A method for suppressing dynamic wind disturbance of unmanned aerial vehicles

Abstract

The invention discloses a method for suppressing dynamic wind disturbance of unmanned aerial vehicles (UAVs), and belongs to the technical field of power control. The invention constructs a difference sequence between an actual moving speed of the UAV and an expected moving speed, and simultaneously collects and constructs an environmental wind speed sequence; secondly, by calculating a change factor and combining a moving speed prediction model, the moving speeds of the UAV at multiple moments are processed, so as to achieve accurate prediction of the moving speed of the UAV at the next moment; then, based on a double-layer wind disturbance control model, the difference between the predicted speed and the expected speed and the difference between the expected speed and the target speed are comprehensively considered, and the required wind disturbance suppression power is calculated; finally, after applying the wind disturbance suppression power, the double-layer wind disturbance control model is dynamically adjusted according to the difference between the actual moving speed of the UAV and the target speed, so as to achieve accurate suppression of wind disturbance and improve the flight accuracy of the UAV in complex meteorological environments.

Description

Technical Field

[0001] The present invention relates to the technical field of power control, and in particular to a method for suppressing dynamic wind disturbance of an unmanned aerial vehicle. Background Art

[0002] During the flight of a UAV, the ambient wind speed is a key factor affecting flight stability and accuracy. When a UAV encounters wind disturbances of different intensities and directions, significant speed deviations will occur. Traditional wind disturbance suppression methods mainly rely on simple linear compensation algorithms or experience-based adjustment strategies, which are often difficult to respond to complex and changing wind environments in real time and accurately. Existing technologies usually use fixed-gain feedback control or simple feedforward compensation. This method cannot effectively and dynamically adjust the flight parameters of the UAV when faced with sudden wind disturbances or rapidly changing wind fields. The main limitations of these technologies are: the lack of real-time perception of the dynamic characteristics of the wind field, the inability to accurately predict the instantaneous impact of wind on the UAV speed, resulting in a significant decrease in the flight accuracy of the UAV under complex meteorological conditions. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, a method for suppressing dynamic wind disturbance of an unmanned aerial vehicle provided by the present invention solves the problem of low flight accuracy of unmanned aerial vehicles under complex meteorological conditions.

[0004] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for suppressing dynamic wind disturbance of an unmanned aerial vehicle, comprising the following steps:

[0005] S1. According to the difference between the moving speed of the UAV and the expected moving speed, a speed difference sequence is constructed, and the ambient wind speed is collected to construct an ambient wind speed sequence;

[0006] S2. Calculate the change factor for the speed difference sequence and the ambient wind speed sequence, use the mobile speed prediction model to process the UAV's mobile speed at multiple moments, and combine the change factor to predict the UAV's mobile speed at the next moment;

[0007] S3, according to the difference between the moving speed of the UAV at the next moment and the expected moving speed, and the difference between the expected moving speed and the target speed, based on the double-layer wind disturbance control model, the wind disturbance suppression power is obtained;

[0008] S4. After applying wind disturbance suppression power to the UAV, the double-layer wind disturbance control model is adjusted according to the difference between the UAV moving speed and the target speed.

[0009] Further, S2 includes the following sub-steps:

[0010] S21. Calculate the speed difference variation factor for the speed difference sequence, where the speed difference sequence is Sd={s t ,…,s t-τ ,…,st-T}, where Sd is the speed difference sequence, s t is the speed difference at time t, s t-τ is the speed difference at the t-τth moment, s t-T is the speed difference at the tTth moment, t is the current moment, τ is the historical moment, and T is the length of the historical moment;

[0011] S22. Calculate the wind speed variation factor for the environmental wind speed sequence, where the environmental wind speed sequence ES = {v w,t ,…,v w,t-τ ,…,v w,t-T}, where ES is the ambient wind speed sequence, v w,t is the ambient wind speed at time t, v w,t is the ambient wind speed at the t-τ moment, v w,t-T is the ambient wind speed at the moment tT;

[0012] S23. Use the mobile speed prediction model to process the drone's mobile speed at multiple moments, combine the speed difference change factor and the wind speed change factor, and predict the drone's mobile speed at the next moment.

[0013] Furthermore, the formula for calculating the speed difference variation factor in S21 is: , where ε Sd is the speed difference variation factor, s c is the element mean of the speed difference sequence, | | is the absolute value operation;

[0014] The formula for calculating the wind speed variation factor in S22 is: , where ε ES is the wind speed variation factor, v w,c is the element mean of the ambient wind speed series.

[0015] Furthermore, the mobile speed prediction model includes: an LSTM layer, a speed feature extraction unit, a speed feature fusion unit, a wind speed feature fusion unit and a prediction layer;

[0016] The input end of the LSTM layer is used to input the moving speed of the UAV at multiple moments, and its output end is connected to the input end of the speed feature extraction unit; the output end of the speed feature extraction unit is respectively connected to the first input end of the speed feature fusion unit and the first input end of the wind speed feature fusion unit; the second input end of the speed feature fusion unit is used to input the speed difference change factor; the second input end of the wind speed feature fusion unit is used to input the wind speed change factor; the input end of the prediction layer is respectively connected to the output end of the speed feature fusion unit and the output end of the wind speed feature fusion unit, and its output end serves as the output end of the moving speed prediction model.

[0017] Furthermore, a Cell unit in the LSTM layer inputs the moving speed of the drone at a moment;

[0018] The expression of the speed feature extraction unit is: , where H is the velocity characteristic value, h i is the i-th feature output by the LSTM layer, ω i h i The weight term, b i h i The bias term, N is the number of output features of the LSTM layer, and i is a positive integer;

[0019] The expression of the velocity feature fusion unit is: , where y Sd is the velocity feature fusion value, ω 1,H is the first weight term of H, b 1,H is the first bias term of H, ε Sd is the speed difference variation factor, ω εSd is Sd The weight term, b εSd is Sd The bias term, f is the activation function;

[0020] The expression of wind speed feature fusion unit is: , where y ES is the wind speed feature fusion value, ω 2,H is the second weight term of H, b 2,H is the second bias term of H, ε ES is the wind speed variation factor, ω εES is ES The weight term, b εES is ES The bias term.

[0021] Furthermore, the expression of the prediction layer is: , , where Y t+1 is the moving speed of the drone at the t+1th moment, y Sd is the velocity feature fusion value, y ES is the wind speed feature fusion value, ω ySd for y Sd The weight term, ω yES for y ES The weight term, b y is the bias term in the prediction layer, g is the characteristic value of the moving speed, tanh is the hyperbolic tangent function, ln is the logarithmic function, and e is a natural constant.

[0022] Further, the double-layer wind disturbance control model in S4 includes: a first difference calculation unit, a second difference calculation unit, a first feedback control unit, a second feedback control unit, an addition unit and a power adjustment unit;

[0023] The first difference calculation unit is used to calculate the difference between the moving speed of the UAV at the next moment and the expected moving speed at the next moment to obtain the first difference; the second difference calculation unit is used to calculate the difference between the expected moving speed at the current moment and the target speed to obtain the second difference; the first feedback control unit is used to input the first difference to obtain the first power component; the second feedback control unit is used to input the second difference to obtain the second power component; the addition unit is used to add the first power component and the second power component to obtain the initial power; the power adjustment unit is used to adjust the initial power to obtain the wind disturbance suppression power, wherein the first feedback control unit and the second feedback control unit are PID models.

[0024] Furthermore, the expression of the power regulation unit is: , where P w is the wind disturbance suppression power, P o is the initial power, μ is the adjustment coefficient.

[0025] Further, S4 includes the following sub-steps:

[0026] S41. After applying wind disturbance suppression power to the UAV, collect the moving speed of the UAV and calculate the speed change value: v b =v t -v t-r , v b is the velocity change value, v t is the moving speed of the drone at the tth moment, v t-r is the moving speed of the drone at the trth moment, t is the current moment, and r is the selected time length;

[0027] S42, calculating the difference between the moving speed of the UAV at the trth moment and the target speed, and obtaining the target speed difference;

[0028] S43, calculate the adjustment speed according to the speed change value and the target speed difference: θ=|v b | / |d tar |, where θ is the adjustment speed, | | is the absolute value operation, d tar is the target speed difference;

[0029] S44, when the adjustment speed is greater than or equal to 0.7, the adjustment coefficient in the double-layer wind disturbance control model is set to 0;

[0030] S45. When the adjustment speed is less than 0.7, the adjustment coefficient in the double-layer wind disturbance control model is calculated according to the difference between the moving speed of the UAV and the target speed.

[0031] Furthermore, the formula for calculating the adjustment coefficient in S45 is: , where μ is the adjustment coefficient, v t is the moving speed of the drone at the tth moment, v tar is the target speed, v d is the speed difference threshold, and || is the absolute value operation.

[0032] The beneficial effects of the present invention are as follows: the present invention calculates the change factor based on the speed difference sequence and the ambient wind speed sequence, and can sense the change of the ambient wind speed and the change of the influence on the speed. The mobile speed prediction model is adopted, combined with the adjustment of the change factor, and the moving speed of the drone at the next moment can be accurately predicted, thereby accurately estimating the instantaneous influence of the wind on the speed.

[0033] The present invention adopts a double-layer wind disturbance control model based on the speed difference at the next moment and the difference between the expected speed and the target speed, and can more accurately calculate the required wind disturbance suppression power. After applying the wind disturbance suppression power, the present invention can adjust the double-layer wind disturbance control model by continuously monitoring the difference between the moving speed of the drone and the target speed, so as to achieve the rapid approach of the moving speed of the drone to the target speed and improve the flight accuracy of the drone under complex meteorological conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of a method for suppressing dynamic wind disturbance of UAV;

[0035] Figure 2 It is a structural diagram of the moving speed prediction model;

[0036] Figure 3 It is a structural diagram of the double-layer wind disturbance control model. DETAILED DESCRIPTION

[0037] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0038] like Figure 1 As shown, a method for suppressing dynamic wind disturbance of an unmanned aerial vehicle comprises the following steps:

[0039] S1. According to the difference between the moving speed of the UAV and the expected moving speed, a speed difference sequence is constructed, and the ambient wind speed is collected to construct an ambient wind speed sequence;

[0040] S2. Calculate the change factor for the speed difference sequence and the ambient wind speed sequence, use the mobile speed prediction model to process the UAV's mobile speed at multiple moments, and combine the change factor to predict the UAV's mobile speed at the next moment;

[0041] S3, according to the difference between the moving speed of the UAV at the next moment and the expected moving speed, and the difference between the expected moving speed and the target speed, based on the double-layer wind disturbance control model, the wind disturbance suppression power is obtained;

[0042] S4. After applying wind disturbance suppression power to the UAV, the double-layer wind disturbance control model is adjusted according to the difference between the UAV moving speed and the target speed.

[0043] The drone moving speed is the actual moving speed of the drone collected. The expected moving speed is the moving speed that the drone can theoretically (in a windless state) stably reach at a given rotation speed. The expected moving speed is a known quantity. The target speed is the moving speed that the drone needs to reach, taking into account the impact of the current environmental wind speed on the drone speed. That is, in the present invention, the expected moving speed is the moving speed that the drone should reach at a given rotation speed, and the target speed is the set moving speed that needs to be reached.

[0044] In this embodiment, S2 includes the following sub-steps:

[0045] S21. Calculate the speed difference variation factor for the speed difference sequence, where the speed difference sequence is Sd={s t ,…,s t-τ ,…,s t-T}, where Sd is the speed difference sequence, s t is the speed difference at time t, s t-τ is the speed difference at the t-τth moment, s t-T is the speed difference at the tTth moment, t is the current moment, τ is the historical moment, and T is the length of the historical moment;

[0046] S22. Calculate the wind speed variation factor for the environmental wind speed sequence, where the environmental wind speed sequence ES = {v w,t ,…,v w,t-τ ,…,v w,t-T}, where ES is the ambient wind speed sequence, v w,t is the ambient wind speed at time t, v w,t is the ambient wind speed at the t-τ moment, v w,t-T is the ambient wind speed at the moment tT;

[0047] S23. Use the mobile speed prediction model to process the drone's mobile speed at multiple moments, combine the speed difference change factor and the wind speed change factor, and predict the drone's mobile speed at the next moment.

[0048] In this embodiment, the formula for calculating the speed difference variation factor in S21 is: , where ε Sd is the speed difference variation factor, s c is the element mean of the speed difference sequence, | | is the absolute value operation;

[0049] The formula for calculating the wind speed variation factor in S22 is: , where ε ES is the wind speed variation factor, v w,c is the element mean of the ambient wind speed series.

[0050] The present invention calculates the speed difference s at the tth moment in the speed difference sequence t , that is, the speed difference s at the latest moment t The difference between the speed difference sequence and the mean value reflects the change of the speed difference. If the speed difference change factor is large, it means that the speed difference fluctuates greatly, and the influence of wind speed on the moving speed of the drone during this period of time fluctuates greatly. If the speed difference change factor is small, it means that the influence of wind speed on the moving speed of the drone during this period of time is relatively consistent. The present invention calculates the ambient wind speed v at the tth moment in the ambient wind speed sequence w,t , that is, the ambient wind speed v at the latest moment w,t The difference from the mean of the ambient wind speed sequence reflects the change of the ambient wind speed. A larger wind speed change factor indicates a larger fluctuation of the ambient wind speed. The present invention can calculate the corresponding change factors by analyzing the historical change trends of the speed difference sequence and the ambient wind speed sequence. These change factors can reflect the dynamic characteristics of the drone speed and the ambient wind speed.

[0051] In this embodiment, the ambient wind speed can be collected by setting a wind speed sensor on the drone. If the wind speed collected by the wind speed sensor on the drone also includes the wind speed generated by its own rotation, the wind speed generated by its own rotation needs to be subtracted. For the drone at a certain rotation speed, the wind speed generated is a known quantity.

[0052] like Figure 2 As shown, the mobile speed prediction model includes: an LSTM layer, a speed feature extraction unit, a speed feature fusion unit, a wind speed feature fusion unit and a prediction layer;

[0053] The input end of the LSTM layer is used to input the moving speed of the UAV at multiple moments, and its output end is connected to the input end of the speed feature extraction unit; the output end of the speed feature extraction unit is respectively connected to the first input end of the speed feature fusion unit and the first input end of the wind speed feature fusion unit; the second input end of the speed feature fusion unit is used to input the speed difference change factor; the second input end of the wind speed feature fusion unit is used to input the wind speed change factor; the input end of the prediction layer is respectively connected to the output end of the speed feature fusion unit and the output end of the wind speed feature fusion unit, and its output end serves as the output end of the moving speed prediction model.

[0054] When predicting the moving speed of the unmanned aerial vehicle at the next moment, the present invention adopts an LSTM layer to process the moving speed of the unmanned aerial vehicle at multiple moments, extracts hidden time series features, and then obtains a speed feature value H through a speed feature extraction unit. The speed feature fusion unit considers the speed difference change factor to realize the consideration of the influence of wind speed on the moving speed under the current situation. If the speed difference change factor is large, the disturbance of the wind speed on the moving speed has a large fluctuation. If the speed difference change factor is small, the disturbance of the wind speed on the moving speed is relatively consistent during this period of time. Then, the wind speed change factor is considered through the wind speed feature fusion unit, and the change of wind speed is considered. Then, the prediction layer combines the outputs of the wind speed feature fusion unit and the speed feature fusion unit to improve the prediction of the moving speed of the unmanned aerial vehicle at the next moment.

[0055] The mobile speed prediction model of the present invention combines the speed difference change factor and the wind speed change factor with the historical UAV mobile speed, so as to incorporate the disturbance of the ambient wind speed on the UAV into the prediction model.

[0056] In this embodiment, a Cell unit of the LSTM layer inputs the moving speed of the drone at a moment;

[0057] The expression of the speed feature extraction unit is: , where H is the velocity characteristic value, h i is the i-th feature output by the LSTM layer, ω i h i The weight term, b i h i The bias term, N is the number of output features of the LSTM layer, and i is a positive integer.

[0058] The expression of the velocity feature fusion unit is: , where y Sd is the velocity feature fusion value, ω 1,H is the first weight term of H, b 1,H is the first bias term of H, ε Sd is the speed difference variation factor, ω εSd is Sd The weight term, bεSd is Sd The bias term of , and f is the activation function.

[0059] The present invention converts the speed characteristic value H and the speed difference change factor ε Sd The speed feature fusion value is obtained by fusion, and the disturbance of the ambient wind speed on the UAV is added, ε Sd If the disturbance is large, the fluctuation of the disturbance during this period is large and there is a large randomness. Sd If the disturbance is smaller, the fluctuation of disturbance during this period is smaller and the degree of disturbance is more consistent.

[0060] The expression of wind speed feature fusion unit is: , where y ES is the wind speed feature fusion value, ω 2,H is the second weight term of H, b 2,H is the second bias term of H, ε ES is the wind speed variation factor, ω εES is ES The weight term, b εES is ES The bias term.

[0061] The present invention combines the velocity characteristic value H with the wind speed variation factor ε ES The wind speed characteristic fusion value is obtained by fusion, and the ambient wind speed is added, ε ES If it is large, the ambient wind speed changes greatly during this period of time, and there is a large randomness. ES If the value is smaller, the ambient wind speed changes less during this period of time.

[0062] In this embodiment, the activation function f may be a Sigmoid function.

[0063] In this embodiment, the expression of the prediction layer is: , , where Y t+1 is the moving speed of the drone at the t+1th moment, y Sd is the velocity feature fusion value, y ES is the wind speed feature fusion value, ω ySd for y Sd The weight term, ω yES for y ES The weight term, b y is the bias term in the prediction layer, g is the characteristic value of the moving speed, tanh is the hyperbolic tangent function, ln is the logarithmic function, and e is a natural constant.

[0064] The prediction layer of the present invention integrates the speed feature fusion value and the wind speed feature fusion value to obtain the moving speed feature value, and then uses the activation function to map the moving speed feature value to the moving speed of the drone at the t+1th moment.

[0065] In this embodiment, the prediction layer may also be implemented using a classifier such as a fully connected layer.

[0066] In this embodiment, the weight term and the bias term in the moving speed prediction model can be obtained by training using the existing gradient descent method.

[0067] like Figure 3 As shown, the double-layer wind disturbance control model in S4 includes: a first difference calculation unit, a second difference calculation unit, a first feedback control unit, a second feedback control unit, an addition unit and a power adjustment unit;

[0068] The first difference calculation unit is used to calculate the difference between the moving speed of the UAV at the next moment and the expected moving speed at the next moment to obtain the first difference; the second difference calculation unit is used to calculate the difference between the expected moving speed at the current moment and the target speed to obtain the second difference; the first feedback control unit is used to input the first difference to obtain the first power component; the second feedback control unit is used to input the second difference to obtain the second power component; the addition unit is used to add the first power component and the second power component to obtain the initial power; the power adjustment unit is used to adjust the initial power to obtain the wind disturbance suppression power, wherein the first feedback control unit and the second feedback control unit are PID models.

[0069] The expected moving speed at the next moment in the present invention can be obtained by looking up a table according to the rotation speed of the drone at the next moment, or can be calculated according to the expected moving speed at the current moment and the moving acceleration at the current moment.

[0070] The present invention predicts the moving speed of the drone at the next moment and compensates for the amount of interference from the environmental wind in advance.

[0071] The present invention calculates a first difference to reflect the degree of interference by the ambient wind speed at the next moment, calculates a second difference to reflect the gap between the current expected moving speed and the target speed, calculates the power for resisting wind disturbance through the first feedback control unit, and then calculates the power for reaching the target speed through the second feedback control unit, combines the two parts of power to obtain the initial power, and the power adjustment unit is used to adjust the initial power so that the moving speed of the UAV can quickly approach the target speed.

[0072] In this embodiment, the expression of the power adjustment unit is: , where P w is the wind disturbance suppression power, P ois the initial power, μ is the adjustment coefficient, and the initial value of the adjustment coefficient is 0.

[0073] In this embodiment, S4 includes the following sub-steps:

[0074] S41. After applying wind disturbance suppression power to the UAV, collect the moving speed of the UAV and calculate the speed change value: v b =v t -v t-r , v b is the velocity change value, v t is the moving speed of the drone at the tth moment, v t-r is the moving speed of the drone at the trth moment, t is the current moment, and r is the selected time length;

[0075] S42, calculating the difference between the moving speed of the UAV at the trth moment and the target speed, and obtaining the target speed difference;

[0076] S43, calculate the adjustment speed according to the speed change value and the target speed difference: θ=|v b | / |d tar |, where θ is the adjustment speed, | | is the absolute value operation, d tar is the target speed difference;

[0077] S44, when the adjustment speed is greater than or equal to 0.7, the adjustment coefficient in the double-layer wind disturbance control model is set to 0;

[0078] S45. When the adjustment speed is less than 0.7, the adjustment coefficient in the double-layer wind disturbance control model is calculated according to the difference between the moving speed of the UAV and the target speed.

[0079] In this embodiment, the threshold corresponding to the adjustment speed can be set to 0.7 or can be adjusted according to demand, and the range of the threshold is 0.5-1.

[0080] Since the first feedback control unit and the second feedback control unit in the present invention are PID models, PID is a feedback control system that measures the deviation between the actual output value of the system and the expected output value, and adjusts the control quantity according to the three parameters of proportion (P), integration (I) and differentiation (D), so that the system output is as close to the expected value as possible. When the external environment changes rapidly, such as a sudden increase in wind speed, the dynamic characteristics of the system will also change rapidly. PID cannot adjust its parameters in time to adapt to such rapid changes, which will cause the control deviation to increase sharply, and eventually cause the system to lose control and crash. In order to prevent PID from failing to converge or even crashing, the present invention calculates the adjustment speed after applying the wind disturbance suppression power to reflect the adjustment of the double-layer wind disturbance control model. If the adjustment speed meets the requirements, the adjustment coefficient is set to 0. If the adjustment speed does not meet the requirements, the adjustment coefficient in the double-layer wind disturbance control model is calculated according to the difference between the moving speed of the drone and the target speed.

[0081] In this embodiment, the formula for calculating the adjustment coefficient in S45 is: , where μ is the adjustment coefficient, v t is the moving speed of the drone at the tth moment, v tar is the target speed, v d is the speed difference threshold, and || is the absolute value operation.

[0082] In the present invention, the speed difference threshold is Set the threshold.

[0083] When the difference between the UAV's moving speed and the target speed is less than the speed difference threshold, that is, the UAV's moving speed is close to the target speed, the adjustment coefficient is 0. When it is greater than the speed difference threshold, the adjustment coefficient is calculated based on the difference between the target speed and the UAV's moving speed. When the target speed is greater than the UAV's moving speed, the power needs to be increased, and the adjustment coefficient is greater than 0, so that the power increases rapidly. When the target speed is less than the UAV's moving speed, the power needs to be reduced, and the adjustment coefficient is less than 0, which reduces the initial power, reduces the speed rapidly, and promotes the convergence of the PID model. When the value is less than 0, the adjustment coefficient returns to 0 to avoid affecting the stability of the PID model.

[0084] The present invention calculates the change factor based on the speed difference sequence and the ambient wind speed sequence, and can sense the change of ambient wind speed and the change of the impact on speed. The mobile speed prediction model is adopted, combined with the adjustment of the change factor, and the moving speed of the drone at the next moment can be accurately predicted, thereby accurately estimating the instantaneous impact of wind on speed.

[0085] The present invention adopts a double-layer wind disturbance control model based on the speed difference at the next moment and the difference between the expected speed and the target speed, and can more accurately calculate the required wind disturbance suppression power. After applying the wind disturbance suppression power, the present invention can adjust the double-layer wind disturbance control model by continuously monitoring the difference between the moving speed of the drone and the target speed, so as to achieve the rapid approach of the moving speed of the drone to the target speed and improve the flight accuracy of the drone under complex meteorological conditions.

[0086] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for suppressing dynamic wind disturbance of an unmanned aerial vehicle, characterized in that: The following steps are involved: S1. According to the difference between the moving speed of the UAV and the expected moving speed, a speed difference sequence is constructed, and the ambient wind speed is collected to construct an ambient wind speed sequence; S2. Calculate the change factor for the speed difference sequence and the ambient wind speed sequence, use the mobile speed prediction model to process the UAV's mobile speed at multiple moments, and combine the change factor to predict the UAV's mobile speed at the next moment; S3, according to the difference between the moving speed of the UAV at the next moment and the expected moving speed, and the difference between the expected moving speed and the target speed, based on the double-layer wind disturbance control model, the wind disturbance suppression power is obtained; S4, after applying wind disturbance suppression power to the UAV, adjust the double-layer wind disturbance control model according to the difference between the UAV moving speed and the target speed; The double-layer wind disturbance control model in S4 includes: a first difference calculation unit, a second difference calculation unit, a first feedback control unit, a second feedback control unit, an addition unit and a power adjustment unit; The first difference calculation unit is used to calculate the difference between the moving speed of the drone at the next moment and the expected moving speed at the next moment to obtain a first difference; the second difference calculation unit is used to calculate the difference between the expected moving speed at the current moment and the target speed to obtain a second difference; the first feedback control unit is used to input the first difference to obtain a first power component; the second feedback control unit is used to input the second difference to obtain a second power component; the addition unit is used to add the first power component and the second power component to obtain an initial power; the power adjustment unit is used to adjust the initial power to obtain a wind disturbance suppression power, wherein the first feedback control unit and the second feedback control unit are PID models; The expression of the power adjustment unit is: , where P w is the wind disturbance suppression power, P o is the initial power, μ is the adjustment coefficient; The S4 comprises the following sub-steps: S41. After applying wind disturbance suppression power to the UAV, collect the moving speed of the UAV and calculate the speed change value: v b =v t -v t-r , v b is the velocity change value, v t is the moving speed of the drone at the tth moment, v t-r is the moving speed of the drone at the trth moment, t is the current moment, and r is the selected time length; S42, calculating the difference between the moving speed of the UAV at the trth moment and the target speed, and obtaining the target speed difference; S43, according to the speed change value and the target speed difference, calculate the adjustment speed: θ=|v b | / |d tar |, where θ is the adjustment speed, | | is the absolute value operation, d tar is the target speed difference; S44, when the adjustment speed is greater than or equal to 0.7, the adjustment coefficient in the double-layer wind disturbance control model is set to 0; S45, when the adjustment speed is less than 0.7, the adjustment coefficient in the double-layer wind disturbance control model is calculated according to the difference between the moving speed of the UAV and the target speed; The formula for calculating the adjustment coefficient in S45 is: , where μ is the adjustment coefficient, v t is the moving speed of the drone at the tth moment, v tar is the target speed, v d is the speed difference threshold, and || is the absolute value operation.

2. The method for suppressing dynamic wind disturbance of an unmanned aerial vehicle according to claim 1, characterized in that: The S2 comprises the following sub-steps: S21. Calculate the speed difference variation factor for the speed difference sequence, where the speed difference sequence is Sd={s t ,…,s t-τ ,…,s t-T }, where Sd is the speed difference sequence, s t is the speed difference at time t, s t-τ is the speed difference at the t-τth moment, s t-T is the speed difference at the tTth moment, t is the current moment, τ is the historical moment, and T is the length of the historical moment; S22. Calculate the wind speed variation factor for the environmental wind speed sequence, where the environmental wind speed sequence ES = {v w,t ,…,v w,t-τ ,…,v w,t-T }, where ES is the ambient wind speed sequence, v w,t is the ambient wind speed at time t, v w,t is the ambient wind speed at the t-τth moment, v w,t-T is the ambient wind speed at the moment tT; S23. Use the mobile speed prediction model to process the drone's mobile speed at multiple moments, combine the speed difference change factor and the wind speed change factor, and predict the drone's mobile speed at the next moment.

3. The method for suppressing dynamic wind disturbance of an unmanned aerial vehicle according to claim 2, characterized in that: The formula for calculating the speed difference variation factor in S21 is: , where ε Sd is the speed difference variation factor, s c is the element mean of the speed difference sequence, | | is the absolute value operation; The formula for calculating the wind speed variation factor in S22 is: , where ε ES is the wind speed variation factor, v w,c is the element mean of the ambient wind speed series.

4. The method for suppressing dynamic wind disturbance of an unmanned aerial vehicle according to claim 2, characterized in that: The mobile speed prediction model in S23 includes: an LSTM layer, a speed feature extraction unit, a speed feature fusion unit, a wind speed feature fusion unit and a prediction layer; The input end of the LSTM layer is used to input the moving speed of the drone at multiple moments, and its output end is connected to the input end of the speed feature extraction unit; the output end of the speed feature extraction unit is respectively connected to the first input end of the speed feature fusion unit and the first input end of the wind speed feature fusion unit; the second input end of the speed feature fusion unit is used to input the speed difference change factor; the second input end of the wind speed feature fusion unit is used to input the wind speed change factor; the input end of the prediction layer is respectively connected to the output end of the speed feature fusion unit and the output end of the wind speed feature fusion unit, and its output end serves as the output end of the moving speed prediction model.

5. The method for suppressing dynamic wind disturbance of a UAV according to claim 4, characterized in that: A Cell unit of the LSTM layer inputs the moving speed of the drone at a moment; The expression of the speed feature extraction unit is: , where H is the velocity characteristic value, h i is the i-th feature output by the LSTM layer, ω i h i The weight term, b i h i The bias term, N is the number of output features of the LSTM layer, and i is a positive integer; The expression of the velocity feature fusion unit is: , where y Sd is the velocity feature fusion value, ω 1,H is the first weight term of H, b 1,H is the first bias term of H, ε Sd is the speed difference variation factor, ω εSd is Sd The weight term, b εSd is Sd The bias term, f is the activation function; The expression of the wind speed feature fusion unit is: , where y ES is the wind speed feature fusion value, ω 2,H is the second weight term of H, b 2,H is the second bias term of H, ε ES is the wind speed variation factor, ω εES is ES The weight term, b εES is ES The bias term.

6. The method for suppressing dynamic wind disturbance of an unmanned aerial vehicle according to claim 4, characterized in that: The expression of the prediction layer is: , , where Y t+1 is the moving speed of the drone at the t+1th moment, y Sd is the velocity feature fusion value, y ES is the wind speed feature fusion value, ω ySd for y Sd The weight term, ω yES for y ES The weight term, b y is the bias term in the prediction layer, g is the characteristic value of the moving speed, tanh is the hyperbolic tangent function, ln is the logarithmic function, and e is a natural constant.

Images