Tailstock type vertical take-off and landing unmanned aerial vehicle data driving model prediction control method and system based on T-S fuzzy approximation

Abstract

The invention discloses a T-S fuzzy approximation-based tailstock type vertical take-off and landing unmanned aerial vehicle data driving model prediction control method and system, and relates to the technical field of attitude control of tailstock type vertical take-off and landing unmanned aerial vehicles. In order to solve the problems that an existing tailstock type vertical take-off and landing unmanned aerial vehicle control algorithm highly depends on an accurate system model and parameters, the calculation cost needed by system modeling and parameter identification is high, and the design cost of a controller is obviously increased; and a traditional control algorithm generally lacks a systematic processing mechanism for state variables and control input constraints, and multi-target collaborative optimization is difficult to realize in a complex flight task. According to the technical key points, a T-S fuzzy system model is introduced, a system state quantity and a control input quantity are defined, constraint conditions are established, and a performance cost function is designed; collecting data and representing a system matrix as a quadratic matrix inequality set through the data; designing a linear matrix inequality condition capable of ensuring stability, recursion feasibility and system constraint, and calculating a control performance upper bound; and establishing a rolling horizon optimization problem on the basis of the cost function and the linear matrix inequality condition, and selecting a proper solver to solve the problem to obtain a fuzzy state feedback control gain and act on a control system. According to the method, the model prediction control law is directly designed through measurement data, and control law design without an identification step is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of attitude control of tail-sitter VTOL UAV, and particularly relates to a tail-sitter VTOL UAV data-driven model predictive control method based on T-S fuzzy approximation. BACKGROUND

[0002] The tail-sitter VTOL UAV is a new type of unmanned flight platform that combines fixed-wing high-speed cruising and rotor vertical take-off and landing capabilities. It has high space efficiency, long endurance, strong environmental adaptability, and great task flexibility, making it widely used in military reconnaissance, emergency rescue, industrial inspection, and other fields. In recent years, domestic and foreign scholars have conducted extensive research on its control methods, such as dynamic inverse control, sliding mode control, and active disturbance rejection control.

[0003] For example, the prior art with document number CN116088549B discloses a tail-sitter VTOL UAV attitude control method, which includes the following steps: constructing a mathematical model of the tail-sitter VTOL UAV. A feedforward compensator is designed to reduce the uncertainty bound. Based on the linearized model, a nominal controller is designed to provide basic control performance for the nominal system without uncertainty. An L1 adaptive controller is designed to compensate for unmatched uncertainty. The feedforward controller, nominal controller, and L1 adaptive controller together form the overall controller. The tail-sitter VTOL UAV attitude control method provided by the invention not only achieves good attitude control effect of the tail-sitter VTOL UAV, but also considers the influence of input constraints on the attitude control effect. The prior art with document number CN118363394A discloses a tail-sitter VTOL UAV vertical mode attitude control method, which relates to flight control law design technology. The technical solution includes using pitch angle, roll angle, yaw angle, pitch angle speed, roll angle speed, yaw angle speed, and height as control variables to control the attitude of the tail-sitter VTOL UAV in vertical mode. The method includes the following steps: pitch angle maintenance control through pitch control; roll angle maintenance control through roll control; yaw angle maintenance control through yaw control; height maintenance control through engine throttle. The invention has the advantages of solving the problems of static instability of the tail-sitter VTOL UAV, strong coupling between attitude control and throttle control, and internal disturbance influence.

[0004] It is worth noting that existing control methods are generally highly dependent on system models based on first principles and accurately identified parameters, but the high-precision identification of aerodynamic parameters in engineering practice faces severe challenges. To break this bottleneck, data-driven control methods become a natural choice, thanks to the increasing data resources and continuously improving computing power. For linear systems, data-driven theories based on available data to directly design end-to-end controllers have become mature: on the one hand, the parameterization framework of linear feedback systems based on Willems' fundamental lemma has derived a series of data-driven parameterization controller design methods; on the other hand, some studies have constructed a data information control framework based on disturbed data and designed data-driven robust controllers for linear systems under this framework. However, there are inherent control difficulties in the vertical flight phase of tail-sitter unmanned aerial vehicles: the strong coupling effect of rotor slipstream and wing aerodynamic force leads to highly nonlinear dynamics. In this context, T-S (Takagi-Sugeno) fuzzy systems provide an effective solution due to their strong approximation ability for nonlinear systems. However, how to develop data-driven control methods for nonlinear systems based on the T-S fuzzy model approximation framework remains an open problem.

[0005] On the other hand, to ensure the safe flight of tail-sitter vertical take-off and landing unmanned aerial vehicles in complex flight missions and fully release their performance potential, the handling of constraints on state variables and control inputs and the optimization of system performance goals become core challenges. Model predictive control technology is an ideal solution to meet this demand as it can handle multiple optimization and constraint conditions simultaneously. However, existing model predictive control design of T-S fuzzy systems still relies on accurate system models.

[0006] Therefore, for the vertical take-off and landing process of tail-sitter vertical take-off and landing unmanned aerial vehicles, it is imperative to propose a model predictive control algorithm that approximates nonlinear dynamics through a T-S fuzzy model and designs a controller solely relying on input-state data. SUMMARY

[0007] The technical problems to be solved by the present application are:

[0008] Existing control algorithms for tail-sitter vertical take-off and landing unmanned aerial vehicles are highly dependent on accurate system models and parameters, but the high computational cost required for system modeling and parameter identification significantly increases the cost of controller design. In addition, traditional control algorithms generally lack systematic handling mechanisms for state variable and control input constraints, and it is difficult to achieve multi-objective collaborative optimization in complex flight missions, which severely limits the performance ceiling and mission reliability of unmanned aerial vehicles. To address these issues, the present application proposes an innovative data-driven model predictive control method for tail-sitter vertical take-off and landing unmanned aerial vehicles based on T-S fuzzy approximation, which directly designs a model predictive control law through measured data to achieve control law design without identification steps.

[0009] The technical scheme adopted by the present application to solve the above technical problems is:

[0010] The present application provides a tail seat type vertical take-off and landing unmanned aerial vehicle data-driven model predictive control method based on T-S fuzzy approximation, which is used for controlling the attitude control system of the tail seat type vertical take-off and landing unmanned aerial vehicle (formula 3) to realize stable control of the tail seat type vertical take-off and landing unmanned aerial vehicle attitude, characterized in that it comprises the following steps:

[0011] Step one, introduce the T-S fuzzy system model, define the state quantity (state vector) and control input quantity of the tail seat type vertical take-off and landing unmanned aerial vehicle attitude control system, and establish the constraint condition, and design the performance cost function, specifically: first, describe the tail seat type vertical take-off and landing unmanned aerial vehicle attitude control system as a T-S fuzzy system model with structural uncertainty, then define the attitude angle and angular acceleration as the system state quantity, and define the propeller tension difference and control rudder deflection as the control input quantity, and establish the constraint condition according to the actual scene, and finally design the cost function of multi-performance index fusion according to the control target;

[0012] Step two, collect data and represent the system matrix as a set of quadratic matrix inequalities through data, specifically: first, randomly give an initial state within the state constraint range, randomly generate an input sequence within the input constraint range and act on the controlled object, and collect the system state response, second, calculate the fuzzy membership degree at each time, then integrate the state sequence, input sequence and membership degree sequence into corresponding matrices, and on the basis of the matrices, construct a set of quadratic matrix (formula 14) inequalities representing the system matrix according to the uncertainty bound;

[0013] Step three, design linear matrix inequality conditions that can guarantee stability, recursive feasibility and system constraints, and calculate the control performance upper bound, specifically: model the fuzzy-dependent Lyapunov function, and according to the Lyapunov stability theorem, derive the linear matrix inequality conditions that guarantee the stability of the closed-loop system; second, establish the linear matrix inequality conditions that can guarantee the recursive feasibility and system constraints according to the positive invariant set theory; determine the control performance cost upper bound of the tail seat type vertical take-off and landing unmanned aerial vehicle;

[0014] Step four: the specific steps of establishing the receding horizon optimization problem are as follows: on the basis of the above cost function and linear matrix inequality conditions, establish the receding horizon optimization problem, select a suitable solver to solve the problem, obtain the fuzzy state feedback control gain and act on the control system. The technical effects of the present application are simulated and verified, in the simulation, the control effect of the system under the action of the designed controller is observed. The simulation effect is evaluated to verify the applicability and superiority of the algorithm.

[0015] The present application has the following beneficial technical effects:

[0016] The present application is aimed at the attitude control problem of tail-sitter VTOL UAV in vertical flight process, and a data-driven control framework based on T-S fuzzy approximation is proposed, and the system constraints are processed and the control performance is optimized by the method of receding horizon optimization. The present application is aimed at the vertical take-off and landing process of tail-sitter VTOL UAV, and a model predictive control algorithm is proposed, which approximates the nonlinear dynamics by T-S fuzzy model and realizes the design of controller only relying on input-state data. The control algorithm can effectively ensure the stability of the controlled system and the continuous feasibility of the optimization problem without the accurate model and system parameters of the system. The related research results have important scientific value and engineering significance for promoting the development of tail-sitter VTOL UAV control technology and expanding the application scenarios.

[0017] Specifically, the beneficial effects of the present application are:

[0018] ① A data-driven controller design method based on T-S fuzzy approximation is proposed, which extends the data-driven control method of traditional linear systems to nonlinear systems through T-S fuzzy approximation. This method directly designs the control law through the collected data, realizes the controller design without system identification, and greatly reduces the cost of UAV controller design.

[0019] ② The hard constraint conditions of the system and the control performance requirements in the task scene are considered, and a model predictive control law is designed. The semi-definite programming optimization problem is solved by rolling to obtain the control input that can guarantee the system constraints and performance optimization.

[0020] In summary, the present patent proposes a data-driven model predictive control method to overcome the defects of the existing tail-sitter VTOL UAV attitude controller design method, which highly depends on the accurate system mechanism model and the parameter identification results, and lacks consideration of system constraints and performance optimization. The data-driven model predictive control method reduces the cost of controller design and significantly improves the autonomy and reliability of UAVs in complex environments, providing key technical support for engineering applications. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings are included to provide a further understanding of the present patent, and constitute a part of the specification, which together with the present patent, are used to explain the application, and do not constitute a limitation of the present patent. In the drawings:

[0022] Figure 1 A flowchart of a tail-sitter VTOL UAV data-driven model predictive control method based on T-S fuzzy approximation;

[0023] Figure 2 A schematic diagram of a tail-sitter VTOL UAV system;

[0024] Figure 3 The attitude angle response curve of a tail-mounted vertical takeoff and landing UAV;

[0025] Figure 4 The attitude angular velocity response curve of a tail-mounted vertical takeoff and landing UAV;

[0026] Figure 5 This is a control input curve diagram for a tail-mounted vertical takeoff and landing (VTOL) UAV. Detailed Implementation

[0027] Specific implementation method one: Applying the compound Figures 1-5 The implementation of the tail-flying vertical takeoff and landing UAV data-driven model predictive control method based on TS fuzzy approximation described in this invention is explained as follows:

[0028] Step 1: Introduce the TS fuzzy system model, define the system state variables and control input variables, establish constraints, and design the performance cost function.

[0029] First, consider the following TS fuzzy system model:

[0030] Rule i: If x1(k) is x2(k) is yes but

[0031]

[0032] in It is a state vector. It is the control input vector, x1(k), x2(k), ..., x n (k) represent the first to nth components of the UAV state vector, u1(k), u2(k), ..., u m (k) represent the 1st to mth components of the UAV control input vector, where n is the dimension of the state vector, m is the dimension of the input vector, and k is the sampling time. The fuzzy system has a total of I fuzzy rules. Let represent the fuzzy sets of the nth state under the i-th rule. and Let represent the matrix of the unknown actual subsystem under the i-th rule. For simplicity, the actual system matrix is ​​defined as .

[0033]

[0034] Through single-value fuzzification, product inference, and center-average defuzzification, the attitude control system of a tail-saddle vertical takeoff and landing UAV can be expressed as follows:

[0035]

[0036] Where hi (k) is the normalized membership function of the ith rule and satisfies 0≤h i (k)≤1 and ΔA(k) and ΔB(k) represent the approximation uncertainty matrices, respectively. The membership degree of the state variable x(k) is defined as There also exist uncertainty upper bounds ∈1 and ∈2 such that ||ΔA(k)||≤ ∈1, ||ΔB(k)||≤ ∈2, therefore, the uncertainty satisfies the following inequality

[0037]

[0038] where is the system uncertainty upper bound.

[0039] Due to the physical limitations of the UAV, the following hard constraints are given in the form of

[0040]

[0041] where Φ x and Φ u are the state and input quadratic constraint weight matrices, respectively.

[0042] On this basis, the cost function of system performance is given as

[0043]

[0044] where x(t|k) and u(t|k) represent the state and input at time k+t predicted at time k, respectively, and Q and R are given weight matrices.

[0045] In this patent, the control gain under the model predictive control framework adopts the following fuzzy-dependent form:

[0046]

[0047] where K i is the control gain corresponding to the ith fuzzy rule.

[0048] Step 2: Collect data and represent the system matrix as a set of quadratic matrix inequalities through data.

[0049] Randomly give an initial state x d (0) within the state constraint range, randomly generate a T d long input sequence {u d (0), u d (1),..., u d (T d -1)} within the input constraint range, and collect the system state response {x d(1), x d (2),..., x d (T d ), and then calculate the membership degree of each time to obtain the membership sequence {h d (0), h d (1),..., h d (T d -1)}, and integrate the state sequence, the control input sequence, and the membership sequence into the following matrix:

[0050]

[0051] where the symbol represents the Kronecker product of two vectors.

[0052] Therefore, the system matrix [AB] compatible with the data (8)-(13) can be represented by the following quadratic matrix inequality form

[0053]

[0054] where is the upper bound of the data approximation deviation, and λ min (·) represents the minimum eigenvalue of the matrix.

[0055] Step three: design a linear matrix inequality condition that can guarantee stability, recursive feasibility, and system constraints, and calculate the control performance upper bound.

[0056] First, define the following fuzzy-dependent Lyapunov function

[0057]

[0058] where P is the weight matrix of the Lyapunov function, and I n is an n-dimensional identity matrix.

[0059] According to the Lyapunov stability criterion, the following linear matrix inequality condition that can guarantee system stability is designed. That is, there exist a positive definite matrix a non-singular matrix L, and positive scalars ζ, α, and γ, such that the following linear matrix inequality holds for any i, j ≤ I

[0060]

[0061] where Θ and E are matrices introduced for convenience of expression, and their definitions are as follows:

[0062]

[0063] ζ is a decision variable to ensure system stability under uncertainty, and α is a decision variable introduced to ensure system stability compatible with data. Therefore, the closed-loop UAV attitude control system is asymptotically stable.

[0064] Then, based on the theory of positive invariant sets, we design linear matrix inequalities that guarantee recursive feasibility and system constraints, i.e., in positive definite matrices... An invertible matrix L, and a positive scalar γ, such that the following linear matrix inequalities hold for any i ≤ I:

[0065]

[0066]

[0067] Where x(k) is the state variable of the system at time k, then the system satisfies the constraints shown in equation (5) and the optimization problem is continuously solvable.

[0068] Under the combined conditions (15)-(18), the upper limit of the performance cost of the UAV attitude control system is γ.

[0069] Step 4: Establish a rolling time-domain optimization problem and perform simulation verification.

[0070] First, the optimization problem of controller design can be established as follows:

[0071]

[0072] In this embodiment, the Yalmip toolbox and Sdpt3 solver are selected to solve the linear matrix inequality. At each sampling time, the semidefinite programming problem shown in equation (19) is solved to obtain the fuzzy-dependent feedback control gain. and control input The attitude control system of the UAV (3) is used to achieve rolling optimization control. Specific Implementation Method Two:

[0074] In this embodiment, for example Figure 2 The vertical flight process of the tail-mounted VTOL UAV shown is subject to attitude control, and the dynamic description of the system is shown in equation (20):

[0075]

[0076] Where φ, θ, and ψ represent the roll angle, pitch angle, and yaw angle, respectively; F1 and F2 represent the pulling forces of the left and right motors, respectively; δ a and δ e These represent the aileron deflection and elevator deflection, respectively; I x =0.025kg·m 2 I y =0.007 kg·m 2, I z = 0.022 kg-m 2 , b = 0.8774 m is the reference wingspan, c = 0.253 m is the reference chord, d = 0.5 m is the distance between left and right motors, p = 1.29 kg / m 3 is the air density, V * = 16.02 m / s is the slipstream velocity of the propeller when the pull force of the motor balances the gravity, V = 2 m / s is the ascending velocity of the UAV, S = 0.222 m 2 is the reference wing area, S e = 0.034 m 2 is the reference area of the elevator region impacted by the propeller airflow, S a = 0.025 m 2 is the reference area of the aileron region impacted by the propeller airflow, u = 0.0204 is the ratio of the propeller's torsion coefficient to the force coefficient, is the roll damping coefficient, is the pitch moment coefficient, is the pitch control derivative, is the pitch damping coefficient, is the yaw control derivative, is the pitch damping coefficient. The sampling period is set to 0.01 s. The maximum amplitude of the attitude angle is 20°, the maximum amplitude of the attitude angular velocity is 5° / s, the maximum difference between the left and right motor pull forces is 4.5 N, and the maximum deflection of the elevator and aileron is 20°.

[0077] The following T-S fuzzy model is introduced based on step one:

[0078] The i-th rule: if is is then

[0079]

[0080] where the dimension of the state vector n = 6, the dimension of the input vector m = 3, the state variables are the input variables are x1(k) represents the roll angle, x2(k) represents the pitch angle, x3(k) represents the yaw angle, x4(k) represents the roll angular velocity, x5(k) represents the pitch angular velocity, and x6(k) represents the yaw angular velocity; u1(k) represents the difference between the left and right motor pull forces, u2(k) represents the aileron deflection, and u3(k) represents the elevator deflection. The fuzzy sets and membership functions are given in Table 1, where is the upper limit of the pitch angular velocity, q = -5 is the lower limit of the pitch angular velocity, is the upper limit of pitch rate, r = -5 is the lower limit of pitch rate.The upper bound of uncertainty is ∈1 = 0.001 and ∈1 = 0, respectively.Then the constraint conditions are set, and the cost function weights are set as Q = diag{10, 10, 10, 1, 1, 1} and R = diag{1, 1, 1}, respectively, and the initial state of the system is x(0) = [0 5 -5 3 -3].

[0081] The fuzzy set and membership function defined in Table 1

[0082]

[0083] According to steps two, three and four, data is collected and the controller is designed.In the simulation results, the system state response curve, the control input curve and the control change curve under the action of the proposed anti-shake variable mode model predictive controller are recorded, as shown in Figure 3 、 Figure 4 and Figure 5 The simulation results verify the effectiveness of the method described in the patent.

[0084] It has been verified that the method proposed in the application solves the technical problems proposed in the application, and the method described in the application has been verified through simulation experiments and practical applications, and the technical effects and practicality claimed in the application have been verified.

[0085] The algorithm (method) proposed in the application is the underlying technical core of the application, and various products can be derived based on the algorithm. Based on the algorithm (method) proposed in the application, a T-S fuzzy approximation based tail seat vertical take-off and landing unmanned aerial vehicle data driven model predictive control system (controller) is developed using a program language, the system has program modules corresponding to the steps of the above technical solutions, and when running, the steps of the above T-S fuzzy approximation based tail seat vertical take-off and landing unmanned aerial vehicle data driven model predictive control method are executed. The computer program of the developed system (software) is stored on a computer readable storage medium, and the computer program is configured to realize the steps of the above T-S fuzzy approximation based tail seat vertical take-off and landing unmanned aerial vehicle data driven model predictive control method when called by a processor. That is, the application is materialized on a carrier to become a computer program product.

[0086] A tail-sitter VTOL UAV data-driven model predictive control device based on T-S fuzzy approximation, the device comprising at least one processor, and a memory connected to the at least one processor in communication, wherein the memory stores instructions executable by the at least one processor, the instructions executed by the at least one processor to enable the at least one processor to perform the above-mentioned tail-sitter VTOL UAV data-driven model predictive control method based on T-S fuzzy approximation, realizing stable control of the control definition tail-sitter VTOL UAV attitude.

[0087] Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0088] Computational procedures (also referred to as programs, software, software applications, or code) of the present application include machine instructions executable by a programmable processor, and can be implemented using a high-level procedural and / or object-oriented programming language, and / or an assembly / machine language. As used in this application, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus and / or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0089] It should be understood that the steps shown in the various forms of flow above can be reordered, added to, or deleted from. For example, the steps recited in the present application can be performed in parallel, in series, or in a different order, as long as the desired results of the technology disclosed in the present application are achieved, and are within the scope of protection of the present application.

Claims

1. A tail-sitter VTOL UAV data-driven model predictive control method based on T-S fuzzy approximation, the model predictive control method is used for controlling a posture control system of a tail-sitter VTOL UAV, and stable control of the tail-sitter VTOL UAV posture is realized, and the method is characterized in that, The method comprises: Step one, introducing a T-S fuzzy system model, defining state variables and control input variables of the tail seat type vertical take-off and landing unmanned aerial vehicle attitude control system and establishing a constraint condition, and designing a performance cost function, specifically: first, the tail seat type vertical take-off and landing unmanned aerial vehicle attitude control system is described as a T-S fuzzy system model with structural uncertainty, then the attitude angle and angular acceleration are defined as system state variables, and the propeller tension difference and control rudder deflection are defined as control input variables, and the constraint condition is established according to the actual scene, and finally a cost function of multiple performance indicators is designed according to the control target; Step two, collecting data and representing the system matrix as a set of quadratic matrix inequalities through the data, specifically: first, randomly give an initial state within the state constraint range, randomly generate an input sequence within the input constraint range and act on the controlled object, and collect the system state response, second, calculate the fuzzy membership degree at each time, then integrate the state sequence, input sequence and membership degree sequence into corresponding matrices, and on the basis of the matrices, construct a set of quadratic matrix inequalities representing the system matrix according to the uncertainty bound; Step three, designing linear matrix inequality conditions capable of guaranteeing stability, recursive feasibility and system constraints, and calculating the control performance upper bound, specifically: modeling a fuzzy-dependent Lyapunov function, and deriving linear matrix inequality conditions capable of guaranteeing the stability of the closed-loop system according to the Lyapunov stability theorem; second, establishing linear matrix inequality conditions capable of guaranteeing recursive feasibility and system constraints according to the positive invariant set theory; determining the control performance cost upper bound of the tail seat type vertical take-off and landing unmanned aerial vehicle; Step four, establishing a receding horizon optimization problem, specifically: on the basis of the above cost function and linear matrix inequality conditions, a receding horizon optimization problem is established, a suitable solver is selected to solve the problem, and a fuzzy state feedback control gain is obtained and applied to the attitude control system.

2. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 1, characterized in that, The specific implementation process of step one is: First, consider the following T-S fuzzy system model: Rule i: if x1(k) is x2(k) is …,x n (k) is then wherein is a state vector, is a control input vector; n is the dimension of the state vector, m is the dimension of the input vector, k is the sampling time, the fuzzy system has I fuzzy rules, respectively represent the fuzzy set of the n-th state under the i-th rule, and represent the system matrix and the control matrix of the unknown actual subsystem under the i-th rule, represents an n-dimensional real number vector; Define the actual system matrix A and the control matrix B as: Through single-valued fuzzification, product reasoning and center average defuzzification, the tail seat type vertical take-off and landing unmanned aerial vehicle attitude control system can be represented as: where h i (k) is the normalized membership function of the ith rule and satisfies 0≤h i (k)≤1 and ΔA(k) and ΔB(k) represent the approximated uncertainty matrices; the membership degree of the state variable x(k) is defined as The uncertainty bounds are ∈1 and ∈2, respectively, such that ||ΔA(k)||≤ ∈1, ||ΔB(k)||≤ ∈2, and the uncertainty satisfies the following inequality wherein is an upper bound on system uncertainty; Based on the physical limitations of the unmanned aerial vehicle, the following hard constraint form is given: where Φ x and Φ u denote the quadratic form constraint weight matrices for states and inputs, respectively; On this basis, the cost function of the system performance is given as: Where x(t|k) and u(t|k) represent the state variable and input variable at k+t time, respectively, Q and R are given weight matrices; The control gain under the model predictive control framework adopts the following fuzzy-dependent form: where K i is the control gain corresponding to the i-th fuzzy rule.

3. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 1 or 2, characterized in that, The specific implementation process of step two is to collect data and represent the system matrix as a set of quadratic matrix inequalities through the data: Randomly give initial state x within state constraint range d (0), randomly generate T within input constraint range d Long input sequence {u d (0), u d (1),..., u d (T d -1)} acts on controlled object, and collects system state response {x d (1), x d (2),..., x d (T d )}; secondly, calculate membership of each time to obtain membership sequence {h d (0), h d (1),..., h d (T d -1)}; integrate state sequence, control input sequence, and membership sequence into following matrix: where the notation represents the Kronecker product of two vectors; Therefore, the system matrix [AB] compatible with the data (8)-(13) can be represented by the following quadratic matrix inequality form wherein is an upper bound on the data approximation bias, λ min denotes the smallest eigenvalue of a matrix, and I denotes the identity matrix.

4. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 3, characterized in that, The design of step three can guarantee stability, recursive feasibility and linear matrix inequality conditions of system constraints, and calculate control performance upper bound, the specific implementation process is: First, define the following fuzzy dependent Lyapunov function where P is the weight matrix of the Lyapunov function, I n is the n-dimensional identity matrix; According to Lyapunov stability criterion, the following linear matrix inequality condition which can guarantee the stability of the system is designed; that is, there exist positive definite matrix invertible matrix L, positive scalar ζ, a and γ, such that the following linear matrix inequality holds for any i, j ≤ I: The symbol * represents the omitted term because of the symmetric matrix; Wherein Θ and E are introduced for the convenience of expression, and the definition is as follows: Zeta is a decision variable for ensuring system stability under uncertainty, and alpha is a decision variable for ensuring system stability compatible with data, so that the closed-loop unmanned aerial vehicle attitude control system is asymptotically stable. Then, linear matrix inequality conditions that guarantee the recursive feasibility and system constraints are designed according to the positive invariance set theory, i.e., there exist a positive definite matrix a nonsingular matrix L and a positive scalar γ such that the following linear matrix inequality holds for any i ≤ I Where x(k) is the state quantity of the system at time k, and the system satisfies the constraint shown in equation (5) and the optimization problem is continuously solvable. Integrating conditions (15)-(18), the performance cost upper bound of the unmanned aerial vehicle attitude control system is γ.

5. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 4, characterized in that, Step four is to establish a rolling horizon optimization problem, which is implemented as follows: First, the optimization problem of the controller design can be established as 6. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 5, characterized in that, The Yalmip toolbox and the Sdpt3 solver are selected to solve the linear matrix inequality, and the fuzzy-dependent feedback control gain is obtained by solving the semi-definite programming problem shown in formula (19) at each sampling time and control input The unmanned aerial vehicle attitude control system (3) realizes the rolling optimization control.

7. The data-driven model predictive control method for tail-sitter vertical take-off and landing unmanned aerial vehicle based on T-S fuzzy approximation according to claim 6, wherein the state vector dimension n = 6, the input vector dimension m = 3, x1(k) represents the roll angle, x2(k) represents the pitch angle, x3(k) represents the yaw angle, x4(k) represents the roll angle velocity, x5(k) represents the pitch angle velocity, and x6(k) represents the yaw angle velocity; u1(k) represents the difference in pulling force of the left and right motors of the unmanned aerial vehicle, u2(k) represents the aileron deflection, and u3(k) represents the elevator deflection. The model predictive control method is used for attitude control of the tail-sitter vertical take-off and landing unmanned aerial vehicle, and the attitude control system of the tail-sitter vertical take-off and landing unmanned aerial vehicle is as shown in equation (20):

8. The data-driven model predictive control method for tail-sitter VTOL UAV based on T-S fuzzy approximation according to claim 7, characterized in that, The system has program modules corresponding to the steps of any one of claims 1-8, and when running, it executes the steps of the data-driven model predictive control method for tail-sitter vertical take-off and landing unmanned aerial vehicle based on T-S fuzzy approximation. where φ, θ and ψ represent roll angle, pitch angle and yaw angle, respectively; F1 and F2 represent the pulling force of the left and right motors, respectively; δ a and δ e represent the aileron deflection and elevator deflection, respectively; I x , I y , I z represent the three-axis rotational inertia of the unmanned aerial vehicle, respectively; b is the reference wingspan, c is the reference chord length, d is the shaft distance of the left and right motors of the unmanned aerial vehicle, p is the air density, V * is the slipstream speed of the propeller when the pulling force of the motor is balanced with the gravity, V is the ascending speed of the unmanned aerial vehicle, S is the reference wing area, S e is the reference area of the elevator region impacted by the propeller airflow, S a is the reference area of the aileron region impacted by the propeller airflow, and u is the ratio of the torsion coefficient to the force coefficient of the propeller, is the roll damping coefficient, is the pitch moment coefficient, is the pitch control derivative, is the pitch damping coefficient, is the yaw control derivative, is the pitch damping coefficient.

9. A data-driven model predictive control system for tail-sitter VTOL UAV based on T-S fuzzy approximation, characterized in that: The computer readable storage medium stores a computer program, and the computer program is configured to realize the steps of the data-driven model predictive control method for tail-sitter vertical take-off and landing unmanned aerial vehicle based on T-S fuzzy approximation according to any one of claims 1-8 when called by the processor.

10. A computer-readable storage medium, characterized in that: ​

Images