A target search method based on probability prediction

Abstract

The application discloses a target searching method based on probability prediction, and belongs to the technical field of target searching, and is characterized in that the method comprises the following steps: a, predicting the existing area of a target at a moment according to the state information of the target, and obtaining a probability function of the target at the moment being located at a point; b, controlling a search party to search for the target according to a search algorithm according to the obtained probability function; c, judging whether the search party searches for the target within a preset time threshold, if yes, entering step e, and if not, entering step d; d, adjusting the prediction parameters of the existing area of the target; e, ending the target searching based on the probability prediction. Through the probability function, the target probability distribution graph can be continuously updated, the search party has better directionality in searching for the target based on the probability distribution graph, the time required for searching can be effectively reduced, the task success rate and the prediction accuracy are improved, and the method has good applicability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of target search, in particular to a target search method based on probability prediction. BACKGROUND

[0002] The existing target search methods are all to analyze the target, judge the possible appearing area, and then search in these areas. However, most of the existing methods directly divide an area and assume that the target is uniformly distributed in the area, and another part of the method only considers the speed information of the target, and considers less about the destination and task time constraints of the target. The prediction accuracy of the existing technology for the possible appearing area of the target is low, and the search efficiency is poor.

[0003] The Chinese patent document with publication number CN110389595A and publication date of October 29, 2019 discloses a double attribute probability graph optimized unmanned aerial vehicle cluster cooperative target search method, which is characterized by at least including the following steps:

[0004] Step 1) According to the initial scene information of the unmanned aerial vehicle, a probability graph flag is introduced, and a double attribute matrix based environment probability graph model to be searched is established according to the probability graph flag; a probability graph update rule is established according to the probability graph model, and the probability graph information of the search process is updated according to the probability graph update rule;

[0005] Step 2) The flight rule of the unmanned aerial vehicle is given in combination with the target scene information, and the unmanned aerial vehicle motion model is established; the target function of the maximum benefit of the heading angle of the unmanned aerial vehicle and the constraint condition are determined;

[0006] Step 3) The optimal value of the heading angle of the unmanned aerial vehicle is obtained, the genetic algorithm is used, the heading angle is encoded into a series of population chromosomes composed of only -1, 0, and 1, the population chromosomes obtained by encoding the initial heading angle are used as initial chromosomes, and the initial population is encoded and decoded by using the improved cooperative evolution genetic algorithm; the genetic algorithm parameters are initialized, the optimal cooperative decision input heading angle set is generated by the improved cooperative evolution genetic algorithm optimization, and the cooperative path is generated.

[0007] The double attribute probability graph optimized unmanned aerial vehicle cluster cooperative target search method disclosed in the patent document overcomes the premature phenomenon of the traditional algorithm, ensures the feasibility of the flight path, and avoids repeated search of the area. However, the probability of any point at any time cannot be calculated, the prediction accuracy is low, and since the unmanned aerial vehicle needs to be solved in real time, the requirement for the computing power of the unmanned aerial vehicle is relatively high, which limits the application range. SUMMARY

[0008] The present application provides a target search method based on probability prediction, which can continuously update the target probability distribution graph through the probability function, and the search direction of the search party is more targeted based on the probability distribution graph, which can effectively reduce the search time, improve the task success rate and prediction accuracy, and has good applicability.

[0009] The present application is realized by the following technical solutions:

[0010] A target search method based on probability prediction, characterized in that it comprises the following steps:

[0011] a. Predicting the target existing area at time t according to the state information of the target, obtaining the probability function of the target located at point (x, y) at time t;

[0012] b. Controlling the search party to search for the target according to the search algorithm based on the obtained probability function;

[0013] c. Judging whether the search party has searched for the target within the preset time threshold, if yes, entering step e, if not, entering step d;

[0014] d. Adjusting the target existing area prediction parameters;

[0015] e. Ending the target search based on probability prediction.

[0016] In step a, the state information of the target includes the speed range of the target, the position of the target at the past time, and the destination position of the target.

[0017] In step b, the search party refers to a UAV.

[0018] In step b, the search algorithm includes algorithm one and algorithm two according to the different paths of the search party searching for the target.

[0019] The algorithm one specifically includes:

[0020] S11, search start;

[0021] S12, calculating the probability distribution graph of the target according to the current time t and the probability function in the target existing area prediction algorithm;

[0022] S13, obtaining the target appearance probability maximum point according to the probability distribution graph, and calculating the time required for the search party to reach the target appearance probability maximum point according to the current position of the search party; ​​​​​​​​

[0023] S14, Calculation Time At that time, the probability distribution diagram of the target, where ;

[0024] S15. Based on the probability distribution map in S14, obtain the point with the highest probability of the target appearing. Calculate the time required for the searcher to reach the point with the highest probability of the target appearing based on the current position of the searcher, and update the time value accordingly. ;

[0025] S16, Judgment If the time is less than a preset time threshold, return to S14; if the time is less than the preset time threshold, proceed to S17.

[0026] S17. The searcher moves to the point with the highest probability and uses the point with the highest probability as the center to search for the target.

[0027] S18, Search ends.

[0028] Algorithm 2 specifically includes:

[0029] S21. Search begins;

[0030] S22, Settings Equal to the current value;

[0031] S23. According to the probability function calculate Probability distribution of the target at any given time;

[0032] S24. Calculate the target probability gradient vector at the location of the searcher based on the probability density distribution map;

[0033] S25. The searcher moves according to the target probability gradient vector. time;

[0034] S26, Judgment If the target is not found within the time limit, proceed to S27; if the target is found, proceed to S29.

[0035] S27. Based on the probability distribution map of the searcher's position and the target, determine whether the searcher has reached the point of maximum probability. If it has, proceed to S29; otherwise, proceed to S28.

[0036] S28, Update The value;

[0037] S29. Search ends.

[0038] In step S28, update The value refers to the value of Get the updated value.

[0039] In the step c, the preset time threshold refers to a time required for the target straight-line heading to reach the destination.

[0040] The beneficial effects of the present application mainly manifest in the following aspects:

[0041] 1. The present application can continuously update the target probability distribution graph through the probability function, and the search for the target by the search party is more directional based on the probability distribution graph, which can effectively reduce the time required for the search, improve the task success rate and prediction accuracy, and has good applicability.

[0042] 2. Compared with the prior art using uniform distribution assumption or only considering target speed information, the present application can more accurately predict the target existence area and effectively improve the search efficiency.

[0043] 3. The present application introduces destination information into the target existence area prediction, which can be well applied to the situation with a large time span compared with the prior art.

[0044] 4. The present application outputs a probability function for the target existence area prediction, which can enable the search party to directly search near the point with the maximum probability through the probability function, making the search more directional and requiring a smaller search area, thereby greatly improving the search efficiency.

[0045] 5. Compared with the Chinese patent document with publication number CN110389595A and publication date October 29, 2019, since the update of the probability graph does not completely depend on the real-time search results, no calculation is required during the process of going to the target point, which can be used for both low-power unmanned aerial vehicles and high-power unmanned aerial vehicles, and has a wider range of applications.

[0046] 6. Compared with the Chinese patent document with publication number CN110389595A and publication date October 29, 2019, which uses genetic algorithm to obtain the optimal value of the unmanned aerial vehicle heading angle, the algorithm is based on maximizing the benefits of the unmanned aerial vehicle, rather than searching for the target as quickly as possible. The unmanned aerial vehicle of the present application can go to the point with the maximum target existence probability as soon as possible, greatly shortening the time required to search for the target. BRIEF DESCRIPTION OF DRAWINGS

[0047] The present application will be further specifically described below in conjunction with the drawings and specific embodiments of the present application:

[0048] Figure 1 The flowchart of the algorithm one of the present application;

[0049] Figure 2 The flowchart of the algorithm two of the present application. DETAILED DESCRIPTION

[0050] Example 1

[0051] A target search method based on probability prediction includes the following steps:

[0052] a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point probability function ;

[0053] b. Based on the obtained probability function The search algorithm controls the searcher to perform the target search.

[0054] c. Determine whether the searcher has found the target within the preset time threshold. If yes, proceed to step e; otherwise, proceed to step d.

[0055] d. Adjust the prediction parameters for the target area;

[0056] e. End the target search based on probability prediction.

[0057] This embodiment is the most basic implementation method. Through the probability function, the target probability distribution map can be continuously updated. Based on the probability distribution map, the searcher's search for the target is more targeted, which can effectively reduce the search time, improve the task success rate and prediction accuracy, and has good applicability.

[0058] Example 2

[0059] A target search method based on probability prediction includes the following steps:

[0060] a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point probability function ;

[0061] b. Based on the obtained probability function The search algorithm controls the searcher to perform the target search.

[0062] c. Determine whether the searcher has found the target within the preset time threshold. If yes, proceed to step e; otherwise, proceed to step d.

[0063] d. Adjust the prediction parameters for the target area;

[0064] e. End the target search based on probability prediction.

[0065] In step a, the target's state information includes the target's velocity range, the target's position at past times, and the target's destination location.

[0066] This embodiment is a preferred implementation method. Compared with existing technologies that use the assumption of uniform distribution or only consider target speed information, it can more accurately predict the area where the target exists and effectively improve search efficiency.

[0067] Example 3

[0068] A target search method based on probability prediction includes the following steps:

[0069] a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point probability function ;

[0070] b. Based on the obtained probability function The search algorithm controls the searcher to perform the target search.

[0071] c. Determine whether the searcher has found the target within the preset time threshold. If yes, proceed to step e; otherwise, proceed to step d.

[0072] d. Adjust the prediction parameters for the target area;

[0073] e. End the target search based on probability prediction.

[0074] In step a, the target's state information includes the target's velocity range, the target's position at past times, and the target's destination location.

[0075] In step b, the search party refers to a drone.

[0076] In step b, the search algorithm includes Algorithm 1 and Algorithm 2, depending on the different paths of the search target.

[0077] This embodiment is another preferred implementation. The target area prediction introduces destination information, which, compared with existing technologies, can be well applied to situations with a large time span.

[0078] Example 4

[0079] See Figure 1 A target search method based on probability prediction includes the following steps:

[0080] a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point a probability function ;

[0081] b, controlling the search party to search the target according to the obtained probability function and a search algorithm;

[0082] c, judging whether the search party searches the target within a preset time threshold, if yes, entering step e, if not, entering step d;

[0083] d, adjusting the target existing area prediction parameter;

[0084] e, ending the target search based on the probability prediction.

[0085] In the step a, the state information of the target includes a speed range of the target, a position of the target at a past time and a destination position of the target.

[0086] In the step b, the search party refers to a UAV.

[0087] In the step b, the search algorithm according to different paths of the search party searching the target includes algorithm one and algorithm two.

[0088] Further, the algorithm one specifically includes:

[0089] S11, starting searching;

[0090] S12, calculating a probability distribution diagram of the target according to a current time and a probability function in a target existing area prediction algorithm ;

[0091] S13, obtaining a maximum point of a target appearance probability according to the probability distribution diagram, and calculating a time required for the search party to reach the maximum point of the target appearance probability according to a current position of the search party ;

[0092] S14, calculating the probability distribution diagram of the target at a time , wherein ;

[0093] S15, obtaining the maximum point of the target appearance probability according to the probability distribution diagram in the S14, calculating a time required for the search party to reach the maximum point of the target appearance probability according to a current position of the search party, and updating the time value ;

[0094] S16, judging whether the time value is less than a preset time threshold, if not, returning to the S14; if yes, entering S17;

[0095] ​S17. The searcher moves to the point with the highest probability and uses the point with the highest probability as the center to search for the target.

[0096] S18, Search ends.

[0097] This embodiment is another preferred implementation. The predicted output of the target area is a probability function. Through the probability function, the searcher can directly go to the vicinity of the point with the highest probability to search, making the search more targeted, requiring a smaller search area, and greatly improving search efficiency.

[0098] Example 5

[0099] See Figure 2 A target search method based on probability prediction includes the following steps:

[0100] a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point probability function ;

[0101] b. Based on the obtained probability function The search algorithm controls the searcher to perform the target search.

[0102] c. Determine whether the searcher has found the target within the preset time threshold. If yes, proceed to step e; otherwise, proceed to step d.

[0103] d. Adjust the prediction parameters for the target area;

[0104] e. End the target search based on probability prediction.

[0105] In step a, the target's state information includes the target's velocity range, the target's position at past times, and the target's destination location.

[0106] In step b, the search party refers to a drone.

[0107] In step b, the search algorithm includes Algorithm 1 and Algorithm 2, depending on the different paths of the search target.

[0108] Furthermore, the second algorithm specifically includes:

[0109] S21. Search begins;

[0110] S22, Settings Equal to the current value;

[0111] S23. According to the probability function calculate Probability distribution of the target at any given time;

[0112] S24. Calculate the target probability gradient vector at the location of the searcher based on the probability density distribution map;

[0113] S25. The searcher moves according to the target probability gradient vector. time;

[0114] S26, Judgment If the target is not found within the time limit, proceed to S27; if the target is found, proceed to S29.

[0115] S27. Based on the probability distribution map of the searcher's position and the target, determine whether the searcher has reached the point of maximum probability. If it has, proceed to S29; otherwise, proceed to S28.

[0116] S28, Update The value;

[0117] S29. Search ends.

[0118] In step S28, update The value refers to the value of Get the updated value.

[0119] In step c, the preset time threshold refers to the time required to reach the destination on the target straight-line course, which is less than or equal to the time required.

[0120] This embodiment is the best implementation method. Compared with the Chinese patent document with publication number CN110389595A and publication date of October 29, 2019, since the update of the probability map does not completely depend on the real-time search results, no calculation is required in the process of going to the target point. It can be used for both low-computing-power drones and high-computing-power drones, and has a wider range of applications.

[0121] Compared to Chinese patent document CN110389595A, published on October 29, 2019, which uses a genetic algorithm to obtain the optimal value of the UAV's heading angle, and whose algorithm is based on maximizing the UAV's benefits rather than finding the target as quickly as possible, the UAV of this invention can reach the point with the highest probability of the target's existence as quickly as possible, greatly shortening the time required to find the target.

[0122] The basic principle of this invention is as follows:

[0123] Based on game theory analysis and probability inference, the objective at time [time] is obtained. Located at point probability function Thus, the target at any given time is obtained. The probability distribution map is used as the basis for target search.

[0124] On the basis of considering the route constraint, the speed constraint and the future speed constraint, the target probability function at time Located at point The probability function , the probability of any point at any time can be calculated, and the accuracy of the probability is improved.

Claims

1. A target search method based on probability prediction, characterized in that, Includes the following steps: a. Based on the target's status information, determine the target's time. Predicting the region of existence to obtain the target at time [time]. Located at point probability function ; b. Based on the obtained probability function The search algorithm controls the searcher to perform the target search. c. Determine whether the searcher has found the target within the preset time threshold. If yes, proceed to step e; otherwise, proceed to step d. d. Adjust the prediction parameters for the target area; e. End the target search based on probability prediction; In step b, the search algorithm includes Algorithm 1 and Algorithm 2 depending on the different paths of the searcher's target. Algorithm 2 specifically includes: S21. Search begins; S22, Settings Equal to the current value; S23. According to the probability function calculate Probability distribution of the target at any given time; S24. Calculate the target probability gradient vector at the location of the searcher based on the probability distribution map; S25. The searcher moves according to the target probability gradient vector. time; S26, Judgment If the target is not found within the time limit, proceed to S27; if the target is found, proceed to S29. S27. Based on the probability distribution map of the searcher's position and the target, determine whether the searcher has reached the point of maximum probability. If it has, proceed to S29; otherwise, proceed to S28. S28, Update The value; S29. Search ends.

2. The target search method based on probability prediction according to claim 1, characterized in that: In step a, the target's state information includes the target's velocity range and the target's past time... The location of the target and the destination location.

3. The target search method based on probability prediction according to claim 1, characterized in that: In step b, the search party refers to a drone.

4. The target search method based on probability prediction according to claim 1, characterized in that: Algorithm 1 specifically includes: S11, Search begins; S12, Based on the current time The probability function in the target area prediction algorithm Calculate the probability distribution of the target; S13. Based on the probability distribution map, find the point with the highest probability of the target appearing, and calculate the time required for the searcher to reach the point with the highest probability of the target appearing based on the searcher's current position. ; S14, Calculation Time At that time, the probability distribution diagram of the target, where ; S15. Based on the probability distribution map in S14, obtain the point with the highest probability of the target appearing. Calculate the time required for the searcher to reach the point with the highest probability of the target appearing based on the current position of the searcher, and update the time value accordingly. ; S16, Judgment If the time is less than a preset time threshold, return to S14; if the time is less than the preset time threshold, proceed to S17. S17. The searcher moves to the point with the highest probability and uses the point with the highest probability as the center to search for the target. S18, Search ends.

5. The target search method based on probability prediction according to claim 1, characterized in that: In step S28, update The value refers to the value of Get the updated value.

6. The target search method based on probability prediction according to claim 1, characterized in that: In step c, the preset time threshold refers to the time required to reach the destination on the target straight-line course, which is less than or equal to the time required.

Images