A hybrid helicopter full-mode adaptive control method

Abstract

The invention discloses a full-mode adaptive control method for a compound helicopter, and belongs to the technical field of calculation, extrapolation or counting. The invention proposes a method for realizing full-mode flight control of a compound helicopter based on an incremental dynamic inversion method, adopts a Pi‑Sigma neural network to adaptively compensate for the incremental dynamic inversion control error, and ensures the stability, rapidity and robustness of the control system through Lyapunov stability analysis.

Description

Technical Field

[0001] The present invention relates to compound helicopter control technology, and specifically discloses a full-mode adaptive control method for a compound helicopter, belonging to the technical field of calculation, reckoning or counting. Background Art

[0002] In recent years, in the command and tracking control of complex multi-modal aircraft such as tilt-rotor aircraft, near-space aircraft, hypersonic aircraft, etc., there have been various excellent methods for reference to solve similar problems. Among these advanced control methods, artificial neural network (ANN) adaptive control is an effective method for dealing with complex disturbances and model uncertainties of multi-modal aircraft. The artificial neural network has the ability to approximate continuous non-linear functions. Compared with the simple table lookup method, one advantage of the neural network is that it reduces the required memory and computing time. In addition, the neural network can provide interpolation between training points without additional computational effort. The Pi-Sigma neural network (PSNN) realizes fast non-linear approximation by introducing summation neurons and multiplication neurons. The PSNN has a simple structure, few hyperparameters, and high convergence efficiency. At present, PSNN has been widely applied in various fields and is more suitable for solving the control problems of compound helicopters.

[0003] However, some factors limit the popularization of traditional neural network adaptive control. The existing neural networks in the control field, including single-layer perceptrons and radial basis function networks, are too simple to effectively process uncertain non-linear functions in complex systems online. In addition, although PSNN has excellent performance, the research on PSNN adaptive control is still in its infancy. The PSNN adaptive fuzzy controllers proposed by experts and scholars are most commonly used to handle tracking problems. These fuzzy controllers require an offline training process to approximate the optimal membership function, but this offline training cannot adapt to the inevitable complex environment and unknown disturbances in aerospace engineering. Other methods mainly conduct research on PSNN for relatively simple systems such as hydraulic control systems. The simple gradient descent adaptive law used in these methods cannot guarantee the convergence robustness in more complex systems.

[0004] In order to ensure the robustness against partial uncertain cross-coupling and model errors, the present application aims to propose a PSNN adaptive algorithm based on Lyapunov theory. Summary of the Invention

[0005] The object of the present invention is to address the deficiencies in the background art and propose a full-mode adaptive control method for a compound helicopter. Based on the incremental dynamic inversion (INDI) control framework, a Pi-Sigma neural network is used to adaptively compensate for the incremental dynamic inversion control error, ensuring the stability, rapidity, and robustness of the control system, and solving the technical problem that the existing PSNN adaptive fuzzy controller cannot guarantee the convergence robustness under the full-mode flight of the compound helicopter.

[0006] The present invention adopts the following technical solutions to achieve the above object:

[0007] A full-mode adaptive control method for a compound helicopter includes the following four major steps.

[0008] Step 1: Establish a non-affine nonlinear dynamic system of the compound helicopter, which is represented by two subsystems:

[0009]

[0010]

[0011] In Equations (1) and (2), x 1 is the state vector of the speed loop of the non-affine nonlinear dynamic system, and x 2 is the state vector of the attitude loop of the non-affine nonlinear dynamic system. x 1 = [u, v, w], where u, v, and w represent the three velocity components in the body coordinate axis system. x 2 = [p, q, r], where p, q, and r represent the three angular velocity components in the body coordinate axis system. represents the derivative of the state vectors x 1 , x 2 . u 1 , u 2 represent the input signals of the non-affine nonlinear dynamic system. f(.), g(.) represent the system functions of the non-affine nonlinear dynamic system. f 1 (.) is the state function of the speed loop, and g 1 (.) is the control function of the speed loop. f 2 (.) is the state function of the attitude loop, and g 2 (.) is the control function of the attitude loop.

[0012] Step 2: The compound helicopter has three flight modes, including the helicopter flight mode, the transition flight mode, and the fixed-wing flight mode. By designing the full-mode control method of the speed loop incremental dynamic inversion, the flight control law can be obtained as follows:

[0013]

[0014] In Equation (3), represents the velocity loop control function g 1 (.) is the gradient in the direction of the change in u 1 and represents the derivative of the desired control command state vector of the velocity loop, v L1 is the pseudo-linear control signal obtained by compensating the incremental dynamic inverse full-mode control system of the velocity loop to have a linear transfer relationship and achieve decoupling, is the derivative of the state vector of the velocity loop of the non-affine nonlinear dynamic system, u 1 is the velocity virtual control command, u 1 0 is the sampled signal of the velocity virtual control command, K P1 and K I1 are the linear control coefficients of the velocity loop, V ad1 is the output of the neural network for compensating the model error;

[0015] From the designed attitude loop incremental dynamic inverse full-mode control method, the flight control law is:

[0016]

[0017] In Equation (4), represents the attitude loop control function g 2 (.) is the gradient in the direction of the change in u 2 and represents the derivative of the desired control command state vector of the attitude loop, v L2 is the pseudo-linear control signal obtained by compensating the incremental dynamic inverse full-mode control system of the attitude loop to have a linear transfer relationship and achieve decoupling, is the derivative of the state vector of the attitude loop of the non-affine nonlinear dynamic system, u 2 is the angular velocity virtual control command, u 2 0 is the sampled signal of the angular velocity virtual control command, K P2 and K I2 are the linear control coefficients of the attitude loop, V ad2 is the output of the neural network for compensating the model error;

[0018] Step 3: Under the action of the inaccuracy of the compound helicopter model and various disturbances, the incremental dynamic inverse controller has a certain error. Design a Pi-Sigma neural network to compensate the incremental dynamic inverse model error V ad . The input of the neural network where x c represents the system state command signal, and x represents the system state vector, denote the derivative of, b x denote the neural network bias, v L denote the linear pseudo-control signal, V ad denote the compensated incremental dynamic inverse model error value denote the two-norm of the estimated neural network weights. The neural network is rapidly approximated by introducing summation neurons and multiplication neurons. The output of the Pi-Sigma neural network is V ad ;

[0019] Step 4. Real-time dynamically detect the speeds u, v, w, attitude angles θ, ψ and attitude angular velocities p, q, r of the compound helicopter, and repeat Steps 1 to 4;

[0020] Furthermore, in a full-mode adaptive control method for a compound helicopter

[0021] The neural network compensation value V in Step 3 ad is designed as follows:

[0022] Step 3.1. The Pi-Sigma neural network is a single-hidden-layer feedforward network. The output layer is a product unit. The weights from the input layer to the hidden layer are adjustable, and the weights from the hidden layer to the output layer are fixed at 1. The input-output of the Pi-Sigma neural network can be expressed as:

[0023]

[0024]

[0025] In Equations (5) and (6), w ij is the adjustable weight of the j-th input node of the i-th hidden layer of the neural network, x j is the scalar input of the j-th input node, K is the number of hidden layers, N is the number of input nodes, y is the scalar output of the Pi-Sigma neural network, h i is the output of the i-th hidden layer, and σ(.) is the non-linear transfer function from the hidden layer to the output layer;

[0026] Step 3.2. The estimated compensation value of the neural network is where is the vector composed of the weights of each input node of the i-th hidden layer of the neural network, is the neural network input, σ(.) is the activation function of the neural network, V r is a high-order term that can be canceled under the Taylor series approximation, that is, the robust adaptive term of the neural network. The above neural network compensation value V ad is:

[0027]

[0028] Step 3.3. Construct the Lyapunov function:

[0029]

[0030] In Equation (8), W * represents the ideal neural network weights, represents the estimated ideal neural network weights, Γ w represents the learning rate of the neural network, tr(.) represents the trace of the matrix, e is the error between the system input and output. According to the Lyapunov criterion, is the estimated value of e, the matrix P is the solution of the Lyapunov equation A T P + PA = -Q, A is the state matrix of the speed loop control system or the attitude loop control system, and Q is taken as the identity matrix.

[0031]

[0032] In Equation (9), ε is the reconstruction error of the neural network, δ is the total uncertainty of the system, b is the matrix used to construct the Lyapunov function, Let ξ = e T Pb.

[0033] The derivative of the Lyapunov function with respect to time is:

[0034]

[0035] In Equation (10), A T P + PA = -Q and Q = Q T > 0, λ is the design parameter of the system, which can determine the balance between control performance and robustness, is a higher-order infinitesimal of. From Equation (10), the adaptive estimation weight of the neural network based on the Lyapunov principle can be derived as the update law:

[0036]

[0037] In Equation (11), is the weight from the estimated neural network input to the hidden layer, is the estimated ideal neural network weight of the output of the hidden layer below, is the ideal value of the weight of the j-th input node of the i-th hidden layer, Γ w represents the neural network learning rate, W0 denotes the initial weights of the neural network, is the compensation value estimated by the neural network In the direction of the gradient, represents the e correction of the PSNN adaptive control, which is used to ensure the robustness when approaching the error. λ is a parameter that determines the balance between the control performance and the robustness.

[0038] The derivative of the Lyapunov function with respect to time is:

[0039]

[0040] In Equation (12),

[0041] In Equation (13), ρ(P) represents the spectral radius of the positive definite matrix. Take Q = I, where I is the identity matrix.

[0042] In Equation (12),

[0043] The neural network input can be maximally limited in terms of the tracking performance as:

[0044]

[0045]

[0046] In Equation (15) and Equation (16), c′ 0 , c′ 1 , c′ 2 , c′ 3 , c′ 4 are arbitrarily given constants.

[0047] According to Equation (15) and Equation (16), it can be seen that

[0048]

[0049] According to Equation (13), Equation (14), Equation (15), and Equation (16), it can be seen that

[0050]

[0051] The robust adaptive term V of the neural network can be deduced as: r is:

[0052]

[0053] In Equation (18), K r1 , K r2 are the robust gains.

[0054] The derivative of the Lyapunov function with respect to time is as follows:

[0055]

[0056]

[0057] When λ satisfies then is less than 0. Therefore, the error generated by the system will converge to zero, ensuring the stability of the flight system.

[0058] The present invention adopts the above technical solutions and has the following beneficial effects:

[0059] (1) The present invention proposes a new adaptive control scheme, regarding the aerodynamic cross-coupling between control surfaces as partial uncertain disturbances. An incremental dynamic inversion framework composed of an inner attitude control loop and an outer velocity control loop is proposed to decouple the overactuated system. The weight update law of the Pi-Sigma neural network is designed using the Lyapunov principle to achieve the adaptive correction of the PSNN for errors, maintain the robustness to weight perturbations in the Pi-Sigma neural network, obtain the adaptive term that meets the robustness requirements according to the input tracking performance of the neural network, and introduce the adaptive term into the compensation terms of the input increments of the inner attitude control loop and the outer velocity control loop, so that the compound helicopter can complete system command tracking within the set finite time, and the convergence is better.

[0060] (2) The attitude loop control system based on the incremental dynamic inversion model of the present invention can decouple each channel of the compound helicopter and enhance the stability of the flight system; the velocity loop control system based on the incremental dynamic inversion model takes the velocity quantity as the design target, improving the command tracking ability and anti-interference ability of the flight system.

[0061] (3) The present invention introduces a robust adaptive term and e correction to correct potential parameter drifts, ensuring the robustness of the convergence process. Compared with the existing adaptive laws, this design can better balance the stability and non-linear mapping ability of the PSNN. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a structural block diagram of a full-mode adaptive control method for a compound helicopter according to the present invention.

[0063] Figures 2 to 4 is a time history diagram of the F norm of the weight matrix of the present invention for velocity control.

[0064] Figures 5 to 7 is a time history diagram of the F norm of the weight matrix of the present invention for attitude control. DETAILED DESCRIPTION OF THE INVENTION

[0065] For the convenience of those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0066] As Figure 1 shown in the structural block diagram of a composite helicopter full-mode adaptive control method, it includes the following four steps.

[0067] Step 1: Establish a composite helicopter non-affine nonlinear dynamic system, which is represented by two subsystems:

[0068]

[0069]

[0070] In Equations (1) and (2), x 1 is the state vector of the speed loop of the non-affine nonlinear dynamic system, and x 2 is the state vector of the attitude loop of the non-affine nonlinear dynamic system. x 1 = [u, v, w], where u, v, and w represent the three velocity components in the body coordinate axis system, and x 2 = [p, q, r], where p, q, and r represent the three angular velocity components in the body coordinate axis system. represents the derivative of the state vector x 1 , x 2 of the non-affine nonlinear dynamic system. u 1 , u 2 represents the input signal of the non-affine nonlinear dynamic system, and f(.), g(.) represent the system functions of the non-affine nonlinear dynamic system.

[0071] Step 2: The composite helicopter has three flight modes, including the helicopter flight mode, the transition flight mode, and the fixed-wing flight mode. By designing the speed loop incremental dynamic inverse full-mode control method, the flight control law can be obtained as:

[0072]

[0073] In Equation (3), represents the derivative of the desired control command state vector of the speed loop. K P1 and K I1 are the linear control coefficients of the speed loop. V ad1 is the output of the neural network, which is used to compensate for the model error;

[0074] By designing the attitude loop incremental dynamic inverse full-mode control method, the flight control law can be obtained as:

[0075]

[0076] In Equation (4), represents the derivative of the desired control command state vector of the attitude loop, K P2 and K I2 are the linear control coefficients of the attitude loop, V ad2 is the output of the neural network, used to compensate for the model error.

[0077] Step 3: Under the action of the inaccuracy of the compound helicopter model and various disturbances, the incremental dynamic inverse controller has a certain error. Design a Pi-Sigma neural network to compensate for the incremental dynamic inverse model error V ad , the input of the neural network where x c represents the system state command signal, x represents the system state vector, represents the derivative of, b x represents the neural network bias, v L represents the linear pseudo-control signal, V ad represents the compensated incremental dynamic inverse model error value, represents the two-norm of the estimated neural network weights. The neural network is approximated rapidly by introducing summing neurons and multiplication neurons. The output of the Pi-Sigma neural network is V ad ;

[0078] Step 3.1: The Pi-Sigma neural network is a single-hidden-layer feedforward network. The output layer is a product unit. The weights from the input layer to the hidden layer are adjustable, and the weights from the hidden layer to the output layer are fixed at 1. The input-output of the Pi-Sigma neural network can be expressed as:

[0079]

[0080]

[0081] In equations (5) and (6), w ij is the adjustable weight of the j-th input node of the i-th hidden layer of the neural network, x j is the scalar input of the j-th input node, K is the number of hidden layers, N is the number of input nodes, y is the scalar output of the Pi-Sigma neural network, h i is the output of the i-th hidden layer, and σ(.) is the non-linear transfer function from the hidden layer to the output layer;

[0082] Step 3.2: Estimate the compensation value of the neural network where is the vector composed of the weights of each input node of the i-th hidden layer of the neural network, is the input of the neural network, σ(.) is the activation function of the neural network, V ris a high-order term that can be canceled under the Taylor series approximation, which is the robust adaptive term of the neural network. The above neural network compensation value V ad is as follows:

[0083]

[0084] Step 3.3. Construct the Lyapunov function:

[0085]

[0086] In equation (8), W * represents the ideal neural network weight, represents the estimated ideal neural network weight, Γ w represents the learning rate of the neural network, tr(.) represents the trace of the matrix, e is the error between the system input and output. According to the Lyapunov criterion, the matrix P is the solution of the Lyapunov equation A T P + PA = -Q, A is the state matrix of the system, and Q is taken as the identity matrix.

[0087]

[0088] In equation (9), ε is the reconstruction error of the neural network, δ is the total uncertainty of the system. Let ξ = e T Pb.

[0089] The derivative of the Lyapunov function with respect to time is:

[0090]

[0091] In equation (10), A T P + PA = -Q and Q = Q T > 0, λ is the design parameter of the system, which can determine the balance between control performance and robustness, is a higher-order infinitesimal of. From equation (10), the adaptive prediction weight of the neural network based on the Lyapunov principle can be deduced as the update law:

[0092]

[0093] In equation (11), Γ w represents the neural network learning rate, W 0 represents the initial weight of the neural network, represents the e correction of the PSNN adaptive control, which is used to ensure the robustness when approaching the error.

[0094] The derivative of the Lyapunov function with respect to time is:

[0095]

[0096] In Equation (12),

[0097] In Equation (13), ρ(P) represents the spectral radius of a positive definite matrix. Let Q = I, where I is the identity matrix.

[0098] In Equation (12),

[0099] The neural network input can be maximally limited in terms of tracking performance as:

[0100]

[0101]

[0102] In Equation (15) and Equation (16), c′ 0 , c 1 ′, c′ 2 , c 3 ′, c′ 4 are arbitrarily given constants.

[0103] According to Equation (15) and Equation (16), it can be seen that

[0104]

[0105] According to Equation (13), Equation (14), Equation (15), and Equation (16), it can be seen that

[0106]

[0107] The robust adaptive term V of the neural network can be deduced as: r is:

[0108]

[0109] In Equation (18), K r1 , K r2 are robust gains.

[0110] The derivative of the Lyapunov function with respect to time is:

[0111]

[0112]

[0113] When λ satisfies then, Less than 0, so the error generated by the system will converge to zero, ensuring the stability of the flight system.

[0114] Step Four: Dynamically detect the speed u, v, w, attitude angles θ, ψ and attitude angular velocities p, q, r of the compound helicopter in real time, and repeat Steps One to Four.

[0115] The present invention conducts command tracking simulation on the compound helicopter. As Figures 2 to 7 can be seen, the weight matrix norm converges within 10 seconds, showing strong robustness. Although inevitable oscillation phenomena occur in the weights during the network convergence process, the designed robust adaptive term and damping term still help maintain the stability of the PSNN. During the first maneuver, after a series of small impacts, the convergence of the weights is completed. In addition, during the second maneuver, although the process is more intense, the norm of the weight matrix still remains stable, which means that the PSNN has successfully eliminated the uncertain nonlinear perturbations and successfully avoided local minima.

Claims

1. A full-mode adaptive control method for a compound helicopter, characterized in that: A non-affine nonlinear dynamic model of a composite helicopter including a speed loop subsystem and an attitude loop subsystem is established. The non-affine nonlinear dynamic model of the composite helicopter updates the state vector of the nonlinear dynamic system according to the input signal of the non-affine nonlinear dynamic system and the real-time flight state of the composite helicopter. In the non-affine nonlinear dynamic model of the composite helicopter, the speed loop subsystem is The attitude loop subsystem is Among them, x1 is the state vector of the velocity loop of the nonlinear dynamic system, x2 is the state vector of the attitude loop of the nonlinear dynamic system, x1 = [u, v, w], u, v, w are the three velocity components in the body coordinate axis system, x2 = [p, q, r], p, q, r are the three angular velocity components in the body coordinate axis system, is the derivative of the state vector x1, x2 of the nonlinear dynamic system, u1 is the speed virtual control instruction, u2 is the angular velocity virtual control instruction, f1(.) is the state function of the speed loop, g1(.) is the control function of the speed loop, f2(.) is the state function of the attitude loop, and g2(.) is the control function of the attitude loop; Establish a speed loop incremental dynamic inverse full-mode control system based on the incremental dynamic inverse model, calculate the input increment of the speed loop subsystem based on the incremental dynamic inverse model according to the pseudo-linear control signal after Pi-Sigma neural network compensation, and update the input signal of the speed loop subsystem; Establish an incremental dynamic inverse full-mode control system for the attitude loop based on the incremental dynamic inverse model, calculate the input increment of the attitude loop subsystem based on the incremental dynamic inverse model according to the pseudo-linear control signal after Pi-Sigma neural network compensation, and update the input signal of the attitude loop subsystem; A Pi-Sigma neural network is used to compensate for the pseudo-linear control signal of a speed loop incremental dynamic inverse full-mode control system and the pseudo-linear control signal of an attitude loop incremental dynamic inverse full-mode control system, wherein a robust adaptive term is accumulated on the output value of the Pi-Sigma neural network as a compensation value, the weight update law of the Pi-Sigma neural network is designed based on the Lyapunov principle, the robust adaptive term is obtained by deriving the derivative of the Lyapunov function with respect to time with the goal of maximally tracking the input of the Pi-Sigma neural network, and the robust adaptive term is Among them, V r is the robust adaptive term, K r1 , K r2 is the robust gain, ξ=e T Pb,e is the error between the input and output of the Pi-Sigma neural network, P is the matrix that satisfies the Lyapunov equation, and b is the matrix used to construct the Lyapunov function. is the estimated neural network weight, for The second norm of is the difference between the estimated neural network weights and the ideal neural network weights, To estimate the ideal neural network weights The output of the next hidden layer, the weight update law of the Pi-Sigma neural network is in, for The estimated value of is the estimated weight of the neural network input to the hidden layer, is the vector of estimated weights of the i-th hidden layer, Γ w is the learning rate of the Pi-Sigma neural network, is the input of the Pi-Sigma neural network, b x is the Pi-Sigma neural network bias, x c is the subsystem's expected control instruction state vector, x is the subsystem's state vector, for The derivative of L is the linear pseudo control signal of the incremental dynamic inverse full-mode control system, V ad is the error value of the linear pseudo control signal of the Pi-Sigma neural network compensation incremental dynamic inverse full-mode control system, The estimated weights of the neural network input to the hidden layer The output of the next hidden layer, K is the number of hidden layers, N is the number of input nodes, is the estimated ideal value of the weight of the jth input node in the ith hidden layer, σ(.) is the nonlinear transfer function from the hidden layer to the output layer, W0 is the initial weight of the Pi-Sigma neural network, Compensation value estimated for Pi-Sigma neural network exist Gradient in direction, ξ=e T Pb,e is the error between the input and output of the neural network, and the matrix P is the Lyapunov equation A T The solution of P+PA=-Q, A is the state matrix of the speed loop control system or the attitude loop control system, Q is taken as the unit matrix, and b is the matrix used to construct the Lyapunov function. λ is a parameter that determines the balance between control performance and robustness. The expression of the pseudo-linear control signal of the incremental dynamic inversion full-mode control system using the Pi-Sigma neural network compensation speed loop is: Among them, v L1 The speed loop increment dynamic inverse full-mode control system is compensated to become a pseudo-linear control signal with a linear transfer relationship and complete decoupling, K P1 and K I1 is the speed loop linear control coefficient, u c is the forward flight speed, v c is the lifting speed, w c is the yaw velocity, (u, v, w) is the real-time three-axis velocity component in the body coordinate system, V ad1 is the compensation value output by the Pi-Sigma neural network, is the output value of the Pi-Sigma neural network, V r is a robust adaptive term, and the expression of the pseudo-linear control signal of the incremental dynamic inversion full-mode control system using the Pi-Sigma neural network to compensate for the attitude loop is: Among them, v L2 In order to compensate the attitude loop incremental dynamic inverse full-mode control system into a pseudo-linear control signal with a linear transfer relationship and complete decoupling, K P2 and K I2 is the linear control coefficient of the attitude loop, (p, q, r) is the real-time three-axis angular velocity component in the body coordinate system, [p c ,q c ,r c ] is the three-axis angular velocity component obtained according to the expected control command of the attitude loop, V ad2 is the compensation value output by the Pi-Sigma neural network, is the output value of the Pi-Sigma neural network, V r is a robust adaptive term.

2. A hybrid helicopter full-mode adaptive control method according to claim 1, characterized in that: The expression for calculating the input increment of the speed loop subsystem based on the incremental dynamic inverse model according to the pseudo-linear control signal after Pi-Sigma neural network compensation is: Among them, Δu1 is the increment of the speed virtual control instruction u1, is the gradient of the speed loop control function g1(.) in the direction of u1 change, is the speed loop desired control command state vector [u c ,v c ,w c ] is the first-order derivative of, is the derivative of the state vector of the velocity loop of the nonlinear dynamic system.

3. A hybrid helicopter full-mode adaptive control method according to claim 2, characterized in that: The input signal of the update speed loop subsystem is expressed as u1=u1 0 +Δu1, where u1 0 It is the sampling signal of speed virtual control instruction.

4. A hybrid helicopter full-mode adaptive control method according to claim 1, characterized in that: The expression for calculating the input increment of the attitude loop subsystem based on the incremental dynamic inverse model according to the pseudo-linear control signal compensated by the Pi-Sigma neural network is: Among them, u2 0 is the sampling signal of the angular velocity virtual control instruction, Δu2 is the increment of the angular velocity virtual control instruction u2, is the gradient of the attitude inner loop control function g2(.) in the direction of u2 change, is the attitude loop desired control command state vector [φ c ,θ c ,ψ c ], φ c is the pitch angle, θ c is the yaw angle, ψ c is the roll angle, is the state vector of the attitude loop of the nonlinear dynamic system.

5. A hybrid helicopter full-mode adaptive control method according to claim 4, characterized in that: The input signal of the attitude update loop subsystem is expressed as u2=u2 0 +Δu2, where u2 0 It is the sampling signal of the angular velocity virtual control instruction.

Images