A multi-station passive cooperative localization method for sea targets based on multi-hypothesis prediction

Abstract

The present invention relates to a multi-station passive cooperative positioning method for sea targets based on multi-hypothesis prediction. According to the relative distances between each observation point and the sea target and the angular measurement error characteristics of each observation point, the distribution area of the sea target is divided in the angular dimension, and the distribution probability of the sea target in each area is calculated; according to the pre-set probability distribution coefficient of the heading and speed change of the sea target, with the latest track point of each predicted track as the center and combining the observed time interval, the distribution areas of the target motion attributes are demarcated respectively from the distance and azimuth dimensions, and the distribution probability of the sea target in each area is calculated; the probability scores of each predicted track are calculated by using the overlapping relationship between the angular dimension distribution area and the motion attribute distribution area, and the track point with the highest current track probability score is output as the positioning result of the sea target.

Description

Technical Field

[0001] The present invention relates to the field of positioning, and particularly to the field of passive cooperative positioning of maritime radiation source targets by multiple observation stations. Background Art

[0002] In modern warfare, it is becoming increasingly important to accurately locate and stably track targets. Active detection methods represented by radar are easily detected, interfered with, and struck. Therefore, under specific conditions, it is more meaningful to locate and track radiation source targets through passive positioning methods based on electronic reconnaissance. Passive positioning means that the observation station does not actively emit signals, and by receiving the angle information, time information, or Doppler information of the target radiation source, one or more of the information such as angle, time difference, and Doppler frequency are used to solve the equations to achieve positioning.

[0003] Passive positioning technology has good concealment. According to the number of observation stations, it can be divided into: multi-station passive cooperative positioning and single-station passive positioning. The information received among multiple observation stations often has complementary and redundant phenomena. Therefore, multi-station passive cooperative positioning is often faster and more accurate than single-station passive positioning, and has also been widely applied. In multi-station passive cooperative positioning, using the angle measurement information of the observation station for target angle measurement and positioning is one of the most commonly used passive cooperative positioning technologies.

[0004] When performing multi-station passive cooperative positioning of sea targets, due to factors such as sea clutter and over-the-horizon detection, there are differences in the signal-to-noise ratios of the target signals received by each observation station, resulting in large differences in the measurement accuracy of the target, and ultimately leading to a decrease in the target positioning accuracy. And too low target positioning accuracy leads to poor track quality of the target, and cannot provide effective track prediction ability for multi-station passive cooperative positioning. Summary of the Invention

[0005] The present invention provides a multi-station passive cooperative positioning method for sea targets based on multi-hypothesis prediction. According to the relative distances between each observation point and the sea target and the angular measurement error characteristics of each observation point, the distribution area of the sea target is divided in the angular dimension, and the distribution probability of the sea target in each area is calculated; according to the pre-set probability distribution coefficient of the heading and speed change of the sea target, with the latest track point of each predicted track as the center, combined with the observed time interval, the distribution areas of the target motion attributes are delimited respectively from the distance and azimuth dimensions, and the distribution probability of the sea target in each area is calculated; the probability scores of each predicted track are calculated by using the overlapping relationship between the angular dimension distribution area and the motion attribute distribution area, and the track point with the highest current track probability score is output as the positioning result of the sea target. The specific implementation process is as follows:

[0006] Step 1: Roughly calculate the distance of the target relative to each observation point by directly intersecting the passive angle measurement results of multiple observation points for the same target;

[0007] Step 2: According to the relative distances between each observation point and the sea target and the angle measurement error characteristics of each observation point, divide the distribution area of the sea target in the angular dimension, and calculate the distribution probability of the sea target in each area;

[0008] Step 3: Determine whether the target has established a track by the measured parameters observed by the observation point; if not, establish initial tracks one by one according to the divided angular dimension areas, and the centroid of each divided area is the initial track point of the corresponding track; if it has established a track, calculate the time interval from the previous observation, and read each predicted track information of the target;

[0009] Step 4: According to the pre-set probability distribution coefficient of the course and speed change of the sea target, with the latest track point of each track as the center, combined with the observation time interval, delimit the distribution areas of the target motion attributes from the distance and azimuth dimensions respectively, and calculate the distribution probability of the sea target in each area;

[0010] Step 5: Calculate the probability scores of each predicted track by using the overlapping relationship between the angular dimension distribution area and the motion attribute distribution area;

[0011] Step 6: Delete the invalid track information with a probability score of zero, and deeply maintain the valid tracks;

[0012] Step 7: Update the multiple predicted track information of the sea target, and output the track point with the highest track probability score as the positioning result of the sea target.

[0013] Further, the calculation process of the track probability score in Step 5 is as follows:

[0014] (1) Poll each track and read the position, course, speed and track probability score of its latest track point;

[0015] (2) For each track, poll each angular dimension distribution area and read the angular dimension distribution probability of this area, calculate the centroid of the current polled area and the coordinates of each vertex;

[0016] (3) Calculate the distance between each vertex of each angular dimension distribution area and the previous track point, and calculate the relative azimuth between the centroid of the angular dimension area and the previous track point;

[0017] (4) According to the minimum distance between each vertex of the angular dimension distribution area and the previous track point, and the relative azimuth information between the area centroid and the previous track point, calculate the motion attribute distribution probability of the current track in this angular dimension distribution area;

[0018] (5) Calculate the new probability score of the current track, which is the product of the previous probability score of the current track, the angular dimension distribution probability, and the motion dimension distribution probability;

[0019] (6) Check whether all angular dimension distribution regions have been polled for a single track; if not, poll the next angular dimension distribution region; if so, proceed with the polling for the next track;

[0020] (7) Check whether all predicted tracks have been polled for a single sea target; if not, poll the next predicted track; if so, proceed to the next step;

[0021] (8) Normalize the new probability scores of all predicted tracks.

[0022] According to the pre-set probability distribution coefficient of the course and speed change of the sea target, with the latest track point of each predicted track as the center, combined with the observed time interval, the target motion attribute distribution regions are delimited respectively from the distance and azimuth dimensions, and the distribution probability of the sea target in each region is calculated. Furthermore, the present invention completes the hypothesis prediction of multiple tracks of a single sea target from the aspect of the motion attributes of the sea target; according to the relative distance between each observation point and the sea target and the angular measurement error characteristics of each observation point, the distribution regions of the sea target are divided in the angular dimension, and the distribution probability of the sea target in each region is calculated. Furthermore, the present invention completes the hypothesis prediction of a single sea target in multiple distribution regions from the aspect of the angular measurement error characteristics of multiple observation stations; by correlating the hypothesis predictions between multiple tracks and multiple target distribution regions, and selecting the track point with the highest track probability score as the single positioning result; the present invention can preserve all possible sea target tracks to the greatest extent, and can make the positioning result converge quickly when the angular measurement accuracy of the observation station is poor. Description of the Drawings

[0023] Figure 1 Schematic diagram of the angular measurement error distribution characteristics of a single observation station.

[0024] Figure 2 Schematic diagram of multi-hypothesis prediction based on the angular measurement error distribution characteristics of a double observation station.

[0025] Figure 3 Schematic diagram of multi-hypothesis prediction based on a single track of a low-speed stationary sea target.

[0026] Figure 4 Schematic diagram of multi-hypothesis prediction based on a single track of a medium-low speed sea target.

[0027] Figure 5 Schematic diagram of the association between multiple predicted tracks and multiple predicted distribution regions of a single sea target.

[0028] Figure 6Schematic diagram of the multi-station passive cooperative positioning process for sea targets based on multi-hypothesis prediction.

[0029] Figure 7 Schematic diagram of the track probability score calculation process. Specific implementation manners

[0030] The present invention provides a multi-station passive cooperative positioning method for sea targets based on multi-hypothesis prediction. When performing passive detection on sea targets, due to factors such as sea clutter and over-the-horizon detection, the measured angle values of each observation station are not necessarily absolute accurate values, but the angle values with the highest probability. The specific probability values are closely related to the angular measurement error distribution characteristics of the observation station. The angular measurement error distribution characteristics of a single observation station are as Figure 1 shown. Figure 1 In [figure], 101 represents the direction finding result given by the observation station. Assuming that the angular measurement error distribution of the observation station is a normal distribution, the possible existence area of the target to be measured in the figure is the three-sigma area of the angular measurement error, and this area is divided into 2N equal parts; the angular areas represented by A(1) and A(-1) have the darkest color and are closest to the actual measured angle, indicating that the probability of the target in these two areas is the highest; while the angular areas represented by A(N) and A(-N) have the lightest color and are farthest from the actual measured angle, indicating that the probability of the target in these two areas is the lowest.

[0031] When performing multi-station passive cooperative positioning on sea targets, the distribution area of the sea target can be divided in the angular dimension according to the relative distance between each observation point and the sea target and the angular measurement error characteristics of each observation point, and the distribution probability of the sea target in each area can be calculated. The multi-hypothesis prediction schematic diagram based on the angular measurement error distribution characteristics of two observation stations is as Figure 2 shown. In the figure, A and B represent two observation stations, 201 represents the direction finding result given by observation station A at time t, and 202 represents the direction finding result given by observation station B at time t; assuming that the angular measurement error distributions of the two observation stations are both normal distributions, the possible existence area of the target to be measured in the figure is the three-sigma area of the angular measurement error, and this area is divided into 4N 2 quadrilateral areas; the areas represented by A(1,1), A(1,-1), A(-1,1), and A(-1,-1) have the darkest color and are closest to the actual measured angles of the two observation stations, indicating that the probability of the target in these four areas is the highest; while the areas represented by A(N,N), A(N,-N), A(-N,N), and A(-N,-N) have the lightest color and are farthest from the actual measured angles of the two observation stations, indicating that the probability of the target in these two areas is the lowest.

[0032] Sea targets have a large inertia due to their heavy weight, so their motion characteristics are relatively stable. Certain probability distribution characteristics of the course and speed changes can be summarized. Generally speaking, the greater the speed, the smaller the course change rate. According to the pre-set probability distribution coefficient of the course and speed changes of sea targets, with the latest track point of each predicted track as the center, combined with the observed time interval, the distribution areas of the target motion attributes are delimited respectively from the distance and azimuth dimensions, and the distribution probabilities of the sea targets in each area are calculated. The multi-hypothesis prediction schematic diagram based on a single track of a low-speed stationary sea target is as shown in Figure 3 Figure; in the figure, M represents the divided area, which can be divided into N + 1 gears from s0 to sN in the distance dimension; s0 is the low-speed stationary gear, which is no longer divided in the angle dimension; as the distance dimension increases, the division in the angle dimension becomes finer, and the highest speed gear sN is divided into M gears from d1 to dM in the angle dimension. Figure 3 For a sea target in the low-speed stationary gear at the previous moment, in the figure, as the distance value increases in the distance dimension, the color of area M gradually fades, indicating that the distribution probability of the sea target gradually decreases as the distance value increases at the next observation moment. That is, the distribution probability of area M(s0) is the largest, the distribution probabilities of areas M(s1,d1) to M(s1, d4) are the second largest, and the distribution probabilities of areas M(sN,d1) to M(sN, dM) are the smallest. Figure 4 Figure is the multi-hypothesis prediction schematic diagram of a single track of a medium-low speed sea target. The speed of the sea target at the previous observation moment is s1 and the course is d1; in the figure, the areas M(s1,d1), M(s1,d2), and M(s1, d4) have the darkest colors, indicating that the probabilities of the sea target in areas M(s1,d1), M(s1,d2), and M(s1, d4) are the largest at the next observation moment; the probabilities in M(s0), M(s2,d1), M(s2,d2), M(s2,d3), M(s2,d4), M(s2,d7), and M(s2,d8) are the second largest; the probabilities outside the above areas are smaller.

[0033] The present invention calculates the probability scores of each predicted track by using the overlapping relationship between the angular dimension distribution area and the motion attribute distribution area, and outputs the track point with the highest current track probability score as the positioning result of the sea target. The schematic diagram of the association between multiple predicted tracks and multiple predicted distribution areas of a single sea target is as shown in Figure 5 Figure. The large quadrilateral, i.e., the area shown as 501 in the figure, is the predicted distribution interval of the sea target at time t1 according to the angular measurement error characteristics of two observation stations; the four circular areas, i.e., the areas shown as 502, are the 4 target distribution intervals predicted based on four groups of effective track information at time t0; the shaded part is the overlapping part of the two, i.e., the effective association area.

[0034] Based on the above key points, the schematic diagram of the processing flow of the present invention is as shown in Figure 6 Figure, and the specific implementation process is as follows:

[0035] (1) In the first step, the passive angle measurement results of multiple observation points for the same target are used to roughly calculate the distance of the target relative to each observation point by direct intersection.

[0036] (2) In the second step, according to the relative distances between each observation point and the sea target and the angle measurement error characteristics of each observation point, the sea target is divided into distribution regions in the angular dimension, and the distribution probability of the sea target in each region is calculated.

[0037] (3) In the third step, it is judged whether the target has established a track through the measured parameters observed by the observation point; if not, initial tracks are established one by one according to the divided angular dimension regions, and the centroid of each sub-region is the initial track point of the corresponding track; if it has established a track, the time interval from the previous observation is calculated, and each predicted track information of the target is read.

[0038] (4) In the fourth step, according to the pre-set probability distribution coefficient of the course and speed change of the sea target, with the latest track point of each track as the center and combined with the observation time interval, the distribution regions of the target motion attributes are delimited respectively from the distance and azimuth dimensions, and the distribution probability of the sea target in each region is calculated.

[0039] (5) In the fifth step, the probability scores of each predicted track are calculated by using the overlapping relationship between the angular dimension distribution region and the motion attribute distribution region.

[0040] (6) In the sixth step, the invalid track information with a probability score of zero is deleted, and the valid tracks are deeply maintained.

[0041] (7) In the seventh step, multiple predicted track information of the sea target is updated, and the track point with the highest track probability score is output as the positioning result of the sea target.

[0042] Among them, the calculation process of the track probability score in the fifth step is as Figure 7 shown, and the specific process is as follows:

[0043] (1) In the first step, poll each track and read the position, course, speed and track probability score of its latest track point.

[0044] (2) In the second step, for each track, poll each angular dimension distribution region and read the angular dimension distribution probability of this region, and calculate the centroid and the coordinates of each vertex of the current polled region.

[0045] (3) In the third step, calculate the distances between each vertex of each angular dimension distribution region and the previous track point, and calculate the relative azimuth between the centroid of the angular dimension region and the previous track point.

[0046] (4) The fourth step is to calculate the distribution probability of the motion attributes of the current track in the angular dimension distribution region based on the minimum distances between the vertices of the angular dimension distribution region and the previous track point, as well as the relative azimuth information between the region centroid and the previous track point.

[0047] (5) The fifth step is to calculate the new probability score of the current track, which is the product of the previous probability score of the current track, the angular dimension distribution probability, and the motion dimension distribution probability.

[0048] (6) The sixth step is to check whether all angular dimension distribution regions have been polled for a single track; if not, poll the next angular dimension distribution region; if so, proceed with the polling for the next track.

[0049] (7) The seventh step is to check whether all predicted tracks have been polled for a single sea target; if not, poll the next predicted track; if so, proceed to the next step.

[0050] (8) The eighth step is to normalize the new probability scores of all predicted tracks.

Claims

1. A multi - station passive collaborative positioning method for sea targets based on multi - hypothesis prediction, characterized in that: Step 1: Roughly calculate the distance of the target relative to each observation point by directly intersecting the passive angle - measuring results of multiple observation points for the same target. Step 2: Divide the detection area into multiple angular - dimension distribution regions by dividing the angles of multiple observation points in the azimuth; calculate the distribution probability of the sea target in each region according to the angular deviation between each region and the measurement values of each observation point, combined with the angular - measurement error distribution characteristics of the observation stations. Step 3: Determine whether the target has established a track through the measured parameters observed by the observation points; if not, establish initial tracks one by one according to the divided angular - dimension regions, and the centroid of each distribution region is the initial track point of the corresponding track; if it has established a track, calculate the time interval from the last observation, and read the information of each predicted track of the target. Step 4: First, with the latest track point of each track as the center, select a distance interval according to the speed information of the latest track and the observation time interval, and divide the prediction area into multiple distance segments; finally, divide the angles of each distance segment in the prediction area to obtain the target motion - attribute distribution region. Step 5: Calculate the probability scores of each predicted track by using the overlapping relationship between the angular - dimension distribution region and the motion - attribute distribution region, including: (1) Poll each track and read the position, course, speed and track probability score of its latest track point. (2) For each track, poll each angular - dimension distribution region and read the angular - dimension distribution probability of this region, calculate the centroid of the current polled region and the coordinates of each vertex. (3) Calculate the distance between each vertex of each angular - dimension distribution region and the previous track point, and calculate the relative azimuth between the centroid of the angular - dimension region and the previous track point. (4) Calculate the motion - attribute distribution probability of the current track in this angular - dimension distribution region according to the minimum distance between each vertex of the angular - dimension distribution region and the previous track point, and the relative azimuth information between the region centroid and the previous track point. (5) Calculate the new probability score of the current track, which is the product of the previous probability score of the current track, the angular - dimension distribution probability and the motion - dimension distribution probability. (6) Check whether all angular - dimension distribution regions have been polled for a single track; if not, poll the next angular - dimension distribution region; if so, perform the polling work for the next track. (7) Check whether all predicted tracks have been polled for a single sea target; if not, poll the next predicted track; if so, proceed to the next step. (8) Normalize the new probability scores of all predicted tracks. Step 6: Delete the invalid track information with a probability score of zero, and deeply maintain the valid tracks. Step 7: Update the information of multiple predicted tracks of the sea target, and output the track point with the highest track probability score as the positioning result of the sea target.

Images