Anti-interference control method, system and equipment of photovoltaic panel cleaning unmanned aerial vehicle and medium

Abstract

The invention relates to the technical field of unmanned aerial vehicle anti-interference control, and provides an anti-interference control method, system and device for a photovoltaic panel cleaning unmanned aerial vehicle, and a medium, and the method comprises the steps: constructing an unmanned aerial vehicle dynamic linearization control model based on the external unknown disturbance of the unmanned aerial vehicle and the dynamic characteristics of a four-rotor unmanned aerial vehicle; respectively carrying out system input control analysis and pseudo partial derivative estimated value control analysis by taking a balance tracking error and an input quantity change as a principle and taking a balance modeling error and a pseudo partial derivative estimated value change as a principle to obtain a corresponding model-free adaptive control law and a model-free adaptive parameter estimation law; and acquiring an external unknown disturbance monitoring quantity based on a preset disturbance observer, and generating a control signal in combination with the dynamic linearization control model of the unmanned aerial vehicle to perform attitude regulation and control. On the basis of model-free self-adaptive control scheme design considering unknown disturbance, the control modeling workload can be reduced, self-adaptive change adjustment can be achieved, the self-adaptive anti-interference and vibration suppression capacity is improved, and the attitude control precision is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) anti-interference control, and in particular to an anti-interference control method, system, equipment and medium for a photovoltaic panel cleaning UAV. Background Art

[0002] With the development of drone technology, using drones for photovoltaic panel cleaning has gradually become a new solution. Drones offer advantages such as high flexibility, ease of operation, and low cost, enabling them to quickly and efficiently remove debris from the surfaces of photovoltaic panels. However, in practice, using drones to clean photovoltaic panels faces several challenges: 1) Drones are susceptible to interference from the external environment during flight, such as wind and airflow. These disturbances can cause the drone's posture to become unstable, affecting the cleaning effect; 2) The reaction force of the drone's onboard high-efficiency water pump nozzles can cause disturbances and even vibrations to the drone, which not only affects the cleaning effect but also threatens the drone's flight stability and safety.

[0003] Existing UAV flight control methods mainly include linear quadratic regulator control, model predictive control, and PID control. However, flight control based on linear quadratic regulators or model predictive control requires first constructing a mathematical model of the aircraft, and then implementing controller design and verification based on the aircraft mathematical model. When the UAV is disturbed, the aircraft mathematical model is difficult to construct, resulting in difficulty in fully and accurately grasping the UAV's real-time status and operational risks in practical applications, and poor robustness. Although PID control does not rely on system model information, it requires a trade-off between dynamic performance and steady-state performance, and it is not applicable to control systems with strong nonlinearity and parameter uncertainty. Therefore, there is an urgent need to provide an anti-disturbance and vibration suppression control method for UAVs for photovoltaic panel cleaning that takes into account unknown external disturbances, reduces design workload, and improves attitude control accuracy. Summary of the Invention

[0004] The purpose of the present invention is to provide an anti-disturbance control method for a photovoltaic panel cleaning drone. The method is based on a model-free adaptive control scheme design that takes unknown disturbances into consideration, avoids the use of complex drone dynamics models, greatly reduces the workload of control modeling, and realizes adaptive adjustment based on drone parameters or structural changes. At the same time, it can also effectively enhance the drone's adaptive anti-disturbance and vibration suppression capabilities and improve attitude control accuracy based on the introduction of an unknown disturbance estimation compensation mechanism based on a disturbance observer, thus providing reliable guarantees for better completion of photovoltaic panel cleaning work.

[0005] In order to achieve the above objectives, an anti-interference control method, system, device and medium for a photovoltaic panel cleaning drone are provided.

[0006] In a first aspect, an embodiment of the present invention provides an anti-disturbance control method for a photovoltaic panel cleaning drone, the method comprising the following steps:

[0007] Based on the external unknown disturbances of the UAV and the dynamic characteristics of the quadrotor UAV, a dynamic linearization control model of the UAV is constructed; the external unknown disturbances of the UAV include the reaction force of the onboard water pump operation; the dynamic linearization control model of the UAV includes multiple dynamic linearization sub-models of the attitude controller;

[0008] Based on the principle of balancing tracking error and input variation, the system input control analysis of the UAV dynamic linearization control model is carried out to obtain a model-free adaptive control law.

[0009] Based on the principle of balancing modeling errors and pseudo-partial derivative estimation changes, the pseudo-partial derivative estimation control analysis of the UAV dynamic linearization control model is performed to obtain a model-free adaptive parameter estimation law.

[0010] According to the model-free adaptive control law and the model-free adaptive parameter estimation law, an external unknown disturbance monitoring quantity is obtained based on a preset disturbance observer, and a control signal is generated according to the external unknown disturbance monitoring quantity and the UAV dynamic linearization control model to perform attitude control on the photovoltaic panel cleaning UAV.

[0011] Furthermore, the step of constructing a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV includes:

[0012] Obtaining an attitude dynamics system model corresponding to the dynamic characteristics of the quadrotor drone;

[0013] The external unknown disturbance of the UAV is regarded as an unmodeled term, and the attitude dynamics system model is corrected and analyzed according to the unmodeled term to obtain a nonlinear system model of the UAV;

[0014] Splitting the nonlinear control system corresponding to the nonlinear system model of the UAV into multiple attitude controllers, and determining the control input and control output of each attitude controller respectively;

[0015] The nonlinear sub-control systems corresponding to the various attitude controllers are respectively equivalent to single-input single-output system models with disturbances, and based on the differential mean value theorem, each single-input single-output system model with disturbances is converted into the corresponding dynamic linearized sub-model;

[0016] The dynamic linearization sub-models corresponding to each attitude controller are aggregated to obtain the dynamic linearization control model of the UAV.

[0017] Furthermore, the attitude controller includes an altitude controller, a yaw angle controller, a roll angle controller and a pitch angle controller;

[0018] The step of splitting the nonlinear control system corresponding to the nonlinear system model of the UAV into multiple attitude controllers and respectively determining the control input and control output of each attitude controller includes:

[0019] The nonlinear control system is divided into a position control loop and an attitude control loop; the position control loop determines the position of the drone based on the pitch and roll motions of the drone; the attitude control loop performs attitude control based on the desired altitude, desired yaw angle, desired roll angle, and desired pitch angle;

[0020] A position controller is designed based on the position control loop, and multiple UAV motion types are obtained based on the attitude control loop; the UAV motion types include vertical motion, yaw motion, roll motion, and pitch motion;

[0021] According to various UAV motion types and UAV motion control targets, the nonlinear control system is subjected to attitude control decoupling and splitting to obtain multiple attitude controllers and corresponding control inputs and outputs.

[0022] Furthermore, the dynamic linearization sub-model is expressed as:

[0023] y m (k+1)=y m (k)+φ m (k)Δu m (k)+ε m (k)

[0024] Where,

[0025] ε m (k) = f pump (y m (k),u m (k))+ω m (k)

[0026] Δu m (k)=u m (k)-u m (k-1)

[0027] Among them, y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k, respectively, m = 1, 2, 3, 4; u m (k) represents the system input of the m-th attitude controller; f pump,m (ym (k),u m (k)) represents the change in the system output disturbance of the m-th attitude controller caused by the reaction force of the water pump cleaning the photovoltaic panels at time k; ω m (k) represents the disturbance change of the m-th attitude controller at time k, excluding the reaction force of the water pump cleaning the photovoltaic panel; ε m (k) represents the total unknown disturbance change of the m-th attitude controller at time k; Δu m (k) represents the system input deviation of the m-th attitude controller at time k and time k-1; φ m (k) represents the pseudo partial derivative of the m-th attitude controller at time k.

[0028] Furthermore, the model-free adaptive control law includes input quantity control laws of each dynamic linearized sub-model;

[0029] The step of performing system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input variation to obtain a model-free adaptive control law includes:

[0030] Based on the principle of balancing tracking error and input change, input control analysis is performed on each dynamic linearization sub-model to obtain the control input criterion function corresponding to each dynamic linearization sub-model; the control input criterion function is expressed as:

[0031]

[0032] Where β represents the weight factor for balancing tracking error and input change; J m,u represents the control input criterion function of the m-th attitude controller, m = 1, 2, 3, 4; and y m (k+1) represents the expected output signal and actual output signal of the m-th attitude controller at time k+1; Δu m (k) represents the degree of change of the input of the m-th attitude controller at time k;

[0033] Substitute each dynamic linearization sub-model into the corresponding control input criterion function, and perform partial derivative analysis on the updated control input criterion function based on the system input deviation at adjacent moments to obtain the initial input control law corresponding to each dynamic linearization sub-model;

[0034] Based on the first preset step size factor, the initial input quantity control law corresponding to each dynamic linearization sub-model is corrected respectively to obtain the input quantity control law corresponding to each dynamic linearization sub-model;

[0035] The model-free adaptive control law is obtained according to the input quantity control law corresponding to each dynamic linearization sub-model.

[0036] Furthermore, the step of performing pseudo-partial derivative estimation control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation changes to obtain a model-free adaptive parameter estimation law includes:

[0037] Based on the principle of balancing modeling errors and changes in pseudo partial derivative estimates, pseudo partial derivative estimate control analysis is performed on each dynamic linearization sub-model to obtain the parameter estimation criterion function corresponding to each dynamic linearization sub-model; the parameter estimation criterion function is expressed as:

[0038]

[0039] where μ represents the weight factor for balancing the modeling error and the change in the pseudo-partial derivative estimate; J m,φ represents the parameter estimation criterion function of the m-th attitude controller, m=1,2,3,4; y m (k) and y m (k-1) represents the actual output signal of the m-th attitude controller at time k and k-1 respectively; φ m (k) represents the actual value of the pseudo partial derivative of the m-th attitude controller at time k; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k-1; u m (k-1) and u m (k-2) represents the input of the m-th attitude controller at time k-1 and k-2; ε m (k-1) represents the total unknown disturbance change of the m-th attitude controller at time k-1;

[0040] Based on the pseudo partial derivative estimation values ​​of each dynamic linearized sub-model, the corresponding parameter estimation criterion function is subjected to partial derivative analysis to obtain the initial adaptive parameter estimation law corresponding to each dynamic linearized sub-model;

[0041] Based on the second preset step size factor, the initial adaptive parameter estimation law corresponding to each dynamic linearization sub-model is corrected respectively to obtain the adaptive parameter estimation law corresponding to each dynamic linearization sub-model;

[0042] The model-free adaptive parameter estimation law is obtained according to the adaptive parameter estimation law corresponding to each dynamic linearization sub-model.

[0043] Furthermore, the preset disturbance observer is designed based on an error feedback mechanism that simultaneously considers external disturbance changes and pseudo partial derivative estimation deviations; the output of the preset disturbance observer is expressed as:

[0044]

[0045] Where, and They represent the estimated value of the external unknown disturbance change of the m-th attitude controller at time k+1 and time k, respectively, m=1,2,3,4; y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k respectively; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k; Δu m (k) represents the system input deviation of the m-th attitude controller at time k and time k-1; Represents ε m (k) is the estimated value; α represents the gain of the preset disturbance observer; ε m (k) represents the total unknown disturbance variation of the m-th attitude controller at time k; ε m,0 represents the limiting constant of the m-th attitude controller.

[0046] In a second aspect, an embodiment of the present invention provides an anti-disturbance control system for a photovoltaic panel cleaning drone, the system comprising:

[0047] A model building module is used to build a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV; the external unknown disturbance of the UAV includes the reaction force of the operation of the onboard water pump; the dynamic linearization control model of the UAV includes multiple dynamic linearization sub-models of the attitude controller;

[0048] A control law acquisition module is used to perform system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input change to obtain a model-free adaptive control law;

[0049] an estimation law acquisition module for performing pseudo-partial derivative estimation value control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation value changes to obtain a model-free adaptive parameter estimation law;

[0050] The anti-disturbance control module is used to obtain an external unknown disturbance monitoring quantity based on a preset disturbance observer according to the model-free adaptive control law and the model-free adaptive parameter estimation law, and to generate a control signal according to the external unknown disturbance monitoring quantity and the UAV dynamic linearization control model to perform attitude control on the photovoltaic panel cleaning UAV.

[0051] In a third aspect, an embodiment of the present invention further provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0052] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0053] The present invention provides an anti-disturbance control method, system, computer equipment and storage medium for a photovoltaic panel cleaning drone. The method realizes a technical solution of obtaining an external unknown disturbance monitoring quantity of the drone and the dynamic characteristics of a quad-rotor drone based on the external unknown disturbance of the drone including the reaction force of the operation of an onboard water pump and the dynamic characteristics of the quad-rotor drone, constructing a dynamic linearized control model of the drone including a plurality of dynamic linearized sub-models of attitude controllers, performing system input control analysis on the dynamic linearized control model of the drone based on the principle of balancing tracking error and input quantity change to obtain a model-free adaptive control law, and performing pseudo-partial derivative estimation value control analysis on the dynamic linearized control model of the drone based on the principle of balancing modeling error and pseudo-partial derivative estimation value change to obtain a model-free adaptive parameter estimation law, and then obtaining an external unknown disturbance monitoring quantity based on a preset disturbance observer according to the model-free adaptive control law and the model-free adaptive parameter estimation law, and generating a control signal according to the external unknown disturbance monitoring quantity and the dynamic linearized control model of the drone to perform attitude control on the photovoltaic panel cleaning drone. Compared with the existing technology, the anti-disturbance control method of the photovoltaic panel cleaning drone is based on the design of a model-free adaptive control scheme taking into account unknown disturbances. It greatly reduces the workload of control modeling and can realize adaptive adjustment control based on changes in drone parameters or structures. At the same time, it can also introduce an unknown disturbance estimation and compensation mechanism based on the introduction of a disturbance observer, effectively improving the drone's adaptive anti-disturbance and vibration suppression capabilities, improving attitude control accuracy, and providing reliable guarantees for better completion of photovoltaic panel cleaning work. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 1 is a flow chart of an anti-disturbance control method for a photovoltaic panel cleaning drone according to an embodiment of the present invention;

[0055] Figure 2 This is a schematic diagram of the attitude disassembly control of the UAV in an embodiment of the present invention;

[0056] Figure 3 Schematic diagram of the control flow of a single dynamic linearization sub-model in an embodiment of the present invention;

[0057] Figure 4 2 is a schematic structural diagram of an anti-disturbance control system for a photovoltaic panel cleaning drone according to an embodiment of the present invention;

[0058] Figure 5 is an internal structural diagram of a computer device according to an embodiment of the present invention;

[0059] Explanation of the reference numerals: 1-model building module; 2-control law acquisition module; 3-estimation law acquisition module; 4-anti-disturbance control module. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and beneficial effects of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the embodiments described below are part of the embodiments of the present invention and are only used to illustrate the present invention, but are not used to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0061] The anti-disturbance control method for the photovoltaic panel cleaning drone provided by the present invention can be understood as an application defect based on the difficulty of existing drone control methods in comprehensively and reliably constructing a drone control model in the presence of unknown disturbances, resulting in insufficient anti-disturbance and vibration suppression capabilities of the drone. Instead, a four-rotor drone anti-disturbance and vibration suppression control method based on model-free adaptive control is proposed for photovoltaic panel cleaning, which can reduce the workload of control scheme design, can adaptively adjust based on drone parameters and structure, and can achieve reliable compensation for unknown disturbances.

[0062] In one embodiment, Figure 1 As shown, an anti-disturbance control method for a photovoltaic panel cleaning drone is provided, comprising the following steps:

[0063] S11. Based on the external unknown disturbances of the UAV and the dynamic characteristics of the quad-rotor UAV, a dynamic linearization control model of the UAV is constructed; wherein, the external unknown disturbances of the UAV can be understood as external disturbances that may disturb the attitude of the UAV and affect the cleaning effect of the photovoltaic panels during the cleaning of the UAV, including but not limited to the reaction force of the onboard water pump (the reaction force brought to the UAV when the UAV water pump is started); the dynamic characteristics of the quad-rotor UAV can be understood as that from a dynamic point of view, the center of mass position vector and Euler angle vector of the UAV determine its spatial position and attitude, the total mass of the UAV and the acceleration of gravity will affect the vertical motion of the UAV, and the angular velocity, moment of inertia, proportional coefficient, air resistance coefficient and other parameters of the UAV rotors The two factors work together to determine the flight state of the UAV; in order to avoid the situation in which the attitude control does not meet the actual needs due to the complexity of the UAV attitude control model or incomplete consideration of the modeling factors in actual applications, while also meeting the application requirements of accurately controlling the height and attitude of the UAV in the actual photovoltaic panel cleaning scenario to ensure the accuracy and safety of the photovoltaic panel cleaning operation, this embodiment preferably performs an attitude decoupling analysis on the UAV attitude control model based on the external unknown disturbance of the UAV and the dynamic characteristics of the four-rotor UAV, uses a single-input and output model-free adaptive control algorithm to design multiple attitude controllers (model-free adaptive controllers), and constructs a UAV dynamic linearization control model including a dynamic linearization sub-model of multiple attitude controllers based on this.

[0064] Specifically, the steps of constructing a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV include:

[0065] Obtain an attitude dynamics system model corresponding to the dynamic characteristics of the quadrotor drone; wherein the attitude dynamics system model can be understood as the following dynamics system model (six-degree-of-freedom control system) constructed based on the three position variables and three attitude variables of the quadrotor drone:

[0066]

[0067] Where,

[0068]

[0069] Where x, y, and z represent the center of mass positions of the photovoltaic panel cleaning drone on the x-axis, y-axis, and z-axis, respectively; and represent the acceleration and velocity of the photovoltaic panel cleaning drone in the x-axis respectively; and represent the acceleration and velocity of the photovoltaic panel cleaning drone in the y-axis respectively; and denote the acceleration and velocity of the photovoltaic panel cleaning drone in the z-axis, respectively; φ, θ, and ψ denote the pitch angle, roll angle, and yaw angle of the photovoltaic panel cleaning drone, respectively; and represent the acceleration and velocity of the photovoltaic panel cleaning drone’s pitch angle, respectively; and represent the acceleration and velocity of the roll angle of the photovoltaic panel cleaning drone, respectively; and They represent the acceleration and velocity of the yaw angle of the photovoltaic panel cleaning drone, respectively; m represents the total mass of the photovoltaic panel cleaning drone; g represents the acceleration of gravity; Ω r Indicates the comprehensive value of the rotor speed of the photovoltaic panel cleaning drone; Ω n and F n They represent the angular velocity of the nth rotor and the lift generated when the rotor is working, n = 1, 2, 3, 4; K f represents the lift coefficient; I x ,I y ,I z Represents the moment of inertia corresponding to the x-axis, y-axis, and z-axis respectively; J r represents the moment of inertia of the rotor; C represents the proportional coefficient; K irepresents the air resistance coefficient corresponding to the i-th degree of freedom, i = 1, 2, …, 6; U1 represents the total lift of the rotor of the photovoltaic panel cleaning drone; U2 represents the pitch moment of the photovoltaic panel cleaning drone; U3 represents the rolling moment of the photovoltaic panel cleaning drone; U4 represents the yaw moment of the photovoltaic panel cleaning drone; l represents the distance between the center point of each propeller and the center of mass of the drone; λ represents a constant coefficient.

[0070] The external unknown disturbance of the UAV is taken as an unmodeled term, and the attitude dynamics system model is corrected and analyzed according to the unmodeled term to obtain the nonlinear system model of the UAV. The unmodeled term can be understood as an external unknown disturbance that is not considered in the above attitude dynamics system model but will affect the attitude control when the actual UAV performs the photovoltaic panel cleaning task. The attitude dynamics system model is corrected and analyzed by adding the unmodeled term to the 6-degree-of-freedom equation of the attitude dynamics system model, so that the corrected attitude dynamics system model is more in line with reality. The corresponding nonlinear system model of the UAV can be expressed as:

[0071]

[0072] Among them, d i represents the unmodeled term added to the i-th degree of freedom term, and i=1,2,…,6; it should be noted that the interpretation of other variable symbols in the model can refer to the relevant explanations in the above-mentioned attitude dynamics system model; in addition, in actual application, the modeling uncertainty term can be used as an unmodeled term to modify the attitude dynamics system model according to the needs, that is, d in the model i In addition to including the reaction forces from the operation of the onboard water pump, modeling uncertainties are also included.

[0073] The nonlinear control system corresponding to the nonlinear system model of the UAV is split into multiple attitude controllers, and the control input and control output of each attitude controller are determined respectively; wherein, the multiple attitude controllers can be understood as the characteristics of the UAV attitude control speed being fast and being able to be individually controlled by intermediate control quantities, and the flight control decoupling analysis of the nonlinear system model of the UAV is performed, and the model-free adaptive control (MFAC) controller for controlling different operation types is determined; wherein, the attitude controller includes an altitude controller, a yaw angle controller, a roll angle controller and a pitch angle controller, and the cooperation between the various attitude controllers can control the altitude and attitude of the quadrotor UAV, so as to realize the quadrotor UAV to respond to the desired signal (including the desired x-axis center of mass position x d , expected y-axis center of mass position y d 、Expected z-axis center of mass position z d and the desired yaw angle ψd ) while ensuring that the pitch angle φ and roll angle θ remain stable within a small angle range.

[0074] Specifically, such as Figure 2 As shown, the steps of splitting the nonlinear control system corresponding to the nonlinear system model of the UAV into multiple attitude controllers and respectively determining the control input and control output of each attitude controller include:

[0075] The nonlinear control system is divided into a position control loop and an attitude control loop; the position control loop determines the position of the drone based on the pitch and roll motions of the drone; the attitude control loop performs attitude control based on the desired altitude, desired yaw angle, desired roll angle, and desired pitch angle;

[0076] According to the position control loop, a position controller is designed, and according to the attitude control loop, a variety of UAV motion types are obtained; wherein, the position controller is a PID controller, and its input is the UAV in the xoy plane coordinate (x, y) and the corresponding desired coordinate (x d ,y d ), the output is the desired roll angle θ d and the desired pitch angle φ d ; Correspondingly, the UAV motion types include vertical motion, yaw motion, roll motion and pitch motion;

[0077] According to various UAV motion types and UAV motion control targets, the nonlinear control system is subjected to attitude control decoupling and splitting to obtain multiple attitude controllers and corresponding control inputs and control outputs; wherein the UAV motion control target may include the desired height z d , desired yaw angle ψ d , desired roll angle θ d and the desired pitch angle φ d The corresponding control input and control output of each attitude controller are: 1) The control input of the height controller is the current height z of the photovoltaic panel cleaning drone and the desired height z d , the corresponding control output is the total lift U1 of the rotor of the photovoltaic panel cleaning drone; 2) The control input of the yaw angle controller is the current yaw angle ψ of the photovoltaic panel cleaning drone and the desired yaw angle ψ d , the corresponding control output is the yaw moment U4 of the photovoltaic panel cleaning drone; 3) The control input of the roll angle controller is the current roll angle θ of the photovoltaic panel cleaning drone and the desired roll angle θ d , the corresponding control output is the rolling torque U3 of the photovoltaic panel cleaning drone; 4) The control input of the pitch angle controller is the current pitch angle φ of the photovoltaic panel cleaning drone and the desired pitch angle φ d, and the corresponding control output is the pitch torque U2 of the photovoltaic panel cleaning drone; it should be noted that the control logic of the four attitude controllers is the same, and the only difference is the different control parameters.

[0078] In practical applications, considering that the determination of the center of mass position of the UAV requires the control of pitch and roll motion, the desired pitch angle and desired roll angle required by the attitude controller can be obtained by connecting the attitude controller and the position controller in series; that is, Figure 2 As shown in the figure, based on the xoy plane coordinates (x, y) of the drone, the position controller (PID controller) is used to calculate the desired pitch angle and the desired roll angle, and then four attitude controllers are used to control the vertical motion, yaw, roll and pitch respectively to generate corresponding component forces.

[0079] The nonlinear sub-control systems corresponding to each attitude controller are equivalent to single-input single-output system models considering disturbances. Based on the differential mean value theorem, each single-input single-output system model considering disturbances is converted into the corresponding dynamic linearized sub-model. The single-input single-output system model considering disturbances can be expressed as:

[0080] y m (k+1)=f m (y m (k),…,y m (kn y ),u m (k),…,u m (kn u ))

[0081] +ε m (k)

[0082] Where y m (k+1),y m (k) and y m (kn y ) represent the m-th attitude controller at time k+1, time k and time kn respectively. y System output at time u m (k) and u m (kn u ) is the m-th attitude controller at time k and kn u System input at time f m (·) represents the unknown nonlinear function of the m-th attitude controller; ε m (k) represents the total unknown disturbance change of the m-th attitude controller at time k, which is bounded; n y and n y Is a positive integer.

[0083] Based on the differential mean value theorem, the process of converting each single-input single-output system model considering disturbance into the corresponding dynamic linearized sub-model is as follows:

[0084] According to the mean value theorem of differential calculus, for a function F(x) that is continuous on the interval [a,b] and differentiable in (a,b), then there is at least one point ξ in (a,b) such that F(b)-F(a)=F′(ξ)(ba). In the above single-input single-output system model, let x1=(y m (k),…,y m (kn y ),u m (k),…,u m (kn u ), assuming that between time k and k+1, the system satisfies the conditions of the differential mean value theorem, then there must exist a time-varying pseudo partial derivative φ(k) (similar to F′(ξ)) such that:

[0085] y m (k+1)-y m (k)=φ m (k)(u m (k)-u m (k-1))+ε m (k)

[0086] That is y m (k+1)=y m (k)+φ m (k)αu m (k)+ε m (k), we get the dynamic linearization model corresponding to the m-th attitude controller, where Δu m (k)=u m (k)-u m (k-1);φ m (k) is a time-varying pseudo partial derivative, which is a parameter introduced in the process of converting the nonlinear system of the UAV into a dynamic linearization model. It is used to approximately describe the degree of influence of the system input change on the output change. It changes with the change of system state and time to adapt to the control requirements of the UAV under different flight conditions. For any time k, φ m (k) is bounded, which is completely equivalent to the actual system model and does not contain unmodeled dynamics.

[0087] Considering that the water pump will bring reaction force to the drone when cleaning the photovoltaic panels, this part of the interference is considered separately, and its impact on the system output is assumed to be f pump (y m (k),u m(k)), the final dynamic linearized sub-model can be expressed as:

[0088] y m (k+1)=y m (k)+φ m (k)Δu m (k)+ε m (k)

[0089] Where,

[0090] ε m (k) = f pump (y m (k),u m (k))+ω m (k)

[0091] Δu m (k)=u m (k)-u m (k-1)

[0092] Among them, y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k, respectively, m = 1, 2, 3, 4; u m (k) represents the system input of the m-th attitude controller; f pump,m (y m (k),u m (k)) represents the change in the system output disturbance of the m-th attitude controller caused by the reaction force of the water pump cleaning the photovoltaic panels at time k; ω m (k) represents the disturbance change of the m-th attitude controller at time k, excluding the reaction force of the water pump cleaning the photovoltaic panel; ε m (k) represents the total unknown disturbance change of the m-th attitude controller at time k; Δu m (k) represents the system input deviation of the m-th attitude controller at time k and time k-1; φ m (k) represents the pseudo partial derivative of the m-th attitude controller at time k.

[0093] The dynamic linearization sub-models corresponding to each attitude controller are aggregated to obtain the dynamic linearization control model of the UAV.

[0094] The dynamic linearized sub-models of each attitude controller constructed through the above method steps can ensure that they are completely equivalent to the actual physical model. While avoiding the complex modeling process and reducing the workload of control modeling, they can also effectively consider nonlinear factors such as the reaction force of the water pump, providing a reliable analysis basis for the subsequent control law design.

[0095] S12. Based on the principle of balancing tracking error and input variation, the system input control analysis of the UAV dynamic linearization control model is performed to obtain a model-free adaptive control law; wherein, based on the principle of balancing tracking error and input variation, it can be understood that the system input quantity control principle is determined taking into account the characteristics that in actual photovoltaic panel cleaning operations, the UAV needs to accurately track the desired trajectory while avoiding drastic changes in the control input quantity to ensure the stability and safety of the UAV flight; since the UAV dynamic linearization control model includes dynamic linearization sub-models of multiple attitude controllers, the model-free adaptive control law obtained by performing input control analysis on each dynamic linearization sub-model in the UAV dynamic linearization control model includes the input quantity control law of each dynamic linearization sub-model.

[0096] Specifically, the step of performing system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input variation to obtain a model-free adaptive control law includes:

[0097] Based on the principle of balancing tracking error and input variation, input control analysis is performed on each dynamic linearization sub-model to obtain the corresponding control input criterion function of each dynamic linearization sub-model. The control input criterion function can be understood as a comprehensive analysis function of tracking error and control input variation, which is expressed as:

[0098]

[0099] Where β represents the weight factor for balancing tracking error and input change; J m,u represents the control input criterion function of the m-th attitude controller; and y m (k+1) represents the expected output signal and actual output signal of the m-th attitude controller at time k+1 respectively; Measures the difference between the actual output and the expected output of the m-th attitude controller; Δu m (k) represents the degree of change in the input of the mth attitude controller at time k, reflecting the degree of change in the control command. During the photovoltaic panel cleaning process, if the control input changes too much, it will cause the drone's attitude to change suddenly, affecting the cleaning effect and even causing the drone to lose balance.

[0100] Substitute each dynamic linearization sub-model into the corresponding control input criterion function, and perform partial derivative analysis on the updated control input criterion function based on the system input deviation at adjacent moments to obtain the initial input control law corresponding to each dynamic linearization sub-model; wherein, the initial input control law can be understood as the optimal solution of the control input obtained based on the control input criterion function. The specific acquisition process can be understood as substituting the dynamic linearization sub-model of each attitude controller into the corresponding control input criterion function, and then performing the control on J. m,u About Δu m (k) Take the derivative, set it equal to 0, and then perform a series of mathematical derivations to obtain:

[0101] In order to simplify the derivation process, we will not consider the ε in the dynamic linearization sub-model. m (k) is analyzed and ε is considered later. m The impact of item (k) is to first m,u Converted to the following expression:

[0102]

[0103] Where, Indicates that ε in the dynamic linearization sub-model is not considered m (k) The control input criterion function expression; About Δu m (k) Derivative:

[0104] make B=Δu m (k), then About Δu m (k) Taking the partial derivative we get:

[0105]

[0106] make It can be deduced that:

[0107]

[0108] And considering f pump (y m (k),u m (k)) and ω m (k), Δu m (k) may be updated to:

[0109]

[0110] Where Δu m (k) represents the initial input control law of the m-th attitude controller at time k.

[0111] Based on the first preset step size factor, the initial input control law corresponding to each dynamic linearization sub-model is modified to obtain the input control law corresponding to each dynamic linearization sub-model. The first preset step size factor can be understood as a tuning parameter introduced to improve the flexibility of the control law application and can be set according to actual application requirements. The corresponding input control law can be expressed as:

[0112]

[0113] Wherein, η represents the first preset step size factor.

[0114] The model-free adaptive control law is obtained according to the input quantity control law corresponding to each dynamic linearization sub-model.

[0115] S13. Based on the principle of balancing modeling errors and changes in pseudo-partial derivative estimates, the pseudo-partial derivative estimate value control analysis is performed on the UAV dynamic linearization control model to obtain a model-free adaptive parameter estimation law; wherein, based on the principle of balancing modeling errors and changes in pseudo-partial derivative estimates, it can be understood that the pseudo-partial derivatives in each dynamic linearization sub-model are unknown, and there are certain errors in the modeling process. If the pseudo-partial derivative estimate is inaccurate or fluctuates too much, the control law will fail, thereby affecting the flight stability of the UAV. The pseudo-partial derivative estimation control principle is designed to improve the control accuracy; similarly, the model-free adaptive parameter estimation law obtained by performing pseudo-partial derivative estimation control analysis on each dynamic linearization sub-model in the UAV dynamic linearization control model also includes the adaptive parameter estimation law of each dynamic linearization sub-model.

[0116] Specifically, the steps of performing pseudo-partial derivative estimation control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation changes to obtain a model-free adaptive parameter estimation law include:

[0117] Based on the principle of balancing the modeling error and the change of the pseudo partial derivative estimate, the pseudo partial derivative estimate control analysis is performed on each dynamic linearization sub-model respectively, and the parameter estimation criterion function corresponding to each dynamic linearization sub-model is obtained. Among them, the parameter estimation criterion function can be understood as a comprehensive analysis function of the modeling error and the change of the pseudo partial derivative estimate, which is expressed as:

[0118]

[0119] where μ represents the weight factor for balancing the modeling error and the change in the pseudo-partial derivative estimate; J m,φ represents the parameter estimation criterion function of the m-th attitude controller, m=1,2,3,4; y m (k) and y m(k-1) represents the actual output signal of the m-th attitude controller at time k and k-1 respectively; φ m (k) represents the actual value of the pseudo partial derivative of the m-th attitude controller at time k; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k-1; u m (k-1) and u m (k-2) represents the input of the m-th attitude controller at time k-1 and k-2; ε m (k-1) represents the total unknown disturbance change of the m-th attitude controller at time k-1;

[0120] Based on the pseudo partial derivative estimated values ​​of each dynamic linearized sub-model, the corresponding parameter estimation criterion function is subjected to partial derivative analysis to obtain the initial adaptive parameter estimation law corresponding to each dynamic linearized sub-model; wherein, the initial adaptive parameter estimation law can be understood as the optimal solution of the pseudo partial derivative obtained based on the parameter estimation criterion function. The specific acquisition process can be understood as the optimal solution of each parameter estimation criterion function J m,φ about Take the derivative, set the derivative equal to 0, and then perform a series of mathematical derivations to obtain:

[0121]

[0122] make have to

[0123]

[0124] Based on the second preset step size factor, the initial adaptive parameter estimation law corresponding to each dynamic linearization sub-model is modified to obtain the adaptive parameter estimation law corresponding to each dynamic linearization sub-model; wherein the second preset step size factor can be understood as a regulation parameter introduced to improve the flexibility of the application of the adaptive parameter estimation law, which can be set according to actual application requirements; the corresponding adaptive parameter estimation law can be expressed as:

[0125]

[0126] Wherein, γ represents the second preset step size factor.

[0127] In addition, in order to improve the tracking capability of pseudo partial derivatives, this embodiment preferably introduces a pseudo partial derivative reset algorithm: if or |Δy m (k)|<δ or |Δu m (k)|<ρ (which means that the currently estimated pseudo-partial derivative may be inaccurate, or the system state changes slightly and the previous estimate is no longer applicable), then in yes The initial value of ∈, δ, and ρ are thresholds set according to actual application requirements. This method can adjust the estimated value of the pseudo partial derivative in time when the system state changes significantly, so that the pseudo partial derivative does not change its sign and the change of the controlled quantity is not zero. This can avoid the failure of the control law due to inaccurate pseudo partial derivative estimation, enable the algorithm to adapt to the changes of the system, and maintain effective control of the system.

[0128] The model-free adaptive parameter estimation law is obtained according to the adaptive parameter estimation law corresponding to each dynamic linearization sub-model.

[0129] After completing the design of the parameter estimation law through the above method steps, it is also necessary to estimate the unknown nonlinear term f through the following method pump (y m (k),u m (k)) to provide support for disturbance compensation in subsequent actual attitude adjustment.

[0130] S14. According to the model-free adaptive control law and the model-free adaptive parameter estimation law, an external unknown disturbance monitoring variable is obtained based on a preset disturbance observer, and a control signal is generated based on the external unknown disturbance monitoring variable and the UAV dynamic linearized control model to perform attitude control on the photovoltaic panel cleaning UAV. The preset disturbance observer can be understood as a disturbance observer that takes into account the characteristics that the reaction force caused by the water pump when the UAV cleans the photovoltaic panel is slowly changing and the disturbance such as airflow to which it is subjected is relatively small. The nonlinear term based on the modeling error correction is introduced to achieve estimation and compensation of unknown disturbances, thereby further improving the anti-disturbance and vibration suppression capabilities of the UAV. In principle, the preset disturbance observer only needs to estimate the external disturbance change (including the external unknown disturbance of the water pump reaction force). However, considering that the estimation of pseudo-partial derivatives may have deviations in actual applications, this embodiment preferably designs a preset disturbance observer based on the input-output relationship and error feedback mechanism of the system to simultaneously achieve parameter uncertainty and external disturbance estimation. That is, the preset disturbance observer is designed based on an error feedback mechanism that simultaneously considers the external disturbance change and the pseudo partial derivative estimation deviation; the output of the preset disturbance observer is expressed as:

[0131]

[0132] Where, and They represent the estimated value of the external unknown disturbance change of the m-th attitude controller at time k+1 and time k respectively; y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k respectively; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k; Δu m(k) represents the system input deviation of the m-th attitude controller at time k and time k-1; α represents the gain of the preset disturbance observer, which determines the tracking speed and accuracy of the observer to the disturbance. If α is too large, the observer may be too sensitive to noise. Otherwise, the observer's response speed will be very slow and it will not be able to compensate for the disturbance in time; ε m (k) represents the total unknown disturbance variation of the m-th attitude controller at time k; Represents ε m The estimated value of (k) is used in practical applications. It can be initialized to 0 or an initial value set based on experience. In each control cycle, according to the current system input u m (k), output y m (k) and the estimated pseudo-partial derivatives Update using the above formula By continuously updating this estimate, the preset disturbance observer can monitor the changes of external disturbances in real time. For example, when the reaction force of the water pump suddenly increases, the system output y m (k) will change accordingly, and the preset disturbance observer can be used to obtain y m (k),u m (k) and Adjust the value of The value of can monitor the changes of external disturbances and update the estimated value It will be fed back into the control law to compensate for the impact of the disturbance on the system, that is, in the subsequent calculation of the control quantity Δu m (k) will be taken into consideration The estimated value of is used to adjust the control input to offset the disturbance effects including the pump reaction force.

[0133] It should be noted that in practical applications, when there is a deviation in the pseudo partial derivative estimation, it will lead to errors in the prediction of the system output. This error can be adjusted by presetting the gain α of the disturbance observer so that It can estimate the actual disturbance more accurately; at the same time, in order to avoid system instability due to the estimated value being too large or too small, the limiting constant ε can also be set m,0 (The unknown disturbance estimate value limit constant of the preset observer of the mth attitude controller) is used to limit the unknown disturbance estimate value of the preset observer: when When Let ε m,0 ;when When Set to -ε m,0 ;when Directly take the corresponding estimated value; that is, in practical applications, the observer gain α and the limiting constant ε m,0The selection of is crucial; in the actual photovoltaic panel cleaning scenario, each flight sub-controller can continuously adjust these parameters so that the corresponding preset disturbance observer can better adapt to the operating state of the UAV, achieve effective estimation and compensation for unknown disturbances, and further improve the UAV's anti-disturbance and vibration suppression capabilities; for example, for the altitude controller, the altitude error caused by the influence of unknown disturbances such as the water pump reaction force and airflow on the UAV's altitude can be estimated and compensated; for the yaw angle controller, roll angle controller and pitch angle controller, the interference errors caused by unknown disturbances such as the water pump reaction force and airflow on the yaw angle, roll angle and pitch angle can be estimated and compensated.

[0134] like Figure 3 As shown, first, the pseudo-partial derivative estimate is corrected based on the data of the previous moment according to the adaptive parameter estimation law, then the disturbance change at the current moment is estimated based on the modeling error of the previous moment, and finally the corresponding control signal is generated in combination with the tracking error. The generated output signal and control signal are used in the calculation of the next moment to achieve trajectory tracking in the time domain; in the actual photovoltaic panel cleaning process, the altitude controller ensures that the UAV maintains an appropriate distance from the photovoltaic panel, the yaw angle controller controls the steering of the UAV, and the roll angle controller and pitch angle controller work together to ensure the stability of the UAV's attitude during the cleaning process; through the careful design and parameter adjustment of these controllers, combined with the compensation effect of the preset disturbance observer, the UAV can operate stably in complex environments and efficiently complete the photovoltaic panel cleaning task. It should be noted that the control logic of the altitude controller, yaw angle controller, roll angle controller and pitch angle controller is designed based on the existing model-free adaptive controller; for example, for the pitch angle controller, it can be set:

[0135]

[0136] Among them, e pitch represents the pitch angle error (the difference between the actual pitch angle and the expected pitch angle), K p , K i , K d Represent the proportional coefficient, integral coefficient and differential coefficient of the controller respectively. The proportional coefficient K p It is used to quickly respond to pitch angle errors. The larger the value, the more sensitive the response to the error. It can quickly adjust the control amount to make the pitch angle of the drone close to the desired angle. However, too large K p This may cause system oscillation. During the photovoltaic panel cleaning process, when the pitch angle of the drone deviates from the expected angle, K p A corresponding control amount will be generated according to the size of the error, so that the UAV can quickly adjust its pitch attitude; the integral coefficient K i It is used to eliminate steady-state errors. By integrating the errors, the error information is continuously accumulated. When there are continuous small errors, Ki The control amount will be gradually increased until the error is eliminated. For example, during the long cleaning process of the drone, due to various factors, the small pitch angle error, K i It can ensure that the UAV can finally reach the accurate pitch angle; differential coefficient K d The control amount is adjusted according to the rate of change of the error. It can predict the trend of error change, control in advance, and enhance the stability of the system. When the pitch angle of the drone changes too quickly, K d A reverse control quantity will be generated to suppress the excessive change of pitch angle. In practical applications, it is necessary to adjust K according to the dynamic characteristics of the UAV, the requirements of the cleaning task and the interference situation. p , K i , K d Make fine adjustments to achieve good pitch angle control and ensure the drone maintains a stable attitude when cleaning photovoltaic panels.

[0137] The embodiment of the present invention provides a UAV dynamic linearization control model based on the external unknown disturbance of the UAV including the reaction force of the operation of the airborne water pump and the dynamic characteristics of the quad-rotor UAV, which includes a dynamic linearization sub-model of multiple attitude controllers. The UAV dynamic linearization control model is subjected to system input control analysis based on the principle of balancing tracking error and input change to obtain a model-free adaptive control law, and the UAV dynamic linearization control model is subjected to pseudo-partial derivative estimation value control analysis based on the principle of balancing modeling error and pseudo-partial derivative estimation value change to obtain a model-free adaptive parameter estimation law. According to the model-free adaptive control law and the model-free adaptive parameter estimation law, based on the preset disturbance observation The technical solution of obtaining external unknown disturbance monitoring quantities through a detector and generating a control signal according to the external unknown disturbance monitoring quantities and the dynamic linearization control model of the UAV to perform attitude control on the photovoltaic panel cleaning UAV can not only design the UAV attitude control scheme based on the model-free adaptive algorithm, effectively avoid the establishment of a complex UAV dynamics model, greatly reduce the workload, ensure that there will be no unmodeled dynamics, and realize adaptive control when the UAV parameters or structure change, but also effectively improve the UAV's adaptive anti-disturbance and vibration suppression capability and attitude control accuracy based on the unknown disturbance estimation compensation mechanism introduced by the disturbance observer, providing reliable guarantee for better completion of photovoltaic panel cleaning work.

[0138] It should be noted that although the steps in the above flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders.

[0139] In one embodiment, Figure 4As shown, an anti-disturbance control system for a photovoltaic panel cleaning drone is provided, the system comprising:

[0140] Model construction module 1 is used to construct a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV; the external unknown disturbance of the UAV includes the reaction force of the operation of the onboard water pump; the dynamic linearization control model of the UAV includes multiple dynamic linearization sub-models of the attitude controller;

[0141] A control law acquisition module 2 is used to perform system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input change, and obtain a model-free adaptive control law;

[0142] an estimation law acquisition module 3, configured to perform pseudo-partial derivative estimation value control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation value changes, and obtain a model-free adaptive parameter estimation law;

[0143] The anti-disturbance control module 4 is used to obtain an external unknown disturbance monitoring quantity based on a preset disturbance observer according to the model-free adaptive control law and the model-free adaptive parameter estimation law, and generate a control signal according to the external unknown disturbance monitoring quantity and the UAV dynamic linearization control model to perform attitude control on the photovoltaic panel cleaning UAV.

[0144] For the specific definition of the anti-interference control system of the photovoltaic panel cleaning drone, please refer to the definition of the anti-interference control method of the photovoltaic panel cleaning drone above. The corresponding technical effects can also be obtained equivalently, so we will not go into details here. Each module in the anti-interference control system of the above-mentioned photovoltaic panel cleaning drone can be implemented in whole or in part through software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0145] Figure 5 FIG. 1 shows an internal structure diagram of a computer device in one embodiment, which may be a terminal or a server. Figure 5As shown, the computer device includes a processor, memory, network interface, display, camera, and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it can implement an anti-interference control method for a photovoltaic panel cleaning drone. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or it can be buttons, a trackball, or a touchpad provided on the computer device housing, or it can be an external keyboard, touchpad, or mouse.

[0146] It can be understood by those skilled in the art that Figure 5 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have the same component arrangement.

[0147] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0148] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0149] In summary, the embodiments of the present invention provide an anti-disturbance control method and system for a photovoltaic panel cleaning drone. The anti-disturbance control method for the photovoltaic panel cleaning drone can not only design a drone attitude control scheme based on a model-free adaptive algorithm, effectively avoid establishing a complex drone dynamics model, greatly reduce the workload, ensure that there will be no unmodeled dynamics, and realize adaptive control when the drone parameters or structure change, but also can introduce an unknown disturbance estimation and compensation mechanism based on the introduction of a disturbance observer, effectively improve the drone's adaptive anti-disturbance and vibration suppression capabilities, improve the attitude control accuracy, and provide reliable guarantees for better completing the photovoltaic panel cleaning work.

[0150] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] The above-described embodiments merely represent several preferred implementations of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art can make several improvements and substitutions without departing from the technical principles of the present invention, and such improvements and substitutions should also be considered within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be based on the scope of protection of the claims.

Claims

1. An anti-disturbance control method for a photovoltaic panel cleaning drone, characterized in that: The method comprises the following steps: Based on the external unknown disturbances of the UAV and the dynamic characteristics of the quadrotor UAV, a dynamic linearization control model of the UAV is constructed; the external unknown disturbances of the UAV include the reaction force of the onboard water pump operation; the dynamic linearization control model of the UAV includes multiple dynamic linearization sub-models of the attitude controller; Based on the principle of balancing tracking error and input variation, the system input control analysis of the UAV dynamic linearization control model is carried out to obtain a model-free adaptive control law. Based on the principle of balancing modeling errors and pseudo-partial derivative estimation changes, the pseudo-partial derivative estimation control analysis of the UAV dynamic linearization control model is performed to obtain a model-free adaptive parameter estimation law. According to the model-free adaptive control law and the model-free adaptive parameter estimation law, an external unknown disturbance monitoring quantity is obtained based on a preset disturbance observer, and a control signal is generated according to the external unknown disturbance monitoring quantity and the UAV dynamic linearization control model to perform attitude control on the photovoltaic panel cleaning UAV.

2. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 1, wherein: The steps of constructing a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV include: Obtaining an attitude dynamics system model corresponding to the dynamic characteristics of the quadrotor drone; The external unknown disturbance of the UAV is regarded as an unmodeled term, and the attitude dynamics system model is corrected and analyzed according to the unmodeled term to obtain a nonlinear system model of the UAV; Splitting the nonlinear control system corresponding to the nonlinear system model of the UAV into multiple attitude controllers, and determining the control input and control output of each attitude controller respectively; The nonlinear sub-control systems corresponding to the various attitude controllers are respectively equivalent to single-input single-output system models with disturbances, and based on the differential mean value theorem, each single-input single-output system model with disturbances is converted into the corresponding dynamic linearized sub-model; The dynamic linearization sub-models corresponding to each attitude controller are aggregated to obtain the dynamic linearization control model of the UAV.

3. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 2, wherein: The attitude controller includes an altitude controller, a yaw angle controller, a roll angle controller and a pitch angle controller; The step of splitting the nonlinear control system corresponding to the nonlinear system model of the UAV into multiple attitude controllers and respectively determining the control input and control output of each attitude controller includes: The nonlinear control system is divided into a position control loop and an attitude control loop; the position control loop determines the position of the drone based on the pitch and roll motions of the drone; the attitude control loop performs attitude control based on the desired altitude, desired yaw angle, desired roll angle, and desired pitch angle; A position controller is designed based on the position control loop, and multiple UAV motion types are obtained based on the attitude control loop; the UAV motion types include vertical motion, yaw motion, roll motion, and pitch motion; According to various UAV motion types and UAV motion control targets, the nonlinear control system is subjected to attitude control decoupling and splitting to obtain multiple attitude controllers and corresponding control inputs and outputs.

4. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 2, wherein: The dynamic linearization sub-model is expressed as: and m (k+1)=y m (k)+φ m (k)Δu m (k)+ε m (k) Where, e m (k)=f pump (y m (k),u m (k))+ω m (k) Δu m (k)=u m (to m (k-1) Among them, y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k, respectively, m = 1, 2, 3, 4; u m (k) represents the system input of the m-th attitude controller; f pump,m (y m (k),u m (k)) represents the change in the system output disturbance of the m-th attitude controller caused by the reaction force of the water pump cleaning the photovoltaic panels at time k; ω m (k) represents the disturbance change of the m-th attitude controller at time k, excluding the reaction force of the water pump cleaning the photovoltaic panel; ε m (k) represents the total unknown disturbance change of the m-th attitude controller at time k; Δu m (k) represents the system input deviation of the m-th attitude controller at time k and time k-1; φ m (k) represents the pseudo partial derivative of the m-th attitude controller at time k.

5. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 1, wherein: The model-free adaptive control law includes input quantity control laws of each dynamic linearized sub-model; The step of performing system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input variation to obtain a model-free adaptive control law includes: Based on the principle of balancing tracking error and input change, input control analysis is performed on each dynamic linearization sub-model to obtain the control input criterion function corresponding to each dynamic linearization sub-model; the control input criterion function is expressed as: Where β represents the weight factor for balancing tracking error and input change; J m,u represents the control input criterion function of the m-th attitude controller, m = 1, 2, 3, 4; and y m (k+1) represents the expected output signal and actual output signal of the m-th attitude controller at time k+1; Δu m (k) represents the degree of change of the input of the m-th attitude controller at time k; Substitute each dynamic linearization sub-model into the corresponding control input criterion function, and perform partial derivative analysis on the updated control input criterion function based on the system input deviation at adjacent moments to obtain the initial input control law corresponding to each dynamic linearization sub-model; Based on the first preset step size factor, the initial input quantity control law corresponding to each dynamic linearization sub-model is corrected respectively to obtain the input quantity control law corresponding to each dynamic linearization sub-model; The model-free adaptive control law is obtained according to the input quantity control law corresponding to each dynamic linearization sub-model.

6. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 1, wherein: The step of performing pseudo-partial derivative estimation value control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation value changes to obtain a model-free adaptive parameter estimation law includes: Based on the principle of balancing modeling errors and changes in pseudo partial derivative estimates, pseudo partial derivative estimate control analysis is performed on each dynamic linearization sub-model to obtain the parameter estimation criterion function corresponding to each dynamic linearization sub-model; the parameter estimation criterion function is expressed as: where μ represents the weight factor for balancing the modeling error and the change in the pseudo-partial derivative estimate; J m,φ represents the parameter estimation criterion function of the m-th attitude controller, m=1,2,3,4; y m (k) and y m (k-1) represents the actual output signal of the m-th attitude controller at time k and k-1 respectively; φ m (k) represents the actual value of the pseudo partial derivative of the m-th attitude controller at time k; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k-1; u m (k-1) and u m (k-2) represents the input of the m-th attitude controller at time k-1 and k-2; ε m (k-1) represents the total unknown disturbance change of the m-th attitude controller at time k-1; Based on the pseudo partial derivative estimation values ​​of each dynamic linearized sub-model, the corresponding parameter estimation criterion function is subjected to partial derivative analysis to obtain the initial adaptive parameter estimation law corresponding to each dynamic linearized sub-model; Based on the second preset step size factor, the initial adaptive parameter estimation law corresponding to each dynamic linearization sub-model is corrected respectively to obtain the adaptive parameter estimation law corresponding to each dynamic linearization sub-model; The model-free adaptive parameter estimation law is obtained according to the adaptive parameter estimation law corresponding to each dynamic linearization sub-model.

7. The anti-disturbance control method for a photovoltaic panel cleaning drone according to claim 1, wherein: The preset disturbance observer is designed based on an error feedback mechanism that considers both external disturbance changes and pseudo partial derivative estimation deviations. The output of the preset disturbance observer is expressed as: Where, and They represent the estimated value of the external unknown disturbance change of the m-th attitude controller at time k+1 and time k, respectively, m=1,2,3,4; y m (k+1) and y m (k) represents the system output of the m-th attitude controller at time k+1 and time k respectively; represents the estimated value of the pseudo partial derivative of the m-th attitude controller at time k; Δu m (k) represents the system input deviation of the m-th attitude controller at time k and time k-1; Represents ε m (k) is the estimated value; α represents the gain of the preset disturbance observer; ε m (k) represents the total unknown disturbance variation of the m-th attitude controller at time k; ε m,0 represents the limiting constant of the m-th attitude controller.

8. An anti-disturbance control system for a photovoltaic panel cleaning drone, characterized in that: The system comprises: A model building module is used to build a dynamic linearization control model of the UAV based on the external unknown disturbance of the UAV and the dynamic characteristics of the quadrotor UAV; the external unknown disturbance of the UAV includes the reaction force of the operation of the onboard water pump; the dynamic linearization control model of the UAV includes multiple dynamic linearization sub-models of the attitude controller; A control law acquisition module is used to perform system input control analysis on the UAV dynamic linearization control model based on the principle of balancing tracking error and input change to obtain a model-free adaptive control law; an estimation law acquisition module for performing pseudo-partial derivative estimation value control analysis on the UAV dynamic linearization control model based on the principle of balancing modeling errors and pseudo-partial derivative estimation value changes to obtain a model-free adaptive parameter estimation law; The anti-disturbance control module is used to obtain an external unknown disturbance monitoring quantity based on a preset disturbance observer according to the model-free adaptive control law and the model-free adaptive parameter estimation law, and to generate a control signal according to the external unknown disturbance monitoring quantity and the UAV dynamic linearization control model to perform attitude control on the photovoltaic panel cleaning UAV.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images