PTZ control path search method based on big data

Abstract

The present invention discloses a pan-tilt control path search method based on big data, and the method includes data collection, data preprocessing, establishing a pan-tilt path planning system, adjusting the parameters of the pan-tilt path planning system, and pan-tilt control path search. The present invention belongs to the field of path search technology, and specifically refers to a pan-tilt control path search method based on big data. The scheme generates an effective path planning strategy by initializing multiple commentator networks and an actor network; designs a nonlinear piecewise function, and adjusts the exploration rate based on the time step, introduces experience playback and designs loss calculation; to enhance the certainty and stability of path planning; the initialized candidate solution position can cover a wider search space, taking into account the total loss value of the system and the fitness of the distance calculation with the optimal solution, and guiding the search process to be closer to the global optimal solution; designs a nonlinear convergence factor and a dual update strategy to ensure that the system can complete parameter adjustment within a reasonable time.

Description

Technical Field

[0001] The present invention relates to the field of path search technology, and in particular to a pan / tilt control path search method based on big data. Background Art

[0002] The big data-based gimbal control path search method uses a large amount of historical path data and machine learning algorithms to achieve path planning and control of the gimbal system in a complex environment through data analysis, modeling, and optimization. However, the general gimbal control path search method has a single angle that cannot accurately evaluate the value corresponding to the state action, improper reward function design leads to weak path planning performance, and poor path planning stability; the general gimbal control path search method has improper parameter settings that lead to poor path search efficiency, improper candidate solution initialization during the parameter search process, resulting in low diversity of initial parameters, and the inability to accurately measure candidate solutions. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a pan-tilt control path search method based on big data. In view of the problems that the general pan-tilt control path search method has a single angle and cannot accurately evaluate the value corresponding to the state action, the improper design of the reward function leads to weak path planning performance, and the path planning has poor stability, this scheme initializes multiple critic networks and an actor network, evaluates the value corresponding to the state action from multiple angles, and generates an effective path planning strategy to enhance the robustness and decision-making ability of the system; designs a nonlinear piecewise function, adjusts the exploration rate based on the time step, introduces experience replay and designs loss calculation; to enhance the certainty and stability of path planning, And it can effectively converge and optimize during the learning process; the general gimbal control path search method has the problems of improper parameter setting resulting in poor path search efficiency, improper candidate solution initialization in the parameter search process resulting in low diversity of initial parameters, and inability to accurately measure candidate solutions. The candidate solution positions initialized by this scheme can cover a wider search space, ensure the diversity of initial parameters, consider the system's total loss value and the distance calculation fitness with the optimal solution, and guide the search process to be closer to the global optimal solution; design nonlinear convergence factors and dual update strategies to improve the search efficiency and the accuracy of the results, take into account both global search and local optimization, and ensure that the system can complete parameter adjustment within a reasonable time.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a pan / tilt control path search method based on big data, the method comprising the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Establishing a PTZ control path search system;

[0008] Step S4: PTZ control path search system parameter adjustment;

[0009] Step S5: PTZ control path search.

[0010] Furthermore, in step S1, the data collection is to collect historical path planning data; the historical path planning data is the actions taken by the gimbal control path search system in different states and the corresponding reward feedback, and the position and attitude data are obtained based on the electronic compass and the RNSS Beidou receiver.

[0011] Furthermore, in step S2, the data preprocessing is to perform data cleaning, data conversion and standardization on the collected data; the data cleaning is to remove noise and outliers; the data conversion is to convert the data into a vector form; and the standardization is to normalize the data to the range of [0,1].

[0012] Furthermore, in step S3, the establishment of the PTZ control path search system specifically includes the following steps:

[0013] Step S31: Initial parameters, initialize k reviewer networks Q i and the parameters of the actor network μ and ;; and initialize the critic network and actor network based on the network parameters; initialize the playback buffer R, the maximum flight time T, the parameters α, β, η, λ, ζ, τ and the number of tracking memories Tm; the critic network is used to estimate the value of performing a specific action in a given state; the actor network is responsible for generating the strategy of the path planning system;

[0014] Step S32: Design a reward function, and design the reward function as a nonlinear piecewise function to adapt to the behavior decision-making and obstacle avoidance requirements of the gimbal control path search system in a continuous environment; the formula used is as follows:

[0015] ;

[0016] In the formula, Rt is the reward value; S is the current state; R obstacle is the obstacle area, which is a negative reward used to punish the gimbal control path search system for entering the obstacle area; R goal is the target area, which is a positive reward used to reward the PTZ control path search system to reach the target position; R safe It is a safe area; It is a proportional factor used to adjust the impact of the distance from the current state to the target state on the reward value; is the distance from the current state to the target state;

[0017] Step S33: Dynamically adjust the exploration rate, the formula used is as follows:

[0018] ;

[0019] In the formula, is the exploration rate at time t; and are the lower and upper bounds of the exploration rate respectively; T is the maximum time step; is a parameter that controls the speed at which the exploration rate decreases;

[0020] Step S34: Select an action, receive the observed state s, and if the current time step is not greater than the tracking memory number, select an action based on the state noise. The formula used is as follows:

[0021] ;

[0022] Step S35: After the experience is returned and the action is executed to obtain the reward, the current state s t 、Action a t , the reward Rt obtained and the new state s t+1 Stored in the experience replay memory R, ​​randomly extract N mini-batch conversion samples (s i ,a i ,Ri,s i+1 ), used for subsequent network parameter adjustment;

[0023] Step S36: Calculate the loss and update the K critic networks by minimizing the loss function in the equation. The formula used is as follows:

[0024] ;

[0025] Where, L mc is the total loss function of the PTZ control path search system, which is used to update the critic network; α and β are weight coefficients used to control the loss impact; L avg is the average loss, which is used to measure the average error of all critic networks; L is the loss of a single critic network; is the coefficient of the impact of the loss term; The parameter is The i-th critic network is in state s t Next select action a t Q value; The parameter is All reviewers in the network at state s t Next select action a t The average Q value of; γ is the discount factor; It’s a strategy Entropy of

[0026] Step S37: Update the actor network based on the policy gradient. The formula used is as follows:

[0027] ;

[0028] In the formula, is the policy gradient of the actor network, which is used to maximize the expected return; E S~ ρ π [·] is the expected value of the state S generated according to the current strategy π; is the gradient of the critic network with respect to action a, where a is determined by the actor network; is the actor network policy function About parameters The gradient of is the policy function of the actor network, which outputs the probability distribution of action a given state s;

[0029] Step S38: Update network parameters, the formula used is as follows:

[0030] ;

[0031] In the formula, and are the parameters of the updated critic network and actor network, respectively; and is the network parameter before updating; τ is the update rate; is the policy gradient of the critic network, which is calculated in the same way as the policy gradient of the actor network; is the learning rate;

[0032] Step S39: The system determines that a loss threshold is set in advance. If the total loss of the gimbal control path search system is lower than the loss threshold, the gimbal control path search system is established. If the maximum number of iterations is reached, the initial parameters are adjusted and the process returns to step S1. Otherwise, the process returns to step S33 to continue the iteration.

[0033] Further, in step S4, the PTZ control path search system parameter adjustment specifically includes the following steps:

[0034] Step S41: Initialization: Establish a search space based on the initial parameters of step S3 and initialize the position of the candidate solution. The formula used is as follows:

[0035] ;

[0036] Where Y i1+1,j and Y i1,j are the positions of the jth dimension of the i1+1th candidate solution and the i1th candidate solution respectively;

[0037] Step S42: Calculate the fitness value, combining the fitness of the candidate solution and the distance to the optimal solution

[0038] ;

[0039] In the formula, It is a candidate solution The fitness value of It is the total system loss value after k iterations of the gimbal control path search system established based on the candidate solution; It is a candidate solution The distance to the optimal solution; w is the fitness weight;

[0040] Step S43: Design a nonlinear convergence factor, the formula used is as follows:

[0041] ;

[0042] Where a1 is the nonlinear convergence factor, t1 and T1 are the current search times and the maximum search times respectively;

[0043] Step S44: Update the candidate solution position, the formula used is as follows:

[0044] ;

[0045] Where Y i1,j (t1+1) and Y i1,j (t1) is the position of the j-th dimension of the i1-th candidate solution in the t1+1-th search and the t1-th search respectively; Y rand,j (t1) and Y best,j (t1) are the positions of the random individual and the optimal individual respectively; A is the coefficient variable; c is a random number ranging from 0 to 1;

[0046] Step S45: Search judgment. A fitness threshold is set in advance. If there is a candidate solution whose fitness value is higher than the fitness threshold, the position of the candidate solution is the initial parameter of the gimbal control path search system. At this time, the gimbal control path search system is established. If the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

[0047] Furthermore, in step S5, the gimbal control path search is based on the established gimbal control path search system, which collects and inputs position and posture data in real time, and the gimbal control path search system outputs a path planning strategy to achieve the gimbal control path search.

[0048] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0049] (1) In view of the problems that the general gimbal control path search method has a single angle and cannot accurately evaluate the value corresponding to the state action, the improper design of the reward function leads to weak path planning performance, and the path planning has poor stability, this scheme initializes multiple critic networks and an actor network to evaluate the value corresponding to the state action from multiple angles and generate an effective path planning strategy to enhance the robustness and decision-making ability of the system; designs a nonlinear piecewise function, adjusts the exploration rate based on the time step, introduces experience replay and designs loss calculation; in order to enhance the certainty and stability of path planning, and enable effective convergence and optimization during the learning process.

[0050] (2) In view of the problems that the general gimbal control path search method has improper parameter settings resulting in poor path search efficiency, improper candidate solution initialization during the parameter search process resulting in low initial parameter diversity, and the inability to accurately measure the candidate solutions, the candidate solution positions initialized by this scheme can cover a wider search space, ensure the diversity of initial parameters, consider the system's total loss value and the distance calculation fitness from the optimal solution, and guide the search process closer to the global optimal solution; design a nonlinear convergence factor and a dual update strategy to improve the search efficiency and the accuracy of the results, take into account both global search and local optimization, and ensure that the system can complete parameter adjustment within a reasonable time. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of the process of the pan / tilt control path search method based on big data provided by the present invention;

[0052] Figure 2 is a schematic flow chart of step S3;

[0053] Figure 3 It is a schematic diagram of the process of step S4.

[0054] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0056] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0057] Example 1, see Figure 1 The present invention provides a pan / tilt control path search method based on big data, the method comprising the following steps:

[0058] Step S1: data collection, collecting historical path planning data;

[0059] Step S2: Data preprocessing, data cleaning, data conversion and standardization of the collected data;

[0060] Step S3: Establish a pan-tilt control path search system, initialize multiple critic networks and an actor network, design a nonlinear piecewise function, adjust the exploration rate based on the time step, introduce experience replay and design loss calculation, and then realize the establishment of a pan-tilt control path search system;

[0061] Step S4: PTZ control path search system parameter adjustment, calculate the fitness based on the total system loss value and the distance to the optimal solution, design a nonlinear convergence factor and a dual update strategy, and then realize the PTZ control path search system parameter adjustment;

[0062] Step S5: PTZ control path search.

[0063] Example 2, see Figure 1 ,This embodiment is based on the above embodiment. In step S1, the historical path planning data is the ,actions taken by the gimbal control path search system in different states and ,corresponding reward feedback, and the position and attitude data are ,obtained based on the electronic compass and the RNSS Beidou receiver.

[0064] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, data cleaning is to remove noise and outliers; data conversion is to convert data into vector form; and standardization is to normalize the data to the range of [0,1].

[0065] Example 4, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S3, establishing a pan / tilt control path search system specifically includes the following steps:

[0066] Step S31: Initial parameters, initialize k reviewer networks Qi and the parameters of the actor network μ and ;; and initialize the critic network and actor network based on the network parameters; initialize the playback buffer R, the maximum flight time T, the parameters α, β, η, λ, ζ, τ and the number of tracking memories Tm; the critic network is used to estimate the value of performing a specific action in a given state; the actor network is responsible for generating the strategy of the path planning system;

[0067] Step S32: Design a reward function, and design the reward function as a nonlinear piecewise function to adapt to the behavior decision-making and obstacle avoidance requirements of the gimbal control path search system in a continuous environment; the formula used is as follows:

[0068] ;

[0069] In the formula, Rt is the reward value; S is the current state; R obstacle is the obstacle area, which is a negative reward used to punish the gimbal control path search system for entering the obstacle area; R goal is the target area, which is a positive reward used to reward the PTZ control path search system to reach the target position; R safe It is a safe area; It is a proportional factor used to adjust the impact of the distance from the current state to the target state on the reward value; is the distance from the current state to the target state;

[0070] Step S33: Dynamically adjust the exploration rate, the formula used is as follows:

[0071] ;

[0072] In the formula, is the exploration rate at time t; and are the lower and upper bounds of the exploration rate respectively; T is the maximum time step; is a parameter that controls the speed at which the exploration rate decreases;

[0073] Step S34: Select an action, receive the observed state s, and if the current time step is not greater than the tracking memory number, select an action based on the state noise. The formula used is as follows:

[0074] ;

[0075] Step S35: After the experience is returned and the action is executed to obtain the reward, the current state s t 、Action a t , the reward Rt obtained and the new state s t+1 Stored in the experience replay memory R, ​​randomly extract N mini-batch conversion samples (s i,a i ,Ri,s i+1 ), used for subsequent network parameter adjustment;

[0076] Step S36: Calculate the loss and update the K critic networks by minimizing the loss function in the equation. The formula used is as follows:

[0077] ;

[0078] Where, L mc is the total loss function of the PTZ control path search system, which is used to update the critic network; α and β are weight coefficients used to control the loss impact; L avg is the average loss, which is used to measure the average error of all critic networks; L is the loss of a single critic network; is the coefficient of the impact of the loss term; The parameter is The i-th critic network is in state s t Next select action a t Q value; The parameter is The network of all reviewers in state s t Next select action a t The average Q value of; γ is the discount factor; It’s a strategy Entropy of

[0079] Step S37: Update the actor network based on the policy gradient. The formula used is as follows:

[0080] ;

[0081] In the formula, is the policy gradient of the actor network, which is used to maximize the expected return; E S~ ρ π [·] is the expected value of the state S generated according to the current strategy π; is the gradient of the critic network with respect to action a, where a is determined by the actor network; is the actor network policy function About parameters The gradient of is the policy function of the actor network, which outputs the probability distribution of action a given state s;

[0082] Step S38: Update network parameters, the formula used is as follows:

[0083] ;

[0084] In the formula, and are the parameters of the updated critic network and actor network, respectively; and is the network parameter before updating; τ is the update rate; is the policy gradient of the critic network, which is calculated in the same way as the policy gradient of the actor network; is the learning rate;

[0085] Step S39: The system determines that a loss threshold is set in advance. If the total loss of the gimbal control path search system is lower than the loss threshold, the gimbal control path search system is established. If the maximum number of iterations is reached, the initial parameters are adjusted and the process returns to step S1. Otherwise, the process returns to step S33 to continue the iteration.

[0086] By performing the above operations, in view of the problems that the general gimbal control path search method has a single angle and cannot accurately evaluate the value corresponding to the state action, the improper design of the reward function leads to weak path planning performance, and the path planning has poor stability, this scheme initializes multiple critic networks and an actor network, evaluates the value corresponding to the state action from multiple angles, and generates an effective path planning strategy to enhance the robustness and decision-making ability of the system; designs a nonlinear piecewise function, adjusts the exploration rate based on the time step, introduces experience replay and designs loss calculation; to enhance the certainty and stability of path planning, and can effectively converge and optimize during the learning process.

[0087] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S4, the PTZ control path search system parameter adjustment specifically includes the following steps:

[0088] Step S41: Initialization: Establish a search space based on the initial parameters of step S3 and initialize the position of the candidate solution. The formula used is as follows:

[0089] ;

[0090] Where Y i1+1,j and Y i1,j are the positions of the jth dimension of the i1+1th candidate solution and the i1th candidate solution respectively;

[0091] Step S42: Calculate the fitness value, combining the fitness of the candidate solution and the distance to the optimal solution

[0092] ;

[0093] In the formula, It is a candidate solution The fitness value of It is the total system loss value after k iterations of the gimbal control path search system established based on the candidate solution; It is a candidate solution The distance to the optimal solution; w is the fitness weight;

[0094] Step S43: Design a nonlinear convergence factor, the formula used is as follows:

[0095] ;

[0096] Where a1 is the nonlinear convergence factor, t1 and T1 are the current search times and the maximum search times respectively;

[0097] Step S44: Update the candidate solution position, the formula used is as follows:

[0098] ;

[0099] Where Y i1,j (t1+1) and Y i1,j (t1) is the position of the j-th dimension of the i1-th candidate solution in the t1+1-th search and the t1-th search respectively; Y rand,j (t1) and Y best,j (t1) are the positions of the random individual and the optimal individual respectively; A is the coefficient variable; c is a random number ranging from 0 to 1;

[0100] Step S45: Search judgment. A fitness threshold is set in advance. If there is a candidate solution whose fitness value is higher than the fitness threshold, the position of the candidate solution is the initial parameter of the gimbal control path search system. At this time, the gimbal control path search system is established. If the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

[0101] By performing the above operations, the candidate solution positions initialized by this scheme can cover a wider search space and ensure the diversity of initial parameters, considering the total loss value of the system and the distance calculation fitness from the optimal solution, guiding the search process closer to the global optimal solution; designing nonlinear convergence factors and dual update strategies to improve the search efficiency and the accuracy of the results, taking into account both global search and local optimization, and ensuring that the system can complete parameter adjustment within a reasonable time.

[0102] Example 6, see Figure 1This embodiment is based on the above embodiment. In step S5, the gimbal control path search is based on the established gimbal control path search system, which collects and inputs position and posture data in real time. The gimbal control path search system outputs a path planning strategy to realize the gimbal control path search.

[0103] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0104] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0105] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A pan / tilt control path search method based on big data, characterized in that: The method comprises the following steps: Step S1: data collection; Step S2: data preprocessing; Step S3: Establish a pan-tilt control path search system, initialize multiple critic networks and an actor network, design a nonlinear piecewise function, adjust the exploration rate based on the time step, introduce experience replay and design loss calculation, and then realize the establishment of a pan-tilt control path search system; Step S4: PTZ control path search system parameter adjustment, calculate the fitness based on the total system loss value and the distance to the optimal solution, design a nonlinear convergence factor and a dual update strategy, and then realize the PTZ control path search system parameter adjustment; Step S5: PTZ control path search; In step S3, the establishment of the PTZ control path search system specifically includes the following steps: Step S31: Initial parameters, initialize k reviewer networks Q i and the parameters of the actor network μ and ; Initialize the critic network and actor network based on the network parameters; Initialize the playback buffer R, the maximum flight time T, the parameters α, β, η, λ, ζ, and tracking memory number Tm; the critic network is used to estimate the value of performing a specific action in a given state; the actor network is responsible for generating the strategy of the path planning system; Step S32: Design a reward function, and design the reward function as a nonlinear piecewise function to adapt to the behavior decision-making and obstacle avoidance requirements of the gimbal control path search system in a continuous environment; the formula used is as follows: ； In the formula, Rt is the reward value; S is the current state; R obstacle is the obstacle area, which is a negative reward used to punish the gimbal control path search system for entering the obstacle area; R goal is the target area, which is a positive reward used to reward the PTZ control path search system to reach the target position; R safe It is a safe area; It is a proportional factor used to adjust the impact of the distance from the current state to the target state on the reward value; is the distance from the current state to the target state; Step S33: Dynamically adjust the exploration rate, the formula used is as follows: ； In the formula, is the exploration rate at time t; and are the lower and upper bounds of the exploration rate, respectively; is a parameter that controls the speed at which the exploration rate decreases; Step S34: Select an action, receive the observed state s, and if the current time step is not greater than the tracking memory number, select an action based on the state noise. The formula used is as follows: ； Step S35: After the experience is returned and the action is executed to obtain the reward, the current state s t 、Action a t , the reward Rt obtained and the new state s t+1 Stored in the experience replay memory R, ​​N mini-batch conversion samples (s i ,a i ,Ri,s i+1 ), used for subsequent network parameter adjustment; Step S36: Calculate the loss and update the K critic networks by minimizing the loss function in the equation. The formula used is as follows: ； Where, L mc is the total loss function of the PTZ control path search system, which is used to update the critic network; α and β are weight coefficients used to control the loss impact; L avg is the average loss, which is used to measure the average error of all critic networks; L is the loss of a single critic network; is the coefficient of the impact of the loss term; The parameter is The i-th critic network is in state s t Next select action a t Q value; The parameter is All reviewers in the network at state s t Next select action a t The average Q value of; γ is the discount factor; It’s a strategy Entropy of Step S37: Update the actor network based on the policy gradient. The formula used is as follows: ； In the formula, is the policy gradient of the actor network, which is used to maximize the expected return; is the expected value of the state S generated according to the current strategy π; is the gradient of the critic network with respect to action a, where a is determined by the actor network; is the actor network policy function About parameters The gradient of is the policy function of the actor network, which outputs the probability distribution of action a given state s; Step S38: Update network parameters, the formula used is as follows: ； In the formula, and are the parameters of the updated critic network and actor network, respectively; and are the network parameters before updating; is the update rate; is the policy gradient of the critic network, which is calculated in the same way as the policy gradient of the actor network; is the learning rate; Step S39: the system determines that a loss threshold is set in advance. If the total loss of the PTZ control path search system is lower than the loss threshold, the PTZ control path search system is established. If the maximum number of iterations is reached, the initial parameters are adjusted and the process returns to step S1. Otherwise, the process returns to step S33 and continues the iteration. In step S4, the PTZ control path search system parameter adjustment specifically includes the following steps: Step S41: Initialization: Establish a search space based on the initial parameters of step S3 and initialize the position of the candidate solution. The formula used is as follows: ； Where Y i1+1,j and Y i1,j are the positions of the jth dimension of the i1+1th candidate solution and the i1th candidate solution respectively; Step S42: Calculate the fitness value, the formula used is as follows: ； In the formula, It is a candidate solution The fitness value of It is the total system loss value after k iterations of the gimbal control path search system established based on the candidate solution; It is a candidate solution The distance to the optimal solution; w is the fitness weight; Step S43: Design a nonlinear convergence factor, the formula used is as follows: ; Where a1 is the nonlinear convergence factor, t1 and T1 are the current search times and the maximum search times respectively; Step S44: Update the candidate solution position, the formula used is as follows: ； Where Y i1,j (t1+1) and Y i1,j (t1) is the position of the j-th dimension of the i1-th candidate solution in the t1+1-th search and the t1-th search respectively; Y rand,j (t1) and Y best,j (t1) are the positions of the random individual and the optimal individual respectively; A is the coefficient variable; c is a random number ranging from 0 to 1; Step S45: Search judgment. A fitness threshold is set in advance. If there is a candidate solution whose fitness value is higher than the fitness threshold, the position of the candidate solution is the initial parameter of the gimbal control path search system. At this time, the gimbal control path search system is established. If the maximum number of iterations is reached, return to step S41; otherwise, return to step S42.

2. The PTZ control path search method based on big data according to claim 1, characterized in that: In step S1, the data collection is to collect historical path planning data; the historical path planning data is the actions taken by the gimbal control path search system in different states and the corresponding reward feedback, and the position and posture data are obtained based on the electronic compass and the RNSS Beidou receiver.

3. The PTZ control path search method based on big data according to claim 1, characterized in that: In step S5, the gimbal control path search is based on the established gimbal control path search system, which collects and inputs position and posture data in real time, and the gimbal control path search system outputs a path planning strategy to achieve the gimbal control path search.

4. The PTZ control path search method based on big data according to claim 1, characterized in that: In step S2, the data preprocessing is to clean, convert and standardize the collected data; the data cleaning is to remove noise and outliers; the data conversion is to convert the data into vector form; and the standardization is to normalize the data to the range of [0,1].

Images