Control method for trajectory tracking of inspection unmanned aerial vehicle

Abstract

The invention provides an inspection unmanned aerial vehicle trajectory tracking control method, and relates to the technical field of power inspection unmanned aerial vehicle trajectories, and the method comprises the steps: building a variable-load inspection unmanned aerial vehicle kinetic model, building a dual-ring active-disturbance-rejection controller based on an event triggering mechanism according to the variable-load inspection unmanned aerial vehicle kinetic model, and carrying out the control of the trajectory tracking of an inspection unmanned aerial vehicle. The double-loop active-disturbance-rejection controller comprises a position controller and an attitude controller, when the unmanned aerial vehicle executes trajectory tracking, expected information of a trajectory, unmanned aerial vehicle parameters of the unmanned aerial vehicle, disturbance and noise are obtained, current expected displacement information is input into the position controller, first control information is obtained through calculation based on the position controller, and the first control information is sent to the attitude controller; according to the first control information and the actual value of the expected yaw angle, an expected rolling angle and an expected pitch angle are obtained through calculation, and parameter self-adaptive adjustment is conducted on parameters of a position controller and parameters of an attitude controller through a particle swarm algorithm. According to the invention, it can be ensured that the unmanned aerial vehicle has high system precision and efficiency when coping with variable loads and external disturbance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power inspection unmanned aerial vehicle trajectory, in particular to a control method for tracking the trajectory of an inspection unmanned aerial vehicle. BACKGROUND

[0002] In today's power system, transmission lines as a key infrastructure, its safe and stable operation plays a decisive role in ensuring reliable power supply. Transmission lines usually stretch in complex and diverse geographical environment, across mountains and vast plains, and even face adverse weather conditions, such as strong winds, heavy rain, snow and so on. This makes the transmission line exposed for a long time, vulnerable to various natural and man-made factors, and thus causes line faults, seriously threatening the normal operation of the power system. The traditional manual inspection method has many drawbacks when faced with such complex and widely distributed transmission lines. Manual inspection not only needs to consume a lot of manpower, material resources and time cost, and the efficiency is extremely low, but also in complex terrain such as remote mountainous areas, marshland, and high-voltage line inspection scenarios, there is a high safety risk, and it is difficult to ensure the safety of the staff.

[0003] With the rapid development of science and technology, unmanned aerial vehicle technology gradually emerges and shows great application potential in the field of power inspection. Unmanned aerial vehicle, with its excellent maneuverability, flexibility and high-resolution image acquisition capability, can quickly reach the areas that are difficult to access by traditional manual inspection, and realize efficient inspection of transmission lines. It can cover a large area of transmission lines in a short time, greatly improving the inspection efficiency, and can carry various sensors such as high-definition cameras, infrared thermal imagers, laser radars, etc., to obtain rich line state information, providing strong support for accurate detection of line faults.

[0004] However, although the UAV has achieved certain results in power inspection, it still faces many challenges in practical application, especially in trajectory tracking control. On the one hand, the hardware conditions of the UAV itself limit its endurance, so that when performing long-distance inspection tasks, it needs to return to the charging base frequently for charging, which not only increases the additional time cost, but also affects the continuity and efficiency of the inspection task. How to reasonably plan the flight trajectory of the UAV under the condition of considering the power limitation of the UAV, and ensure that the scheduled inspection task is completed within the limited power, has become a key problem to be solved. On the other hand, in a complex power inspection environment, the UAV is easily disturbed by various external disturbances such as strong wind and air flow disturbance. These disturbances will cause the UAV flight state to be unstable, making it difficult to accurately track the preset inspection trajectory, thereby affecting the accuracy and integrity of the inspection data collection. In addition, due to the complexity of the dynamics of the UAV itself and the uncertainty of the model parameters, the trajectory tracking control is also very difficult. For example, the change of the load of the UAV and the fluctuation of the motor performance will have a significant impact on its flight trajectory. At the same time, in the actual inspection process, the UAV also needs to adjust the flight trajectory in real time according to the direction of the power transmission line, the position of the tower and the area that needs to be detected, so as to realize the comprehensive and accurate inspection of the power transmission line, which puts higher requirements on the real-time performance and adaptability of the trajectory tracking control algorithm of the UAV.

[0005] At present, although there are many researches and methods for UAV trajectory tracking control, there are still many deficiencies in dealing with this special and complex application scene of power inspection. Some existing trajectory planning algorithms often fail to fully consider the key problems such as actual disturbance factors in the power inspection environment and power limitation of the UAV, resulting in that the planned trajectory lacks feasibility and effectiveness in actual application. Some control methods have poor robustness when facing complex and variable disturbances, and cannot guarantee that the UAV can stably and accurately track the expected trajectory. Therefore, it is necessary to design a control method for trajectory tracking of inspection UAV. SUMMARY

[0006] In order to overcome the deficiencies of the prior art, the purpose of the present application is to provide a control method for trajectory tracking of inspection UAV.

[0007] In order to achieve the above-mentioned purpose, the present application provides the following solutions:

[0008] The present application also provides a control method for trajectory tracking of inspection UAV, comprising:

[0009] Step 1: Establishing a variable load inspection UAV dynamics model;

[0010] Step 2: constructing a double-loop active disturbance rejection controller based on an event-triggered mechanism according to the variable load inspection UAV dynamic model, the double-loop active disturbance rejection controller comprising a position controller and an attitude controller;

[0011] Step 3: when the UAV performs trajectory tracking, obtaining expected information of the trajectory, UAV parameters of the UAV, and disturbances and noises;

[0012] Step 4: inputting the current expected displacement information into the position controller, and calculating first control information based on the position controller;

[0013] Step 5: calculating expected roll angle and pitch angle according to the first control information and an actual value of the expected yaw angle;

[0014] Step 6: inputting the expected attitude angle into the attitude controller to obtain second control information output by the attitude controller;

[0015] Step 7: using a particle swarm algorithm to perform adaptive adjustment on parameters of the position controller and the attitude controller.

[0016] Preferably, in step 1, the variable load inspection UAV dynamic model is established, specifically as follows:

[0017] The body coordinate system and the inertial coordinate system are determined by the right-hand rule;

[0018] A load-free UAV model is established through kinematic characteristic analysis of the quadrotor UAV;

[0019] A variable load UAV model is established on the basis of the load-free UAV model by introducing load mass variation and center of gravity variation.

[0020] Preferably, the position controller comprises a first tracking differentiator and a first extended state observer, the first tracking differentiator being used for processing a target signal and obtaining a differential signal, wherein the differential signal is used for designing a state error feedback; and the first extended state observer being used for observing and compensating total disturbances in real time.

[0021] Preferably, the design process of the first extended state observer is as follows:

[0022] An equivalent formula is generated according to the UAV parameters and the expected information, wherein the equivalent formula comprises UAV position state variables;

[0023] An extended state observer of a position subsystem, an event-triggered mechanism of the extended state observer of the position subsystem, an extended state observer of a velocity subsystem, and a corresponding event-triggered mechanism are constructed according to the equivalent formula, the extended state observer of the position subsystem and the extended state observer of the velocity subsystem together constituting the first extended state observer.

[0024] Preferably, the time-triggered mechanism of the position controller is as follows:

[0025]

[0026] In the formula, the variable σ x σ y and σ z It is based on the control variable U1(t) at the instant the event is triggered. The values ​​were calculated from the sampled values. It is the yaw angle at the moment the event is triggered. The sampled values, It is the roll angle at the moment the event is triggered. The sampled values, It is the pitch angle at the moment the event is triggered. The sampled values, Threshold extremum factor, As an error sensitivity factor, the controller trigger threshold is designed as a time-varying value related to the control error, introducing trajectory estimation error. for:

[0027]

[0028] In the formula, and State variables α x1 α y1 and α z1 The estimated value, x d y d and z d The expected value of the corresponding trajectory.

[0029] Preferably, the attitude controller includes:

[0030] The second extended state observer and the first state error feedback for the roll and pitch channels;

[0031] The second tracking differentiator, the third extended state observer, and the second state error feedback of the yaw channel.

[0032] Preferably, the design process for the second extended state observer in the roll and pitch channels and the third extended state observer in the yaw channel is as follows:

[0033] The expected values ​​of roll and pitch angles are calculated based on the unique channel virtual control variable output by the position controller and the actual value of the yaw angle. The second extended state observer for the roll and pitch channels and the third extended state observer for the yaw channel, as well as the corresponding event triggering mechanism, are constructed based on the calculated expected values.

[0034] Preferably, the attitude controller event triggering mechanism is as follows:

[0035]

[0036] In the formula, is an estimated value of the state variable alpha φ1 is an expected value of the roll angle. d

[0037] Preferably, in step 7, the parameters of the position controller and the attitude controller are adaptively adjusted by using a particle swarm algorithm, specifically:

[0038] The total number of parameters to be tuned is reduced by using a bandwidth method.

[0039] The inertia weight is dynamically adjusted based on a linearly decreasing inertia weight strategy.

[0040] The fitness function is designed according to the trajectory tracking error and the swing degree of the attitude angle.

[0041] An event triggering mechanism is designed according to the particle swarm algorithm.

[0042] According to the specific embodiments provided by the present application, the following technical effects are disclosed:

[0043] The present application provides a control method for trajectory tracking of a patrol unmanned aerial vehicle, which comprises establishing a variable load patrol unmanned aerial vehicle dynamics model, constructing a double-loop active disturbance rejection controller based on an event triggering mechanism according to the variable load patrol unmanned aerial vehicle dynamics model, the double-loop active disturbance rejection controller comprising a position controller and an attitude controller, when the unmanned aerial vehicle performs trajectory tracking, obtaining expected information of the trajectory, unmanned aerial vehicle parameters of the unmanned aerial vehicle, and disturbances and noises, inputting the current expected displacement information into the position controller, calculating first control information based on the position controller, calculating expected roll angle and pitch angle according to the first control information and the actual value of the expected yaw angle, and adaptively adjusting the parameters of the position controller and the attitude controller by using a particle swarm algorithm. The present application can ensure that the unmanned aerial vehicle has high system accuracy and efficiency when coping with variable load and external disturbances. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 The control method for trajectory tracking of a patrol unmanned aerial vehicle provided by the embodiments of the present application is shown in the flowchart.

[0046] Figure 2 ​The four-rotor position response effect diagram under the control method of the trajectory tracking of the inspection unmanned aerial vehicle in the embodiment of the present application is shown in the figure.

[0047] Figure 3 The comparison diagram of the trajectory tracking error of the four-rotor under the control method of the trajectory tracking of the inspection unmanned aerial vehicle in the embodiment of the present application and the ADRC control method using the traditional ESO is shown in the figure. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0049] The purpose of the present application is to provide a control method of trajectory tracking of an inspection unmanned aerial vehicle, which can ensure that the unmanned aerial vehicle has high system accuracy and efficiency when coping with variable loads and external disturbances.

[0050] In order to make the above-mentioned purposes, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail in combination with the drawings and specific embodiments.

[0051] Figure 1 The flowchart of the control method of trajectory tracking of the inspection unmanned aerial vehicle provided in the embodiment of the present application is shown in the figure, and the present application provides a control method of trajectory tracking of an inspection unmanned aerial vehicle, which comprises: Figure 1

[0052] Step 1: establishing a variable load inspection unmanned aerial vehicle dynamics model;

[0053] Step 2: constructing a double-loop active disturbance rejection controller based on an event trigger mechanism according to the variable load inspection unmanned aerial vehicle dynamics model, wherein the double-loop active disturbance rejection controller comprises a position controller and an attitude controller;

[0054] Step 3: when the unmanned aerial vehicle performs trajectory tracking, obtaining expected information of the trajectory, unmanned aerial vehicle parameters of the unmanned aerial vehicle, and disturbances and noises;

[0055] Step 4: inputting the current expected displacement information into the position controller, and obtaining first control information based on the position controller;

[0056] Step 5: calculating expected roll angle and pitch angle according to the first control information and the actual value of the expected yaw angle;

[0057] Step 6: inputting the expected attitude angle into the attitude controller to obtain second control information output by the attitude controller;

[0058] ​Step 7: The parameters of the position controller and the attitude controller are self-adaptively adjusted by using a particle swarm algorithm.

[0059] In step 1, a variable load inspection unmanned aerial vehicle dynamics model is established, specifically as follows:

[0060] The body coordinate system and the inertial coordinate system are determined by the right-hand rule;

[0061] Through the kinematic characteristic analysis of the quadrotor unmanned aerial vehicle, a no-load unmanned aerial vehicle model is established;

[0062] By introducing the load mass change and the center of gravity change, a variable load unmanned aerial vehicle model is established on the basis of the no-load unmanned aerial vehicle model.

[0063] The position controller comprises a first tracking differentiator and a first extended state observer, the first tracking differentiator is used for processing a target signal and obtaining a differential signal, wherein the differential signal is used for designing a state error feedback; and the first extended state observer is used for observing and compensating a total disturbance in real time.

[0064] The design process of the first extended state observer is as follows:

[0065] An equivalent formula is generated according to unmanned aerial vehicle parameters and expected information, wherein the equivalent formula comprises unmanned aerial vehicle position state variables;

[0066] An extended state observer of a position subsystem, an event trigger mechanism of the extended state observer of the position subsystem, an extended state observer of a speed subsystem and a corresponding event trigger mechanism are constructed according to the equivalent formula, and the extended state observer of the position subsystem and the extended state observer of the speed subsystem jointly constitute the first extended state observer.

[0067] The time trigger mechanism of the position controller is as follows:

[0068]

[0069] In the formula, the variables σ x , σ y and σ z are calculated according to the sampling value of the control variable U1(t) at the event trigger moment , is the sampling value of the yaw angle at the event trigger moment , is the sampling value of the roll angle at the event trigger moment , is the sampling value of the pitch angle at the event trigger moment , is a threshold extreme value factor, is the error sensitivity factor, the controller triggering threshold is designed as a time-varying value related to the control error, and the trajectory estimation error is introduced is:

[0070]

[0071] In the formula, and are the estimated values of state variables α x1 , α y1 and α z1 , x d , y d and z d correspond to the expected values of the trajectory.

[0072] The attitude controller comprises:

[0073] a second extended state observer and a first state error feedback of the roll and pitch channels;

[0074] a second tracking differentiator, a third extended state observer and a second state error feedback of the yaw channel.

[0075] The design process of the second extended state observer of the roll and pitch channels and the third extended state observer of the yaw channel is:

[0076] The expected values of the roll angle and the pitch angle are calculated according to the unique channel virtual control variable output by the position controller and the actual value of the yaw angle, and the second extended state observer of the roll and pitch channels and the third extended state observer of the yaw channel are constructed according to the calculated expected values, and the corresponding event triggering mechanism is constructed.

[0077] The event triggering mechanism of the attitude controller is:

[0078]

[0079] In the formula, is the estimated value of the state variable α φ1 , and φ d is the expected value of the roll angle.

[0080] In step 6, the particle swarm algorithm is used to adaptively adjust the parameters of the position controller and the attitude controller, specifically:

[0081] The total number of parameters to be tuned is reduced by the bandwidth method;

[0082] The inertia weight is dynamically adjusted based on the linearly decreasing inertia weight strategy;

[0083] The fitness function is designed according to the trajectory tracking error and the swing degree of the attitude angle;

[0084] An event-triggering mechanism is designed according to a particle swarm algorithm.

[0085] Next, the present application will be described in detail in combination with specific embodiments.

[0086] Step 1: Establishing a variable load inspection unmanned aerial vehicle dynamics model:

[0087] Modeling and hierarchical decoupling of the quadrotor system:

[0088] For the height control channel of the quadrotor aircraft, the dynamics model can be simplified as:

[0089]

[0090] In the formula, f is other parts in system modeling, d z represents the part of the system that is not modeled inside the system and the external disturbance of the system, and a new virtual control variable h z =z2, the dynamic relationship expressed in formula (4) can be rewritten as:

[0091] Position subsystem:

[0092]

[0093] Velocity subsystem:

[0094]

[0095] In the formula, ω z1 ,ω z2 respectively represent the disturbance and the influence of system nonlinearity in the two subsystems, and the position subsystem S z1 The virtual control variable h z obtained can be used as the reference input of the velocity subsystem S z2 .

[0096] Step 2: Constructing a double-loop active disturbance rejection controller based on an event-triggering mechanism according to the variable load inspection unmanned aerial vehicle dynamics model, the double-loop active disturbance rejection controller comprising a position controller and an attitude controller:

[0097] The cascade active disturbance rejection controller design includes two steps:

[0098] For the position subsystem S z1 , its extended state form can be written as:

[0099]

[0100] Its extended state observer is:

[0101]

[0102] In the formula, Let α be the state variable zi The estimated value of (i = 1, 2), the observer gain l zi The design of (i = 1, 2) needs to satisfy the matrix For the Hurwitz matrix, the variable ρ z1 Indicates the instant the observer event is triggered. At that time, the sampled value of the system state z1(t) will be used as the state value of the current control cycle. The event triggering mechanism of this observer is designed as follows:

[0103]

[0104] Observer trigger threshold γ z1 Designed to be a fixed value;

[0105] Based on the given desired trajectory z d Virtual control variable h z The design is as follows:

[0106]

[0107] Where, k z1 These are the controller parameters that need to be tuned;

[0108] For the velocity subsystem S z2 Its extended state can be written as:

[0109]

[0110] Its extended state observer is:

[0111]

[0112] In the formula, For state variable β zi The estimated value of (i = 1, 2), the observer gain l zi The design of (i = 3, 4) needs to satisfy the matrix For the Hurwitz matrix, the variable ρ z2 Indicates the instant the observer formula event is triggered. At that time, the sampled value of the system state z2(t), the event triggering mechanism of this observer is designed as follows:

[0113]

[0114] Observer trigger threshold γ z2 Designed to be a fixed value;

[0115] Based on the virtual control variable h zvirtual control variable U z is designed as follows:

[0116]

[0117] where k z2 is the controller parameter to be tuned;

[0118] The cascade structure separates the position and velocity into two subsystems (S1, S2), so that the disturbance can be suppressed in a targeted manner. The outer loop ESO estimates the disturbance related to the position, while the inner loop ESO focuses on the uncertainty at the velocity level. This double-layer design can achieve the optimization of the system between transient response and noise suppression performance by adjusting the bandwidth (ω1, ω2). This design overcomes the inherent contradiction between dynamic performance and disturbance rejection ability of the traditional single-ESO system, and significantly improves the robustness of the system.

[0119] The control design of the x and y channels in the position controller is similar to that of the z channel. In these channels, the control variables U x and U y are calculated by the displacement expectation values x d and y d through the controller.

[0120] In the position controller, the virtual control variables U x , U y , U z and the yaw angle ψ are the sampled values at the event triggering moment

[0121] , which are used to calculate the expectation values of the roll angle and the pitch angle, as well as the control variable U1.

[0122]

[0123]

[0124] The variables σ x , σ y and σ z are calculated according to the sampled values of the control variable U1(t) at the event triggering moment . The design of the controller event triggering mechanism is as follows:

[0125]

[0126] where, is the sampled value of the yaw angle at the event triggering moment , σ is the sampled value of the roll angle at the event triggering moment , and σ is the sampled value of the pitch angle at the event triggering moment The sampled values, Threshold extremum factor, As an error sensitivity factor, the controller trigger threshold is designed as a time-varying value related to the control error, and trajectory estimation error is introduced into the design. for:

[0127]

[0128] In the formula, and State variables α x1 α y1 and α z1 The estimated value, x d y d and z d The expected value of the corresponding trajectory.

[0129] In this design, the event-triggered mechanism of the controller incorporates a trajectory tracking error threshold, which is time-varying and dynamically adjusted according to the error magnitude: as the tracking error increases, the trigger threshold decreases and the update frequency increases; conversely, the smaller the error, the larger the threshold and the lower the trigger update frequency. For a quadcopter UAV with strongly coupled dynamic characteristics, the tracking error should be calculated comprehensively as a 3D trajectory deviation, rather than independently. In the three-dimensional error equation, due to the event-triggered mechanism, the controller cannot obtain the actual state value of the UAV in real time. Therefore, the estimated state of ESO is used to calculate the event triggering conditions and the trajectory tracking error. This is based on the following two considerations: First, it will be proven later that the estimated error of ESO is eventually uniformly bounded, theoretically ensuring that the estimated value can effectively approximate the true state of the UAV; second, the core objective of event-triggered control is to reduce unnecessary computation and communication overhead. Using real-time state information in the event-triggered mechanism would violate this objective. Therefore, using ESO to estimate the state is more in line with the design intent of event-triggered control.

[0130] The control design of the attitude channel inner loop controller is similar to that of the position channel, with control variables U2, U3, and U4 determined by the desired angle value φ. d θ d and ψ d It is found that the controller event triggering mechanism is slightly different. Taking the roll channel as an example, the event triggering mechanism of this channel is as follows:

[0131]

[0132] In the formula, Let α be the state variable φ1 The estimated value, φ d This is the expected value of the roll angle.

[0133] The designed controller also uses the particle swarm algorithm for online adaptive parameters. The cascade sub-controller has six parameters to be tuned [l1, l2, l3, l4, k1, k2], and the total controller of the quadrotor has 36 parameters to be tuned. The bandwidth of the observer and the bandwidth of the controller are introduced to simplify the parameters to be tuned. The controller parameters can be determined systematically through the following bandwidth-based relationship:

[0134]

[0135] In the attitude and position controller, the control target of the sub-controller is of the same order of magnitude, so the same parameter set is considered in the attitude controller as in the position controller. After simplification, the adjustable parameters will be reduced from a total of 36 to 6, of which 3 are used for attitude control and 3 are used for position control, thereby ensuring the effectiveness of the control. The parameters to be adjusted are [ω oa1 , ω oa2 , ω ca , ω op1 , ω op2 , ω cp ], wherein the particle swarm algorithm is applied to the algorithm of the cascade controller;

[0136] In this embodiment, to ensure that the control system still has good convergence and robustness in the presence of uncertain disturbances and modeling errors, the stability of the method is theoretically analyzed and it is ensured that the event-triggered mechanism will not cause the Zeno phenomenon;

[0137] First, the convergence of the cascade extended state observer is analyzed:

[0138] Define the estimation error of the extended state observer as The dynamic equation of the estimation error system is:

[0139]

[0140]

[0141] Define δ1=ρ1-α1, δ2=ρ2-β1, then we get:

[0142]

[0143] wherein,

[0144] In order to analyze the convergence of the extended state observer estimation error system, the following assumptions and theorems are proposed:

[0145] Assumption: All external disturbances and system unmodeled parts and their derivatives are bounded, and there are positive constants as constants;

[0146] Theorem: Consider the variable load quadrotor system, its extended state observer design and event-triggered condition are shown in the above formula, there exists observer gain through proper gain selection, which can ensure that all estimation errors converge to a compact set of arbitrarily small;

[0147] The proof is as follows:

[0148] Define Lyapunov function as:

[0149]

[0150] Where the matrix P1, P2 is symmetric positive definite matrix, and satisfies The derivative of V1 is:

[0151]

[0152] Since And the trigger condition forces δ i ≤γ i (i = 1, 2), we can get:

[0153]

[0154] Let

[0155] When the error is true, the time derivative of Lyapunov function satisfies It shows that the ESO estimation error converges, and the estimation error And is eventually uniformly bounded, and its convergence threshold can be arbitrarily adjusted by adjusting the observer gain l i , so that the error boundary can be constrained to an arbitrarily given compact set range;

[0156] Before analyzing the stability of the controller, give the assumption:

[0157] Use v1, v2 to represent v、 The controller tracking error is defined as e1 = α1-v1, e2 = β1-h. The dynamic model of the tracking error system is:

[0158]

[0159] Theorem: Consider the variable load quadrotor system with controller and event-triggered condition, there exists controller gain k1, k2 that can ensure the tracking error is bounded;

[0160] The proof is as follows:

[0161] Define Lyapunov function as:

[0162]

[0163] Its derivative is:

[0164]

[0165] According to Young's inequality, we have:

[0166]

[0167] where ν i is a positive constant, we have:

[0168]

[0169] where,

[0170]

[0171] According to the design of event-triggered condition, it is known that [U(τ q )-U(t)] is bounded, by properly choosing the control gains k1, k2, ensuring κ1>0, κ2>0, we can get that the tracking errors are ultimately uniformly bounded, and the closed-loop system realizes asymptotic stability in the sense of Lyapunov;

[0172] Since the event-triggered mechanism is introduced in the observer and controller, it is necessary to prove that the design will not trigger Zeno phenomenon, that is, to ensure that the system will not trigger an infinite number of events in a finite time, thus causing the inability to execute;

[0173] Theorem: Considering the variable load quadrotor system, the design of the event-triggered condition is formula (9), formula (13), formula (18) and formula (20) in the system containing the event-triggered mechanism, there are positive parameters to ensure that the system has no Zeno behavior;

[0174] For the event-triggered mechanism of the observation channel, the trigger conditions of the speed and position state are similar. Taking the z channel as an example, the actual output z1(t) of the system is differentiable, and has where L is a positive finite constant, defined as Then we have:

[0175]

[0176] When the event-triggered condition is met, we have:

[0177]

[0178] Therefore, we can get:

[0179]

[0180] The minimum triggering interval of the event-triggered mechanism can be denoted as ι1, which is a positive constant, and can ensure that there is no Zeno phenomenon in the observer event-triggered mechanism;

[0181] For the event-triggered mechanism of the control channel, taking U2 as an example, the controller output U2(t) is differentiable, so that Where K is a positive finite constant, and is defined as It can be obtained that:

[0182]

[0183] When the event-triggered condition is met, there is:

[0184]

[0185] Therefore, it can be obtained that:

[0186]

[0187] The minimum triggering interval can be written as ι2, which is a positive constant, and can ensure that there is no Zeno phenomenon in the controller event-triggered mechanism;

[0188] In one possible embodiment, it is assumed that the initial position and attitude of the unmanned aerial vehicle are [x y z φ θ ψ] T = [0 0 0 0 0] T The parameters of the quadrotor and particle swarm algorithm are selected as shown in Table 1, and the event-triggered mechanism parameters are selected as shown in Table 2.

[0189] Table 1 Parameters of quadrotor and particle swarm algorithm

[0190]

[0191]

[0192] Table 2 Selected event-triggered mechanism parameters

[0193]

[0194] Taking the x-axis as an example, the desired trajectory is set as:

[0195]

[0196] The desired trajectories of the y-axis and z-axis are the same as the x-axis, and the yaw angle ψ d = 0 ° The trajectory requires the quadrotor aircraft to maintain a constant speed for t≤3s, and then stay at the target point. It is assumed that the change of the load L in flight is:

[0197]

[0198] When t≤6s, the load mass is 1.5, and the center of gravity is continuously changed. At t=6s, the load is released from the body, and the system mass and moment of inertia change in steps. The x-axis disturbance is W x =[w x1 w x2 ] T which is designed as:

[0199]

[0200] The disturbances in the y-axis and z-axis are the same as those in the x-axis, and in addition, there is zero-mean Gaussian noise with a standard deviation of 1 in the attitude channel and the position channel as disturbances. The position response of the quadrotor UAV is as shown in Figure 2 , and the trajectory tracking error is as shown in Figure 3 It should be noted that the above examples are only exemplary and do not limit the specific values.

[0201] Step 3: When the UAV performs trajectory tracking, the expected information of the trajectory, the UAV parameters of the UAV, and the disturbances and noises are obtained:

[0202] In practical applications, the premise of realizing trajectory tracking control is to obtain the expected information of the trajectory, the UAV parameters of the UAV, and the disturbances and noises. The expected information of the trajectory includes the expected three-dimensional displacement and the expected pose angle. The UAV parameters of the UAV include the initial state of the quadrotor UAV and the variable load parameters. The disturbances and noises include external disturbances (such as wind, air flow interference model, etc.) and internal sensor noise.

[0203] This step ensures that the control system has real scene applicability under modeling errors and noise environment, and solves the problem that traditional algorithms often assume "no disturbance / constant parameters" which leads to large deviation in practical applications;

[0204] Step 4: Input the current expected displacement information into the position controller, and obtain the first control information based on the position controller:

[0205] In order to effectively cope with the nonlinear disturbances and dynamic uncertainty factors that may occur during flight, the dynamics system of the quadrotor UAV is first decoupled and modeled, divided into "position subsystem" and "speed subsystem" two levels, respectively designed by using extended state observer (ESO) and hierarchical controller. This method realizes effective perception and estimation of external disturbances, and provides a theoretical basis for subsequent hierarchical control strategy.

[0206] By introducing an event-triggered mechanism to reduce the calculation frequency of the controller, unlike the traditional periodic update, the mechanism decides whether to update the control signal according to whether the state error exceeds the set threshold, and only triggers the controller to calculate the output when the error is significant, thereby effectively reducing unnecessary calculation and communication overhead, in addition, by reasonably designing the error threshold, the system improves the resource utilization efficiency while ensuring the control performance, and improves the response ability to disturbance and energy saving of the system;

[0207] Step 5: calculate the expected roll angle and pitch angle according to the first control information and the actual value of the expected yaw angle:

[0208] The virtual control quantity of the displacement channel calculated by the position controller and the actual value of the expected yaw angle can be used to further calculate the expected roll angle and pitch angle, which are used as the input of the subsequent pose controller to further solve the control information. In addition, this step also has the function of solving part of the control information;

[0209] Step 6: input the expected attitude angle into the attitude controller to obtain the second control information output by the attitude controller:

[0210] The attitude controller uses an extended state observer and a hierarchical controller to solve the second control information. After calculating the expected attitude angle, the attitude controller related to the expected attitude angle introduces an event-triggered mechanism to reduce the calculation frequency of the controller;

[0211] Step 7: use the particle swarm optimization algorithm to adaptively adjust the parameters of the position controller and the attitude controller:

[0212] The particle swarm optimization algorithm is used to online self-tune the controller parameters to realize dynamic optimization and update of the controller gain. This method significantly improves the adaptability of the system when facing dynamic load changes and external disturbances, so that the controller can maintain optimal performance response in different working states, effectively enhancing the robustness and flexibility of the system;

[0213] The event-triggered CADRC control strategy is verified through simulation experiments, and the results show that this method has better tracking accuracy, faster disturbance recovery speed and lower control update frequency in the trajectory tracking task of the quadrotor unmanned aerial vehicle, and the comprehensive performance is better than that of the traditional control method, which has good engineering application prospect;

[0214] A cascade active disturbance rejection control method based on event-triggered mechanism is proposed for trajectory tracking control of variable load quadrotor unmanned aerial vehicle in this example, the system is designed hierarchically, the extended state observer and controller of position and velocity subsystems are constructed respectively, the accurate estimation and compensation of disturbance are realized, the control accuracy is improved significantly, the introduction of event-triggered mechanism makes the system update the system state and control signal only when necessary, the simulation results show that the data transmission amount is reduced by 55.6% and 68.4% respectively, the performance is guaranteed while the consumption of computing resources is greatly reduced;

[0215] Compared with the traditional ADRC, the CADRC shows stronger robustness under dynamic load changes and external disturbances, the root mean square error of position is reduced by 60.3%, in addition, the online parameter tuning of PSO algorithm further optimizes the adaptive ability of the controller, so that it can quickly adapt to the system parameter changes, the stability of the method is proved by theoretical analysis, and it is ensured that the event-triggered mechanism will not cause Zeno phenomenon.

[0216] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same or similar parts between various embodiments can be referred to each other.

[0217] The principle and implementation mode of the present application are described by applying specific examples in the present application, and the above embodiment is only used to help understand the method and core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In view of the above, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for controlling trajectory tracking of a patrol unmanned aerial vehicle, characterized in that, include: Step 1: Establish a dynamic model of the variable load inspection UAV; Step 2: Construct a dual-loop active disturbance rejection controller based on the dynamic model of the variable load inspection UAV, which includes a position controller and an attitude controller. Step 3: When the drone performs trajectory tracking, acquire the expected trajectory information, the drone's parameters, and disturbances and noise; Step 4: Input the current desired displacement information into the position controller, and calculate the first control information based on the position controller; Step 5: Calculate the desired roll angle and pitch angle based on the first control information and the actual value of the desired yaw angle; Step 6: Input the desired attitude angle into the attitude controller to obtain the second control information output by the attitude controller; Step 7: Use the particle swarm optimization algorithm to adaptively adjust the parameters of the position controller and attitude controller.

2. The method of claim 1, wherein, In step 1, a dynamic model of the variable load inspection UAV is established, specifically as follows: The body coordinate system and the inertial coordinate system are determined by the right-hand rule; Kinematic characteristics analysis of multiple quadcopter UAVs, and establishment of a no-load UAV model; By introducing changes in payload mass and center of gravity, a variable payload UAV model is established based on the unloaded UAV model.

3. The method of claim 2, wherein, The position controller includes a first tracking differentiator and a first extended state observer. The first tracking differentiator is used to process the target signal and acquire the differential signal, wherein the differential signal is used to design state error feedback. The first extended state observer is used to observe and compensate for the total disturbance in real time.

4. The method of claim 3, wherein, The design process of the first extended state observer is as follows: Equivalent equations are generated based on UAV parameters and desired information, whereby the equivalent equations include UAV position and state variables. Based on the equivalent formula, we construct the extended state observer of the position subsystem, the event triggering mechanism of the extended state observer of the position subsystem, the extended state observer of the velocity subsystem, and the corresponding event triggering mechanism. The extended state observers of the position subsystem and the extended state observers of the velocity subsystem together constitute the first extended state observer.

5. The method of claim 4, wherein, The time-triggered mechanism of the position controller is as follows: In the formula, the variable σ x σ y and σ z It is based on the control variable U1(t) at the instant the event is triggered. The values ​​were calculated from the sampled values. It is the yaw angle at the moment the event is triggered. The sampled values, It is the roll angle at the moment the event is triggered. The sampled values, It is the pitch angle at the moment the event is triggered. The sampled values, Threshold extremum factor, As an error sensitivity factor, the controller trigger threshold is designed as a time-varying value related to the control error, introducing trajectory estimation error. for: where and are the estimated values of the state variables a x1 , a y1 and a z1 , and x d , y d and z d correspond to the expected values of the trajectory.

6. The method of claim 5, wherein, The attitude controller includes: The second extended state observer and the first state error feedback for the roll and pitch channels; The second tracking differentiator, the third extended state observer, and the second state error feedback of the yaw channel.

7. The method of claim 6, wherein, The design process for the second extended state observer in the roll and pitch channels and the third extended state observer in the yaw channel is as follows: The expected values ​​of roll and pitch angles are calculated based on the unique channel virtual control variable output by the position controller and the actual value of the yaw angle. The second extended state observer for the roll and pitch channels and the third extended state observer for the yaw channel, as well as the corresponding event triggering mechanism, are constructed based on the calculated expected values.

8. The method of claim 7, wherein, The attitude controller event triggering mechanism is as follows: wherein is an estimate of the state variable a φ1 φ d is the expected value of the roll angle.

9. The method of claim 8, wherein, In step 7, the parameters of the position controller and attitude controller are adaptively adjusted using the particle swarm optimization algorithm, specifically as follows: By using the bandwidth method, the total number of parameters to be tuned can be reduced; Dynamically adjust the inertia weight based on a linearly decreasing inertia weight strategy; The fitness function is designed based on the trajectory tracking error and the degree of attitude angle oscillation. An event triggering mechanism is designed according to a particle swarm algorithm. An event triggering mechanism is designed according to a particle swarm algorithm.

Images