Direct / gas composite aircraft model-free attitude control method considering model uncertainty

Abstract

The invention provides a direct / gas composite aircraft model-free attitude control method considering model uncertainty. According to the method, in a control law design process, an LSTM neural network is utilized to fit system items, so that the system items do not depend on prior information of a model any more. On the basis of a Lyapunov method, an online updating rule of a neural network weight is deduced, and the system stability under the dual effects of a control law and a network weight updating law is proved. In the simulation process, ASMFC is applied to an aircraft attitude control scene with model uncertainty and time-varying parameters. A simulation result shows that the performance of the ASMFC is superior to that of other existing adaptive sliding mode control methods.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft attitude control, and in particular to a model-free attitude control method for a vertical / air composite aircraft taking model uncertainty into consideration. Background Art

[0002] Nonlinear model uncertainties and control model parameter perturbations are one of the reasons why the design of aircraft attitude controllers is difficult, which can lead to a decrease in the control quality of the controller or even divergence. For direct / air composite control aircraft, due to the jet interference effect and the drastic change of the aircraft state, the impact of such problems is more serious, making the controller design difficult. With the development of neural network technology, algorithms that combine neural networks with traditional control methods are increasingly used in the design of control systems. Considering that external interference and model parameter perturbations can lead to problems such as difficulty in controller design and decreased control quality, artificial neural networks are combined with sliding mode control to improve the quality of the controller and reduce the difficulty of control design. Summary of the invention

[0003] The purpose of the present invention is to solve the problem of attitude control of existing tangential / pneumatic composite control aircraft with model uncertainty, and proposes a model-free attitude control method for tangential / pneumatic composite aircraft considering model uncertainty.

[0004] The present invention is implemented by the following technical scheme. The present invention proposes a model-free attitude control method for a straight / air composite aircraft considering model uncertainty, and the method comprises the following steps:

[0005] Step 1: Establish a three-channel attitude dynamics model of the direct / air compound control aircraft considering the aerodynamic parameter perturbation;

[0006] Step 2: Design a sliding mode controller based on the nominal system without considering external disturbances and parameter perturbations;

[0007] Step 3: Use LSTM neural network to fit the system terms in the sliding mode control law and design an adaptive sliding mode model-free controller ASMMFC;

[0008] Step 4: Based on the Lyapunov method, design the weight update law of the LSTM neural network and ensure the system stability under the dual constraints of the sliding mode control law and the weight update law;

[0009] Step 5: Combining the models and control strategies designed in the previous steps, conduct simulation experimental analysis of the three-channel attitude control of the direct / air composite control aircraft.

[0010] Furthermore, in step 1, the aerodynamic parameters of the aircraft are specifically defined as follows:

[0011] a4=Y α / (mV), a6=-(J x -J y ) / J z , b4=-Z β / (mV), b6=-(J z -J x ) / J y , k y =1 / (mV),k z =-1 / (mV),l y =l / J y ,l z =-l / J z ,

[0012] Where:

[0013]

[0014] Using the above aerodynamic parameters, the three-channel attitude dynamics model of the direct / air composite control aircraft can be modeled as:

[0015]

[0016] Where:

[0017] X=[γω x ω y ω z n y n z δ x δ y δ z F z F y ] T is the system state variable; u=[δ xc δ yc δ zc F zc F yc ] T It is the system control input.

[0018] Furthermore, in step 2, a multi-input multi-output nonlinear system is considered:

[0019]

[0020] Where: Defined as the system state vector, Defined as the system control input, f(X)∈R N×NDefined as the system state function, B∈R N×u Defined as the control matrix, d∈R N Defined as external interference;

[0021] The system control command tracking error is defined as:

[0022] e=XX d (3)

[0023] By taking the derivative of the instruction tracking error with respect to time t, we can obtain:

[0024]

[0025] Design a sliding surface as:

[0026] S=Ce(5).

[0027] Furthermore, in step 2, it is assumed that the matrix CB is reversible and the control matrix B is known for a specific control system; this assumption can be satisfied by selecting a suitable sliding mode parameter matrix C;

[0028] By taking the derivative of the sliding mode function with respect to time t, we can obtain:

[0029]

[0030] set up Assuming d = 0, the equivalent control rate of sliding mode control can be obtained as follows:

[0031]

[0032] In order to realize the exponential sliding mode reaching law, the switching control law is designed as:

[0033]

[0034] in is the right pseudo-inverse of the control matrix B, ie

[0035] Combining the equivalent control rate (7) and the switching control law (8), the control law of the nonlinear system can be obtained:

[0036] u=u eq +u sw (9)

[0037] Substituting the control law (9) into formula (6), we can obtain:

[0038]

[0039] Where: Matrix and external disturbance d are bounded, so we can select appropriate sliding mode parameter matrices K and P so that Negative definite; that is, existence Sliding mode accessibility is satisfied.

[0040] Furthermore, in step 2, when S=0, the Lyapunov equation is designed as:

[0041]

[0042] By taking the derivative of formula (11) with respect to time t and introducing control law (9), we can obtain:

[0043]

[0044] For the system function f(X) at point X=X d Doing a first-order Taylor expansion at , we can get:

[0045]

[0046] Define ψ=Δf(X)-B(CB) -1 CΔf(X), where Substituting formula (13) into formula (12), we can obtain:

[0047]

[0048] Guarantee the linear matrix inequality ψ=(Δf(X)-B(CB) -1 CΔf(X))<0 holds, that is, we can get Negative definiteness, according to Lyapunov stability theory, the system is asymptotically stable, that is, there exists And the value of the sliding mode parameter matrix C can be obtained by solving this linear matrix inequality.

[0049] Furthermore, in the process of LSTM neural network fitting the nonlinear equation y=f(X), there is an optimal weight W * , standard bias term and the optimal gating unit weights The optimal output of the neural network can be expressed as y * =W * h * +ε, where h * represents the optimal value of h, and ε is the bounded fitting error, that is, ||ε||≤ε b , ε b is a positive constant, considering that all optimal weights are constants or slowly varying parameters.

[0050] Further, define:

[0051] y * =W * h * +ε=f(X) (15)

[0052] The actual output value of the LSTM neural network during the fitting process is defined as:

[0053]

[0054] in: is the actual scaling matrix, is the actual hidden state of the LSTM neural network, is a nonlinear function with 13 independent variables, and its implicit expression can be written as:

[0055]

[0056] By bringing the LSTM neural network fitting value into the control law, we can get the new control law:

[0057]

[0058] The fitting error of the LSTM neural network for nonlinear functions can be defined as:

[0059]

[0060] Where: Defined as the approximation error, Expand the weight matrix of the LSTM neural network gate unit into a vector form, that is:

[0061]

[0062] According to Taylor expansion formula, we can get:

[0063]

[0064] Where: O h represents the approximation error, represents the Jacobian matrix.

[0065] Furthermore, the weight update law of the LSTM neural network is designed as:

[0066]

[0067]

[0068] Where: W ci represents the i-th column of the matrix W.

[0069] The present invention also proposes an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for model-free attitude control of a straight / air composite aircraft considering model uncertainty are implemented.

[0070] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of a model-free attitude control method for a straight / air composite aircraft taking into account model uncertainty.

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] Aiming at the attitude control problem of direct force / aerodynamic composite aircraft with model uncertainty, the present invention proposes an adaptive sliding mode model-free control (ASMMFC) algorithm based on a long-short-term memory (LSTM) neural network. Compared with the existing neural network sliding mode algorithm, ASMMFC relaxes the basic assumptions, thereby broadening its scope of application. In the process of control law design, the LSTM neural network is used to fit the system terms so that it no longer depends on the prior information of the model. Based on the Lyapunov method, the online update law of the neural network weights is derived, and the system stability under the dual effects of the control law and the network weight update law is proved. In the simulation process, ASMMFC is applied to the aircraft attitude control scenario with model uncertainty and time-varying parameters. The simulation results show that the performance of ASMMFC is better than other existing adaptive sliding mode control methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0074] Figure 1 It is a schematic diagram of the body coordinate system and velocity coordinate system of the vertical / air compound control aircraft;

[0075] Figure 2 It is a schematic diagram of the structure of a direct / air compound control aircraft;

[0076] Figure 3 This is the LSTM neural network structure diagram;

[0077] Figure 4It is the structure diagram of the adaptive model-free sliding mode control algorithm;

[0078] Figure 5 is a schematic diagram of the aircraft angle of attack response curve;

[0079] Figure 6 It is a schematic diagram of the aircraft sideslip angle response curve;

[0080] Figure 7 It is a schematic diagram of the aircraft roll angle response curve;

[0081] Figure 8 It is a schematic diagram of the overload response curve of the aircraft pitch channel;

[0082] Fig. 9 It is a schematic diagram of the overload response curve of the aircraft yaw channel;

[0083] Fig.10 It is a schematic diagram of the aircraft pitch channel control input curve;

[0084] Fig.11 It is a schematic diagram of the control input curve of the aircraft yaw channel;

[0085] Fig.12 It is a schematic diagram of the aircraft roll channel control input curve. DETAILED DESCRIPTION

[0086] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0087] Combination Figure 1-Figure 12 The present invention proposes a model-free attitude control method for a straight / air composite aircraft considering model uncertainty, the method comprising the following steps:

[0088] Step 1: Establish a three-channel attitude dynamics model of the direct / air compound control aircraft considering the aerodynamic parameter perturbation;

[0089] Step 2: Design a sliding mode controller based on the nominal system without considering external disturbances and parameter perturbations;

[0090] Step 3: Use LSTM neural network to fit the system terms in the sliding mode control law and design an adaptive sliding mode model-free controller ASMMFC;

[0091] Step 4: Based on the Lyapunov method, design the weight update law of the LSTM neural network and ensure the system stability under the dual constraints of the sliding mode control law and the weight update law;

[0092] Step 5: Combining the models and control strategies designed in the previous steps, conduct simulation experimental analysis of the three-channel attitude control of the direct / air composite control aircraft.

[0093] In step 1, the aerodynamic parameters of the aircraft are specifically defined as follows:

[0094] a4=Y α / (mV), a6=-(J x -J y ) / J z , b4=-Z β / (mV), b6=-(J z -J x ) / J y , k y =1 / (mV),k z =-1 / (mV),l y =l / J y ,l z =-l / J z ,

[0095] Where:

[0096]

[0097] Using the above aerodynamic parameters, the three-channel attitude dynamics model of the direct / air composite control aircraft can be modeled as:

[0098]

[0099] Where:

[0100] X=[γω x ω y ω z n y n z δ x δ y δ z F z F y ] T is the system state variable; u=[δ xc δ yc δ zc F zc F yc ] T It is the system control input.

[0101] In step 2, consider a class of multi-input and multi-output nonlinear systems:

[0102]

[0103] Where: Defined as the system state vector, Defined as the system control input, f(X)∈R N×N Defined as the system state function, B∈R N×u Defined as the control matrix, d∈R N Defined as external interference;

[0104] The system control command tracking error is defined as:

[0105] e=XX d (3)

[0106] By taking the derivative of the instruction tracking error with respect to time t, we can obtain:

[0107]

[0108] Design a sliding surface as:

[0109] S=Ce(5).

[0110] In step 2, it is assumed that the matrix CB is reversible and the control matrix B is known for a specific control system; this assumption can be satisfied by selecting a suitable sliding mode parameter matrix C.

[0111] By taking the derivative of the sliding mode function with respect to time t, we can obtain:

[0112]

[0113] set up Assuming d = 0, the equivalent control rate of sliding mode control can be obtained as follows:

[0114]

[0115] In order to realize the exponential sliding mode reaching law, the switching control law is designed as:

[0116]

[0117] in is the right pseudo-inverse of the control matrix B, ie

[0118] Combining the equivalent control rate (7) and the switching control law (8), the control law of the nonlinear system can be obtained:

[0119] u=u eq +u sw (9)

[0120] Substituting the control law (9) into formula (6), we can obtain:

[0121]

[0122] Where: Matrix and external disturbance d are bounded, so we can select appropriate sliding mode parameter matrices K and P so that Negative definite; that is, existence Sliding mode accessibility is satisfied.

[0123] In step 2, when S = 0, the Lyapunov equation is designed as:

[0124]

[0125] By taking the derivative of formula (11) with respect to time t and introducing control law (9), we can obtain:

[0126]

[0127] For the system function f(X) at point X=X d Doing a first-order Taylor expansion at , we can get:

[0128]

[0129] Define ψ=Δf(X)-B(CB) -1 CΔf(X), where Substituting formula (13) into formula (12), we can obtain:

[0130]

[0131] Guarantee the linear matrix inequality ψ=(Δf(X)-B(CB) -1 CΔf(X))<0 holds, that is, we can get Negative definiteness, according to Lyapunov stability theory, the system is asymptotically stable, that is, there exists And the value of the sliding mode parameter matrix C can be obtained by solving this linear matrix inequality.

[0132] In actual application scenarios, the system usually contains uncertain parameters and unmodeled parts, so it is very difficult to obtain an accurate system model. When designing a control law, if the accurate model of the system cannot be obtained, the control effect of the control law will be reduced, and even the system will diverge. Therefore, the present invention designs an adaptive sliding mode control law that does not rely on model prior information, and uses LSTM neural network to fit the system terms in the control law.

[0133] In the process of LSTM neural network fitting the nonlinear equation y=f(X), there is an optimal weight W * , standard bias term and the optimal gating unit weights The optimal output of the neural network can be expressed as y * =W * h * +ε, where h * represents the optimal value of h, and ε is the bounded fitting error, that is, ||ε||≤ε b , ε b is a small enough positive constant, considering that all optimal weights are constants or slowly varying parameters.

[0134] definition:

[0135] y * =W * h * +ε=f(X) (15)

[0136] The actual output value of the LSTM neural network during the fitting process is defined as:

[0137]

[0138] in: is the actual scaling matrix, is the actual hidden state of the LSTM neural network, is a nonlinear function with 13 independent variables, and its implicit expression can be written as:

[0139]

[0140] By bringing the LSTM neural network fitting value into the control law, we can get the new control law:

[0141]

[0142] The fitting error of the LSTM neural network for nonlinear functions can be defined as:

[0143]

[0144] Where: Defined as the approximation error, Expand the weight matrix of the LSTM neural network gate unit into a vector form, that is:

[0145]

[0146] According to Taylor expansion formula, we can get:

[0147]

[0148] Where: O h represents the approximation error, represents the Jacobian matrix.

[0149] The weight update law of the LSTM neural network is designed as:

[0150]

[0151]

[0152] Where: W ci represents the i-th column of the matrix W.

[0153] In the neural network training process, the update laws (22) to (24) are used to update the neural network weights, and the control law (18) is used to ensure that the network weight training error and the system tracking error are consistent and asymptotically stable. Specifically:

[0154] Design a Lyapunov function:

[0155]

[0156] V=V1+V2 (27)

[0157] Where:

[0158] Taking the derivative of the Lyapunov function with respect to time, we can get:

[0159]

[0160] Substituting the control rate and the network weight update law, the above formula can be rewritten as:

[0161]

[0162] Substituting formula (15) into the above formula, we can get:

[0163]

[0164] Substituting into formula (19), we can get:

[0165]

[0166] set up The update law of the scaling matrix W can be obtained. Expanding the left and right sides of the equation yields:

[0167]

[0168] By combining similar terms, we can obtain the scaling matrix update law (22).

[0169] set up The update law of the gating unit weights can be obtained (23).

[0170] Similarly, suppose The update law (24) of the bias term of each gating unit can be obtained.

[0171] Since ε0, O h , d and X d are all bounded, we can assume that ||Cε0||≤ε E , ||Cd||≤d E , Substituting the network weight and bias update law into the Lyapunov function (27), we can obtain:

[0172]

[0173] From the above formula, we can get To get Negative definite, according to Lyapunov stability theory, there exists According to formulas (11) to (14), we can get That is, the system tracking error eventually converges uniformly.

[0174] Simulation part

[0175] The aircraft attitude dynamics equation (1) is used in the simulation process to verify the tracking ability of ASMMFC to the overload command. All aerodynamic parameters are designed with reference to the aircraft operating at an altitude of 30km.

[0176] When the aircraft is operating at an altitude of 30km, the atmospheric density decreases, resulting in a decrease in dynamic pressure, which in turn causes a decrease in aerodynamic force and a decrease in the effectiveness of the aerodynamic rudder. The pure aerodynamic control mode can no longer meet the control requirements, so it is necessary to adopt a direct force aerodynamic composite control strategy to improve the dynamic response speed of the aircraft attitude control. The direct force engine designed by the present invention is an attitude control engine, that is, the direct force does not directly generate overload, but ensures rapid tracking of overloads in each channel by quickly establishing the angle of attack and the angle of sideslip. In order to track sufficiently large overload instructions, the aircraft needs to operate at a large angle of attack or angle of sideslip. At this time, it will become unrealistic to assume that the aerodynamic parameters are time-invariant or slowly variable.

[0177] Therefore, taking the aerodynamic parameters a2 and b2 as examples, consider the two as functions of the angle of attack and the sideslip angle, respectively, namely:

[0178]

[0179] Other aerodynamic parameters are considered as perturbation parameters, namely:

[0180]

[0181] Since the roll angle and roll rate are both very small, c1 and c3 are considered as perturbation parameters. The external interference d is considered as Gaussian white noise.

[0182] Substituting formulas (37) to (39) into the aircraft three-channel attitude dynamics model (1), we can obtain the system model used in this simulation.

[0183] The initial value of the neural network weight is set to be randomly selected in the interval [-1,1]. Considering the actual application scenario, the aerodynamic rudder deflection angle and the direct force are bounded, that is, the control input has a saturation constraint, so |δ xc |≤30°, |δ yc |≤30°, |δ zc |≤30°,|F z |≤3000N,|F y |≤3000N. The initial value of the system state is defined as X=[3 / 57.3 0 0 0 00 0 0] T .

[0184] Sine maneuver is a common aircraft maneuver mode. Therefore, in the simulation experiment, the pitch channel and yaw channel overload commands are set to be in the form of sine functions; the roll angle command is set to 0°.

[0185] The sliding mode control parameter matrix is ​​defined as follows:

[0186] K=diag(9.5 9.5 9.5 5.5 5.5)

[0187] eta1=0.01, eta2=eta3=...=eta9=0.1

[0188] Table 1 Aircraft parameters

[0189]

[0190] The aerodynamic parameters of the yaw channel are the same as those of the pitch channel.

[0191] The present invention also proposes an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for model-free attitude control of a straight / air composite aircraft considering model uncertainty are implemented.

[0192] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of a model-free attitude control method for a straight / air composite aircraft taking into account model uncertainty.

[0193] The memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus RAM (DRRAM). It should be noted that the memory of the method described in the present invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0194] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a high-density digital video disc (DVD)), or a semiconductor medium (eg, a solid state disc (SSD)).

[0195] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software. The steps of the method disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software modules in a processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it is not described in detail here.

[0196] It should be noted that the processor in the embodiment of the present application can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor can be combined and performed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0197] The above is a detailed introduction to the model-free attitude control method for a straight / air composite aircraft considering model uncertainty proposed in the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A model-free attitude control method for a vertical / air composite aircraft considering model uncertainty, characterized in that: The method comprises the following steps: Step 1: Establish a three-channel attitude dynamics model of the direct / air compound control aircraft considering the aerodynamic parameter perturbation; Step 2: Design a sliding mode controller based on the nominal system without considering external disturbances and parameter perturbations; Step 3: Use LSTM neural network to fit the system terms in the sliding mode control law and design an adaptive sliding mode model-free controller ASMMFC; Step 4: Based on the Lyapunov method, design the weight update law of the LSTM neural network and ensure the system stability under the dual constraints of the sliding mode control law and the weight update law; Step 5: Combining the models and control strategies designed in the previous steps, conduct simulation experimental analysis of the three-channel attitude control of the direct / air composite control aircraft.

2. The method according to claim 1, characterized in that In step 1, the aerodynamic parameters of the aircraft are specifically defined as follows: a6=-(J x -J y ) / J z , b4-Z β / (mV), b6=-(J z -J x ) / J y , k y =1 / (mV),k z =-1 / (mV),l y =l / J y ,L z =-l / J z , Where: Using the above aerodynamic parameters, the three-channel attitude dynamics model of the direct / air composite control aircraft can be modeled as: Where: X=[γω x ω y ω z n y n z δ x δ y δ z F z F y ] T is the system state variable; u=[δ xc δ yc δ zc F zc F yc ] T It is the system control input.

3. The method according to claim 2, characterized in that In step 2, consider a class of multi-input and multi-output nonlinear systems: Where: Defined as the system state vector, Defined as the system control input, f(X)∈R N×N Defined as the system state function, B∈R N×u Defined as the control matrix, d∈R N Defined as external interference; The system control command tracking error is defined as: e=X-X d (3) By taking the derivative of the instruction tracking error with respect to time t, we can obtain: Design a sliding surface as: S=Ce(5).

4. The method according to claim 3, characterized in that In step 2, it is assumed that the matrix CB is reversible and the control matrix B is known for a specific control system; this assumption can be satisfied by selecting a suitable sliding mode parameter matrix C. By taking the derivative of the sliding mode function with respect to time t, we can obtain: set up Assuming d = 0, the equivalent control rate of sliding mode control can be obtained as follows: In order to realize the exponential sliding mode reaching law, the switching control law is designed as: in is the right pseudo-inverse of the control matrix B, ie Combining the equivalent control rate (7) and the switching control law (8), the control law of the nonlinear system can be obtained: in=in eq +in sw (9) Substituting the control law (9) into formula (6), we can obtain: Where: Matrix and external disturbance d are bounded, so we can select appropriate sliding mode parameter matrices K and P so that Negative definite; that is, existence Sliding mode accessibility is satisfied.

5. The method according to claim 4, characterized in that In step 2, when S = 0, the Lyapunov equation is designed as: By taking the derivative of formula (11) with respect to time t and introducing control law (9), we can obtain: For the system function f(X) at point X=X d Doing a first-order Taylor expansion at , we can get: Define ψ=Δf(X)-B(CB) -1 CΔf(X), where Substituting formula (13) into formula (12), we can obtain: Guarantee the linear matrix inequality ψ=(Δf(X)-B(CB) -1 CΔf(X))<0 holds, that is, we can get Negative definiteness, according to Lyapunov stability theory, the system is asymptotically stable, that is, there exists And the value of the sliding mode parameter matrix C can be obtained by solving this linear matrix inequality.

6. The method according to claim 5, characterized in that In the process of LSTM neural network fitting the nonlinear equation y=f(X), there is an optimal weight W * , standard bias term and the optimal gating unit weights The optimal output of the neural network can be expressed as y * =W * h * +ε, where h * represents the optimal value of h, and ε is the bounded fitting error, that is, ||ε||≤ε b , ε b is a positive constant, considering that all optimal weights are constants or slowly varying parameters.

7. The method according to claim 6, characterized in that definition: y * =W * h * +ε=f(X) (15) The actual output value of the LSTM neural network during the fitting process is defined as: in: is the actual scaling matrix, is the actual hidden state of the LSTM neural network, is a nonlinear function with 13 independent variables, and its implicit expression can be written as: By bringing the LSTM neural network fitting value into the control law, we can get the new control law: The fitting error of the LSTM neural network for nonlinear functions can be defined as: Where: Defined as the approximation error, Expand the weight matrix of the LSTM neural network gate unit into a vector form, that is: According to Taylor expansion formula, we can get: Where: O h represents the approximation error, represents the Jacobian matrix.

8. The method according to claim 7, characterized in that The weight update law of the LSTM neural network is designed as: Where: W ci represents the i-th column of the matrix W.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium for storing computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images