Inspection of drone flight control methods, equipment, programs and storage media

Abstract

This application relates to the field of unmanned aerial vehicle (UAV) control technology, and discloses a flight control method, device, program, and storage medium for an inspection UAV. The method includes: stacking standard operating condition time-series data of the UAV under the current environment based on a sliding time window to output a multi-dimensional disturbance perception matrix to determine the core feature set of the UAV; determining the surrounding wind intensity category under the current environment by combining a preset feature threshold set; if it is a strong wind, determining the actual disturbance torque vector based on the core feature set and a preset torque data table; determining a feedforward flight compensation vector based on a preset dynamic model and the actual disturbance torque vector; generating a feedback correction vector based on the multi-dimensional disturbance perception matrix, a preset basic neural network model, a preset expected disturbance torque vector, the core feature set, the actual disturbance torque vector, and a feedback control algorithm; and controlling the UAV flight based on the feedforward flight compensation vector and the feedback correction vector. This application can improve the UAV's resistance to strong wind interference.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) control technology, specifically to a flight control method, equipment, program, and storage medium for an inspection UAV. Background Technology

[0002] With the rapid development of power grid construction, drones have been widely used in the inspection of power transmission lines due to their advantages such as maneuverability, flexibility, and low cost. However, power transmission lines are often erected in complex terrain areas such as mountains, hills, or wind gaps. The meteorological environment in these areas is extremely complex, often accompanied by nonlinear sudden strong crosswind interference such as sudden increases or decreases in wind speed and abrupt changes in wind direction.

[0003] Current inspection drone flight control systems typically employ a traditional feedback control architecture based on inertial measurement units (IMUs) and global positioning systems (GPS) when dealing with complex wind fields. This architecture relies heavily on a passive logic of "detecting deviations and then correcting them," meaning the controller only calculates and outputs correction commands after sensors detect a substantial tilt or positional shift in the aircraft. This delayed perception and response mechanism often results in the drone failing to generate sufficient counter-torque to offset sudden strong winds, leading to significant flight overshoot.

[0004] Furthermore, most existing flight control algorithms employ fixed-parameter linear PID control strategies, whose control gain is difficult to change once set, making them unable to detect changes in external wind intensity. When encountering extreme strong winds exceeding design conditions, the fixed control parameters often fail to match the current dynamic requirements, leading to insufficient system adjustment capabilities or severe oscillations. Simultaneously, existing obstacle avoidance strategies are mostly based on static obstacle detection, lacking consideration for maneuverability boundaries under strong wind conditions. When a drone is blown off course by strong winds and approaches a live power line, it often fails to generate an effective avoidance path that balances attitude stability and rapid escape. These shortcomings result in poor wind resistance for existing inspection drones in adverse weather conditions, easily leading to flight path deviations, gimbal vibrations, and even collisions with power lines, severely restricting the efficiency and safety of drone power line inspections. Summary of the Invention

[0005] The purpose of this application is to provide a flight control method, device, program, and storage medium for inspection drones, in order to solve the problem of poor resistance to strong wind interference in existing inspection drones.

[0006] To achieve the above objectives, the first aspect of this application provides a flight control method for an inspection drone, comprising:

[0007] In response to receiving multiple operating condition time series data of the UAV in the current environment, the system standardizes each operating condition time series data and outputs the corresponding standard operating condition time series data.

[0008] By stacking time series data for each standard operating condition through a sliding time window, a multidimensional disturbance sensing matrix is ​​output.

[0009] The core feature set of the UAV is determined based on the multi-dimensional disturbance perception matrix;

[0010] The intensity category of the surrounding wind in the current environment is determined based on the core feature set and the preset feature threshold set;

[0011] If the surrounding wind intensity category is strong wind, the actual disturbance torque vector of the UAV is determined based on the core feature set and the preset torque data table; the preset torque data table stores the mapping relationship between the core features and the disturbance torque.

[0012] The feedforward flight compensation vector of the UAV is determined based on the preset dynamic model and the actual disturbance moment vector.

[0013] The feedback correction vector of the UAV is generated based on the multidimensional disturbance perception matrix, the preset basic neural network model, the preset expected disturbance torque vector, the core feature set, the actual disturbance torque vector, and the feedback control algorithm.

[0014] The UAV is controlled to fly based on the feedforward flight compensation vector and the feedback correction vector.

[0015] In this embodiment, the core feature set includes the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate. The step of generating the UAV's feedback correction vector based on the multidimensional disturbance perception matrix, the preset basic neural network model, the preset expected disturbance moment vector, the core feature set, the actual disturbance moment vector, and the feedback control algorithm includes: adjusting the preset basic neural network model according to the multidimensional disturbance perception matrix and minimizing the training loss function, and outputting the adjusted neural network model; determining the UAV's state deviation according to the preset expected disturbance moment vector and the actual disturbance moment vector; inputting the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate into the neural network model, and outputting the optimal feedback control parameters under the current environment; generating an adaptive function according to the optimal feedback control parameters and the feedback control algorithm; and generating the UAV's feedback correction vector according to the state deviation, the adaptive function, and the actual disturbance moment vector.

[0016] In this embodiment of the application, determining the core feature set of the UAV based on the multidimensional perturbation perception matrix includes: determining the maximum attitude change rate according to the following formula:

[0017]

[0018] The course deviation acceleration is determined using the following formula:

[0019]

[0020] The rate of change of wind speed gradient is determined using the following formula:

[0021]

[0022] in, For the first The first moment attitude angle; This refers to the sensor sampling period for the drone. For drones in the The distance of the flight path deviation at any given time; For drones in the Wind speed at any given moment; This represents the maximum attitude change rate. This refers to the acceleration due to the deviation from the flight path. The rate of change of wind speed gradient; This represents the total number of samples within the sliding time window.

[0023] In this embodiment of the application, generating the UAV's feedback correction vector based on the state deviation, the adaptive function, and the actual disturbance moment vector includes determining the feedback correction vector according to the following formula:

[0024]

[0025] in, For feedback correction vector; It is an adaptive function; This is the actual disturbance moment vector; For the first State deviation at any given moment.

[0026] In this embodiment of the application, the method further includes: acquiring the set of deployment coordinate points of the transmission line to be inspected and the UAV in three-dimensional space and the current position coordinates; determining the shortest distance between the UAV and the transmission line to be inspected based on the current position coordinates and the set of deployment coordinate points; determining that the shortest distance is less than the sum of a preset safety boundary warning threshold and a preset safety inspection zone radius; constructing an avoidance path objective function based on multiple maneuvering parameters of the UAV, the constraints corresponding to each maneuvering parameter, and the shortest distance; determining the optimal values ​​of each of the multiple maneuvering parameters with the aim of minimizing the function value of the avoidance path objective function; and controlling the UAV to fly according to the optimal values ​​of each of the multiple maneuvering parameters.

[0027] In this embodiment, the preset feature threshold set includes an attitude change rate threshold, a flight path deviation acceleration threshold, and a wind speed gradient change rate threshold. Determining the surrounding wind intensity category based on the core feature set and the preset feature threshold set includes: determining the surrounding wind intensity category according to the following formula:

[0028]

[0029] in, When the value is 1, the surrounding wind intensity category is determined to be strong wind; When the value is 0, the surrounding wind intensity category is determined to be conventional wind; This is the threshold for the attitude change rate; The flight path deviation acceleration threshold; The threshold for the rate of change of wind speed gradient; For logical OR.

[0030] In this embodiment, multiple maneuver parameters include roll rate, pitch rate, and yaw rate; the avoidance path objective function is constructed based on the multiple maneuver parameters of the UAV, the constraints corresponding to each maneuver parameter, and the shortest distance, including: the avoidance path objective function satisfies the following formula:

[0031]

[0032] The constraint conditions for each kinematic parameter satisfy the following formula:

[0033]

[0034] in, It is the roll angular velocity; This is the minimum value among the constraints corresponding to the roll angular velocity; This is the maximum value among the constraints corresponding to the roll angular velocity; It is the pitch angular velocity; This is the minimum value among the constraints corresponding to the pitch angular velocity; This represents the maximum value among the constraints corresponding to the pitch angular velocity; Yaw angular velocity; This is the minimum value among the constraints corresponding to the yaw rate; This represents the maximum value among the constraints corresponding to the yaw rate. In the first The shortest distance at any given moment; This is the safety weighting coefficient; To stabilize the weighting coefficients; To avoid the function value of the objective function of the path.

[0035] A second aspect of this application provides a computer device, comprising:

[0036] The memory is configured to store instructions;

[0037] And the processor, configured to retrieve instructions from memory and to implement the methods described above when executing instructions.

[0038] A third aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0039] A fourth aspect of this application provides a machine-readable storage medium storing instructions that cause a machine to perform the methods described above.

[0040] The above technical solution first utilizes standardized processing and sliding time window stacking to transform the multi-source heterogeneous instantaneous operating data of the UAV into a multi-dimensional disturbance perception matrix containing temporal dynamic laws, thereby accurately capturing the evolution trend of environmental disturbances over time. Next, based on the comparative analysis of the core feature set and the preset feature threshold set, accurate identification of the surrounding wind intensity category is achieved. Furthermore, through a preset torque data table, the abstract strong wind disturbance is further quantified into a specific actual disturbance torque vector, providing precise physical input for the control system. On this basis, a composite control strategy of "feedforward + feedback" is adopted. On the one hand, it utilizes a feedforward flight compensation vector based on a preset dynamic model to quickly respond to and offset the initial impact of sudden strong winds. On the other hand, it uses a feedback control algorithm combined with an online fine-tuning neural network to generate a feedback correction vector, and adaptively optimizes control parameters according to real-time operating conditions to eliminate residual deviations. Finally, the UAV is driven by the total control vector generated by fusing the feedforward and feedback, which takes into account both the dynamic response speed against disturbances and the steady-state control accuracy. This effectively solves the problem of lag or poor adaptability of single control methods in complex and variable wind fields, significantly enhances the stability of flight attitude, and thus improves the inspection UAV's ability to resist strong winds.

[0041] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0042] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0043] Figure 1 A flowchart illustrating a flight control method for an inspection drone according to an embodiment of this application is shown schematically.

[0044] Figure 2 This illustration schematically shows another flight control method for an inspection drone according to an embodiment of this application;

[0045] Figure 3 The schematic diagram illustrates a structural diagram of a computer device according to an embodiment of this application. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0047] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0048] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0049] The acquisition, transmission, storage, use, and processing of data in this application comply with relevant laws and regulations. Furthermore, it should be noted that certain software, components, models, and other existing industry solutions may be mentioned in the embodiments of this application. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0050] It should be noted that all data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are information and data authorized by the client or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0051] Figure 1 A flowchart illustrating a flight control method for an inspection drone according to an embodiment of this application is shown schematically. Figure 1 As shown in the figure, this application provides a flight control method for an inspection drone, which may include the following steps.

[0052] Step 101: In response to receiving multiple operating condition timing data of the UAV in the current environment, standardize each operating condition timing data and output the corresponding standard operating condition timing data.

[0053] In this embodiment, a drone refers to an unmanned aerial vehicle capable of performing automatic inspection tasks in specific scenarios such as power transmission lines. Its fuselage is equipped with various detection devices for sensing its own motion state and external environmental information, including an inertial measurement unit, barometer, anemometer, and visual sensors. The current environment refers to the actual three-dimensional physical space in which the drone operates, including meteorological conditions, terrain, and facilities such as power transmission lines that may affect flight stability. Operating condition time-series data refers to the set of raw physical quantity values ​​collected in real-time by the aforementioned sensors during drone operation, reflecting multiple dimensions such as its roll and pitch flight attitudes, flight path position deviations, relative distances to obstacles, and surrounding wind speed and direction. Standardization processing refers to the mathematical transformation process of removing dimensions from various data based on their numerical characteristics, aiming to eliminate the incomparability caused by significant differences in measurement units, orders of magnitude, or dynamic ranges of different physical quantities. Standard operating condition time-series data refers to a normalized data sequence with uniform and comparable dimensions after the above transformation, where the values ​​are mapped to a unified interval or dimension. This step effectively unifies the measurement standards of multi-source heterogeneous sensor data, eliminates interference from differences in dimensions between data, and provides a high-quality data foundation for subsequent accurate capture of flight status changes and environmental disturbance characteristics.

[0054] Step 102: Stack the time series data of each standard operating condition using a sliding time window to output a multidimensional disturbance sensing matrix.

[0055] In this embodiment, the sliding time window refers to a data extraction logic unit with a preset time span or number of sampling points that dynamically shifts along the time axis. It is used to continuously extract standard operating condition time-series data of the UAV within a recent historical period to capture the continuous evolution of flight status over time. Stacking refers to the data processing operation of orderly splicing and recombining data sequences collected by different sensors or data sources within the aforementioned time window according to a unified time index and feature dimensions, thereby constructing a structured data set containing temporal correlation information. The multidimensional disturbance perception matrix refers to a two-dimensional array or high-dimensional tensor generated based on the recombined data, capable of comprehensively characterizing the dynamic coupling features of UAV flight attitude and environmental disturbances in the temporal domain. It contains the temporal distribution patterns of key information such as sudden wind speed changes and attitude jitter. Through this step, discrete instantaneous observation values ​​can be transformed into an integrated matrix description with temporal depth and multidimensional features, effectively preserving dynamic trend information during strong wind disturbances. This provides a data foundation containing rich spatiotemporal correlation information for subsequent accurate identification of complex wind conditions and efficient anti-disturbance control, thereby improving the inspection UAV's ability to withstand strong winds.

[0056] Step 103: Determine the core feature set of the UAV based on the multidimensional disturbance perception matrix.

[0057] Step 104: Determine the category of surrounding wind intensity in the current environment based on the core feature set and the preset feature threshold set.

[0058] In this embodiment, the core feature set refers to a set of key quantitative indicators extracted from the multi-dimensional disturbance perception matrix, which characterize the degree of external interference experienced by the UAV within a specific time period. This set specifically includes parameters such as the maximum attitude change rate (reflecting the degree of drastic fluctuation in the aircraft's attitude angle within a short period), the flight path deviation acceleration (reflecting the rate of change of the flight trajectory deviating from the preset flight path), and the wind speed gradient change rate (reflecting the degree of drastic differences in the spatial distribution of surrounding airflow speed). The preset feature threshold set refers to a critical numerical reference system pre-set and stored in the control system, used to define the safety boundaries of flight status or the severity of meteorological conditions. It includes standard judgment limits such as the attitude change rate threshold, flight path deviation acceleration threshold, and wind speed gradient change rate threshold, each corresponding to one of the aforementioned key indicators. The surrounding wind intensity category refers to the qualitative classification result, determined by the system through logical operations by comparing and analyzing the real-time calculated values ​​in the core feature set with the preset feature threshold set, describing the threat level of the current meteorological environment to flight stability, such as classifying it as a strong wind state or a normal wind state. This step enables the rapid extraction of the most representative disturbance features from high-dimensional and complex sensing data. Through a multi-parameter joint verification mechanism, it achieves accurate identification and classification of extreme weather conditions such as sudden strong winds, avoiding false alarms or missed alarms that may be caused by a single indicator. This provides an accurate decision-making basis for timely triggering of targeted wind-resistant control strategies, thereby improving the wind-resistant capability of inspection drones.

[0059] Step 105: If the surrounding wind intensity category is strong wind, then determine the actual disturbance torque vector of the UAV based on the core feature set and the preset torque data table; the preset torque data table stores the mapping relationship between the core features and the disturbance torque.

[0060] In this embodiment, strong wind refers to a specific state within the surrounding wind intensity category that characterizes a severe meteorological environment and poses a significant threat to flight safety. Specifically, it manifests as any index value in the core feature set exceeding its corresponding preset safety limit. The preset torque data table refers to a structured data set pre-built and stored in the system memory. It establishes a digital mapping relationship between the values ​​of each index in the core feature set and the aerodynamic torque values ​​(disturbance torque) acting on the aircraft axis through offline experiments or simulation calculations. The actual disturbance torque vector refers to a physical quantity vector that specifically quantifies the three-dimensional torsional force generated by the external strong airflow on the UAV body at the current moment, obtained by retrieving or interpolating the aforementioned data table based on the currently monitored core feature data. This vector typically includes components in three dimensions: roll, pitch, and yaw. Through this step, abstract sensor feature readings can be transformed into specific physical torque values, realizing the digital quantification of the invisible wind field interference force. This provides a direct physical model input for subsequent calculation of accurate wind resistance compensation commands, thereby improving the inspection UAV's ability to withstand strong winds.

[0061] Step 106: Determine the feedforward flight compensation vector of the UAV based on the preset dynamic model and the actual disturbance torque vector.

[0062] Step 107: Generate the feedback correction vector of the UAV based on the multidimensional disturbance perception matrix, the preset basic neural network model, the preset expected disturbance torque vector, the core feature set, the actual disturbance torque vector, and the feedback control algorithm.

[0063] In this embodiment, the preset dynamics model refers to a set of mathematical relationships pre-established and stored in the system that describes the motion response of the UAV when subjected to external torque. In the specific implementation scenario of this application, the core of this model includes a feedforward gain matrix, which is a parameter array composed of a series of proportional coefficients used to directly and linearly map the input physical torque values ​​to the corresponding offset control quantities. The feedforward flight compensation vector refers to an open-loop control command vector generated by performing matrix operations on the actual disturbance torque vector using the aforementioned feedforward gain matrix, aiming to directly generate a reverse torque to balance external wind interference. The preset basic neural network model refers to a deep learning network structure pre-trained in a plain wind field environment, capable of mapping the input state to control parameters. It serves as an initial intelligent model for subsequent parameter fine-tuning or optimal parameter generation for specific environments. The preset expected disturbance torque vector refers to the set of theoretical resultant torque values ​​that the UAV should experience under ideal flight conditions or a predetermined trajectory, usually used as a benchmark target for measuring flight stability. Feedback control algorithms refer to classic feedback control logic that dynamically adjusts the system output by calculating the proportional, integral, and derivative terms of the deviation. These can be PID, PI, PD, or other algorithms based on error feedback. The feedback correction vector is a closed-loop control command vector generated based on the state deviation, used to correct residual attitude deviations, combining the adaptive adjustment capability of neural networks with the error elimination mechanism of feedback control algorithms. Through this step, a composite control system combining feedforward prediction and feedback correction can be constructed. The feedforward stage quickly suppresses the initial impact of sudden strong winds, the feedback stage accurately eliminates accumulated errors, and the neural network enhances the system's adaptability to complex nonlinear wind fields, thereby ensuring flight attitude stability and improving the inspection UAV's resistance to strong winds.

[0064] Step 108: Control the UAV flight based on the feedforward flight compensation vector and the feedback correction vector.

[0065] In this embodiment, a total control vector can be generated based on the feedforward flight compensation vector and the feedback correction vector. Based on the superposition principle, this vector is a final set of execution signals containing multiple dimensions of control commands, including roll, pitch, and yaw, generated by linearly combining the feedforward flight compensation vector (aimed at offsetting initial disturbances) and the feedback correction vector (aimed at eliminating residual deviations). Controlling the UAV's flight involves decoding the total control vector into specific motor speed adjustment signals or control surface deflection angle signals, driving the UAV's power system and actuators to perform corresponding wind-resistant maneuvers. This step fully integrates the advantages of fast response speed of feedforward control and high steady-state accuracy of feedback control, enabling rapid response and precise maintenance of flight attitude under sudden strong wind interference. It effectively solves the problem that a single control strategy cannot simultaneously address dynamic performance and steady-state error, thus ensuring the operational safety and trajectory maintenance capability of the UAV under adverse weather conditions, and ultimately improving the inspection UAV's resistance to strong winds.

[0066] The above technical solution first utilizes standardized processing and sliding time window stacking to transform the multi-source heterogeneous instantaneous operating data of the UAV into a multi-dimensional disturbance perception matrix containing temporal dynamic laws, thereby accurately capturing the evolution trend of environmental disturbances over time. Next, based on the comparative analysis of the core feature set and the preset feature threshold set, accurate identification of the surrounding wind intensity category is achieved. Furthermore, through a preset torque data table, the abstract strong wind disturbance is further quantified into a specific actual disturbance torque vector, providing precise physical input for the control system. On this basis, a composite control strategy of "feedforward + feedback" is adopted. On the one hand, it utilizes a feedforward flight compensation vector based on a preset dynamic model to quickly respond to and offset the initial impact of sudden strong winds. On the other hand, it uses a feedback control algorithm combined with an online fine-tuning neural network to generate a feedback correction vector, and adaptively optimizes control parameters according to real-time operating conditions to eliminate residual deviations. Finally, the UAV is driven by the total control vector generated by fusing the feedforward and feedback, which takes into account both the dynamic response speed against disturbances and the steady-state control accuracy. This effectively solves the problem of lag or poor adaptability of single control methods in complex and variable wind fields, significantly enhances the stability of flight attitude, and thus improves the inspection UAV's ability to resist strong winds.

[0067] In this embodiment, the core feature set includes the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate. The step of generating the UAV's feedback correction vector based on the multidimensional disturbance perception matrix, the preset basic neural network model, the preset expected disturbance moment vector, the core feature set, the actual disturbance moment vector, and the feedback control algorithm includes: adjusting the preset basic neural network model according to the multidimensional disturbance perception matrix and minimizing the training loss function, and outputting the adjusted neural network model; determining the UAV's state deviation according to the preset expected disturbance moment vector and the actual disturbance moment vector; inputting the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate into the neural network model, and outputting the optimal feedback control parameters under the current environment; generating an adaptive function according to the optimal feedback control parameters and the feedback control algorithm; and generating the UAV's feedback correction vector according to the state deviation, the adaptive function, and the actual disturbance moment vector.

[0068] In this embodiment, adjusting the preset basic neural network model refers to a transfer learning process that uses a multi-dimensional disturbance perception matrix to minimize the training loss function through an optimization algorithm, and then updates and corrects the internal weight parameters of the preset basic neural network model online. This aims to enable the model to quickly adapt from a general state and converge to the current specific weather and flight environment. The adjusted neural network model refers to a dedicated deep learning network obtained after the above-mentioned online adaptive training, possessing the ability to perform accurate reasoning based on the characteristics of the current environment. It establishes a nonlinear mapping relationship between core features such as the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate, and control parameters. State deviation refers to the error signal quantified by calculating the vector difference between the preset expected disturbance moment vector and the actual disturbance moment vector, representing the degree to which the current force state of the UAV deviates from the ideal equilibrium state. Optimal feedback control parameters refer to the proportional, integral, and differential gain values ​​specifically adapted to the current wind field disturbance characteristics, calculated in real time by the fine-tuned neural network model based on the input core feature set. The adaptive function refers to the specific mathematical expression formed by loading the above-mentioned optimal feedback control parameters into a standard feedback control algorithm, capable of dynamically adjusting the control law and response characteristics according to environmental changes. This step enables the control system to have environmental perception and self-learning capabilities through online fine-tuning technology, transforming fixed parameter control into adaptive control that evolves in real time with environmental characteristics. This ensures that when facing complex and ever-changing mountain wind fields, the system can accurately generate feedback correction vectors based on instantaneous state deviations, effectively eliminating the impact of environmental uncertainties on flight stability and thus improving the inspection drone's ability to withstand strong winds.

[0069] In this embodiment of the application, determining the core feature set of the UAV based on the multidimensional perturbation perception matrix includes: determining the maximum attitude change rate according to the following formula:

[0070]

[0071] The course deviation acceleration is determined using the following formula:

[0072]

[0073] The rate of change of wind speed gradient is determined using the following formula:

[0074]

[0075] in, For the first The first moment attitude angle; This refers to the sensor sampling period for the drone. For drones in the The distance of the flight path deviation at any given time; For drones in the Wind speed at any given moment; This represents the maximum attitude change rate. This refers to the acceleration due to the deviation from the flight path. The rate of change of wind speed gradient; This represents the total number of samples within the sliding time window.

[0076] In one embodiment, for example, the window size is set to process data from the past 10 sampling points each time, then =10.

[0077] In this embodiment of the application, generating the UAV's feedback correction vector based on the state deviation, the adaptive function, and the actual disturbance moment vector includes determining the feedback correction vector according to the following formula:

[0078]

[0079] in, For feedback correction vector; It is an adaptive function; This is the actual disturbance moment vector; For the first State deviation at any given moment.

[0080] In this embodiment of the application, the method further includes: acquiring the set of deployment coordinate points of the transmission line to be inspected and the UAV in three-dimensional space and the current position coordinates; determining the shortest distance between the UAV and the transmission line to be inspected based on the current position coordinates and the set of deployment coordinate points; determining that the shortest distance is less than the sum of a preset safety boundary warning threshold and a preset safety inspection zone radius; constructing an avoidance path objective function based on multiple maneuvering parameters of the UAV, the constraints corresponding to each maneuvering parameter, and the shortest distance; determining the optimal values ​​of each of the multiple maneuvering parameters with the aim of minimizing the function value of the avoidance path objective function; and controlling the UAV to fly according to the optimal values ​​of each of the multiple maneuvering parameters.

[0081] In this embodiment, the power transmission line to be inspected refers to the target object of the UAV operation, namely, the high-voltage power transmission lines and towers erected in complex terrain environments that require status monitoring. The set of coordinate points refers to a series of geometric coordinate data that precisely describes the specific direction, suspension shape, and spatial distribution of the line in three-dimensional geographic space in digital form. The current position coordinates refer to the three-dimensional vector data of the UAV's own centroid in the same spatial coordinate system, calculated in real time by the UAV using its onboard high-precision positioning module. The shortest distance refers to the straight-line Euclidean distance from the UAV's current position to the nearest point on the transmission line trajectory, calculated based on spatial geometric algorithms. The preset safety boundary warning threshold refers to an additional distance buffer value pre-set to trigger emergency avoidance mechanisms in response to potential position drift or control response delays caused by sudden strong winds. The preset safe inspection zone radius refers to the minimum safe space limit allowed to approach energized equipment during normal inspection operations, determined based on power industry safety operating procedures and the UAV's own positioning error. Multiple maneuver parameters refer to the kinematic control variables that directly determine the UAV's flight attitude and trajectory changes, specifically covering roll rate, pitch rate, and yaw rate. Constraints refer to the upper and lower limits of the numerical range that each maneuvering parameter must strictly adhere to when performing actions, limited by the drone's motor power limits, airframe structural strength, and aerodynamic characteristics. The avoidance path objective function is a mathematical optimization model designed to balance avoidance efficiency and flight stability. It transforms the requirement to maximize safety by moving away from the route and minimize the need for aggressive maneuvering into a quantifiable cost, typically expressed as a weighted combination of the reciprocal of distance and the amplitude of maneuvering parameters. The optimal value refers to the set of best maneuvering parameters determined by an optimization algorithm that minimizes the objective function value while strictly satisfying the physical constraints. This step enables the forced triggering of a path planning mechanism based on physical limits in extreme emergencies where conventional wind-resistant compensation control cannot effectively maintain a safe distance. It automatically generates an escape trajectory that balances rapid escape from the hazard source with maintaining a stable attitude, providing a final safety net when strong winds cause the drone to approach the route, thereby improving the inspection drone's ability to withstand strong winds.

[0082] In this embodiment, the preset feature threshold set includes an attitude change rate threshold, a flight path deviation acceleration threshold, and a wind speed gradient change rate threshold. Determining the surrounding wind intensity category based on the core feature set and the preset feature threshold set includes: determining the surrounding wind intensity category according to the following formula:

[0083]

[0084] in, When the value is 1, the surrounding wind intensity category is determined to be strong wind; When the value is 0, the surrounding wind intensity category is determined to be conventional wind; This is the threshold for the attitude change rate; The flight path deviation acceleration threshold; The threshold for the rate of change of wind speed gradient; For logical OR.

[0085] In this embodiment, multiple maneuver parameters include roll rate, pitch rate, and yaw rate; the avoidance path objective function is constructed based on the multiple maneuver parameters of the UAV, the constraints corresponding to each maneuver parameter, and the shortest distance, including: the avoidance path objective function satisfies the following formula:

[0086]

[0087] The constraint conditions for each kinematic parameter satisfy the following formula:

[0088]

[0089] in, It is the roll angular velocity; This is the minimum value among the constraints corresponding to the roll angular velocity; This is the maximum value among the constraints corresponding to the roll angular velocity; It is the pitch angular velocity; This is the minimum value among the constraints corresponding to the pitch angular velocity; This represents the maximum value among the constraints corresponding to the pitch angular velocity; Yaw angular velocity; This is the minimum value among the constraints corresponding to the yaw rate; This represents the maximum value among the constraints corresponding to the yaw rate. In the first The shortest distance at any given moment; This is the safety weighting coefficient; To stabilize the weighting coefficients; To avoid the function value of the objective function of the path.

[0090] The above technical solution first utilizes standardized processing and sliding time window stacking to transform the multi-source heterogeneous instantaneous operating data of the UAV into a multi-dimensional disturbance perception matrix containing temporal dynamic laws, thereby accurately capturing the evolution trend of environmental disturbances over time. Next, based on the comparative analysis of the core feature set and the preset feature threshold set, accurate identification of the surrounding wind intensity category is achieved. Furthermore, through a preset torque data table, the abstract strong wind disturbance is further quantified into a specific actual disturbance torque vector, providing precise physical input for the control system. On this basis, a composite control strategy of "feedforward + feedback" is adopted. On the one hand, it utilizes a feedforward flight compensation vector based on a preset dynamic model to quickly respond to and offset the initial impact of sudden strong winds. On the other hand, it uses a feedback control algorithm combined with an online fine-tuning neural network to generate a feedback correction vector, and adaptively optimizes control parameters according to real-time operating conditions to eliminate residual deviations. Finally, the UAV is driven by the total control vector generated by fusing the feedforward and feedback, which takes into account both the dynamic response speed against disturbances and the steady-state control accuracy. This effectively solves the problem of lag or poor adaptability of single control methods in complex and variable wind fields, significantly enhances the stability of flight attitude, and thus improves the inspection UAV's ability to resist strong winds.

[0091] Figure 2 This illustration schematically depicts another flight control method for an inspection drone according to an embodiment of this application. For example... Figure 2 As shown in the embodiment of this application, firstly, the sensors of the UAV are controlled to collaboratively collect multiple operating condition time-series data in the current environment, and wind disturbance features (i.e., core feature set) including attitude change rate, flight path deviation acceleration, and wind speed gradient change rate are extracted in real time. Subsequently, feedforward compensation of the UAV is performed based on the quantized disturbance moment vector, and feedback correction of the UAV is performed in combination with feedback control algorithm and neural network model. During this process, the system adaptively optimizes the compensation parameters (i.e., adjusts the basic neural network model). If the parameter calculation is inaccurate, it is iteratively optimized until the optimal feedback control parameters are obtained. When the parameters are accurate, the system reads the image information of the transmission line TL to be inspected to construct three-dimensional layout coordinates, calculates the safe area including the radius of the preset safe inspection area, and calculates the shortest distance between the flight path of the UAV and the line in real time. Finally, it is determined whether the flight path is close to the safety boundary (i.e., whether the shortest distance is less than the warning threshold). If it is not close, the current judgment ends and normal flight is maintained. If it is determined to be close to the safety boundary, emergency attitude control of the UAV is immediately executed and flight path replanning is initiated to generate the optimal avoidance trajectory to ensure that the UAV stays away from the transmission line.

[0092] The following is an example of this application:

[0093] 1. Sensing sudden strong wind disturbances:

[0094] To address the characteristics of sudden strong winds during transmission line (TL) inspections, such as rapid increases in wind speed, abrupt changes in wind direction, and uncertain disturbances, a multi-source sensor fusion disturbance perception system for UAVs integrates the UAV's inertial measurement unit, barometer-built-in sensors, visual sensors, and an additional miniature ultrasonic anemometer to construct a multi-dimensional data acquisition architecture for the UAV. This architecture collects time-series data on multiple operating conditions of the UAV under the current environment. This data is then processed through feature extraction and pattern recognition to achieve real-time and accurate identification of strong wind disturbances. This overcomes the limitations of single sensors in terms of limited sensing dimensions and weak anti-interference capabilities, providing reliable data support for UAV strong wind disturbance compensation and control.

[0095] Different UAV sensors have different output dimensions (i.e., there are dimensional differences between time-series data under different operating conditions). Therefore, these data are first standardized to eliminate the influence of dimensions. The first time the data was collected The sensor (i.e. the first) Standard operating condition time series data (for similar operating conditions) for:

[0096]

[0097] in, For the first Time of the first The original values ​​of time-series data for similar operating conditions; and These are the extreme values ​​of this type of data.

[0098] By sliding time window (length is T The time series data for each standard operating condition are stacked to output a multi-dimensional disturbance sensing matrix. X :

[0099]

[0100] Where T is the total number of samples within the sliding time window; in the matrix Multi-dimensional sensing data corresponding to the same sampling time; By using time-series data corresponding to the same type of operating conditions, a comprehensive characterization of the drone's flight status and strong wind disturbances can be achieved.

[0101] The core feature set of the UAV is determined based on the multi-dimensional disturbance perception matrix to quantitatively characterize strong wind features. The core feature set includes the maximum attitude change rate, flight path deviation acceleration, and wind speed gradient change rate. The calculation logic is as follows:

[0102] In the extraction of key features for wind resistance of UAVs, the attitude change rate reflects the degree of change in the UAV attitude angle over time, and is the direct impact of strong winds on UAV disturbances. It is defined as the maximum value of the first derivative of the attitude angle within the sliding time window:

[0103]

[0104] Flight path deviation acceleration reflects the rate of change of the UAV's deviation from the preset flight path. It is defined as the second derivative of the flight path deviation and is determined by the following formula:

[0105]

[0106] The wind speed gradient change reflects the sudden rise and fall characteristics of wind speed in the power transmission line being inspected by the UAV. It is the mean of the first derivative of the wind speed around the UAV within the sliding time window. The rate of change of wind speed gradient is determined according to the following formula:

[0107]

[0108] in, For the first The first moment attitude angle; This refers to the sensor sampling period for the drone. For drones in the The distance of the flight path deviation at any given time; For drones in the Wind speed at any given moment; This represents the maximum attitude change rate. This represents the acceleration due to flight path deviation, with positive and negative values ​​indicating an increase or decrease in the deviation trend, respectively. The wind speed gradient change rate is positive, indicating an increase in wind speed, and negative, indicating a decrease in wind speed. The larger the absolute value, the more drastic the wind speed change around the drone.

[0109] Based on the core feature set and the preset feature threshold set (including , and Determine the category of surrounding wind intensity in the current environment. :

[0110]

[0111] in, When the value is 1, the surrounding wind intensity category is determined to be strong wind; When the value is 0, the surrounding wind intensity category is determined to be conventional wind; This is the threshold for the attitude change rate; The flight path deviation acceleration threshold; The threshold for the rate of change of wind speed gradient; The condition is a logical OR, and a sudden strong wind is considered when any of the above thresholds are met.

[0112] 2. Hybrid disturbance compensation control:

[0113] Hybrid disturbance compensation control achieves rapid cancellation and correction of sudden strong wind disturbances faced by UAVs through feedforward compensation, feedback correction, and transfer optimization. In feedforward compensation, the disturbance torque quantified in real time by the sensing module is used to quickly generate initial compensation commands through a pre-trained preset dynamic model, thus offsetting the wind disturbance before it affects the UAV's flight attitude. In feedback correction, the deviation between the UAV's actual attitude and flight path and the desired state is monitored in real time. An adaptive PID controller dynamically corrects deficiencies or deviations in feedforward compensation, handling compensation errors caused by wind field uncertainties. In transfer optimization, to address the issue of UAV control model failure in wind fields with unfamiliar transmission lines, a model (pre-set basic neural network model) is trained on large datasets of plain wind fields and then rapidly fine-tuned using a small amount of measured data (time-series data from multiple operating conditions) in the current scenario. This allows the model to learn the complex mapping relationship between disturbance characteristics and optimal compensation parameters, and to optimize the initial commands of the feedforward model and the PID parameters of the feedback loop in real time, improving the model's adaptability to wind fields with unfamiliar transmission lines.

[0114] like If the value is 1, the surrounding wind intensity category is determined to be strong wind. At this point, the actual disturbance torque vector of the UAV is determined based on the core feature set and the preset torque data table (which stores the mapping relationship between core features and torques). :

[0115]

[0116] in, , and These are the torque components of the actual disturbance moment vector corresponding to the maximum attitude change rate, the route deviation acceleration, and the wind speed gradient change rate, respectively.

[0117] (1) Feedforward flight compensation:

[0118] The feedforward flight compensation vector of the UAV is determined based on the preset dynamic model (including the feedforward gain matrix) and the actual disturbance torque vector, so as to achieve rapid suppression of the main disturbance:

[0119]

[0120] in, This is the feedforward flight compensation vector; This is the feedforward gain matrix in the preset dynamics model.

[0121] (2) Feedback correction and parameter adaptation:

[0122] In the feedback phase, the feedback correction vector of the UAV is generated based on multiple operating condition time series data, a preset basic neural network model, a preset expected disturbance torque vector, a core feature set, an actual disturbance torque vector, and a feedback control algorithm.

[0123] First, transfer learning is performed on the neural network using operating condition time-series data. This is based on multiple operating condition time-series data and minimizing the training loss function. For the pre-defined basic neural network model Make adjustments and output the adjusted neural network model (parameters are...). This adapts the drone to its current environment (mountainous wind field environment):

[0124]

[0125]

[0126] Secondly, calculate the state deviation. This is based on the preset desired disturbance torque vector. and the actual disturbance moment vector Determine the state deviation of the UAV :

[0127]

[0128] in, For the first The actual disturbance moment vector of the UAV at any given moment; For the first The preset expected disturbance moment vector at any given time.

[0129] Then, an adaptive control law is generated. The maximum attitude change rate, the flight path deviation acceleration, and the wind speed gradient change rate (core feature set) are input into the fine-tuned neural network model, which outputs the optimal feedback control parameters for the current environment. Based on these parameters and the feedback control algorithm, an adaptive function is generated. .

[0130] Finally, based on the state deviation Adaptive function and the actual disturbance moment vector Generate feedback correction vector for the UAV :

[0131]

[0132] (3) Generation of total control vector:

[0133] Based on the feedforward flight compensation vector and feedback correction vector Generate total control vector This information is used to control the drone's flight and ensure its stability in strong winds.

[0134]

[0135] 3. Dynamic evasion trajectory planning:

[0136] When hybrid compensation control is still unable to maintain a safe distance, the system will activate a dynamic avoidance mechanism. This mechanism monitors the relative position of the drone and the transmission line to be inspected in real time and automatically plans the optimal escape path when approaching the safety boundary.

[0137] Suppose there is at least one transmission line to be inspected, and its point in three-dimensional space is... , m This represents the total number of points in three-dimensional space representing multiple transmission lines to be inspected. Each point is composed of vertical, horizontal, and depth coordinates. The set of coordinate points for the target transmission line to be inspected in three-dimensional space is: .

[0138] (1) Calculation of safe zone and distance:

[0139] Obtain the coordinate sets of the transmission lines to be inspected and the drones in three-dimensional space. and current position coordinates The radius of the preset safety inspection zone is set according to industry standards. The shortest distance between the UAV and the power transmission line to be inspected is determined based on the current location coordinates and the set of deployed coordinate points. :

[0140]

[0141]

[0142] (2) Path avoidance generation:

[0143] If it is determined that the shortest distance is less than the preset safety boundary warning threshold radius of the preset safety inspection zone The sum of these factors, i.e., when the following formula is satisfied, indicates that an emergency evasive maneuver is triggered:

[0144]

[0145] At this time, based on multiple maneuver parameters of the drone ( The objective function for constructing the evasion path is based on the constraints and shortest distance for each maneuver parameter. :

[0146]

[0147] The constraint conditions for each kinematic parameter satisfy the following formula:

[0148]

[0149] in, It is the roll angular velocity; This is the minimum value among the constraints corresponding to the roll angular velocity; This is the maximum value among the constraints corresponding to the roll angular velocity; It is the pitch angular velocity; This is the minimum value among the constraints corresponding to the pitch angular velocity; This represents the maximum value among the constraints corresponding to the pitch angular velocity; Yaw angular velocity; This is the minimum value among the constraints corresponding to the yaw rate; This represents the maximum value among the constraints corresponding to the yaw rate. In the first The shortest distance at any given moment; This is the safety weighting coefficient; To stabilize the weighting coefficients; To avoid the function value of the objective function of the path.

[0150] With the aim of minimizing the function value of the objective function of the avoidance path, the optimal values ​​of multiple maneuvering parameters are determined; the UAV is controlled to fly according to the optimal values ​​of the multiple maneuvering parameters, so as to move away from the power transmission line in the fastest and most stable way while ensuring that the maneuvering performance does not exceed the limit.

[0151] Figure 3 A schematic diagram illustrating the structure of a computer device according to an embodiment of this application is provided. Figure 3 As shown, this application also provides a computer device, including:

[0152] Memory 310 is configured to store instructions;

[0153] The processor 320 is configured to retrieve instructions from memory 310 and to implement the methods described above when executing instructions.

[0154] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0155] This application also provides a machine-readable storage medium storing instructions that cause a machine to perform the above-described method.

[0156] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0158] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0159] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0160] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0161] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0162] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0163] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0164] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A flight control method for an inspection drone, characterized in that, include: In response to receiving multiple operating condition time series data of the UAV in the current environment, the system standardizes each operating condition time series data and outputs the corresponding standard operating condition time series data. By stacking time series data for each standard operating condition through a sliding time window, a multidimensional disturbance sensing matrix is ​​output. The core feature set of the UAV is determined based on the multidimensional disturbance perception matrix; the core feature set includes the maximum attitude change rate, the flight path deviation acceleration, and the wind speed gradient change rate. The surrounding wind intensity category in the current environment is determined based on the core feature set and the preset feature threshold set; If the surrounding wind intensity category is strong wind, the actual disturbance torque vector of the UAV is determined according to the core feature set and the preset torque data table; the preset torque data table stores the mapping relationship between core features and disturbance torque. The feedforward flight compensation vector of the UAV is determined based on the preset dynamics model and the actual disturbance moment vector. The preset basic neural network model is adjusted based on the multidimensional perturbation sensing matrix and the minimized training loss function, and the adjusted neural network model is output. The state deviation of the UAV is determined based on the preset expected disturbance moment vector and the actual disturbance moment vector. The maximum attitude change rate, the flight path deviation acceleration, and the wind speed gradient change rate are input into the adjusted neural network model, and the optimal feedback control parameters under the current environment are output. An adaptive function is generated based on the optimal feedback control parameters and the feedback control algorithm. The feedback correction vector of the UAV is generated based on the state deviation, the adaptive function, and the actual disturbance moment vector; The UAV is controlled to fly according to the feedforward flight compensation vector and the feedback correction vector; The feedback correction vector is determined according to the following formula: in, The feedback correction vector; The adaptive function is... This refers to the actual disturbance torque vector; For the first State deviation at any given moment.

2. The method according to claim 1, characterized in that, The process of determining the core feature set of the UAV based on the multidimensional perturbation perception matrix includes: The maximum attitude mutation rate is determined according to the following formula: The trajectory deviation acceleration is determined according to the following formula: The rate of change of the wind speed gradient is determined according to the following formula: in, For the first The first moment attitude angle; The sensor sampling period of the UAV; For the drone in the first The distance of the flight path deviation at any given time; For the drone in the first Wind speed at any given moment; This represents the maximum value of the attitude change rate; The acceleration due to the deviation of the flight path; The wind speed gradient change rate; The total number of samples within the sliding time window.

3. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain the set of coordinate points of the power transmission line to be inspected and the UAV in three-dimensional space, as well as their current position coordinates; The shortest distance between the UAV and the power transmission line to be inspected is determined based on the current location coordinates and the set of deployed coordinate points. The shortest distance is determined to be less than the sum of a preset safety boundary warning threshold and a preset safety inspection zone radius; An evasion path objective function is constructed based on multiple maneuver parameters of the UAV, the constraints corresponding to each maneuver parameter, and the shortest distance. With the aim of minimizing the function value of the objective function of the evasion path, the optimal values ​​of each of the multiple maneuver parameters are determined; The drone is controlled to fly according to the optimal values ​​of each of the multiple maneuvering parameters.

4. The method according to claim 3, characterized in that, The preset feature threshold set includes attitude change rate threshold, flight path deviation acceleration threshold, and wind speed gradient change rate threshold. The step of determining the surrounding wind intensity category in the current environment based on the core feature set and the preset feature threshold set includes: The surrounding wind intensity category is determined according to the following formula: in, When the value is 1, the surrounding wind intensity category is determined to be strong wind; When the value is 0, the surrounding wind intensity category is determined to be conventional wind; The attitude mutation rate threshold; The acceleration threshold for the flight path deviation; The threshold value for the rate of change of the wind speed gradient; Logical OR; This represents the maximum value of the attitude change rate; The acceleration due to the deviation of the flight path; The wind speed gradient change rate is given.

5. The method according to claim 4, characterized in that, The multiple maneuvering parameters include roll rate, pitch rate, and yaw rate; The step of constructing the evasion path objective function based on multiple maneuver parameters of the UAV, the constraints corresponding to each maneuver parameter, and the shortest distance includes: The objective function for the avoidance path satisfies the following formula: The constraint conditions corresponding to each kinematic parameter satisfy the following formula: in, The roll angular velocity; It is the minimum value among the constraints corresponding to the roll angular velocity; The maximum value among the constraints corresponding to the roll angular velocity; The pitch angular velocity; It is the minimum value among the constraints corresponding to the pitch angular velocity; The maximum value among the constraints corresponding to the pitch angular velocity; The yaw rate is mentioned. It is the minimum value among the constraints corresponding to the yaw angular velocity; The maximum value among the constraints corresponding to the yaw angular velocity; In the first The shortest distance at any given moment; This is the safety weighting coefficient; To stabilize the weighting coefficients; Let be the function value of the objective function for the evasion path.

6. A computer device, characterized in that, include: The memory is configured to store instructions; And a processor configured to retrieve the instructions from the memory and, when executing the instructions, to implement the method according to any one of claims 1 to 5.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method according to any one of claims 1 to 5.

8. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform the method according to any one of claims 1 to 5.

Images