Self-balancing control method, device and equipment for unmanned aerial vehicle to flush insulators and medium

Abstract

This application provides a self-balancing control method, apparatus, device, and medium for washing insulators by a drone. The method includes: acquiring attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data; calculating the attitude deviation based on the attitude data and a preset ideal attitude; generating a feedback control command based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data; generating a predicted attitude change based on the water flow pressure data, the water flow direction data, and the ambient wind speed data; and calculating a final control command based on the feedback control command and the predicted attitude change. This application achieves precise, proactive, and coordinated self-balancing control of the drone's attitude by combining the feedback control command and the predicted attitude change to calculate the final control command.

Description

Technical Field

[0001] This application belongs to the field of insulator cleaning, and in particular relates to a self-balancing control method, device, equipment and medium for unmanned aerial vehicle (UAV) washing insulators. Background Technology

[0002] With the improvement of the intelligent operation and maintenance level of high-voltage transmission lines, UAV-borne high-pressure water flushing technology has become a preferred solution for insulator cleaning. Its core requirement is to achieve stable control of the UAV's flight attitude during high-pressure water flushing operations, so as to maintain the optimal cleaning distance and angle with the insulators and ensure operational safety and cleaning effect.

[0003] Existing technologies can be mainly divided into two categories. One category is general-purpose UAV attitude control technology, which specifically involves: relying on an inertial measurement unit (IMU) to collect the UAV's attitude data (such as pitch and roll angles), and employing a traditional fixed-parameter PID (proportional-integral-derivative) control algorithm. Control commands are calculated based on the deviation between attitude sensor data and set values ​​to drive motors and servos, thereby attempting to maintain the aircraft's basic attitude stability. The other category is specialized control technology for rinsing scenarios. Building upon general-purpose technologies, this category adds the application of water pressure sensors to attempt simple compensation for the reaction force generated by high-pressure water jets. Through these methods, existing technologies can achieve a certain level of self-balancing control in relatively ideal, static environments.

[0004] During the research and practice of existing technologies, it was found that when the above-mentioned technical solutions are controlled by "relying on a single or a few sensors and using a fixed parameter PID algorithm", the single source of sensing information fails to effectively integrate multi-dimensional data such as high-altitude environmental wind speed, water flow direction, and visual information of insulator position. As a result, the system cannot fully perceive various dynamic interference factors that affect the balance of the UAV, and cannot predict the risk of attitude deviation in advance. It can only make passive and delayed balance adjustments.

[0005] Meanwhile, because its control algorithm parameters are fixed, it cannot make adaptive dynamic adjustments according to the real-time changing intensity of interference, resulting in slow control response speed, difficulty in real-time cancellation of dynamic interference, and insufficient balance accuracy.

[0006] More importantly, the self-balancing control logic of the existing technology operates independently from the high-pressure water flushing operation, without establishing a coordinated linkage mechanism. When adjusting the flushing pressure or water flow direction, it is impossible to adjust the attitude control command synchronously and predictively, which can easily lead to water flow deviation, cleaning blind spots, or the need to suspend the flushing operation to avoid serious attitude instability, which seriously affects the cleaning efficiency and effect. Summary of the Invention

[0007] The purpose of this application is to overcome the deficiencies in the prior art and provide a self-balancing control method, device, equipment and medium for unmanned aerial vehicle (UAV) flushing insulators.

[0008] This application provides a self-balancing control method for unmanned aerial vehicle (UAV) flushing insulators, including:

[0009] Acquire attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data;

[0010] Based on the posture data and the preset ideal posture, the posture deviation is calculated;

[0011] Based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data, a feedback control command is generated.

[0012] Based on the water flow pressure data, the water flow direction data, and the ambient wind speed data, the expected attitude change is generated;

[0013] The final control command is calculated based on the feedback control command and the expected attitude change.

[0014] Optionally, before acquiring attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data, the following steps are also included:

[0015] The weight of the drone and the water pressure and flow rate data of the high-pressure water washing system are obtained.

[0016] The load characteristic coefficient is determined based on the fuselage weight, the water pressure data, and the water flow rate data.

[0017] The load characteristic coefficient is associated with the calculation of the water flow reaction force data and / or the generation of the feedback control command.

[0018] Optionally, the steps for obtaining the water flow reaction force data include:

[0019] Data on the first water flow reaction force is obtained using a force sensor;

[0020] Calculate the second water flow reaction force data based on the water flow pressure data, water flow rate data, and load characteristic coefficient;

[0021] The water flow reaction force data is determined based on the first water flow reaction force data and the second water flow reaction force data.

[0022] Optionally, based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data, a feedback control command is generated, including:

[0023] Obtain the rate of change of the attitude deviation;

[0024] The feedback control command is generated by the adaptive fuzzy PID controller based on the attitude deviation, the rate of change of the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

[0025] Optionally, based on the water flow pressure data, the water flow direction data, and the ambient wind speed data, the expected attitude change is generated, including:

[0026] The expected attitude change is generated by the collaborative correlation model based on the water flow pressure data, the water flow direction data, and the environmental wind speed data.

[0027] The expected attitude change includes the expected pitch angle change and the expected roll angle change. The expected pitch angle change is associated with the water flow pressure data, the sine function value of the water flow direction data, and the ambient wind speed data. The expected roll angle change is associated with the water flow pressure data, the cosine function value of the water flow direction data, and the ambient wind speed data.

[0028] Optionally, calculating the final control command based on the feedback control command and the expected attitude change includes:

[0029] The expected attitude change is used as the feedforward compensation amount.

[0030] The final control command is calculated based on the feedback control command and the feedforward compensation amount.

[0031] Optionally, correcting the collaboration coefficients in the collaboration association model based on the real-time status information includes:

[0032] The prediction error of the collaborative association model is determined based on the real-time state information and the expected attitude change.

[0033] Based on the prediction error, the synergy coefficients in the synergy association model are corrected.

[0034] This application also provides a self-balancing control device for unmanned aerial vehicle (UAV) flushing insulators, comprising:

[0035] The acquisition module acquires attitude data, environmental wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data.

[0036] The deviation module calculates the attitude deviation based on the attitude data and the preset ideal attitude;

[0037] The instruction module generates feedback control instructions based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

[0038] The prediction module generates the expected attitude change based on the water flow pressure data, the water flow direction data, and the ambient wind speed data.

[0039] The control module calculates the final control command based on the feedback control command and the expected attitude change.

[0040] This application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0041] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the above-described method.

[0042] The beneficial effects of this application are:

[0043] This application provides a self-balancing control method for washing insulators of a drone, comprising: acquiring attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data; calculating attitude deviation based on the attitude data and a preset ideal attitude; generating feedback control commands based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data; generating an expected attitude change based on the water flow pressure data, the water flow direction data, and the ambient wind speed data; and calculating a final control command based on the feedback control command and the expected attitude change. This application achieves precise, proactive, and coordinated self-balancing control of the drone's attitude by comprehensively calculating attitude deviation from multiple sources, generating feedback control commands, and generating an expected attitude change based on washing parameters and ambient wind speed, and finally combining these two methods to calculate the final control command. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the self-balancing control process for unmanned aerial vehicle (UAV) flushing insulators in this application;

[0045] Figure 2 This is a schematic diagram of the self-balancing control device for washing insulators by a drone in this application. Detailed Implementation

[0046] Exemplary embodiments of the present disclosure will now be provided in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, the embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0047] This application provides a self-balancing control method for washing insulators by drones, which is applied in the fields of drone control and power system operation and maintenance. It is suitable for drone-based autonomous high-pressure water washing operations of transmission line insulators and is used to solve the technical problems of low attitude deviation control accuracy, lack of collaborative control, lag in control response, and poor load adaptability caused by insufficient sensor sensing range, limited control algorithm, neglect of coordination in technical design, and lack of load adaptation mechanism during drone-borne high-pressure water washing operations of insulators.

[0048] Please refer to Figure 1 As shown, the self-balancing control method for unmanned aerial vehicle (UAV) flushing insulators described in this application includes:

[0049] S101. Acquire attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data.

[0050] The multi-source sensing module works collaboratively with multiple types of sensors to collect UAV attitude data, environmental interference data, flushing operation data, and insulator position data. It establishes a multi-dimensional data association model to provide data support for self-balancing control and collaborative decision-making, enabling comprehensive perception and prediction of interference factors.

[0051] Attitude data is acquired through a high-precision inertial measurement unit, GPS, gyroscope, and electronic compass. The drone's pitch angle, roll angle, yaw angle, angular velocity, acceleration, and position coordinates are collected in real time. At the same time, the speed of each motor and the angle of the servo motor are collected to obtain the drone's real-time motion status.

[0052] Environmental wind speed data is collected through miniature wind speed sensors and barometers, and high-altitude wind speed, wind direction, and atmospheric pressure are collected in real time to analyze the impact of wind disturbance on the attitude of the UAV.

[0053] Water pressure and flow rate data are obtained through operational information collection.

[0054] The steps for obtaining water flow reaction force data include:

[0055] 1. Obtain the data of the first water flow reaction force through a force sensor;

[0056] 2. Calculate the second water flow reaction force data based on the water flow pressure data, water flow rate data, and load characteristic coefficient.

[0057] The operation information collection also uses electromagnetic sensors to collect the intensity of electromagnetic interference around the insulator.

[0058] Water flow direction data was also obtained through operational information collection.

[0059] Visual information acquisition uses a high-definition camera and a laser rangefinder to collect the insulator position coordinates in real time, which is used to adjust the drone's attitude and ensure the optimal cleaning distance and angle.

[0060] After obtaining the raw data, data preprocessing is required, including data cleaning and time-series calibration, to eliminate noise interference and ensure data consistency and timeliness.

[0061] Data cleaning addresses the issues of excessive noise, disordered timing, and fluctuating accuracy in collected data.

[0062] The preprocessing process is executed, and the 3σ principle is used to identify and remove invalid data. Let a certain type of data sample set be... Calculate the mean:

[0063]

[0064] Standard deviation:

[0065]

[0066] Where X is a sample set of attitude data, wind speed data, or water flow pressure data. Let be the i-th single sample in the sample set, n be the total number of samples in the sample set, μ be the arithmetic mean of the sample set, σ be the standard deviation of the sample set, and 3σ be the outlier threshold.

[0067] like Outliers are identified and removed. At the same time, linear interpolation is used to supplement missing data, locate the missing data position, and confirm the valid data points before and after the missing point. Based on the reasonable assumption that the data changes linearly with the collection time, the time ratio of the missing point to the valid data points before and after it is calculated, and the numerical difference between the valid data before and after is allocated to obtain the missing point's supplementary value, thus ensuring data integrity.

[0068] Timing calibration is used to achieve time synchronization of multi-source data. Let the GPS master clock time be... The local timestamp of the i-th sensor is Define time deviation The control synchronization accuracy meets the following requirements:

[0069]

[0070] in This is the maximum permissible time deviation.

[0071] like This triggers sensor time recalibration, ensuring that multi-source data are fused and analyzed in the same time dimension.

[0072] S102. Calculate the attitude deviation based on the attitude data and the preset ideal attitude.

[0073] Preset ideal attitude parameters include pitch angle Roll angle Yaw angle These ideal angles are determined by the insulator position data obtained by the visual information acquisition module, with the aim of ensuring that the drone maintains the optimal cleaning distance and angle with the insulator.

[0074] The formula for calculating attitude deviation is:

[0075]

[0076] in This is the actual attitude data collected. For the ideal posture angle.

[0077] A threshold for attitude deviation is set. When the calculated deviation exceeds the threshold, subsequent self-balancing control commands will be triggered; when the deviation is below the threshold, the current control parameters will be maintained to keep the attitude stable.

[0078] S103. Generate feedback control commands based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

[0079] The generation of feedback control instructions includes acquiring the rate of change of the attitude deviation, and generating the feedback control instructions by an adaptive fuzzy PID controller based on the attitude deviation, the rate of change of the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

[0080] The generation of feedback control commands involves calculating the instructions to drive the UAV's motors and servos using a control algorithm. This process is specifically implemented through adaptive fuzzy PID control.

[0081] Adaptive fuzzy PID control dynamically adjusts the PID control parameters, including the proportional coefficient. Integral coefficient Differential coefficients This is to achieve rapid elimination of attitude deviations.

[0082] Fuzzy rule design is key to this controller. It takes attitude deviation, deviation change rate, water flow reaction force, and wind speed as inputs, and adjusts them using PID parameters. The output is designed with a fuzzy rule base to ensure that the parameters are adaptively adjusted under different working conditions.

[0083] When the attitude deviation is large, the rate of change of deviation is large, the water flow reaction force is large, and the wind speed is high, increase , reduce Increase This enables rapid response and strong anti-interference capabilities;

[0084] When the attitude deviation is small, the rate of change of deviation is small, the water flow reaction force is small, and the wind speed is low, reduce Increase , reduce This enables precise control and steady-state performance.

[0085] The formula for adjusting PID parameters is:

[0086]

[0087] in, These initial PID parameters are adaptively determined using the load characteristic parameter K. This is the parameter adjustment amount for the output of fuzzy inference.

[0088] Finally, based on the adjusted PID parameters and attitude deviation, feedback control commands are calculated, specifically the speed adjustment amounts of each motor of the UAV. and servo angle adjustment amount The calculation formula is:

[0089]

[0090] in, pitch angle deviation Integral over time, pitch angle deviation Similarly, the roll angle can be obtained by differentiating the time derivative. Relevant calculation formulas.

[0091] The meaning of execution control is to adjust the calculated motor speed. and servo angle adjustment amount The commands are converted into control commands and sent to the UAV motor controller and servo controller to drive the motor speed and servo angle adjustment, thereby achieving attitude balance.

[0092] At the same time, the adjusted attitude data is collected in real time and fed back to the attitude error calculation to ensure that the attitude deviation is always controlled within the threshold range.

[0093] S104. Based on the water flow pressure data, the water flow direction data, and the ambient wind speed data, generate the expected attitude change amount.

[0094] Generating the expected attitude change involves modeling and predicting how much the UAV's attitude will change when the flushing parameters are adjusted. This expected attitude change is generated by a collaborative correlation model based on the water pressure data, the water flow direction data, and the ambient wind speed data.

[0095] The collaborative correlation model establishes the mathematical relationship between flushing parameters and attitude balance parameters. The specific formula is as follows:

[0096]

[0097]

[0098] in, and In this model, the expected pitch angle change and the expected roll angle change are represented. For water flow pressure data, is the water flow direction data, v is the ambient wind speed data, and a, b, c, and d are the coordination coefficients, which are determined through training with experimental data.

[0099] The expected attitude changes include the expected pitch angle change and the expected roll angle change. The expected pitch angle change is a sinusoidal function of the water pressure data and the water direction data. And correlated with the environmental wind speed data, the expected roll angle change is correlated with the cosine function value of the water flow pressure data and the water flow direction data. And it is associated with the environmental wind speed data.

[0100] During the operation, the flushing parameters are adjusted automatically based on the insulator position data and the degree of contamination from the vision acquisition submodule. The adjusted parameters are then input into the model to generate the corresponding expected attitude change.

[0101] S105. Calculate the final control command based on the feedback control command and the expected attitude change.

[0102] The calculation of the final control command involves combining the expected attitude change with the feedback control command to generate a command that directly drives the actuator.

[0103] Specifically, the expected attitude change is used as the feedforward compensation amount, and then the final control command is calculated based on the feedback control command and the feedforward compensation amount.

[0104] The execution control unit converts the calculated motor speed adjustment and servo angle adjustment into specific control commands, which are then sent to the UAV motor controller and servo controller to drive the motor speed and servo angle adjustment, thereby achieving attitude balance.

[0105] Furthermore, before acquiring attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data, a load adaptive processing step is also included.

[0106] This step acquires the fuselage weight of the UAV, the water pressure data and water flow rate data of the high-pressure water flushing system, and determines the load characteristic coefficient based on these data. The load characteristic coefficient is then associated with the calculation of the water flow reaction force data and / or the generation of the feedback control command.

[0107] This process is executed by the load adaptive analysis module, which is deployed on the UAV's airborne controller. It uses a load characteristic identification algorithm to automatically identify the load characteristic parameters corresponding to the UAV model and flushing pressure, establish a load-reaction force correlation model, and achieve adaptive matching of control parameters without manual debugging.

[0108] The specific meaning of load characteristic identification is that the system automatically identifies the load characteristic parameters corresponding to the current drone model and flushing pressure. This is achieved by building a load characteristic sample library in advance. The sample library contains load parameters such as motor output power, fuselage weight, and center of gravity position for different drone models and flushing pressures, and the samples cover common operating scenarios.

[0109] During recognition, let the input feature vector be... ,in Let Q be the water pressure, Q be the water flow rate, and m be the weight of the UAV fuselage.

[0110] This feature vector is mapped to load characteristic categories using a support vector machine algorithm, outputting load characteristic parameters:

[0111]

[0112] in, For the classification hyperplane weight vector, is the feature mapping function, and b is the bias term. These parameters are determined through training with a sample database.

[0113] The larger the load factor K is, the greater the impact of the load on the attitude.

[0114] After determining the load factor K, the water flow reaction force is calculated based on the water pressure and flow rate data, providing a basis for reaction force compensation for self-balancing control. The total water flow reaction force is:

[0115]

[0116] in, The density of water (1000) Q is the water flow rate ( ), The water outlet velocity ( ), due to water flow pressure Calculation K is the load factor.

[0117] The direction of the water flow reaction force is opposite to the direction of the water flow outlet, and it can be decomposed into a component in the pitch direction. and the component of the roll direction This is used for adaptive PID control calculations in the subsequent self-balancing control module.

[0118] Furthermore, the water flow reaction force data acquisition step includes acquiring first water flow reaction force data through a force sensor, calculating second water flow reaction force data based on the water flow pressure data, water flow rate data, and load characteristic coefficient, and then determining the water flow reaction force data based on the first water flow reaction force data and the second water flow reaction force data.

[0119] Furthermore, the method also includes a step of correcting the coordination coefficients in the coordinated correlation model based on real-time status information. This step is executed by the control feedback optimization module, forming a closed-loop control system of acquisition, control, feedback, and optimization, continuously improving the self-balancing control performance.

[0120] The modified synergy coefficient refers to using actual operational results to optimize the prediction model and improve its accuracy.

[0121] Specifically, this includes determining the prediction error of the cooperative association model based on the real-time state information and the expected attitude change, and then correcting the cooperative coefficients a, b, c, and d in the cooperative association model based on the prediction error.

[0122] The control feedback optimization module collects real-time status information of UAV operations. Based on this information, it forms closed-loop control by correcting the fuzzy rules of the adaptive fuzzy PID algorithm and the cooperative coefficients of the cooperative correlation model, thereby continuously improving the self-balancing control performance.

[0123] This application achieves precise and stable attitude control during UAV-borne high-pressure water washing by layering multiple modules such as multi-source data fusion perception, load adaptive analysis, self-balancing control, cleaning operation coordination, and control feedback optimization, thereby simultaneously improving cleaning accuracy and operation efficiency and adapting to complex high-altitude operation conditions.

[0124] Compared to the closest existing technology, this application utilizes a multi-source sensing module to integrate data from inertia, wind speed, water flow, and vision, enabling early prediction of attitude deviation risks and proactive balance adjustments. This precisely controls the attitude deviation of the UAV to an extremely small range, improving balance control accuracy. By designing an adaptive fuzzy PID control algorithm, control parameters can be dynamically adjusted based on real-time interference intensity, significantly improving control response speed and dynamic anti-interference capability. Employing a coordinated flushing and balance control method, flushing parameter adjustments and attitude balance control are synchronized, maintaining attitude stability without interrupting operations and greatly improving operational efficiency. Furthermore, a load characteristic adaptive algorithm automatically identifies the load characteristics and flushing pressure of different UAVs, eliminating the need for manual adjustments and enhancing system versatility and adaptability. In implementation, Python language achieves the best results.

[0125] Please refer to Figure 2 As shown, this application also provides a self-balancing control device for washing insulators by unmanned aerial vehicles, comprising:

[0126] The acquisition module 201 acquires attitude data, environmental wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data.

[0127] The deviation module 202 calculates the attitude deviation based on the attitude data and the preset ideal attitude;

[0128] The instruction module 203 generates feedback control instructions based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

[0129] The prediction module 204 generates the expected attitude change based on the water flow pressure data, the water flow direction data, and the ambient wind speed data.

[0130] The control module 205 calculates the final control command based on the feedback control command and the expected attitude change.

[0131] This application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0132] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the above-described method.

[0133] The above embodiments are provided to enable those skilled in the art to understand and apply this application. Those skilled in the art will readily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without inventive effort. Therefore, this application is not limited to the above embodiments, and any improvements and modifications made to this application based on the disclosure thereof should be within the scope of protection of this application.

Claims

1. A self-balancing control method for flushing insulators by unmanned aerial vehicles (UAVs), characterized in that, include: Acquire attitude data, environmental wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data; Based on the posture data and the preset ideal posture, calculate the posture deviation; Based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data, a feedback control command is generated. Based on the water flow pressure data, the water flow direction data, and the ambient wind speed data, the expected attitude change is generated; The final control command is calculated based on the feedback control command and the expected attitude change.

2. The method according to claim 1, characterized in that, Before acquiring attitude data, ambient wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data, the following is also included: The weight of the drone and the water pressure and flow rate data of the high-pressure water washing system are obtained. The load characteristic coefficient is determined based on the fuselage weight, the water pressure data, and the water flow rate data. The load characteristic coefficient is associated with the calculation of the water flow reaction force data and / or the generation of the feedback control command.

3. The method according to claim 1, characterized in that, The steps for obtaining the water flow reaction force data include: Data on the first water flow reaction force is obtained using a force sensor; Calculate the second water flow reaction force data based on the water flow pressure data, water flow rate data, and load characteristic coefficient; The water flow reaction force data is determined based on the first water flow reaction force data and the second water flow reaction force data.

4. The method according to claim 1, characterized in that, Based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data, feedback control commands are generated, including: Obtain the rate of change of the attitude deviation; The feedback control command is generated by the adaptive fuzzy PID controller based on the attitude deviation, the rate of change of the attitude deviation, the ambient wind speed data, and the water flow reaction force data.

5. The method according to claim 1, characterized in that, Based on the water flow pressure data, the water flow direction data, and the ambient wind speed data, the expected attitude change is generated, including: The expected attitude change is generated by the collaborative correlation model based on the water flow pressure data, the water flow direction data, and the environmental wind speed data. The expected attitude change includes the expected pitch angle change and the expected roll angle change. The expected pitch angle change is associated with the water flow pressure data, the sine function value of the water flow direction data, and the ambient wind speed data. The expected roll angle change is associated with the water flow pressure data, the cosine function value of the water flow direction data, and the ambient wind speed data.

6. The method according to claim 1, characterized in that, The step of calculating the final control command based on the feedback control command and the expected attitude change includes: The expected attitude change is used as the feedforward compensation amount. The final control command is calculated based on the feedback control command and the feedforward compensation amount.

7. The method according to claim 1, characterized in that, The step of correcting the collaboration coefficients in the collaboration association model based on the real-time status information includes: Based on the real-time state information and the expected attitude change, the prediction error of the collaborative association model is determined; Based on the prediction error, the synergy coefficients in the synergy association model are corrected.

8. A self-balancing control device for washing insulators by unmanned aerial vehicles, characterized in that, include: The acquisition module acquires attitude data, environmental wind speed data, water flow reaction force data, water flow pressure data, and water flow direction data. The deviation module calculates the attitude deviation based on the attitude data and the preset ideal attitude; The instruction module generates feedback control instructions based on the attitude deviation, the ambient wind speed data, and the water flow reaction force data. The prediction module generates the expected attitude change based on the water flow pressure data, the water flow direction data, and the ambient wind speed data. The control module calculates the final control command based on the feedback control command and the expected attitude change.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any one of the methods described in claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed in a computer, causes the computer to perform any one of the methods described in claims 1 to 7.

Images