. Lane I includes Wall 5 which is not connected with Lane II and Wall 4 and Wall 2 which are connected with Lane II. Wall 4 intersects with Wall 3 to form Wall Corner 1. Wall 2 intersects with Wall 1 to form Wall Corner 2.

[0009] Assume that the radar is located at the coordinate origin O in Lane II, and the initial phase angle is 0°, and the distance between the radar and wall five is L, the distance between the radar and wall three is d2, and the distance between the radar and wall one is d1; let the target be located between wall two and wall five, and at the non-line-of-sight area P = [x, y] point behind the corner two formed by the intersection of wall two and wall one; assume that all walls are smooth, only consider the multipath signal generated by mirror reflection, and only simplify it to the first-order reflection multipath of millimeter waves, let the reflection point of the one-way primary reflection path at wall three be G2, and the reflection point at wall five be G1; then

[0010] The propagation path of the transmitted signal is:

[0011] O→G1→P, O→G2→P

[0012] The propagation path of the received signal is:

[0013] P→G1→O, P→G2→O

[0014] The transmission and reception paths of the signal electromagnetic waves are a combination of the above four paths, including:

[0015] ① Propagation path I round trip: O→G1→P, P→G1→O, the electromagnetic wave signal is emitted from point O, reflected by point G1 to point P, and returns along the original route, which is recorded as path Φ 11 ;

[0016] ② Propagation path I + propagation path II: O→G1→P, P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G1 and reaches point P. The return path is reflected by point G2 and returns to the radar receiving antenna, which is recorded as path Φ 12 ;

[0017] ③ Propagation path II + propagation path I: O→G2→P and P→G1→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P. The return path is reflected back to the radar receiving antenna via point G1, which is recorded as path Φ 21 ;

[0018] ④ Propagation path II round trip: O→G2→P and P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P, which is recorded as path Φ 22 ;

[0019] Round trip path Φ 11 , Φ 22 About the radar's mirror target point Q 11 , Q 22 Calculated by geometric relationship:

[0020]

[0021] Where: [m,n] represents Q 11 The coordinates of; [p,q] represents Q 22 The coordinates of

[0022] Φ 12 The path echo is reflected from point G2 and reaches the receiving antenna in the same direction as the mirror target point Q. 22 equal, the virtual target point is recorded as Q 12 ; Φ 21 The path echo is reflected from point G1 to the receiving antenna, and its direction is the same as the mirror target point Q 11 equal, the virtual target point is recorded as Q 21 ; Let Q 21 The distance from the radar is R 21 , the azimuth is θ 21 ;Q12 The distance from the radar is R 12 , the azimuth is θ 12 ; The distances between the false target points and the radar corresponding to the four paths are expressed as follows:

[0023]

[0024] According to Q 11 , Q 22 The coordinates of the four paths are used to obtain the azimuths of the false target points as follows:

[0025]

[0026] Combined path virtual target point Q 12 , Q 21 The position is calculated by the following formula:

[0027]

[0028] Step 2: Establish a radar echo model based on the T-shaped tunnel scene:

[0029] Use continuous frequency modulation millimeter wave radar to transmit millimeter waves to locate the target. Suppose the linear frequency modulation signal of the millimeter wave radar is s(t), and its expression is:

[0030]

[0031] Where: f0 represents the starting frequency of the transmitted signal; A represents the amplitude of the transmitted signal; B represents the signal bandwidth, T represents the signal pulse width; μ is the frequency modulation slope of the signal, μ = B / T; is the initial phase angle,

[0032] The received signal is represented as:

[0033] r(t)=K1s(t-τ 11 )+K2s(t-τ 12 )+K3s(t-τ 22 )+K4s(t-τ 44 )+n(t)

[0034] Where: K1~K4 represent the scattering coefficients of the four propagation paths; n(t) represents the environment and system noise; τ 11 , τ 12 , τ 21 , τ 22 Represent the path Φ 11 , Φ 12 , Φ 21 , Φ 22 The propagation delay of

[0035] τ 11 , τ 12 , τ 21 , τ 22 The corresponding delay calculation formula is as follows:

[0036]

[0037] In the formula: c represents the speed of light; || represents the distance operation between two points;

[0038] Step 3, collect radar echoes and complete preprocessing:

[0039] Perform MTI clutter suppression on the sampled echo signal, use pulse compression on the signal after MTI processing to obtain the original range image, perform incoherent superposition on the original range image, and obtain the range image incoherent accumulation signal;

[0040] Step 4, use CA-CFAR detection method to obtain the target distance R for the non-coherent accumulation signal of the range image;

[0041] Step 5, using the minimum mean square distortion-free response (MVDR) method to obtain the target azimuth angle θ;

[0042] Step 6, calculate the false target position point set Q according to the target distance R and the target azimuth θ:

[0043] Convert the polar coordinate system into the Cartesian coordinate system to obtain the initial position point set Q of the multipath false target points.

[0044] Q=[Q1,Q2,....,Q n ]

[0045] Where: Q n represents the solved false target position, Q n =[R n sinθ n ,R n cosθ n ];

[0046] Step 7, perform multipath false target matching elimination on the false target position point set Q to obtain a valid false target position point set Q′;

[0047] Step 8, transform the coordinates of the effective false target point set Q′ to obtain the real target position point set P:

[0048] Combine the distance L and d2 in the prior geometric scene to solve the real target point set P corresponding to the effective false target point and complete the positioning process.

[0049] Furthermore, when MTI clutter is suppressed in Step 3, let the echo signal received by the radar be r, then:

[0050] r i * =r i -r i-1

[0051] Where: r i * represents the signal after MTI processing, and i represents the number of pulses.

[0052] Furthermore, the steps for obtaining the non-coherent accumulation signal of the range image in Step 3 are as follows:

[0053] The range image data of the i-th pulse is defined as:

[0054] x(i)=[x(i,1),x(i,2),...,x(i,j),...,x(i,N d )]

[0055] Where: x(i,j) represents the amplitude of the jth distance unit; N d Represents the number of distance units;

[0056] For the distance image x, N T The number of pulses is one frame for incoherent superposition to obtain the range image incoherent accumulation signal. The process is described as:

[0057]

[0058] Where: |·| represents the absolute value operation; u(h) represents the signal after non-coherent accumulation of the hth frame; N T Represents the number of pulses in each frame.

[0059] Furthermore, the specific steps of Step 4 are as follows:

[0060] For the j-distance unit, the detection threshold is expressed as:

[0061]

[0062] Where: N r represents the number of reference cells; P fa represents the constant false probability;

[0063] Compare the detection threshold η according to the adaptive judgment criterion j The size of the unit u(h,j) to be detected is used to obtain the unit number of the distance where the target is located;

[0064] Apply the adaptive judgment criterion to all distance units of the range image to obtain the target distance R detected in the hth frame.

[0065] Furthermore, the adaptive judgment criteria are as follows:

[0066]

[0067] Where: H1 represents a target hypothesis; H0 represents a non-target hypothesis.

[0068] Furthermore, the specific steps of Step 5 are as follows:

[0069] Define the steering vector a(θ) of the receiving antenna element receiving the signal. The specific expression is:

[0070] a(θ)=[1,e -jφ ,...,e -j(K-1)φ ] T ,φ=(2πdsinθ) / λ

[0071] Where: θ represents the direction of the incoming wave; d represents the receiving array element spacing, d = λ c / 2;λ c Indicates the carrier wavelength;

[0072] The spatial filter output in MVDR beamforming is:

[0073] y(j)=w H x(j)

[0074] Where: w is the weight vector of the spatial domain filter, w = [w1, w2, ..., w K ] T ; x(j) is the target distance detection result signal containing K channels; [·] H represents the conjugate transpose operation;

[0075] The average output power P(θ) is:

[0076] P(θ)=E{|y(j)| 2}=E{w H x(j)x H (j)w}=w H R

[0077] Where: R = E{x(j)x H (j)} is the autocorrelation matrix; E{·} is the mathematical expectation;

[0078] The optimal weight vector of the MVDR beamformer can be expressed as:

[0079]

[0080] Substituting the optimal weight vector into the average power expression, the average power is obtained as:

[0081]

[0082] Change the angle θ in a(θ) in the range [-π,π] to get P MVDR (θ) change curve, for P MVDR (θ) curve to search for the spectrum peak and obtain the corresponding target azimuth angle θ.

[0083] Furthermore, in Step 7, when determining the angle θ′1 corresponding to the false target point and determining the round-trip propagation path to which the false target belongs, the specific determination method is as follows:

[0084]

[0085] The radar's angular detection range is set to be 0° along the positive direction of the y-axis and [-π,π] along the negative half axis to the positive half axis of the x-axis. If θ′1>0, it means that the signal is along Φ 11 Propagation, the corresponding mirror target point is Q 11 Otherwise, it means that the mirror target point corresponding to the echo is Q 22 .

[0086] Furthermore, the specific steps of Step 7 are as follows:

[0087] First, the false target position point set Q is arranged from small to large in terms of distance, and the corresponding distance of the false target point is recorded as R′1, and the angle θ′1 corresponding to the false target point is determined to determine the round-trip propagation path to which the false target belongs;

[0088] Solve and estimate the true target position point P′:

[0089]

[0090] Then use P′ to reversely estimate the corresponding multipath false target point position Q′ 12 , Q′ 21 , Q′ 22 , Q′ 11 , transform the other initial coordinate points from the polar coordinate system to the Cartesian coordinate system Q i =[R′ i sinθ i ′,R′ i cosθ′ i ], i = 2, ..., n, respectively calculate the distance error between the multipath false target position point and other initial position points:

[0091]

[0092] Where: ||·||2 represents the Euclidean norm calculation;

[0093] The distance error is compared with the error threshold respectively. If the distance error is less than the threshold value TH, it means that the initial positioning point compared at this time is a two-way valid false target point. Its coordinate position is added to the valid false target point set Q′. Then, the above matching process is repeated for the retained initial positioning point to eliminate all invalid false target points.

[0094] Compared with the existing technology, this mine T-tunnel target positioning method based on millimeter-wave radar first establishes and analyzes the simplified T-tunnel scene millimeter-wave electromagnetic propagation path, and establishes the corresponding FMCW system radar echo model; secondly, the radar signal is preprocessed using MTI, pulse compression, and incoherent superposition, and the CA-CFAR detection method is used to extract the target distance, and the target azimuth is estimated by the MVDR spatial domain filtering method, and the false target point set is solved by combining the distance and azimuth information; finally, the false target point set is matched and eliminated to obtain the effective false target point set, and the prior scene information is used to transform the coordinates of the effective false target point set to obtain the real target point set, completing the positioning process. This mine T-tunnel target positioning method based on millimeter-wave radar has the following advantages:

[0095] 1. It can effectively solve the problem of target positioning in non-line-of-sight scenes in T-shaped tunnels in coal mines;

[0096] 2. The positioning algorithm has low computational complexity and high positioning accuracy;

[0097] 3. The invalid false target matching elimination method can effectively eliminate useless false target interference and improve positioning accuracy;

[0098] 4. It can be applied to mine disaster rescue, mine unmanned driving and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] Figure 1 This is a simplified schematic diagram of millimeter wave multipath propagation in a T-shaped lane according to the present invention;

[0100] Figure 2 It is a flow chart of the invalid false target matching elimination algorithm of the present invention;

[0101] Figure 3 It is a schematic diagram of a simulation scenario of the present invention;

[0102] Figure 4 yes Figure 3 Illustration of the positioning results obtained from the simulation scenario;

[0103] Figure 5 These are two positioning error diagrams of the simulation scene of the present invention. DETAILED DESCRIPTION

[0104] The present invention will be further described below in conjunction with the accompanying drawings.

[0105] The target positioning method of the mine T-shaped tunnel based on millimeter wave radar specifically includes the following steps:

[0106] Step 1: Analyze and simplify the millimeter wave electromagnetic propagation path in the T-shaped tunnel scenario:

[0107] like Figure 1 As shown, the T-shaped structure tunnel includes tunnel I and tunnel II which is perpendicular to tunnel I and connected with tunnel I. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel II and tunnel II. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel I and tunnel II. T-shaped structure is formed between tunnel I and tunnel II.

[0108] Construct a priori geometric scenarios and simplify the T-shaped lane millimeter wave multipath propagation scenarios, such as Figure 1 As shown, the radar is located at the coordinate origin O in Lane II, and the radar does not rotate at an angle, that is, the initial phase angle of the radar is The target is located between wall 2 and wall 5, and at the non-line-of-sight area P = [x, y] point behind the corner 2 formed by the intersection of wall 2 and wall 1. The distance between the radar and wall 5 is L, the distance between the radar and wall 3 is d2, and the distance between the radar and wall 1 is d1. The characteristic that electromagnetic waves will produce reflections on the wall surface is used to realize the detection of non-line-of-sight targets. Assuming that the wall surface is smooth, only the multipath signal generated by mirror reflection is considered. During the target detection process, due to the existence of wall 2, wall 3 and wall 5, the propagation path will emit primary reflection, secondary reflection and higher-order reflection. However, for millimeter-wave radars, the second-order and higher-order millimeter-wave reflection multipath signals are severely attenuated, so the simplified model of the present invention mainly considers the first-order reflection multipath of millimeter waves. The reflection point of the one-way primary reflection path on wall 3 is G2, and the reflection point on wall 5 is G1.

[0109] Therefore, the propagation path of the transmitted signal is:

[0110] O→G1→P, O→G2→P

[0111] The propagation path of the received signal is:

[0112] P→G1→O, P→G2→O

[0113] In the actual process of detecting targets, the emission and reception paths of signal electromagnetic waves are a combination of the above four paths, including:

[0114] ① Propagation path I round trip: O→G1→P, P→G1→O, the electromagnetic wave signal is emitted from point O, reflected by point G1 to point P, and returns along the original route, which is recorded as path Φ 11 ;

[0115] ② Propagation path I + propagation path II: O→G1→P, P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G1 and reaches point P. The return path is reflected by point G2 and returns to the radar receiving antenna, which is recorded as path Φ 12 ;

[0116] ③ Propagation path II + propagation path I: O→G2→P and P→G1→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P. The return path is reflected back to the radar receiving antenna via point G1, which is recorded as path Φ 21 ;

[0117] ④ Propagation path II round trip: O→G2→P and P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P, which is recorded as path Φ 22 .

[0118] Round trip path Φ 11 , Φ 22 About the radar's mirror target point Q 11 , Q 22 It can be calculated by geometric relationship:

[0119]

[0120] Where: [m,n] represents Q 11 The coordinates of; [p,q] represents Q 22 The coordinates of .

[0121] Depend on Figure 1 It can be seen that the combined path Φ 12 , Φ 12 The distance is exactly the same, and the distance is 1 / 2 of the round trip distance. Φ 12 The path echo is reflected from point G2 and reaches the receiving antenna in the same direction as the mirror target point Q. 22 equal, the virtual target point is recorded as Q 12 . And Φ 21 The path echo is reflected from point G1 to the receiving antenna, and its direction is the same as the mirror target point Q 11 equal, the virtual target point is recorded as Q 21 . Let Q 21 The distance from the radar is R 21 , the azimuth is θ 21 ;Q 12 The distance from the radar is R 12 , the azimuth is θ 12 The distances between the false target points corresponding to the four paths and the radar are expressed as follows:

[0122]

[0123] According to Q11 , Q 22 The coordinates of the four paths can be used to obtain the azimuths of the false target points as follows:

[0124]

[0125] Therefore, the virtual target point Q of the combined path 12 , Q 21 The position can be calculated by the following formula:

[0126]

[0127] Step 2: Establish the FMCW radar echo model based on the T-shaped tunnel scenario:

[0128] For the non-line-of-sight scenario in a T-shaped tunnel, a radar signal echo model is established. A 60GHz millimeter wave is emitted using a continuous frequency modulated millimeter wave radar (FMCW) to locate the target. Assume that the linear frequency modulated signal of the millimeter wave radar is s(t), and its expression is:

[0129]

[0130] Where: f0 represents the starting frequency of the transmitted signal; A represents the amplitude of the transmitted signal; B represents the signal bandwidth, T represents the signal pulse width; μ is the frequency modulation slope of the signal, μ = B / T; is the initial phase angle,

[0131] Considering the four electromagnetic propagation paths in Step 1, the received signal can be expressed as:

[0132] r(t)=K1s(t-τ 11 )+K2s(t-τ 12 )+K3s(t-τ 22 )+K4s(t-τ 44 )+n(t)

[0133] Where: K1~K4 represent the scattering coefficients of the four propagation paths; n(t) represents the environment and system noise; τ 11 , τ 12 , τ 21 , τ 22 Represent the path Φ 11 , Φ 12 , Φ 21 , Φ 22 of propagation delay.

[0134] τ 11 , τ 12 , τ 21 , τ 22 The corresponding delay calculation formula is as follows:

[0135]

[0136] In the formula: c represents the speed of light; || represents the distance operation between two points.

[0137] From the above formula, we can see that the propagation delay of the four paths mainly depends on the mirror target point Q 22 and Q 11 .

[0138] Step 3, collect radar echo r and complete preprocessing:

[0139] The echo signal received by the radar is r, and the sampled echo signal is subjected to MTI clutter suppression. The MTI method is specifically as follows:

[0140] r i * =r i -r i-1

[0141] Where: r i * represents the signal after MTI processing, and i represents the number of pulses.

[0142] Pulse compression is used on the signal after MTI processing to obtain the original range image.

[0143] The range image data of the i-th pulse is defined as:

[0144] x(i)=[x(i,1),x(i,2),...,x(i,j),...,x(i,N d )]

[0145] Where: x(i,j) represents the amplitude of the jth distance unit; N d Represents the number of distance units.

[0146] Then, the distance image x is N T The number of pulses is one frame for incoherent superposition, and the process is described as follows:

[0147]

[0148] Where: |·| represents the absolute value operation; u(h) represents the signal after non-coherent accumulation of the hth frame; N T Represents the number of pulses in each frame.

[0149] Step 4: Use CA-CFAR detection method to obtain the target distance R for the non-coherent accumulation signal of the range image:

[0150] For the j-distance unit, the detection threshold is expressed as:

[0151]

[0152] Where: N r represents the number of reference cells; P fa represents constant false probability.

[0153] Compare the detection threshold η according to the adaptive judgment criterion j The size of the unit to be detected u(h,j) is used to obtain the unit number of the distance where the target is located. The judgment criteria are as follows:

[0154]

[0155] Where: H1 represents a target hypothesis; H0 represents a non-target hypothesis.

[0156] Apply the adaptive judgment criterion to all distance units of the range image to obtain the target distance R detected in the hth frame.

[0157] Step 5, use the minimum mean square distortion-free response (MVDR) method to obtain the target azimuth angle θ:

[0158] Define the steering vector a(θ) of the receiving antenna element receiving the signal. The specific expression is:

[0159] a(θ)=[1,e -jφ ,...,e -j(K-1)φ ] T ,φ=(2πdsinθ) / λ

[0160] Where: θ represents the direction of the incoming wave; d represents the receiving array element spacing, d = λ c / 2;λ c Indicates the carrier wavelength.

[0161] The spatial filter output in MVDR beamforming is:

[0162] y(j)=w H x(j)

[0163] Where: w=[w1,w2,...,w K ] T is the weight vector of the spatial filter; x(j) is the target distance detection result signal containing K channels; [·] H Represents the conjugate transpose operation.

[0164] The average output power P(θ) is:

[0165] P(θ)=E{|y(j)| 2}=E{w H x(j)x H (j)w}=wH R

[0166] Where: R = E{x(j)x H (j)} is the autocorrelation matrix; E{·} is the mathematical expectation.

[0167] The optimal weight vector of the MVDR beamformer can be expressed as:

[0168]

[0169] Substituting the optimal weight vector into the average power expression, the average power is obtained as:

[0170]

[0171] Change the angle θ in a(θ) in the range [-π,π] to get P MVDR (θ) change curve, for P MVDR (θ) curve to search for the spectrum peak and obtain the corresponding target azimuth angle θ.

[0172] Step 6, calculate the false target position point set Q according to the target distance R and the target azimuth θ:

[0173] After obtaining the target distance through CAFAR distance detection and the multipath false target position with the same angle but different distance through MVDR angle estimation, the polar coordinate system is converted into the Cartesian coordinate system to obtain the initial position point set Q of the multipath false target point:

[0174] Q=[Q1,Q2,....,Q n ]

[0175] Where: Q n =[R n sinθ n ,R n cosθ n ] represents the solved false target position.

[0176] Step 7, perform multipath false target matching elimination on the false target position point set Q to obtain the effective false target position point set Q′:

[0177] The algorithm flow is as follows Figure 2 As shown, firstly, the false target position point set Q is arranged from small to large according to the distance, and the corresponding distance of the first false target point is recorded as R′1. Then, the angle θ′1 corresponding to the first false target point is determined to determine the round-trip propagation path to which the false target belongs. The specific determination method is as follows:

[0178]

[0179] The radar's angular detection range is set to be 0° along the positive direction of the y-axis and the angle range from the negative semi-axis to the positive semi-axis [-π, π] along the x-axis, Figure 1 It can be seen that if θ′1>0, it means that the signal is along Φ 11 Propagation, the corresponding mirror target point is Q 11 Otherwise, it means that the mirror target point corresponding to the echo is Q 22 .

[0180] After determining the round-trip propagation path to which a false target point belongs, the true target position point P′ is estimated by solving the prior scene geometry information:

[0181]

[0182] Then use P′ to reversely estimate the corresponding multipath false target point position Q′ 12 , Q′ 21 , Q′ 22 , Q′ 11 , transform the other initial coordinate points from the polar coordinate system to the Cartesian coordinate system Q i =[R′ i sinθ′ i ,R′ i cosθ′ i ], i = 2, ..., n, respectively calculate the distance error between the multipath false target position point and other initial position points:

[0183]

[0184] Where: ||·||2 represents the Euclidean norm calculation.

[0185] The distance error is compared with the error threshold respectively. If the distance error is less than the threshold value TH, it means that the initial positioning point compared at this time is a two-way valid false target point. Its coordinate position is added to the valid false target point set Q′, and then the above matching process is repeated for the retained initial positioning point to eliminate all invalid false target points.

[0186] Step 8: Use the prior scene information to transform the coordinates of the effective false target point set Q′ to obtain the real target position point set P:

[0187] According to the law of mirror reflection of electromagnetic waves, combined with the distance L and d2 in the prior geometric scene, the real target point set P corresponding to the effective false target point is solved to complete the positioning process.

[0188] In order to verify the positioning performance of the target positioning method in the mine T-shaped tunnel based on millimeter-wave radar, simulation experiments were carried out on two scenarios: static single target positioning and moving single target positioning in the mine T-shaped tunnel.

[0189] For the static single target positioning in the T-shaped tunnel of the mine:

[0190] Positioning scenes such as Figure 3 As shown in (a), a rectangular coordinate system is established with the radar as the center origin, and the radar position is (0, 0) m. The target position is (1, 1) m, which is located behind the corner and has no direct line of sight with the radar. The corridor behind the corner is 2.3 m wide. Figure 3 (b) shows the initial target positioning result. It can be seen that the false target points form four clusters. The distance from the radar from near to far corresponds to the false target points Q formed by different paths. 22 , Q 21 , Q 12 , Q 11 , where Q 21 , Q 12 , Q 11 Invalid false target points that need to be eliminated.

[0191] The estimated real target point is calculated using the position coordinates of the initial positioning point, and then the theoretical position of the estimated real target point corresponding to the false target point is calculated through geometric relationship. Then the false target theoretical position is matched with the remaining initial positioning points, and the positioning points matched to the invalid false target points are eliminated. Finally, the positioning point set after the invalid false target points are eliminated can be obtained. The matching elimination result is as follows: Figure 3 (c) As shown. After matching elimination, the positioning result only retains the valid target points of the round-trip path, and the invalid false target positioning points of the combined path are completely eliminated. The position of the real target point is calculated through geometric information. The final positioning result is as follows Figure 3 In this scenario, the positioning error comparison between the present invention and the grid matching method is shown in Figure 5 As shown in (a), the average positioning error of the present invention in 100 cycles is 0.101m, and the positioning accuracy is better than that of the grid matching method.

[0192] For single target positioning in T-shaped tunnel of mine:

[0193] Positioning scenes such as Figure 4 As shown in (a), a rectangular coordinate system is established with the radar as the center origin, and the radar position is (0, 0) m. The target position is behind the corner and there is no direct line of sight between the radar and the front wall. The distance from the radar to the front wall is 3 m. The target moves from (1, 1) m to (1, 4) m at a speed of 1 m per second. Figure 4 (b) shows the initial target positioning result. It can be seen that the false target points form four clusters. The distance from the radar from near to far corresponds to the false target points Q formed by different paths. 22 , Q 21 , Q 12 , Q 11 , where Q 21 , Q 12, Q 11 Invalid false target points that need to be eliminated.

[0194] The invalid false target elimination method is used to calculate the theoretical false target position from the estimated real point, and the invalid false target is matched and eliminated in combination with the initial positioning point. The matching elimination result is as follows: Figure 4 (c) As shown. After matching elimination, the valid target points of the round-trip path are retained, the invalid false target positioning points of the combined path are completely eliminated, and the real target point position is calculated through geometric information. The final positioning result is as follows Figure 4 In this scenario, the positioning error comparison between the present invention and the grid matching method is shown in Figure 5 As shown in (b), the average positioning error of the present invention during the target motion process is 0.103m, and the positioning accuracy is better than that of the grid matching method.

[0195] The above experimental simulation verifies the effectiveness of this millimeter-wave radar-based mine T-type tunnel target positioning method in the T-type tunnel. It can not only effectively solve the problem of non-line-of-sight scene target positioning in the T-type tunnel of the coal mine, but also has low computational complexity and high positioning accuracy. It can be applied to mine disaster rescue, mine unmanned driving and other fields.

Claims

1. A method for positioning a target in a T-shaped tunnel of a mine based on a millimeter-wave radar, wherein the T-shaped tunnel includes a tunnel I and a tunnel II which is perpendicular to the tunnel I and connected to the tunnel I, the tunnel I and the tunnel II form a T-shaped structure, the tunnel II includes a wall surface three and a wall surface one, the tunnel I includes a wall surface five which is not connected to the tunnel II and a wall surface four and a wall surface two which are connected to the tunnel II, the wall surface four intersects with the wall surface three to form a wall corner one, and the wall surface two intersects with the wall surface one to form a wall corner two; It is characterized in that The target positioning method of the mine T-shaped tunnel based on millimeter wave radar specifically includes the following steps: Step 1: Simplify the millimeter wave electromagnetic propagation path in the T-shaped tunnel scenario: Construct a priori geometric scene, let the radar be located at the coordinate origin O in lane II, and the radar initial phase angle is 0°, and the distance between the radar and wall 5 is L, and the distance between the radar and wall 3 is d 2 , the distance from the radar to the wall is d 1 ; Let the target be located between wall 2 and wall 5, and at the non-line-of-sight area P = [x, y] point behind the corner 2 formed by the intersection of wall 2 and wall 1; Assume that all walls are smooth, only consider the multipath signal generated by mirror reflection, and simplify it to the first-order reflection multipath of millimeter wave, let the reflection point of the one-way primary reflection path at wall 3 be G2, and the reflection point at wall 5 be G1; then The propagation path of the transmitted signal is: O→G1→P, O→G2→P The propagation path of the received signal is: P→G1→O, P→G2→O The transmission and reception paths of the signal electromagnetic waves are a combination of the above four paths, including: ① Propagation path I round trip: O→G1→P, P→G1→O, the electromagnetic wave signal is emitted from point O, reflected by point G1 to point P, and returns along the original route, which is recorded as path Φ 11 ; ② Propagation path I + propagation path II: O→G1→P, P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G1 and reaches point P. The return path is reflected by point G2 and returns to the radar receiving antenna, which is recorded as path Φ 12 ; ③ Propagation path II + propagation path I: O→G2→P and P→G1→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P. The return path is reflected back to the radar receiving antenna via point G1, which is recorded as path Φ 21 ; ④ Propagation path II round trip: O→G2→P and P→G2→O. The electromagnetic wave signal is emitted from point O, reflected by point G2 and reaches point P, which is recorded as path Φ 22 ; Round trip path Φ 11 , Φ 22 About the radar's mirror target point Q 11 , Q 22 Calculated by geometric relationship: Where: [m,n] represents Q 11 The coordinates of; [p,q] represents Q 22 The coordinates of Φ 12 The path echo is reflected from point G2 and reaches the receiving antenna in the same direction as the mirror target point Q. 22 equal, the virtual target point is recorded as Q 12 ; Φ 21 The path echo is reflected from point G1 to the receiving antenna, and its direction is the same as the mirror target point Q 11 equal, the virtual target point is recorded as Q 21 ; Let Q 21 The distance from the radar is R 21 , the azimuth is θ 21 ;Q 12 The distance from the radar is R 12 , the azimuth is θ 12 ; The distances between the false target points and the radar corresponding to the four paths are expressed as follows: According to Q 11 , Q 22 The coordinates of the four paths are used to obtain the azimuths of the false target points as follows: Combined path virtual target point Q 12 , Q 21 The position is calculated by the following formula: Step 2: Establish a radar echo model based on the T-shaped tunnel scene: Use continuous frequency modulation millimeter wave radar to transmit millimeter waves to locate the target. Suppose the linear frequency modulation signal of the millimeter wave radar is s(t), and its expression is: Where: f 0 represents the starting frequency of the transmitted signal; A represents the amplitude of the transmitted signal; B represents the signal bandwidth, T represents the signal pulse width; μ is the frequency modulation slope of the signal, μ = B / T; is the initial phase angle, The received signal is represented as: r(t)=K 1 s(t-τ 11 )+K 2 s(t-τ 12 )+K 3 s(t-τ 22 )+K 4 s(t-τ 44 )+n(t) Where: K 1 ~K 4 represents the scattering coefficient of the four propagation paths; n(t) represents the environment and system noise; τ 11 , τ 12 , τ 21 , τ 22 Represent the path Φ 11 , Φ 12 , Φ 21 , Φ 22 The propagation delay of τ 11 , τ 12 , τ 21 , τ 22 The corresponding delay calculation formula is as follows: In the formula: c represents the speed of light; || represents the distance operation between two points; Step 3, collect radar echoes and complete preprocessing: Perform MTI clutter suppression on the sampled echo signal, use pulse compression on the signal after MTI processing to obtain the original range image, perform incoherent superposition on the original range image, and obtain the range image incoherent accumulation signal; Step 4, use CA-CFAR detection method to obtain the target distance R for the non-coherent accumulation signal of the range image; Step 5, using the minimum mean square distortion-free response (MVDR) spatial domain filtering method to obtain the target azimuth angle θ; Step 6, calculate the false target position point set Q according to the target distance R and the target azimuth θ: Convert the polar coordinate system into the Cartesian coordinate system to obtain the initial position point set Q of the multipath false target points. Q=[Q 1 ,Q 2 ,....,Q n ] Where: Q n represents the solved false target position, Q n =[R n sinθ n ,R n cosθ n ]; Step 7, perform multipath false target matching elimination on the false target position point set Q to obtain a valid false target position point set Q′; Step 8, transform the coordinates of the effective false target point set Q′ to obtain the real target position point set P: Combining the distance L and d in the prior geometric scene 2 Solve the real target point set P corresponding to the effective false target points to complete the positioning process.

2. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 1, It is characterized in that When MTI clutter is suppressed in Step 3, let the echo signal received by the radar be r, then: r i * =r i -r i-1 Where: r i * represents the signal after MTI processing, and i represents the number of pulses.

3. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 2, It is characterized in that The steps to obtain the non-coherent accumulation signal of the range image in Step 3 are as follows: The range image data of the i-th pulse is defined as: x(i)=[x(i,1),x(i,2),...,x(i,j),...,x(i,N d )] Where: x(i,j) represents the amplitude of the j-th distance unit; N d Represents the number of distance units; For the distance image x, N T The number of pulses is one frame for incoherent superposition to obtain the range image incoherent accumulation signal. The process is described as: Where: |·| represents absolute value operation; u(h) represents the signal after non-coherent accumulation of the hth frame; N T Represents the number of pulses in each frame.

4. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 1, It is characterized in that Step 4 The specific steps are as follows: For the j-distance unit, the detection threshold is expressed as: Where: N r represents the number of reference cells; P fa represents the constant false probability; Compare the detection threshold η according to the adaptive judgment criterion j The size of the unit u(h,j) to be detected is used to obtain the unit number of the distance where the target is located; Apply the adaptive judgment criterion to all distance units of the range image to obtain the target distance R detected in the hth frame.

5. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 4, It is characterized in that The adaptive judgment criteria are as follows: Where: H 1 represents the target hypothesis; H 0 Represents the no-target hypothesis.

6. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 1, It is characterized in that Step 5 The specific steps are as follows: Define the steering vector a(θ) of the receiving antenna element receiving the signal. The specific expression is: a(θ)=[1,e -jφ ,...,e -j(K-1)φ ] T ,φ=(2πdsinθ) / λ Where: θ represents the direction of the incoming wave; d represents the receiving array element spacing, d = λ c / 2;λ c Indicates the carrier wavelength; The spatial filter output in MVDR beamforming is: y(j)=w H x(j) Where: w is the weight vector of the spatial domain filter, w = [w 1 ,w 2 ,...,w K ] T ; x(j) is the target distance detection result signal containing K channels; [·] H represents the conjugate transpose operation; The average output power P(θ) is: P(θ)E{|y(j)| 2 }E{w H x(j)x H (j)w}jw H Rw Where: R = E{x(j)x H (j)} is the autocorrelation matrix; E{·} is the mathematical expectation; The optimal weight vector of the MVDR beamformer can be expressed as: Substituting the optimal weight vector into the average power expression, the average power is obtained as: Change the angle θ in a(θ) in the range [-π,π] to get P MVDR (θ) change curve, for P MVDR (θ) curve to search for the spectrum peak and obtain the corresponding target azimuth angle θ.

7. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 1, It is characterized in that In Step 7, determine the angle θ′ corresponding to the false target point 1 , When determining the round-trip propagation path to which the false target belongs, the specific judgment method is as follows: The radar's angular detection range is set to be 0° along the positive direction of the y-axis and the angle range from the negative semi-axis to the positive semi-axis [-π, π] along the x-axis. If θ′ 1 >0, indicating that the signal is along Φ 11 Propagation, the corresponding mirror target point is Q 11 Otherwise, it means that the mirror target point corresponding to the echo is Q 22 .

8. The method for locating a target in a mine T-shaped tunnel based on millimeter wave radar according to claim 1, It is characterized in that Step 7 The specific steps are as follows: First, the false target position point set Q is arranged from small to large distance, and the corresponding distance of the false target point is recorded as R 1 ′, and determine the angle θ′ corresponding to the false target point 1 , determine the round-trip propagation path to which the false target belongs; Solve and estimate the true target position point P′: Then use P′ to reversely estimate the corresponding multipath false target point position Q′ 12 , Q′ 21 , Q′ 22 , Q′ 11 , transform the other initial coordinate points from the polar coordinate system to the Cartesian coordinate system Q i =[R′ i sinθ′ i ,R′ i cosθ′ i ], i = 2, ..., n, respectively calculate the distance error between the multipath false target position point and other initial position points: Where: ||·|| 2 Represents Euclidean norm calculation; The distance error is compared with the error threshold respectively. If the distance error is less than the threshold value TH, it means that the initial positioning point compared at this time is a two-way valid false target point. Its coordinate position is added to the valid false target point set Q′. Then, the above matching process is repeated for the retained initial positioning point to eliminate all invalid false target points.