CN106842165A - One kind is based on different distance angular resolution radar centralization asynchronous fusion method - Google Patents

Abstract

The invention discloses a centralized asynchronous fusion method based on radars with different distance and angle resolutions, and relates to research on measurement alignment of radars with different distance resolutions and multi-radar data fusion technology. The present invention first establishes the target movement and measurement equations under the polar coordinate system of the radar; then adopts space grid division, and aligns radar measurement data with different distances and angular resolutions to the same size by means of grid alignment , and then use the centralized asynchronous fusion method to fuse the measurement data of multiple radars, and use the DP-TBD method to process the fused data and restore the target trajectory. The present invention effectively solves the problem that radars with different distances and angular resolutions are difficult to perform centralized fusion using a dynamic programming algorithm based on tracking before detection in practical applications, thereby realizing centralized measurement of radars with different distances and angular resolutions Asynchronous fusion can effectively improve the tracking effect on weak targets.

Description

A centralized asynchronous fusion method based on radars with different range and angle resolutions

technical field

The invention belongs to the field of radar technology, and relates to the research on measurement alignment of radars with different distance resolutions and multi-radar data fusion technology.

Background technique

Track-before-detect (TBD) technology is a technology to detect and track targets under the condition of low signal-to-noise ratio. The difference from the general detection method is that it adopts a new idea: the detection result is not announced in a single frame, but the single-frame echo data information is digitized and stored, and the multi-frame data is processed simultaneously. Announce detection results and track of the target. Its essence is to accumulate multiple frames of the target signal, highlight the target information while suppressing clutter interference, and use time accumulation in exchange for an increase in the signal-to-noise ratio, which solves the problem of processing each frame of data by the traditional tracking algorithm after detection The problem that the useful target information may be lost, effectively improves the radar's ability to detect and track weak targets under strong clutter, strong interference, and low signal-to-noise ratio.

There are many algorithms for implementing TBD technology, including Dynamic Programming, Particle Filter, Hough Transform, and Maximum Likelihood Probability Data Fusion (ML-PDA).

Among them, the tracking-before-detection algorithm based on dynamic programming (DP-TBD) has the characteristics of allowing the target to have a certain mobility and being easy to implement, and has become a hot spot in the field of TBD research. In recent years, as an efficient method for detecting and tracking targets, DP-TBD technology has broad development space in both civilian and military fields.

For the multi-radar fusion method based on DP-TBD, by using the measurement data of multiple radars, the detection and tracking effects can be significantly improved and the spatial gain of the radar reflection cross section (RCS) of the target can be obtained. However, there are not many studies on multi-sensor fusion methods based on DP-TBD, and the problems of how to process a large amount of raw data from multiple sensors and how to fuse asynchronous measurement data are difficult to solve. In the document "An Amplitude Association Dynamic Programming TBD Algorithm with Multistatic Radar, Control Conference (CCC), 35thChinese TCCT, 5076-5079, 2016", the multi-radar fusion based on the DP-TBD algorithm is realized, and the detection of weak targets is effectively improved. The tracking effect and the spatial gain of the radar reflection cross section (RCS) of the target are achieved. However, this model can only solve the fusion problem of radars with the same range and angle resolution, and cannot be applied to solve the fusion problem of radars with different range resolutions.

Contents of the invention

The purpose of the present invention is to address the defects of the existing technology, research and design a centralized fusion method based on DP-TBD that can realize radars with different distances and angular resolutions, and solve the problems of existing radar data with different distances and angular resolutions. Problems that can be addressed by dynamic (DP) programming algorithms based on track-before-detect (TBD).

The solution of the present invention is to first divide the space into a grid according to its distance and angle resolution, and then model it in the polar coordinate system of the real radar. The measurement uses spatial grid alignment to align radar data with different distance resolutions to the same grid size. First, a certain grid size is selected, and the detection area is quantified according to this grid size. When the size of the radar grid is smaller than the grid size, the largest measurement in the grid in the vicinity is taken out as the grid measurement. When the radar grid is larger than the determined grid size, the measurements of these radar grids are respectively corresponding to the positions of the specified grid, and Gaussian distribution noise with a mean of 0 and a variance of 1 is added to all other grid measurements. This method effectively solves the problem that radar data with different distance and angular resolutions cannot be processed by the DP-TBD method in practical applications, and uses the centralized fusion of multiple radar data to improve the detection ability of weak targets .

In order to describe content of the present invention conveniently, at first the following terms are explained:

Term 1: Polar coordinates

The polar coordinate system (polar coordinates) refers to the coordinate system composed of poles, polar axes and polar diameters in a plane. Take a fixed point O on the plane and call it a pole. A ray Ox is induced from O, which is called the polar axis. Then take a unit of length, usually the anticlockwise direction of the specified angle is positive. In this way, the position of any point P on the plane can be determined by the length ρ of the line segment OP and the angle θ from Ox to OP. The ordered number pair (ρ, θ) is called the polar coordinate of point P, and ρ is called the Polar radius, θ is called the polar angle of point P.

Term 2: Grid

Grid refers to dividing the radar detection area into several rectangular grids of specified size, which is the first step of DP-TBD algorithm

Term 3: Centralized Fusion

Centralized fusion means that the measurement data of each radar is directly transmitted to a fusion center for data processing without processing

Term 4: Fusion Center

A data processing center that processes the data transmitted to the fusion center according to the specified algorithm and recovers the motion state estimation of the target

Term 5: Radar range and bearing resolution

Radar distance resolution refers to the minimum actual distance at which the echo data generated by two targets on the radar screen can be distinguished. Angular resolution is the ability of an imaging system or system component to differentiate the smallest distance between two adjacent objects. The resolving power is generally expressed by the angle between the two smallest recognizable targets of the imaging system.

Term 6: Radar Cross Section

Radar cross-section (RCS) refers to the reflection cross-sectional area of the radar. The principle of radar detection is to emit electromagnetic waves to irradiate the surface of the object and then reflect back to the receiving antenna, and the less electromagnetic waves returned by the radar wave to the surface of the object according to the original path , the smaller the radar cross-sectional area, the smaller the signal characteristics of the radar to the target, and the shorter the detection distance.

The present invention proposes a centralized asynchronous fusion method based on radars with different distance and angle resolutions. The method includes:

Step 1: Each radar obtains the echo signal of the measurement space;

Step 2: Each radar divides its own measurement space into grids according to its own angle and distance resolution;

Step 3: Determine a grid template of the measurement space, map the measurement space of each radar according to the selected grid template, so as to unify the size of each radar measurement space;

Step 4: Transmit the measurement space data mapped by each radar to the fusion center, and the fusion center sorts each measurement space data in chronological order;

Step 5: Fuse the measurement data of each radar at the same time;

Step 6: Judging the target trajectory based on the fused data.

Further, the specific method of the step 2 is:

Step 2.1: Determine the radar with the centered resolution among the radars;

Step 2.2: divide the radar measurement space into grids, and determine the number of distance cells included in each grid;

Step 2.3: Carry out grid division for other radar measurement spaces, so that the number of distance cells contained in each divided grid is the same as the number of distance cells in the radar grid in step 2.2.

Further, the specific method of step 3 is:

Step 3.1: Determine the radar grid division method with the centered radar resolution as the grid template;

Step 3.2: When performing grid mapping for high-resolution radars higher than the radar resolution in step 3.1, first determine which grids of the high-resolution radar the measurement space of each grid in the grid template corresponds to, and then Select the grid with the largest measurement value from all the grids of the high-resolution radar corresponding to the same template grid, and map the selected grid to the template grid;

Step 3.3: When performing raster mapping on a low-resolution radar that is lower than the radar resolution in Step 3.1, use the following formula for raster mapping:

Z"(i,j)=Z(m,n), i=n _rl ×m, j=n _θl ×n;

Where Z″(i,j) represents the measurement value of the grid whose position is (i,j) after mapping, Z(m,n) is the measurement value of the grid whose position is (m,n) before mapping, n _rl represents the ratio of the range resolution of the radar to be mapped to the range resolution of the grid template radar, n _θl represents the ratio of the angular resolution of the radar to be mapped to the angular resolution of the grid template radar, and the unmapped grid The grid is filled with Gaussian white noise with a mean of 0 and a variance of 1.

Beneficial effects of the present invention: the method of the present invention utilizes the method of grid alignment and centralized asynchronous fusion of radar space measurement with different distance and angle resolutions, firstly the target is modeled in the direction of distance and azimuth, and then the space is gridded Grid division, the radar measurement data with different distances and angular resolutions are aligned to the grid size, and then the aligned measurement data is sent to the fusion center, and then the measurement data are sorted in time order and performed by the motion DP-TBD algorithm. Processing, thus solving the problem that radars with different distance and angle resolutions cannot achieve data fusion using DP-TBD algorithm. The advantage of the present invention is that the above method can effectively improve the detection and tracking effect of the target and realize the fusion of radar data with different distances and angular resolutions, and the solution process is simple. The invention can be applied to the fields of radar signal processing, early warning radar tracking targets and the like.

Description of drawings

Fig. 1 is a flowchart of the method provided by the present invention.

Fig. 2 is a schematic diagram of different sizes of grid divisions of radars with different distances and angular resolutions in a specific embodiment of the present invention.

FIG. 3 is a schematic diagram of measurement sorting and centralized asynchronous fusion performed by multiple radars according to acquisition measurement time in a specific embodiment of the present invention.

Fig. 4 is the variation of the number of times of accurate target tracking by the first radar with the number of frames in the specific embodiment of the present invention.

Fig. 5 is the variation of the number of times of accurate target tracking by the second radar with the number of frames in the specific embodiment of the present invention.

Fig. 6 shows the variation of the number of accurate target tracking by the third radar with the number of frames in the specific embodiment of the present invention.

Fig. 7 shows the change of the number of accurate target tracking times with the number of frames through centralized fusion in the specific embodiment of the present invention.

detailed description

The present invention mainly adopts the method of simulation experiment to verify, and all steps and conclusions are verified correctly on MatlabR2013a. The present invention will be further described in detail with regard to specific embodiments below.

Step 1: Determine the target motion equation;

It is assumed that there are P radars in the detection area, each radar processes F frames, and the target track is a series of continuous states, defined as follows from the first frame to the P×F frame:

Among them, P×F represents the total number of frames processed by the fusion center for a dynamic planning;

Step 2: Measuring space rasterization;

Consider that the detection area is an area composed of distance and angle; select a measurement space Z"' to determine the grid size; the measurement space is selected as the intermediate resolution radar among multiple radars; this is because when the resolution is relatively small When the value is high, the calculation amount of radar tracking will be greatly increased, and when the radar resolution is low, the detection error will increase;

This determined measurement space is evenly divided into N _r =a grids in the distance direction, and N _θ =b grids in the angle direction; the number of grid divisions here is based on the range resolution and angle of the radar Resolution; for radar with higher range resolution, it is assumed that the number of grids in the range direction is N _r ′=a×n _rh , and the number of grids in the angle direction is N′ _θ =b×n _θh ; n _xh represents the ratio of the measurement space grid size determined in the distance direction and the angle direction to the grid size of the high-resolution radar; for the low-range resolution radar, it is assumed that the number of grids in the r direction is N _r ″ = a/n _rl and N″ _θ = b/n _θl ;

Step 3: Align radar grid numbers with different range and velocity resolutions;

Alignment includes aligning both the high-range-angle-resolution radar and the low-range-angle-resolution radar to the measurement space Z"' with a certain grid size; for high-resolution radar, there are many useless These measurements will only increase the probability of false alarms but not the probability of detection; therefore, for DP iterations, we use a method of dividing the grid size. The most meaningful high-amplitude The measurement values are retained in the new grid. The conversion formula is as follows

Z _k (i, j) represents the measurement value of the (i, j)th resolution unit in the measurement space of the kth frame. Z' _k represents the measurement data of the kth frame after processing. And 1≤i≤a, 1≤j≤b, where M(i,j) represents all the grids corresponding to grid(i,j).

M _i,j ＝{(m,n):m∈(n _p ×in _p +1,...,n _p ×i),n∈(n _p ×jn _p +1,...,n _p ×j)} (3)

For radars with low range resolution, in the DP iteration, we adopt a method of dividing the size of the smaller grid size. The conversion formula from the original measurement space Z _k (m,n) to the measurement space Z _k ″(i,j) is as follows

Z _k "(i, j) = Z _k (m, n) (4)

Z _k ″ represents the measurement of the k-th frame after processing. And 1≤i≤a, 1≤j≤b, 1≤m≤N _r ″, and i=n _rl ×m, j=n _θl ×n. In this scenario, each grid in the measurement _{Zk will correspond to several grids in the measurement Zk″. Therefore, after transformation, many grids in Z″k do not} contain any measurement information . Since these grids do not contain any target information, the measurement information of these grids is filled with Gaussian white noise with a mean value of 0 and a variance of 1.

Step 4: Centralized Fusion

After measurement grid matching, radar measurement grids of different resolutions are matched to the same grid size. These measurements from different radars are then sent to a fusion center. The fusion center receives these measurement data Z′ ₁ …Z′ _k , Z″ ₁ …Z″ _k , Z″′ ₁ …Z″′ _k from the asynchronous radars with different resolutions and different periods, and then these measurement data The data are sorted in the order in which they were obtained. After sorting, these measurement data are Z ₁ , Z ₂ ...Z _P×F , (P represents the total number of radars, F represents the total number of frames processed by each radar DP). Z _i represents the measurement data of the i-th frame sorted by time. Afterwards, the DP algorithm is used to process these measurement data.

Step 5: Value function and state transition function initialization

At the initial moment i=1, the polar coordinate system motion equation of the target and

Among them, I is a value function, Ψ is used to record the state transition relationship, and the current moment is the initial moment, so let it be equal to zero.

Step 6: Dynamic Programming Recursion

When 2≤i≤P×F, for the state x _i of the i-th frame, we have

Set D is the area of state transition.

Step 7: Track Termination

The maximum value of the value function in the last frame is judged as passing the threshold, and if it exceeds the threshold, it is judged as an existing target:

Among them, V _T represents the threshold value, which is obtained by Monte Carlo simulation and satisfies the constant false alarm.

Step 8: Trajectory backtracking:

right When i=P×F-1,P×F-2,...,2,1, let

Get the estimated state sequence:

After steps 1 to 7, the trajectory of the target movement can be recovered. And the tracking effect of the restored track can be obtained by comparing it with the real moving track of the target.

Figures 4 to 7 are the accurate tracking renderings of three radars with different range resolutions adopted in the embodiment and the precise tracking renderings after centralized fusion, respectively, and the corresponding parameter table is Table 1. And the selected grid size is the same as the grid division size of Radar 2.

Table 1

It can be seen from the specific implementation of the present invention that the present invention can well realize centralized asynchronous fusion of radars with different distances and angular resolutions, and use the DP-TBD algorithm to improve the tracking effect on targets.

Those skilled in the art will appreciate that the embodiments described here are to help readers understand the principles of the present invention, and it should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical revelations disclosed in the present invention without departing from the essence of the present invention, and these modifications and combinations are still within the protection scope of the present invention.

Claims (3)

1. A centralized asynchronous fusion method based on different distance angle resolution radars, the method comprising: Step 1: Each radar obtains the echo signal of the measurement space; Step 2: Each radar divides its own measurement space into grids according to its own angle and distance resolution; Step 3: Determine a grid template of the measurement space, map the measurement space of each radar according to the selected grid template, so as to unify the size of each radar measurement space; Step 4: Transmit the measurement space data mapped by each radar to the fusion center, and the fusion center sorts each measurement space data in chronological order; Step 5: Fuse the measurement data of each radar at the same time; Step 6: Judging the target trajectory based on the fused data. 2. a kind of centralized asynchronous fusion method based on different distance angle resolution radars as claimed in claim 1, is characterized in that the concrete method of described step 2 is: Step 2.1: Determine the radar with the centered resolution among the radars; Step 2.2: divide the radar measurement space into grids, and determine the number of distance cells included in each grid; Step 2.3: Carry out grid division for other radar measurement spaces, so that the number of distance cells contained in each divided grid is the same as the number of distance cells in the radar grid in step 2.2. 3. a kind of centralized asynchronous fusion method based on different distance angle resolution radars as claimed in claim 1, is characterized in that the concrete method of described step 3 is: Step 3.1: Determine the radar grid division method with the centered radar resolution as the grid template; Step 3.2: When performing grid mapping for high-resolution radars higher than the radar resolution in step 3.1, first determine which grids of the high-resolution radar the measurement space of each grid in the grid template corresponds to, and then Select the grid with the largest measurement value from all the grids of the high-resolution radar corresponding to the same template grid, and map the selected grid to the template grid; Step 3.3: When performing raster mapping on a low-resolution radar that is lower than the radar resolution in Step 3.1, use the following formula for raster mapping: Z"(i,j)=Z(m,n), i=n _rl ×m, j=n _θl ×n; Where Z″(i,j) represents the measurement value of the grid whose position is (i,j) after mapping, Z(m,n) is the measurement value of the grid whose position is (m,n) before mapping, n _rl represents the ratio of the range resolution of the radar to be mapped to the range resolution of the grid template radar, n _θl represents the ratio of the angular resolution of the radar to be mapped to the angular resolution of the grid template radar, and the unmapped grid The grid is filled with Gaussian white noise with a mean of 0 and a variance of 1.