A Preset-Time Optimal Fault-Tolerant Control Method for Three-DOF Helicopters

Abstract

This invention discloses a three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method, relating to the field of unmanned helicopter flight control. Addressing constrained flight scenarios with both actuator failures and input quantization, this invention first establishes a helicopter dynamics model; secondly, it constructs a preset time function and a switchable bridge hole constraint function to temporally control attitude tracking errors; subsequently, it decouples the system based on backstepping and dynamic surface techniques, and introduces a low-pass filter to eliminate the problem of derivative complexity explosion; finally, it introduces a fuzzy logic system with an execution-evaluation-identification structure and a reinforcement learning framework to compensate for system uncertainties, failures, and quantization errors online, and update the adaptive optimal policy weights. This invention can ensure high-precision trajectory tracking within any preset time while meeting strict physical constraints, and simultaneously achieves an optimal synergistic balance between fault robustness and control energy consumption.

Description

Technical Field

[0001] This invention relates to the field of flight control for unmanned helicopters, and specifically to a three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method that considers bridge hole constraints and input quantization. Background Technology

[0002] In recent years, unmanned helicopters have been widely used in transportation, marine surveillance, and geological exploration due to their unique advantages such as vertical takeoff and landing, maneuverability, and hovering. In these scenarios, achieving high-precision tracking control of a three-degree-of-freedom helicopter along a desired trajectory is crucial for completing complex tasks. However, three-degree-of-freedom helicopter systems exhibit typical high-order, multi-input multi-output nonlinear characteristics, with strong dynamic coupling effects between their channels. Furthermore, the inherent parameter uncertainties, unmodeled dynamics, and susceptibility to external disturbances significantly challenge high-precision trajectory tracking control. Adding to the complexity, in practical engineering applications, small unmanned helicopters often need to perform tasks in complex and constrained environments such as bridge inspection, pipeline detection, and operations in confined spaces. These scenarios involve constraints with specific geometric features, meaning the aircraft's flight path and attitude must meet the physical boundaries of the bridge structure, pier spacing, and aperture size; otherwise, collision risks exist. Moreover, as unmanned aerial vehicle systems develop towards networking and swarming, the transmission of control signals in limited bandwidth communication channels becomes a bottleneck, and quantization effects inevitably introduce control accuracy loss and chattering phenomena. Meanwhile, the actuators may experience bias faults or failures during long-term service, further deteriorating the system's tracking performance and flight safety.

[0003] Therefore, how to achieve high-precision, fast, safe and reliable trajectory tracking control of a three-degree-of-freedom helicopter under the coexistence of multiple adverse factors such as bridge hole constraints, input quantization, and actuator failure has become the current research frontier and challenge. Summary of the Invention

[0004] The present invention provides a three-degree-of-freedom helicopter preset time optimal fault-tolerant control method that considers bridge hole constraints and input quantization, which can solve the above-mentioned problems.

[0005] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0006] A preset-time optimal fault-tolerant control method for a three-degree-of-freedom helicopter includes the following steps:

[0007] S1: Establish a three-degree-of-freedom helicopter nonlinear dynamics model that considers actuator failure and input quantization;

[0008] S2: Construct a preset time function and a switchable bridge hole constraint function, and use the bridge hole constraint function to constrain the attitude tracking error of the three-degree-of-freedom helicopter to ensure that the attitude tracking error remains within the feasible domain set according to the physical size of the bridge hole within a preset time.

[0009] S3: Based on the backstepping control framework, the nonlinear dynamic model is decoupled through coordinate transformation to construct a two-level subsystem; combined with dynamic surface control technology, a first-order low-pass filter is introduced to dynamically process the virtual control law, eliminating the complexity explosion problem caused by repeated differentiation of the virtual control law;

[0010] S4: For the two decoupled subsystems, performance evaluation index functions are constructed respectively, and a fuzzy logic system with an execution-evaluation-identification structure is introduced to perform online approximation and dynamic compensation for system uncertainty, actuator failure, and errors generated by input quantization; an adaptive optimal fault-tolerant control strategy is designed based on a reinforcement learning framework, and the weights and adaptive compensation parameters of the execution network, evaluation network, and identification network in the fuzzy logic system are updated online respectively to achieve a synergistic improvement in the system's optimal control performance and fault tolerance capability.

[0011] Further, in step S1, the nonlinear dynamic model is expressed as:

[0012]

[0013] In the formula: Represents the helicopter's climb and pitch angles; Represents the helicopter's climb rate and pitch rate; For system state variables; For nominal dynamics; This represents system uncertainty; For control gain matrix; External interference; For system output; To control the input signal; The mathematical description of the composite control input, which is affected by both actuator failure and input quantization, is as follows:

[0014]

[0015] Where t is time; Characterize actuator efficiency, These are uncontrollable efficiency-related parameters; For uncontrollable additive actuator failure, For parameters related to additive faults; hysteresis quantizer Defined as a piecewise function, satisfying ,in and , and For quantizer parameters, Design parameters for the quantizer. This represents the dead zone range of the quantizer.

[0016] Furthermore, in step S2, a preset time function is used. for:

[0017]

[0018] in, is a positive design constant; T is the preset convergence time. Preset steady-state accuracy; preset time function Satisfy: Under the condition and Under these conditions, transient performance is satisfied. Steady-state performance meets ,in The first of the attitude tracking errors One portion, and It is a time-varying boundary function.

[0019] Furthermore, in step S2, the bridge hole constraint function for:

[0020]

[0021] in, To enter the bridge opening constraint moment, To exit the bridge hole constraint at the time when the constraint is satisfied ; For attenuation adjustment parameters; It is a positive integer.

[0022] Furthermore, in step S2, the specific method for constraining the attitude tracking error includes:

[0023] Define error transformation Tracking error Through the bridge hole constraint function Air conditioning in progress, among which for The equivalent representation of; System status The One portion, For the desired trajectory;

[0024] Constructing asymmetric barrier functions :

[0025]

[0026] Through coordinate transformation By mapping the constrained tracking error to the real number field, the constrained tracking problem is transformed into an unconstrained stability problem.

[0027] Furthermore, in step S3, the coordinate transformation and the introduction of the first-order low-pass filter include:

[0028] Define the error coordinate transformation of the two-level subsystem:

[0029]

[0030] in For the second-level error variable, For system state variables, This is the filtered output signal;

[0031] Design virtual control law As the desired input to the first-stage subsystem, a first-order low-pass filter is introduced to affect the virtual control law. Perform filtering processing to obtain the filtered output signal. The first-order low-pass filter is in the following form:

[0032]

[0033] in, Let be the filtering time constant; define the filtering error as . .

[0034] Furthermore, in the adaptive optimal fault-tolerant control strategy described in step S4, the virtual control law of the first-level subsystem... Designed as follows:

[0035]

[0036] in, For gain terms; These are partial derivative terms; and For intermediate conversion items; For terms that include the partial derivatives of the barrier function; For the corresponding partial derivative terms; ; Design constant; Weight estimation for the first layer of the execution network; For fuzzy basis function vectors; This is the input vector of the fuzzy logic system.

[0037] Furthermore, in the adaptive optimal fault-tolerant control strategy described in step S4, the actual control law of the second-level subsystem... Designed as follows:

[0038]

[0039] in, Design constant; Weight estimation for the second layer of the execution network; To identify network weight estimates; The estimated value of the adaptive compensation parameter represents the upper bound of the composite uncertainty; and For fuzzy basis function vectors; and The input vector.

[0040] Furthermore, in the fuzzy logic system described in step S4, the online weight update law for the execution network and the evaluation network is as follows:

[0041]

[0042] in, The corresponding level index of the two-level subsystem; and These are respectively performing network weight estimation and evaluating network weight estimation; Let be the fuzzy basis function vector of the j-th level; and The learning rate parameter is positive.

[0043] Furthermore, in the fuzzy logic system described in step S4, the online update law of the weights of the identification network and the online update law of the adaptive compensation parameters are as follows:

[0044]

[0045]

[0046] in, and These are identification network weight estimation and adaptive estimation, respectively; For the corresponding fuzzy basis function vectors of the identification network; , A learning rate parameter >0 is positive. >0 is a positive adjustment parameter.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] (1) This invention constructs a preset time function and a switchable bridge hole constraint function, realizing a smooth dynamic switching between strict constraints on the output state during critical time periods and relaxed constraints during non-critical time periods. This mechanism effectively overcomes the shortcomings of excessive constraints on the system by the monotonic symmetric envelope in traditional preset performance control, avoiding actuator saturation and unnecessary energy consumption; at the same time, it enables the tracking error to converge to the preset accuracy within any time directly specified by the user, and the convergence time is completely independent of the initial state of the system or the control gain, meeting the task requirements with strict time window and spatial physical boundary constraints.

[0049] (2) This invention incorporates dynamic surface control technology into the backstepping control framework. By introducing a first-order low-pass filter to filter the virtual control law, the filtered output signal is used to replace the original direct differentiation operation. This effectively eliminates the complexity explosion problem caused by repeated differentiation of the virtual control law in traditional backstepping control methods, significantly reduces the computational load of the controller, and improves the real-time performance and engineering feasibility of the algorithm.

[0050] (3) This invention innovatively introduces a fuzzy logic system with an execution-evaluation-identification structure and designs an adaptive optimal control strategy based on a reinforcement learning framework. This architecture can not only achieve online approximation and dynamic compensation for system uncertainties, input quantization errors, and actuator failures under unknown disturbances, but also minimize the performance index function, which includes tracking errors and control energy, by solving the Hamilton-Jacobi-Bellman equation. Compared with existing methods that only focus on fault compensation, this invention strictly follows the principle of optimality and achieves a synergistic balance between high fault tolerance accuracy and low energy consumption.

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, embodiments of the present invention are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of the dynamics analysis of a three-degree-of-freedom helicopter, showing the forces acting on the pitch axis, elevator axis, and rotation axis.

[0054] Figure 2 This is a structural diagram of the Quanser three-degree-of-freedom helicopter experimental platform;

[0055] Figure 3 This is a tracking curve of the climb angle versus the desired trajectory under the constraint of the bridge hole of a three-degree-of-freedom helicopter.

[0056] Figure 4 This is a tracking curve of the pitch angle versus the desired trajectory under the constraint of the bridge opening of a three-degree-of-freedom helicopter.

[0057] Figure 5 This is a graph showing the tracking curves of the climb and pitch angles of a three-degree-of-freedom helicopter to the desired trajectory under intermittent constraints.

[0058] Figure 6 The response curve of the tracking error of a three-degree-of-freedom helicopter system is shown.

[0059] Figure 7 A control input curve diagram for a three-degree-of-freedom helicopter system;

[0060] Figure 8 A diagram illustrating the tracking effect of a three-degree-of-freedom helicopter in three-dimensional space.

[0061] Figure 9 Convergence curve of the weight norm for a fuzzy logic system for a three-degree-of-freedom helicopter;

[0062] Figure 10 Convergence curve of the weight norm for evaluating a fuzzy logic system for a three-degree-of-freedom helicopter;

[0063] Figure 11 Convergence curves of the weight norm and adaptive parameter update law for identifying fuzzy logic systems for three-degree-of-freedom helicopters. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0065] Please refer to Figure 1 and Figure 2 This invention provides a three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method considering bridge hole constraints and input quantization, comprising the following steps:

[0066] Step 1, the 3-DOF helicopter system targeted by this invention, such as Figure 1 As shown, the system has three degrees of freedom: pitch, elevation, and travel. Based on the Euler-Lagrange equations, the complete dynamic equations of the system are:

[0067] (1)

[0068] In the formula: , , These are the moments of inertia of the pitch axis, rise axis, and rotation axis, respectively. This is the propeller thrust coefficient;

[0069] , , These are the pitch axis lever arm, the lifting axis lever arm, and the counterweight lever arm;

[0070] , These are the helicopter's main body mass and its counterweight mass, respectively. , These are the masses of the front and rear propellers, respectively.

[0071] It is the acceleration due to gravity; , These are the input voltages for the front and rear motors, respectively.

[0072] , , These are pitch angle, rise angle, and roll angle, respectively.

[0073] , , These represent unknown bounded external disturbances for each axis.

[0074] Detailed parameters of the three-degree-of-freedom helicopter system model are shown in Table 1:

[0075] Table 1: Model Parameters of a Three-Degree-of-Freedom Helicopter System

[0076]

[0077] To facilitate controller design, system state variables are defined. ,in , Therefore, the dynamic model of a three-degree-of-freedom helicopter system can be expressed as follows:

[0078] (2)

[0079] In the formula: Due to system uncertainty, External interference;

[0080] For composite control inputs affected by actuator failure and input quantization, This is the output of the hysteresis quantizer. and The gain matrices for the three-degree-of-freedom helicopter system model are as follows:

[0081]

[0082]

[0083] System Input Due to the combined effects of actuator failure and input quantization, its mathematical description is as follows:

[0084] (3)

[0085] in, Characterize actuator efficiency, This corresponds to an uncontrollable additive actuator failure. The hysteresis quantizer... The complete mathematical definition is:

[0086]

[0087] in, , The parameters satisfy , ;

[0088] Quantization output , This represents the dead zone range of the quantizer. The hysteresis quantizer has the following mathematical properties:

[0089] (4)

[0090] in and satisfy and .

[0091] Step 2: The bridge hole constraint function constructed in this invention is used to achieve a smooth switch between strictly constrained and unconstrained states of the system output within a specific time period. The specific construction is as follows:

[0092] (5)

[0093] Among them, and During certain time periods, the system output is unrestricted; The system output must be strictly limited to a specific time interval. Inside, and It is a time-varying boundary function.

[0094] Define tracking error To embed constraints into the controller design, the following error transformation is introduced:

[0095] (6)

[0096] Constructing asymmetric barrier functions:

[0097] (7)

[0098] right Differentiation yields:

[0099] (8)

[0100] in

[0101]

[0102] To achieve convergence within a preset time, a preset time function is defined. :

[0103] (9)

[0104] in To preset the convergence time, .

[0105] Associate the bridge hole constraint boundary with a preset time function:

[0106] ,

[0107] Thus ensuring when When the tracking error converges to To achieve steady-state accuracy.

[0108] Step 3: To simplify the controller structure and avoid the "complexity explosion" problem in the traditional backstepping method, based on the equivalent error obtained in Step 2 and the dynamic surface control technique, the following error coordinate transformation is designed:

[0109] (10)

[0110] in, For system state variables, This is the filtered output signal. To obtain... Introduce the following first-order low-pass filter:

[0111] (11)

[0112] in, The filter time constant; Let be the virtual control law to be designed. Define the filtering error as... Its dynamics satisfy:

[0113] (12)

[0114] In the formula: ,and It is a positive constant.

[0115] The three-degree-of-freedom helicopter system model is decomposed into two-level subsystems, and the first-level subsystem transformation error variable is introduced. Second-level subsystem transformation error variable .

[0116] Step 4: Based on the coordinate transformation obtained in Step 3, adjust the first-level error variables after the transformation. Taking the derivative, we get:

[0117] (13)

[0118] in: , ,

[0119] and For terms containing the partial derivatives of the barrier function: .

[0120] To achieve optimal control at the preset time, a first-level optimal performance function is constructed for the first-level subsystem. :

[0121] (14)

[0122] in, For including the origin The allowable control set on a compact set, Discount factor;

[0123] Represents the cost function, This is the optimal control virtual controller. The corresponding Hamilton-Jacobi-Bellman equations are:

[0124] (15)

[0125] In the formula, Describes the optimal cost function Regarding state variables The gradient vector.

[0126] By optimality condition The ideal optimal virtual control law can be obtained as follows:

[0127] (16)

[0128] in, .However, Since the value is unknown, it needs to be approximated online using a subsequent AC fuzzy system. Therefore, the gradient term is designed as follows:

[0129] (17)

[0130] in, For design constants, ,and The residual term of the unknown optimal cost function will be estimated using a fuzzy logic system.

[0131] Using the universal approximation property, the FLS approximation is performed. :

[0132] (18)

[0133] in, The input vector; Ideal weights;

[0134] For fuzzy basis function vectors; This is the bounded approximation error.

[0135] Based on this, the virtual control law that can be realized and cost gradient estimation They are designed as follows:

[0136] (19)

[0137] (20)

[0138] in, and These are the weights for the execution network and the evaluation network, respectively. Their update law is:

[0139] (twenty one)

[0140] in, For fuzzy basis function vectors, and The learning rates are for the execution network and the evaluation network, respectively. and for The estimated value.

[0141] For the first-level subsystem of a three-degree-of-freedom helicopter system model, the following Lyapunov function candidate is constructed:

[0142] (twenty two)

[0143] in: These are the first-level state variables after error transformation; To calculate the estimation error of the network weights;

[0144] To evaluate the estimation error of the network weights, we can differentiate the Lyapunov function with respect to time, and obtain:

[0145] (twenty three)

[0146] Dynamics of the first-level subsystem (13) (Including virtual control laws) (19) Filtering error Substituting (12) and the weight update law (21), and using Young's inequality and the properties of fuzzy basis functions, we finally obtain:

[0147] (twenty four)

[0148] Through the Lyapunov analysis described above, the value of the gain of the virtual controller to be designed in the first-level subsystem is obtained during the inequality simplification process.

[0149] Step 5: Based on the coordinate transformation obtained in Step 3, adjust the transformed second-level error variables. Taking the derivative, we get:

[0150] (25)

[0151] in, This represents the lumped uncertainty that includes nominal dynamics, modeling errors, and external disturbances;

[0152] This represents the combined effect of actuator efficiency loss and input quantization deviation.

[0153] Define the performance index function for the second-level subsystem:

[0154] (26)

[0155] in It is a discount factor. Let be the cost function. The corresponding HJB equation is:

[0156] (27)

[0157] By invoking the optimality condition The ideal optimal control law can be obtained. as follows:

[0158] (28)

[0159] in, It is a positive design constant. For subsequent design of intermediate control laws, Let be the positive constants used for smoothing. The adaptive law is designed as follows:

[0160] (29)

[0161] in, Representing unknown parameters The estimated value, and These are the parameters to be designed.

[0162] Similar to the first level, fuzzy logic systems (FLSs) are used to... and Perform online estimation:

[0163] (30)

[0164] in: and The input vector; and The ideal weight vector; and For fuzzy basis function vectors;

[0165] , For bounded approximation error, satisfying , .

[0166] To ensure excellent trajectory tracking performance and compensate for system uncertainties, an adaptive optimal actual control law is used. and the corresponding cost function to estimate the gradient They are designed as follows:

[0167] (31)

[0168] (32)

[0169] in and These are the online update weights for the execution network and the identification network, respectively. Accordingly, the Execution-Evaluation-Identification (ACI) adaptive update law is formulated as follows:

[0170] (33)

[0171] In the formula: and for The estimated value; for The estimated value; , and These are the parameters to be designed.

[0172] For a three-degree-of-freedom helicopter system model, the composite Lyapunov function of the second-stage subsystem is defined as:

[0173] (34)

[0174] In the formula: , , as well as This is for estimating the error.

[0175] Furthermore, the Lyapunov function Taking the derivative with respect to time, we get:

[0176] (35)

[0177] The intermediate control law is designed as follows:

[0178]

[0179] Dynamics of the second-level subsystem (25) Actual control law (31) ACI weight update law (33) Filter error dynamics and adaptive law (29) Substituting the values ​​and using Young's inequality and the properties of fuzzy basis functions, we can obtain the following after simplification:

[0180] (36)

[0181] Through the Lyapunov analysis described above, the value of the virtual controller gain to be designed in the second-level subsystem is obtained during the inequality simplification process.

[0182] Step 6: Verify the results obtained in steps 4 and 5, and construct the global Lyapunov function for the three-degree-of-freedom helicopter system model: ,Will and By adding the upper bounds of the derivatives, we can obtain the derivative of the overall Lyapunov function as:

[0183] (37)

[0184] in, , Let be a bounded constant, and its expression is:

[0185]

[0186] By considering the global Lyapunov function The derivative analysis shows that when appropriate design parameters are selected, At that time, according to Lyapunov's stability theorem, all signals within the closed-loop system... All are semi-globally consistent and eventually bounded.

[0187] This result verifies that, under conditions of bridge hole asymmetry constraints, actuator failure, and input quantization, the control scheme proposed in this invention can ensure that a three-degree-of-freedom helicopter can achieve its predetermined time target. It achieves high-precision trajectory tracking and keeps the output state strictly within the safety constraints, thus solving the robust optimization problem of helicopter control in complex mission environments.

[0188] To illustrate the control effect of the method of this invention in detail, a simulation experiment will be conducted in MATLAB, with the reference trajectory set as follows:

[0189]

[0190] The actuator fault model is designed as follows:

[0191]

[0192] The constraint boundaries are set as follows:

[0193]

[0194] In the simulation verification of this invention embodiment, the motor voltage is limited to... The system workspace constraints are:

[0195] Lift angle Rotation angle Pitch angle .

[0196] The system initial conditions are set as follows: , .

[0197] The preset time function parameters are selected as follows: , , , , .

[0198] The controller gain is shown in Table 2.

[0199] Table 2: Controller Gain

[0200]

[0201] The control scheme proposed in this invention has been verified using the MATLAB simulation platform. Figures 3 to 9 The complete simulation results are presented. Specifically:

[0202] Figures 3 to 5 The output tracking performance of the system under constraints was characterized. Specifically, this was achieved through... Figure 3 and Figure 4 The simulation results were compared and analyzed to verify the effectiveness of the control strategy proposed in this invention in handling bridge hole constraints. Figure 5 This further demonstrates the system's robust tracking performance under intermittent constraints. Figure 6 The tracking error is depicted under a preset time adaptive optimal control scheme. The dynamic evolutionary trajectory. Figure 7 and Figure 8 The control input curves and trajectory tracking performance in three-dimensional space of the system are displayed, intuitively reflecting the control accuracy and spatial convergence of the control system. Regarding the internal characteristics of the controller, Figure 9 and Figure 10 The weight norm variation curves of the fuzzy logic systems for the actuator and evaluator are presented separately. Furthermore, Figure 11 The evolution trajectory of the weight norm of the fuzzy logic system of the identifier and the convergence process of the adaptive parameter update law are shown. Simulation results show that all weights and adaptive parameters can converge quickly to a stationary state, proving the stability and effectiveness of the proposed control algorithm.

[0203] In summary, the simulation results fully demonstrate that the control scheme proposed in this invention exhibits significant robustness and superiority in dealing with actuator failures, input quantization, and complex spatiotemporal constraints. Specifically, under the dynamic adjustment of the adaptive weight update law designed in this invention, the system achieves the expected trajectory tracking control objective; while ensuring control accuracy, it not only minimizes the performance cost function but also strictly adheres to the Bellman optimality principle, effectively reducing system energy consumption and improving task execution safety in complex environments.

[0204] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A preset-time optimal fault-tolerant control method for a three-degree-of-freedom helicopter, characterized in that, Includes the following steps: S1: Establish a three-degree-of-freedom helicopter nonlinear dynamics model that considers actuator failure and input quantization; S2: Construct a preset time function and a switchable bridge hole constraint function, and use the bridge hole constraint function to constrain the attitude tracking error of the three-degree-of-freedom helicopter to ensure that the attitude tracking error remains within the feasible domain set according to the physical size of the bridge hole within a preset time. S3: Based on the backstepping control framework, the nonlinear dynamic model is decoupled through coordinate transformation to construct a two-level subsystem; combined with dynamic surface control technology, a first-order low-pass filter is introduced to dynamically process the virtual control law, eliminating the complexity explosion problem caused by repeated differentiation of the virtual control law; S4: For the two-level subsystems after decoupling, construct performance evaluation index functions respectively, and introduce a fuzzy logic system with an execution-evaluation-identification structure to perform online approximation and dynamic compensation for system uncertainty, actuator failure and error generated by input quantization; An adaptive optimal fault-tolerant control strategy is designed based on a reinforcement learning framework. The weights and adaptive compensation parameters of the execution network, evaluation network, and identification network in the fuzzy logic system are updated online to achieve a synergistic improvement in the system's optimal control performance and fault tolerance capability.

2. The three-degree-of-freedom helicopter preset time optimal fault-tolerant control method according to claim 1, characterized in that, In step S1, the nonlinear dynamic model is expressed as: In the formula: Represents the helicopter's climb and pitch angles; Represents the helicopter's climb rate and pitch rate; For system state variables; For nominal dynamics; This represents system uncertainty; For control gain matrix; External interference; For system output; To control the input signal; The mathematical description of the composite control input, which is affected by both actuator failure and input quantization, is as follows: in, For time; Characterize actuator efficiency, These are uncontrollable efficiency-related parameters; For uncontrollable additive actuator failure, For parameters related to additive faults; hysteresis quantizer Defined as a piecewise function, satisfying ,in and , and For quantizer parameters, Design parameters for the quantizer. This represents the dead zone range of the quantizer.

3. The three-degree-of-freedom helicopter preset time optimal fault-tolerant control method according to claim 2, characterized in that, In step S2, a preset time function is used. for: in, is a positive design constant; T is the preset convergence time. Preset steady-state accuracy; preset time function Satisfy: Under the condition and Under these conditions, transient performance is satisfied. Steady-state performance meets ,in The first of the attitude tracking errors One portion, and It is a time-varying boundary function.

4. The three-degree-of-freedom helicopter preset time optimal fault-tolerant control method according to claim 3, characterized in that, In step S2, the bridge hole constraint function for: in, To enter the bridge hole constraint moment, To exit the bridge hole constraint at the time when the constraint is satisfied ; For attenuation adjustment parameters; It is a positive integer.

5. The three-degree-of-freedom helicopter preset time optimal fault-tolerant control method according to claim 4, characterized in that, In step S2, the specific method for constraining the attitude tracking error includes: Define error transformation Tracking error Through the bridge hole constraint function Air conditioning in progress, among which for The equivalent representation of; System status The One portion, For the desired trajectory; Constructing asymmetric barrier functions : Through coordinate transformation By mapping the constrained tracking error to the real number field, the constrained tracking problem is transformed into an unconstrained stability problem.

6. The three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method according to claim 5, characterized in that, In step S3, the coordinate transformation and the introduction of the first-order low-pass filter include: Define the error coordinate transformation of the two-level subsystem: in For the second-level error variable, For system state variables, This is the filtered output signal; Design virtual control law As the desired input to the first-stage subsystem, a first-order low-pass filter is introduced to affect the virtual control law. Perform filtering processing to obtain the filtered output signal. The first-order low-pass filter is in the following form: in, Let be the filtering time constant; define the filtering error as . .

7. The three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method according to claim 6, characterized in that, In the adaptive optimal fault-tolerant control strategy described in step S4, the virtual control law of the first-level subsystem... Designed as follows: in, For gain terms; These are partial derivative terms; and For intermediate conversion items; For terms that include the partial derivatives of the barrier function; For the corresponding partial derivative terms; ; Design constant; Weight estimation for the first layer of the execution network; For fuzzy basis function vectors; This is the input vector of the fuzzy logic system.

8. The three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method according to claim 7, characterized in that, In the adaptive optimal fault-tolerant control strategy described in step S4, the actual control law of the second-level subsystem... Designed as follows: in, Design constant; Weight estimation for the second layer of the execution network; To identify network weight estimates; The estimated value of the adaptive compensation parameter represents the upper bound of the composite uncertainty; and For fuzzy basis function vectors; and The input vector.

9. A three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method according to claim 8, characterized in that, In the fuzzy logic system described in step S4, the online weight update law for the execution network and the evaluation network is as follows: in, The corresponding level index of the two-level subsystem; and These are respectively performing network weight estimation and evaluating network weight estimation; For the first Level fuzzy basis function vector; and The learning rate parameter is positive.

10. A three-degree-of-freedom helicopter preset-time optimal fault-tolerant control method according to claim 9, characterized in that, In the fuzzy logic system described in step S4, the online update law for the weights of the identification network and the online update law for the adaptive compensation parameters are as follows: in, and These are identification network weight estimation and adaptive estimation, respectively; For the corresponding fuzzy basis function vectors of the identification network; , A learning rate parameter >0 is positive. >0 is a positive adjustment parameter.

Images