Unmanned aerial vehicle flight control method and system based on unmanned aerial vehicle search and rescue platform

Abstract

The present application relates to the field of unmanned aerial vehicle flight control, more particularly, the present application relates to unmanned aerial vehicle flight control method and system based on unmanned aerial vehicle search and rescue platform, the method comprises: based on the preset time window calculates the current attitude stability and wind resistance consumption of unmanned aerial vehicle, and then the excess proportion is obtained and the comprehensive wind resistance index is calculated, according to which the weak machine is judged; Then calculate the next moment of weak machine traction (current traction minus release amount) and strong machine next moment of traction (original traction plus weak machine release amount total sum of increments), complete the reassignment of unmanned aerial vehicle cluster traction. The present application can intelligently evaluate the wind resistance of unmanned aerial vehicle and dynamically adjust the traction distribution, combined with accurate prediction and environmental adaptation, significantly improve the task execution stability and reliability of unmanned aerial vehicle cluster in wind disturbance environment, ensure that the task allocation is more scientific and reasonable.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of unmanned aerial vehicle flight control. More particularly, the present application relates to an unmanned aerial vehicle flight control method and system based on an unmanned aerial vehicle search and rescue platform. BACKGROUND

[0002] Unmanned aerial vehicles play an important role in danger reconnaissance, personnel search and material delivery, can quickly locate, assess disaster conditions, accurately transport materials, and provide support when communication is limited, effectively improving rescue efficiency.

[0003] Prior art such as patent application document CN118915795A discloses a multi-unmanned aerial vehicle cooperative control method and system. The cooperative control method includes: first, defining the observation and action space of the unmanned aerial vehicle, constructing a multi-unmanned aerial vehicle cooperative network framework; then, providing an executor and an evaluator network for each unmanned aerial vehicle, defining a target function; then, constructing a policy network model with multi-channel matrix and adjacent unmanned aerial vehicle state as input, and action probability distribution as output; finally, training the policy network model, initializing the parameters, generating flight actions, interacting with the environment to obtain rewards and punishments and the next state, storing them in the experience pool, randomly sampling samples for training, obtaining the time difference error and policy gradient to update the parameters, optimizing the target function, until the training is completed, thereby realizing the cooperative control of multiple unmanned aerial vehicles.

[0004] However, the above-mentioned cooperative control method cannot reasonably allocate tasks according to the actual wind resistance capability of the unmanned aerial vehicle under complex weather conditions, especially in a wind disturbance environment, affecting the stability and reliability of the overall task execution. SUMMARY

[0005] To solve the technical problem that the above-mentioned prior art cannot reasonably allocate tasks according to the actual wind resistance capability of the unmanned aerial vehicle under a wind disturbance environment, affecting the stability and reliability of the overall task execution, the present application provides solutions in the following aspects.

[0006] In a first aspect, an unmanned aerial vehicle flight control method based on an unmanned aerial vehicle search and rescue platform, comprising:

[0007] calculating the attitude stability and wind resistance consumption of each unmanned aerial vehicle at the current time based on the preset time window at the current time;

[0008] calculating the excess proportion of the current attitude stability of each unmanned aerial vehicle relative to the historical windless environment reference attitude stability as a first proportion, calculating the excess proportion of the current wind resistance consumption of each unmanned aerial vehicle relative to the historical windless environment reference wind resistance consumption as a second proportion, and taking the product of the first proportion and the second proportion as the comprehensive wind resistance index of the unmanned aerial vehicle at the current time;

[0009] Weak wind turbines are identified and marked based on comprehensive wind resistance indicators; the traction force of each weak wind turbine at the current moment is calculated for the next moment, where the traction force at the next moment is the difference between the traction force of the weak wind turbine at the current moment and the calculated traction force release amount of the weak wind turbine at the current moment; the traction force of each strong wind turbine at the current moment is calculated by adding the increment obtained by the average of the total traction force release amount of all weak wind turbines at the current moment to the traction force of each strong wind turbine at the next moment.

[0010] Based on the calculated traction forces of each weak and strong drone in the next moment, the traction forces of the drone swarm are redistributed.

[0011] Preferably, the process of setting the time window includes:

[0012] Calculate the ratio of the wind speed at the drone's current location to the maximum tolerable wind speed, and use it as the first factor;

[0013] The upper and lower limits of the preset time window are used to calculate the difference between the upper and lower limits of the time window as the first difference; the length of the time window is obtained by calculating the difference between the upper limit of the time window and the product of the first factor and the first difference.

[0014] Using the current time as the endpoint, set the corresponding time window for the drone based on the calculated length of the time window.

[0015] Preferably, the process of obtaining the attitude stability includes:

[0016] For a single UAV, acquire the attitude status at all times within the UAV's time window, including pitch angle, roll angle, and yaw angle;

[0017] Calculate the standard deviation of each attitude state separately, and take the square root of the sum of the squares of the standard deviations of all attitude states as the attitude stability of the UAV at the current moment.

[0018] Preferably, the process of obtaining the wind resistance consumption includes:

[0019] For a single drone, calculate the instantaneous power of the drone at each moment within the current time window; integrate the instantaneous power within the drone's time window, and then divide it by the length of the time window to obtain the average power. Use this average power as the wind resistance consumption of the drone at the current moment.

[0020] Preferably, the process for determining the weak machine includes:

[0021] Based on the pre-trained prediction model, the traction force prediction value assigned to each UAV at the current moment is obtained. A traction force threshold is set according to the traction force prediction values ​​of all UAVs, and UAVs with traction force prediction values ​​less than the traction force threshold are judged as weak UAVs.

[0022] Preferably, the prediction model selects a gradient boosting decision tree, the model training input includes the real-time comprehensive wind resistance index of the UAV, the carried weight and the wind speed, and the output is the predicted value of the traction force allocated to the UAV at the current time, and the mean square error is used as the loss function in the model training process.

[0023] Preferably, the setting process of the traction force threshold value comprises:

[0024] calculating the mean and standard deviation of the predicted values of the traction force of all UAVs at the current time;

[0025] taking the product of the ratio of the wind speed at the current time to the maximum bearable wind speed and the standard deviation of the predicted values of the traction force of all UAVs at the current time as a correction term;

[0026] taking the difference between the mean of the predicted values of the traction force of all UAVs at the current time and the correction term as the traction force threshold value.

[0027] Preferably, the calculation process of the traction force release amount of the weak UAV at the current time comprises:

[0028] taking the product of the ratio of the wind speed at the current time to the maximum bearable wind speed and the actual traction force borne by the weak UAV at the current time as the traction force release amount of the weak UAV at the current time.

[0029] Preferably, after the calculated traction force of each weak UAV and strong UAV at the next time is obtained, the method comprises:

[0030] calculating the priority of the secondary distribution of the traction force of the strong UAV;

[0031] updating the traction force of the strong UAV at the next time based on the calculated priority of each strong UAV to obtain the updated traction force of the strong UAV at the next time.

[0032] In a second aspect, a UAV flight control system based on a UAV search and rescue platform comprises a processor and a memory, and the memory stores computer program instructions which, when executed by the processor, implement any of the UAV flight control methods based on the UAV search and rescue platform.

[0033] The present application has the following advantages:

[0034] By scientifically setting a time window, the anti-wind ability of the unmanned aerial vehicle is accurately evaluated from two dimensions of attitude stability and wind resistance consumption, and a comprehensive anti-wind index is generated to provide a key basis for task allocation; the gradient boosting decision tree model is used to accurately predict the real-time traction force demand of the unmanned aerial vehicle in combination with the comprehensive anti-wind index and other factors, and the traction force threshold is reasonably set to accurately determine the weak machine; the cluster traction force is dynamically redistributed based on the weak machine traction force release amount and other data, and the secondary allocation priority of the strong machine is further optimized, which effectively guarantees the stable execution of the task in the complex wind disturbance environment, greatly improves the anti-wind ability and overall reliability of the unmanned aerial vehicle cluster, and finally realizes the efficient and reliable operation of the unmanned aerial vehicle cluster in the complex environment. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a method flowchart of steps S1-S3 in the unmanned aerial vehicle flight control method based on the unmanned aerial vehicle search and rescue platform according to the embodiment of the application.

[0036] Figure 2 is a structural block diagram of the unmanned aerial vehicle flight control system based on the unmanned aerial vehicle search and rescue platform according to the embodiment of the application. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments of the application.

[0038] The application scenario of the application is: multiple unmanned aerial vehicles cooperatively performing a search and rescue transportation task (transporting emergency medical supplies, rescue equipment, etc.).

[0039] Referring to Figure 1 The unmanned aerial vehicle flight control method based on the unmanned aerial vehicle search and rescue platform includes steps S1-S3, and specifically as follows:

[0040] S1: Calculate the attitude stability and wind resistance consumption of each unmanned aerial vehicle at the current time based on the preset time window at the current time.

[0041] In the search and rescue transportation scenario, the environmental wind speed is a key factor affecting the attitude stability and anti-wind ability of the unmanned aerial vehicle. By setting a dynamically adjusted time window, the anti-wind performance of the unmanned aerial vehicle under different wind conditions can be more accurately evaluated according to the real-time wind speed changes. Specifically:

[0042] When the wind speed is small, the attitude angle fluctuation is mainly caused by sensor noise or slight disturbance of the individual, and a longer window can smooth these fluctuations to avoid misjudgment of the UAV as weak wind resistance due to too short window, and ensure that the evaluation results reflect the real wind resistance of the UAV; when the wind speed is large, the influence of the wind on the flight stability of the UAV is rapidly amplified, and shortening the window can identify the UAV that cannot maintain stability in time, improve the timeliness of abnormal situation discovery, and ensure the safety of search and rescue transportation tasks.

[0043] In one embodiment, the maximum tolerable wind speed of the UAV for search and rescue under different flight conditions is determined by looking up the product manual or technical specification.

[0044] Further, in order to reflect the influence of the real-time wind speed, the ratio of the wind speed at the current position of each UAV to the maximum tolerable wind speed is calculated, and the ratio is taken as the first factor.

[0045] Further, the upper and lower limits of the time window are set in combination with the windless condition and the high wind speed condition. For example, the upper limit of the time window is set to 10 seconds according to the windless condition, and the lower limit of the time window is set to 1 second according to the high wind speed condition. The difference between the upper limit and the lower limit of the time window is calculated as the first difference. The difference between the upper limit of the time window and the product of the first factor and the first difference calculated above is calculated to obtain the length of the time window.

[0046] The time window corresponding to the UAV is set according to the length of the time window calculated with the current time as the end point.

[0047] In another embodiment, considering that the wind speed jumps over time, a smoothing factor is introduced to prevent the window from changing dramatically due to temporary fluctuations in wind speed.

[0048] For example, the length of the time window after adding the smoothing factor satisfies the relationship:

[0049]

[0050] In the formula, is the length of the time window at the current time after adding the smoothing factor, is the smoothing factor, is the length of the time window at the current time without adding the smoothing factor, is the length of the time window at the current time .

[0051] Since a stable attitude is the key to ensuring that the UAV flies along the predetermined route, secondly, in order to maintain a stable flight attitude in strong wind, the UAV needs to constantly adjust the attitude, such as changing the inclination angle of the wings, adjusting the deflection of the tail, etc.

[0052] These attitude adjustment operations all require energy. By calculating energy consumption, the attitude adjustment efficiency and energy consumption level of the flight control system under different wind speeds can be evaluated. If the flight control system can achieve effective attitude adjustment with low energy consumption, it indicates that the system has strong wind resistance; conversely, if the energy consumption during attitude adjustment is too high and the attitude stability is still poor, it indicates that the flight control strategy is insufficient in terms of wind resistance.

[0053] Therefore, based on the time window of the UAV at the current moment calculated above, the attitude stability and wind resistance consumption for maintaining attitude of each UAV at the current moment are calculated.

[0054] In one embodiment, for a single UAV, the attitude state at all times within the UAV's time window is acquired, including pitch angle, roll angle, and yaw angle. Then, conventional preprocessing operations such as filtering and noise reduction and outlier removal are performed on the acquired attitude state data.

[0055] Furthermore, the standard deviation of each attitude state is calculated separately, and the square root of the sum of the squares of the standard deviations of all attitude states is taken as the attitude stability of the UAV at the current moment.

[0056] By calculating the aforementioned attitude stability, the degree of attitude fluctuation of the drone under current wind conditions can be reflected. Drones with good wind resistance exhibit small fluctuations in various attitude state data, resulting in low attitude stability values; conversely, drones with poor wind resistance show large fluctuations in certain attitude state data, leading to relatively high attitude stability values.

[0057] In one embodiment, for a single drone, the instantaneous power of the drone at each moment within the current time window is calculated; the instantaneous power is integrated within the drone's time window and then divided by the length of the time window to obtain the average power, which is then used as the wind resistance consumption of the drone at the current moment.

[0058] The above wind resistance consumption satisfies the following relationship:

[0059]

[0060] In the formula, For the current moment Wind resistance consumption for drones to maintain attitude. The current time step after adding a smoothing factor The length of the time window, Within the time window The instantaneous power at a given moment (calculated by obtaining the voltage and current at the drone's battery terminal at that moment).

[0061] By calculating the above wind resistance consumption, the energy cost required for the UAV to maintain the current attitude and load can be reflected, and the higher the energy consumption, the weaker the wind resistance ability.

[0062] According to the above operation, the attitude stability and wind resistance consumption of all UAVs at the current time can be finally calculated.

[0063] S2: Calculate the excess proportion of the current attitude stability of each UAV relative to the historical wind-free environment reference attitude stability as the first proportion; calculate the excess proportion of the current wind resistance consumption of each UAV relative to the historical wind-free environment reference wind resistance consumption as the second proportion; and take the product of the first proportion and the second proportion as the comprehensive wind resistance index of the UAV at the current time.

[0064] The essence of wind resistance performance is the dynamic balance ability of the UAV in a complex wind disturbance environment, which can not only control the attitude fluctuation accurately with less attitude fluctuation, but also avoid excessive energy consumption when resisting wind.

[0065] In one embodiment, the average or standard value of the attitude stability and the wind resistance consumption of the UAV in the historical flight data in the wind-free environment is first calculated as an environmental reference for measuring wind resistance ability.

[0066] Then, by comparing the current attitude stability and wind resistance consumption with the reference value, the part exceeding the reference is quantified as the excess proportion, and the wind resistance ability is comprehensively evaluated in the form of product. That is, the relationship is satisfied:

[0067]

[0068] In the formula, is the current time is the comprehensive wind resistance index of the UAV, is the current time is the attitude stability of the UAV, is the current time is the wind resistance consumption of the UAV to maintain the attitude, is the reference standard value of the attitude stability, is the reference standard value of the wind resistance consumption to maintain the attitude, is a very small constant to prevent any calculation result from being 0; is the first proportion, is the second proportion.

[0069] By calculating the above comprehensive wind resistance index, when ≥ and ≥ , The larger, the more stable the UAV is in a strong wind environment (high consumption) and the better the wind resistance is. If any index is not up to standard (such as unstable attitude or too low consumption), will be significantly reduced due to the influence of , indicating that the wind resistance is insufficient.

[0070] According to the above operation, the comprehensive wind resistance index of all UAVs at the current time can be finally calculated.

[0071] S3: Determine and mark weak machines based on the comprehensive wind resistance index; calculate the traction force of each weak machine at the next moment, which is the difference between the traction force of the weak machine at the current moment and the calculated traction force release amount of the weak machine at the current moment; based on the traction force of each strong machine at the current moment, add the increment obtained by equally distributing the total sum of the traction force release amount of all weak machines at the current moment, to obtain the traction force of each strong machine at the next moment; according to the calculated traction force of each weak machine and strong machine at the next moment, complete the redistribution of the traction force of the UAV cluster.

[0072] In the cooperative transportation task of UAVs, the wind resistance of each UAV changes in real time, and only knowing the wind resistance cannot directly determine the traction force it can bear. Through model prediction, the traction force that each UAV can safely bear under the current conditions can be accurately predicted according to factors such as specific wind conditions (real-time wind speed), load (load weight), and UAV's own wind resistance performance (real-time wind resistance performance), so as to realize the reasonable distribution of the traction force.

[0073] In one embodiment, first, all historical flight data of UAVs are collected, which include comprehensive wind resistance index of UAVs, real-time wind speed, load weight, and actual traction force output, etc. A regression model is trained using these data, the purpose of which is to predict the traction force that each UAV can safely bear under the current wind and load conditions.

[0074] Further, gradient boosting decision tree is selected as the model, which can handle nonlinear relationships in data and has strong robustness to noise data, and is suitable for complex and variable actual situations.

[0075] Further, the historical flight data of the UAVs are used as input to train the model. In the training process, mean square error is used as the loss function to measure the accuracy of the model prediction, and at the same time, the hyperparameters of the model are optimized through cross-validation method. Finally, the output is the predicted value of the traction force allocated to the UAV at the current moment.

[0076] A dynamic threshold is set according to the predicted traction of each UAV, so as to find out the UAVs that cannot undertake the transportation task under the current wind condition. The predicted tractions of the UAVs with good wind resistance performance are similar, and the predicted tractions of the UAVs with poor wind resistance performance are smaller. The dynamic threshold reflects the overall performance of the UAV cluster, and avoids unreasonable threshold caused by individual abnormalities.

[0077] In a UAV cluster, the performances of different UAVs may be different, and the actual performances may be different even if the predicted tractions are the same. By setting a threshold, UAVs that can stably and reliably perform tasks under the current environment can be screened out, and the transportation task is allocated to these “strong machines”, thereby improving the reliability and task completion rate of the entire system.

[0078] In one embodiment, the mean and standard deviation of the predicted traction values of all UAVs at the current time are calculated, the product of the ratio of the current wind speed to the maximum bearable wind speed and the standard deviation of the predicted traction values of all UAVs at the current time is taken as a correction term, and the difference between the mean of the predicted traction values of all UAVs at the current time and the correction term is taken as the traction threshold.

[0079] The mean reflects the average level of the traction of the UAV group, and the correction term adjusts it according to the wind speed and the traction fluctuation. This dynamic adjustment makes the traction threshold more accurately reflect the required traction range of the UAV system under the current environmental conditions, thereby improving the flexibility and response ability of the system.

[0080] Further, the UAVs with predicted traction values less than the traction threshold are determined as weak machines, and the UAVs with predicted traction values greater than or equal to the traction threshold are determined as strong machines. In addition, the weak machines and the strong machines can be marked for easy distinction, and the number of weak machines and strong machines can be counted.

[0081] When there are weak machines in the UAV cluster, in order to avoid sudden changes in the system force, it is necessary to realize smooth traction transfer from the weak machines to the strong machines in a short time, so as to ensure that the system can remain stable when facing sudden conditions such as performance degradation of a UAV.

[0082] In one embodiment, a gradual traction return mechanism is adopted, that is, the weak machines do not transfer all the tractions at once, but gradually return the tractions. At the same time, the return rate is adjusted according to the real-time wind speed. The greater the wind speed, the stronger the environmental disturbance, and the faster the traction transfer needs to be completed. The tractions transferred out are evenly distributed to the strong machine group, so as to ensure that the additional traction increments borne by each strong machine are basically consistent, thereby maintaining the overall stability of the system. Specifically:

[0083] Firstly, the weak machine should release the traction force at the current time, that is, the ratio of the wind speed of the weak machine at the current time to the maximum wind speed and the product of the actual traction force of the weak machine at the current time is taken as the traction force release amount of the weak machine at the current time.

[0084] Then, the traction force of the weak machine at the next time is calculated, that is, the actual traction force of the weak machine at the current time is subtracted from the traction force release amount of the weak machine at the current time calculated above.

[0085] Secondly, the sum of the traction force release amounts of all weak machines at the current time is calculated, and then the sum of the traction force release amounts of all weak machines at the current time is divided by the number of strong machines to obtain the additional traction force that each strong machine should bear at the current time (that is, the increment obtained by averaging the sum of the traction force release amounts of all weak machines at the current time).

[0086] Finally, the traction force of each strong machine at the next time is calculated, that is, the increment obtained by averaging the sum of the traction force release amounts of all weak machines at the current time is added to the traction force of each strong machine at the current time to obtain the traction force of each strong machine at the next time.

[0087] In another embodiment, after the weak machine traction force completes the preliminary transfer, the long-term optimization of the load distribution within the strong machine group is focused on. That is, the energy advantage of the strong machine is fully utilized to avoid individual strong machines from being highly fatigued or running out of power due to long-term high load, thereby improving the overall endurance and long-term stability of the system. Specifically:

[0088] Firstly, the traction forces of all strong machines at each time within the time window corresponding to the current time and the total traction force at each time are calculated, the ratio of the traction force of any strong machine at each time to the total traction force of all strong machines at the time is calculated, and the average historical load proportion of any strong machine based on the time window is obtained by accumulating all times. The difference between 1 and the average historical load proportion is taken as the historical load coefficient.

[0089] Then, the current power of the strong machine is obtained and the dimension is eliminated using the conventional processing, and the product of the historical load coefficient of the strong machine at the current time and the current power is taken as the priority of the strong machine in the secondary distribution.

[0090] Next, the traction force of the strong machine at the current time is updated using the priority calculated above, that is, the ratio of the priority corresponding to any strong machine to the sum of the priorities corresponding to all strong machines is multiplied by the total traction force borne by all strong machines to obtain the updated traction force of the strong machine.

[0091] It should be noted that the "current time" involved in the operation of the secondary distribution after the weak machine traction force completes the preliminary transfer refers to the "next time of the current time" in the operation of the gradual traction force return mechanism.

[0092] Thus far, the redistribution of the cluster's towing force is completed.

[0093] The present application also provides a UAV flight control system based on a UAV search and rescue platform. Figure 2 As shown in the figure, the system comprises a processor and a memory, and the memory stores computer program instructions which, when executed by the processor, implement the UAV flight control method based on the UAV search and rescue platform according to the first aspect of the present application.

[0094] The system also comprises a communication bus and a communication interface and other components well known to those skilled in the art, the settings and functions of which are known in the art, and thus will not be described here.

[0095] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the inventive concept, and these all fall within the scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A method for controlling the flight of a UAV based on a UAV search and rescue platform, characterized in that, The method comprises the following steps: calculating the attitude stability and the wind resistance consumption of each unmanned aerial vehicle at the current time based on a preset time window at the current time; the process of obtaining the attitude stability comprises: for a single unmanned aerial vehicle, obtaining the attitude state at all times within the time window of the unmanned aerial vehicle, including the pitch angle, the roll angle and the yaw angle; calculating the standard deviation of each attitude state respectively, and taking the square root of the sum of the squares of the standard deviations of all attitude states as the attitude stability of the unmanned aerial vehicle at the current time; the process of obtaining the wind resistance consumption comprises: for a single unmanned aerial vehicle, calculating the instantaneous power at each time within the time window of the unmanned aerial vehicle at the current time; integrating the instantaneous power within the time window of the unmanned aerial vehicle, and then dividing the length of the time window to obtain the average power, which is taken as the wind resistance consumption of the unmanned aerial vehicle at the current time; calculating the excess proportion of the current attitude stability of each unmanned aerial vehicle relative to the historical reference attitude stability in a windless environment as a first proportion, and calculating the excess proportion of the current wind resistance consumption of each unmanned aerial vehicle relative to the historical reference wind resistance consumption in a windless environment as a second proportion; taking the product of the first proportion and the second proportion as the comprehensive wind resistance index of the unmanned aerial vehicle at the current time; wherein the average or standard value of the attitude stability and the wind resistance consumption in the historical flight data of the unmanned aerial vehicle in a windless environment is taken as the historical reference in a windless environment; judging and marking the weak machine based on the comprehensive wind resistance index; the process of judging the weak machine comprises: obtaining the predicted value of the traction force allocated to each unmanned aerial vehicle at the current time based on a pre-trained prediction model, setting a traction force threshold value according to the predicted values of the traction forces of all unmanned aerial vehicles, and judging the unmanned aerial vehicle with a predicted traction force less than the traction force threshold value as a weak machine; and judging the unmanned aerial vehicle with a predicted traction force greater than or equal to the traction force threshold value as a strong machine; calculating the traction force of each weak machine at the current time at the next time, which is the difference between the traction force of the weak machine at the current time and the calculated amount of traction force released by the weak machine at the current time; on the basis of the traction force of each strong machine at the current time, adding the increment obtained by equally sharing the total amount of traction force released by all weak machines at the current time to obtain the traction force of each strong machine at the next time at the current time; redistributing the traction force of the unmanned aerial vehicle cluster according to the calculated traction force of each weak machine and strong machine at the next time. the process of setting the time window comprises: 2.The UAV flight control method based on the UAV search and rescue platform of claim 1, wherein, calculating the ratio of the wind speed at the position of the unmanned aerial vehicle at the current time to the maximum bearable wind speed as a first factor; presetting the upper and lower limits of the time window, calculating the difference between the upper limit and the lower limit of the time window as a first difference, and calculating the difference between the upper limit of the time window and the product of the first factor and the first difference to obtain the length of the time window; setting the time window corresponding to the unmanned aerial vehicle according to the calculated length of the time window with the current time as the end point. The prediction model selects a gradient boosting decision tree, the model training input includes the real-time comprehensive wind resistance index of the unmanned aerial vehicle, the carried weight and the wind speed, and the output is the predicted value of the traction force allocated to the unmanned aerial vehicle at the current time, and the mean square error is used as the loss function in the model training process. 3.The UAV flight control method based on the UAV search and rescue platform of claim 1, wherein, the process of setting the traction force threshold value comprises: 4.The UAV search and rescue platform based flight control method of claim 1, wherein, ​ calculating the mean and the standard deviation of the traction force prediction values of all the UAVs at the current time; taking the product of the ratio of the wind speed at the current time to the maximum bearable wind speed and the standard deviation of the traction force prediction values of all the UAVs at the current time as a correction term; taking the difference between the mean of the traction force prediction values of all the UAVs at the current time and the correction term as the traction force threshold. 5.The UAV search and rescue platform based flight control method of claim 1, wherein, The process of calculating the traction force release amount of the weak UAV at the current time comprises: taking the product of the ratio of the wind speed at the current time to the maximum bearable wind speed and the actual traction force borne by the weak UAV at the current time as the traction force release amount of the weak UAV at the current time. 6.The UAV flight control method based on the UAV search and rescue platform according to claim 5, wherein, After the traction forces of the weak UAVs and the strong UAVs at the next time are calculated, the process comprises: calculating the priority of the secondary distribution of the traction force of the strong UAVs; updating the traction force of the strong UAVs at the next time based on the calculated priority of each strong UAV to obtain the updated traction force of the strong UAVs at the next time.

7. A UAV flight control system based on a UAV search and rescue platform, characterized by, The method comprises: a processor and a memory, wherein the memory stores computer program instructions which, when executed by the processor, implement the UAV flight control method based on the UAV search and rescue platform according to any one of claims 1-6.

Images