Landing stage multi-mode control design method for short-distance takeoff and vertical landing aircraft

Abstract

The invention belongs to the technical field of flight control, and relates to a landing stage multi-mode control design method of a short-distance takeoff and vertical landing aircraft. The method comprises the following steps: firstly, respectively designing a control law based on cascade nonlinear dynamic inverse and a control law based on direct force or torque calculation according to a transition mode and a vertical landing mode contained in a landing section; secondly, for a complex control system in which a pneumatic control surface and a thrust vector coexist, a sequential quadratic programming optimization algorithm is adopted for control distribution; and finally, in order to realize smooth switching between modes, an instruction smooth fusion mechanism based on fuzzy logic is introduced, and according to the switching process state, the two-mode instruction is adaptively weighted and fused to generate a final control quantity. According to the method, a complete control framework of'split-mode control law design-multi-control-surface optimal distribution-mode smooth transition 'is formed, and the control problem caused by large difference of mode dynamic characteristics, strong coupling of a control mechanism and discontinuous switching of instructions in the landing process of the STOVL aircraft is effectively solved.

Description

Technical Field

[0001] This invention belongs to the field of flight control technology and relates to a multimodal control design method for the landing phase of short takeoff and vertical landing aircraft. Background Technology

[0002] Short Take-Off and Vertical Landing (STOVL) aircraft, with their unique configuration, can perform missions in confined spaces such as aircraft carrier decks, giving them significant military value. However, their broad flight envelope encompasses multiple modes from short take-off to vertical landing, resulting in extremely complex dynamic characteristics and posing a series of challenges to flight control system design. First, given the significant differences between modes such as short take-off, cruise, transition, and vertical landing, traditional single control laws struggle to adapt to the varying requirements for response speed, accuracy, and robustness across the entire flight envelope. Therefore, control strategies designed independently for each mode can effectively address the problems encountered in that specific mode. Due to the low speed and insufficient aerodynamics in the vertical landing mode, this mode directly solves for the resultant force and torque commands based on the aircraft's kinematic requirements, decoupling pitch angle and forward velocity control, thereby significantly improving system control accuracy and flight safety.

[0003] Secondly, STOVL aircraft are equipped with various heterogeneous control mechanisms, such as aerodynamic control surfaces, vectoring nozzles, lift fans, and roll nozzles. Their functions, effectiveness, and coupling relationships evolve complexly with each flight phase and are subject to strict physical constraints. Traditional control allocation methods based on linear or decoupling assumptions struggle to handle the strong nonlinear coupling and constraint conflicts between multiple control surfaces, easily leading to low allocation efficiency, control surface saturation, or redundant control. Especially during the transition and vertical landing phases, the contribution relationships of each actuator to force and torque undergo fundamental changes. If the allocation strategy cannot dynamically adapt, it will severely impair control accuracy and may trigger torque competition, threatening flight safety. Therefore, multi-control surface allocation design becomes a critical aspect. This invention combines a sequential quadratic programming (SQP) optimization algorithm for control allocation.

[0004] Furthermore, during mission execution, aircraft frequently switch between cruise, transition, and vertical takeoff and landing (VTOL) scenarios. The control laws and allocation strategies corresponding to each mode differ significantly. Directly employing a "hard switch" during mode transitions can easily lead to abrupt attitude changes, thrust misalignment, or even system oscillations due to abrupt changes in control commands, seriously threatening flight safety. Especially when transitioning from aerodynamic control surface-dependent transition modes to thrust vector-dependent VTOL modes, pitch angle commands often exhibit significant abrupt changes, becoming a key challenge for the flight control system. Fuzzy control, due to its independence from precise mathematical models and its ability to handle system nonlinearity and uncertainty, provides an effective solution for smooth mode transitions. It mimics the pilot's continuous decision-making through fuzzy sets and rule bases, achieving stepless interpolation transitions of control commands, thereby avoiding abrupt state changes and ensuring stable and safe full-envelope flight. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-modal control design method for the landing phase of short takeoff and vertical landing aircraft. This method first designs corresponding control strategies for different flight modes to achieve dynamic adaptation of the entire flight envelope and attitude stability; secondly, it coordinates the work allocation of multiple control surfaces through optimization algorithms to efficiently generate control commands under constraints; and finally, it adopts a smooth switching method to fuse control commands of different modes to ensure a smooth transition of flight state during mode transition.

[0006] The technical solution of the present invention is as follows:

[0007] A multimodal control design method for the landing phase of short takeoff and vertical landing (STOVL) aircraft is proposed. First, control laws are designed using a modular approach, employing cascaded NDI and direct force control strategies in the transition and vertical landing modes respectively to achieve dynamic adaptation across the entire control envelope and attitude decoupling. Second, facing a complex control system with both aerodynamic control surfaces and a propulsion system, a sequential quadratic programming (SQP) optimization algorithm is used for real-time control allocation, effectively coordinating the performance of each control surface while satisfying physical constraints. Finally, to address command jumps and state oscillations during mode switching, a smooth switching mechanism based on fuzzy control is introduced. Dual-modal commands are fused in real-time based on the switching time and the pitch angle error rate of change, achieving continuous and stable transition of attitude and control inputs. Specifically, the method includes the following steps:

[0008] Step 1: Dynamic Model Building

[0009] First, the aerodynamic forces acting on the STOVL aircraft during flight are analyzed. The longitudinal aerodynamic forces include drag and lift, while the aerodynamic moments include pitching moments. These forces and moments vary with flight conditions. Second, coordinate system transformation relationships are established, considering the conversion between the inertial and kinematic frames, and the equations of translational motion of the center of mass in the body coordinate system are derived. These equations comprehensively consider the forces generated by the aerodynamic forces and the propulsion system. Next, the dynamic equations of the STOVL aircraft's rotation around its center of mass are established, considering the torques generated by the tail nozzle and lift fan. Based on the angular momentum theorem and the properties of the inertial tensor, expressions for the three-axis angular velocity components are derived. Then, linear displacement equations in the ground coordinate system and angular displacement equations in the body coordinate system are established to fully describe the changes in the aircraft's position and attitude. Finally, the model is simplified for the specific longitudinal motion conditions of the STOVL aircraft. By setting the lateral motion parameters to zero and considering only the deflection of the all-moving horizontal stabilizer and longitudinal attitude changes, a longitudinal dynamic model of the STOVL aircraft suitable for control design is obtained.

[0010] The established STOVL aircraft longitudinal dynamics model can accurately reflect the dynamic characteristics of the aircraft in the longitudinal plane, providing a theoretical basis for the design and optimization of flight control systems.

[0011] Step 2: Design of the reference control law

[0012] The dynamic model provides the mathematical object for control law design, while the control method based on nonlinear dynamic inverse (NDI) linearizes the nonlinear dynamics of the system by solving the pitch motion equations in the model in real time, thereby achieving precise control of the pitch channel. Together, they form a closed loop from object description to control implementation.

[0013] The NDI (Non-Inductively Coupled) control method achieves precise decoupled control of pitch angular velocity by real-time linearization of the system's nonlinear characteristics, thereby generating high-precision pitch angular velocity commands. This method significantly improves the system's adaptability to nonlinear dynamics and external disturbances. In the specific design of the control law, the mathematical model of the pitch angular velocity inner loop must first be established:

[0014] (1)

[0015] In the formula, For pitch angular velocity command, For pitch rate of change command, For pitch moment command, The pitching moment command generated by the aerodynamic control surfaces. The pitch torque command generated by the power system. This represents the pitching moment of inertia.

[0016] Based on the NDI control method and combined with the state-space representation, equation (1) can be reconstructed into the following form:

[0017] (2)

[0018] In the formula, This indicates the current pitch rate command status. To correspond to the pitch rate of change command state, For inner loop control input, and Describe the nonlinear dynamic characteristics of the inner loop of the system.

[0019] When designing the inner-loop control law, the desired pitch rate of change command is... Pitch angular velocity command It is obtained through slow loop calculation, that is:

[0020] (3)

[0021] In the formula, The bandwidth of the pitch angular velocity channel. This represents the current pitch angular velocity.

[0022] According to the NDI control method, in order to obtain the desired form of equation (3), the fast-loop NDI control law expression is:

[0023] (4)

[0024] In the cascaded control architecture, the slow loop is the attitude angle loop, which is the outer loop control loop. It receives attitude angle commands from the command generation module and outputs angular rate commands to the inner loop control loop as input setpoints for the inner loop controller. The following equations can be used to derive the equation set of the slow loop:

[0025] (5)

[0026] In the formula, For pitch angle command, For pitch angular velocity command, This indicates the current pitch angle command status. The pitch angle change rate command status is... For outer loop control input, and Describe the nonlinear dynamic characteristics of the outer loop of the system.

[0027] When designing the outer-loop control law, the pitch angular velocity command External pitch angle command The calculated expression is:

[0028] (6)

[0029] In the formula, The bandwidth of the pitch angle channel. It is the pitch angle.

[0030] According to the NDI control method, in order to obtain the desired form of equation (6), the fast-loop NDI control law expression is:

[0031] (7)

[0032] By implementing time-scale separation and NDI cascaded control design for the pitch angle channel, inner and outer loop control laws were constructed respectively, and the reference pitch moment command of the aircraft was generated by combining them.

[0033] Step 3: Multimodal control allocation

[0034] The intelligent control allocation algorithm based on Sequential Quadratic Programming (SQP) decomposes the torque command generated by the baseline control law obtained in step 2 into various heterogeneous control surfaces, thereby solving the nonlinear allocation problem of the composite control system. This allows the theoretical control command to be transformed into the physical response of each actuator.

[0035] Compared to conventional fixed-wing aircraft, STOVL aircraft have more complex composite control surfaces, including not only conventional aerodynamic control surfaces but also various heterogeneous actuators such as tail nozzles, lift fans, and roll nozzles. These control surfaces exhibit significant nonlinear coupling relationships at different flight phases, and their control effectiveness and physical constraints vary, leading to a more complex control assignment problem for STOVL aircraft. Traditional linear control assignment algorithms struggle to solve this highly nonlinear and constrained control assignment problem; while nonlinear control assignment methods can handle nonlinear problems to some extent, they generally suffer from high computational complexity and slow convergence speed, making it difficult to simultaneously meet the real-time and high-precision control assignment requirements of STOVL aircraft in actual flight missions.

[0036] To address the aforementioned issues, an intelligent control allocation algorithm for the composite control surfaces of STOVL aircraft is proposed. Based on the characteristics of different flight phases of a scaled-down STOVL aircraft and the control law designed in step 2, a corresponding control allocation strategy is given. Furthermore, the SQP optimization algorithm is used to determine how to allocate the desired control commands to the composite control surfaces.

[0037] Step 4: Design of Multimodal Switching Method

[0038] After achieving command allocation to composite control surfaces, the design of the control system needs to further address the issue of smooth transitions between different modes throughout the flight. This requires the underlying high-performance allocation strategy to be combined with upper-level intelligent decision-making logic. Therefore, a multi-modal switching strategy based on fuzzy logic was designed for the switching process between key flight modes such as transition and vertical landing. This method simulates the pilot's decision-making process and adaptively adjusts the control law structure and parameters based on real-time flight status, thereby ensuring that the aircraft maintains overall stability and handling qualities when traversing complex flight envelopes. Ultimately, this forms a complete intelligent flight control system from command generation and allocation to modal adaptive management.

[0039] For the switching between transition mode and vertical landing mode, the core steps are to construct a fuzzy controller, fuzzify the input, and then perform fuzzy inference based on the designed fuzzy rules to obtain the output variable. Finally, the output variable is defuzzified to obtain the final output value.

[0040] A fuzzy logic control system mainly consists of three parts: the input part, the fuzzy controller, and the output part. The input and output parts require defining their respective membership functions. By defining the membership functions of each influencing factor, the numerical values ​​of these factors are normalized and fuzzified using grey relational analysis, and then used as the system input. Fuzzy inference is then performed using custom fuzzy rules, and finally, the fuzziness is resolved using the membership function of the defined optimization level, outputting precise numerical values.

[0041] Receive input variables (switching time) and pitch angle error rate of change This completes data normalization. Each input variable corresponds to multiple Gaussian membership functions. The degree to which the input variable belongs to each fuzzy set is calculated using the following formula:

[0042] (8)

[0043] In the formula, The membership function used for fuzzification. It is an input variable. For the first The first input variable A fuzzy set, The center point of the fuzzy set, The standard deviation of the width of the fuzzy set. Indicates the number of input variables. This represents the number of membership functions for each input.

[0044] The input variables are fuzzified, and membership functions for the input and output variables are established. The switching time is then... After normalization, the fuzzy set is defined as {VS, S, MS, M, ML, L, VL}, where VS represents very short, S represents short, MS represents moderately short, M represents moderate, ML represents moderately long, L represents long, and VL represents very long. This is a fuzzy description of "length" to represent the degree of time duration, with seven consecutive levels from "very short" to "very long". VS indicates extremely short time, close to the minimum; S indicates relatively short time; MS indicates between "short" and "moderate", leaning towards short; M indicates moderate time; ML indicates between "moderate" and "long", leaning towards long; L indicates relatively long time; and VL indicates extremely long time, close to the maximum.

[0045] Rate of change of pitch angle error After normalization, the fuzzy set is defined as {NB, NM, NS, ZE, PS, PM, PB}, where NB represents large negative, NM represents medium negative, NS represents small negative, ZE represents zero, PS represents small positive, PM represents medium positive, and PB represents large positive. A Gaussian function is chosen as the membership function. This is a fuzzy description of the "direction and magnitude of the deviation," used to describe the positive and negative changes in the error rate of change near zero, ranging from "maximum in the negative direction" through "zero" to "maximum in the positive direction" across seven consecutive levels. NB represents a very large negative number (rapidly changing in the negative direction); NM represents a medium negative number; NS represents a small negative number; ZE represents a rate of change close to zero (almost no change); PS represents a small positive number; PM represents a medium positive number; and PB represents a very large positive number (rapidly changing in the positive direction).

[0046] Define the weight coefficients of the output variables. and The fuzzy sets correspond to the current instruction weight and the target instruction weight, respectively.

[0047] The input variable is determined to be the switching time. and pitch angle error rate of change The output is the weighting coefficients for both modes. and Furthermore, we analyze the relationship between different input states and the output, and design a fuzzy rule base.

[0048] Based on the designed fuzzy rule base, the corresponding input conditions will output corresponding variables. The centroid method is used to defuzzify the fuzzy output set. The defuzzified weight coefficients are normalized. According to the system stability requirements, the weight coefficients are limited to prevent sudden changes in instructions.

[0049] The system performs a weighted fusion calculation, combining the weight coefficients. and and transition mode instructions and vertical landing mode target command Perform linear weighting.

[0050] (9)

[0051] The beneficial effects of this invention are:

[0052] The multimodal control design method for the landing phase of an STOVL aircraft proposed in this invention offers several key advantages. First, by designing submodal control laws, cascaded NDI control is employed in the transition mode, while direct force control is used in the vertical landing mode. This significantly improves attitude tracking accuracy and dynamic response capabilities across different flight phases, enhancing the system's adaptability to complex flight conditions and external disturbances. Second, for the complex control system involving both aerodynamic surfaces and a power system, a sequential SQP algorithm is used for real-time control allocation. Under strict physical constraints, this maximizes control efficiency and optimizes energy, effectively avoiding control surface saturation and torque competition, and ensuring high-precision execution of control commands. Finally, by introducing a smooth switching mechanism based on fuzzy logic, dual-modal commands are intelligently fused according to the switching time and pitch angle error change rate. This completely eliminates command jumps and state oscillations caused by traditional "hard switching," achieving continuous and stable transitions between flight attitude and control inputs. This method has a clear overall architecture, high computational efficiency, and good model adaptability and robustness, significantly improving the flight safety, handling quality and mission reliability of STOVL aircraft during mode transitions. Attached Figure Description

[0053] Figure 1 This is the overall flowchart. Figure 2 This is a block diagram of the transition mode control law. Figure 3 This is a block diagram of the vertical landing mode control law. Figure 4 This is a block diagram of the transition mode control allocation strategy. Figure 5 This is a simulation diagram of the transition mode. Figure 6 This is a simulation diagram of the transition mode tail nozzle and lift fan throttle. Figure 7 This is a simulation diagram of the transition mode tail nozzle and lift fan deflection angle. Figure 8 This is a simulation diagram of the transition mode velocity. Figure 9 This is a simulation diagram of the pitch angle in the transition mode. Figure 10 This is a simulation diagram of the elevator in transition mode. Figure 11 This is a simulation diagram of the angle of attack in the transition mode. Figure 12 This is a simulation diagram of the transition mode aileron. Figure 13 This is a block diagram of the vertical landing mode control allocation strategy. Figure 14 This is a simulation diagram of the vertical landing mode altitude. Figure 15It is a simulation diagram of the tail nozzle and lift fan throttle in the vertical landing mode. Figure 16 This is a simulation diagram of the tail nozzle and lift fan deflection angle in the vertical landing mode. Figure 17 This is a simulation diagram of the vertical landing mode altitude. Figure 18 This is a simulation diagram of the pitch angle in the vertical landing mode. Figure 19 This is a simulation diagram of the vertical landing mode angle of attack. Figure 20 It is a multimodal switching fuzzy controller. Figure 21 It is a simulation diagram comparing pitch angles. Figure 22 This is a simulation diagram comparing pitch angle errors. Figure 23 This is a simulation diagram comparing the tail nozzle deflection angle. Figure 24 This is a simulation diagram comparing the deflection difference of the tail nozzle at each step. Figure 25 This is a simulation diagram comparing the deflection angle of the lift fan. Figure 26 This is a simulation diagram comparing the deflection difference of the lift fan at each step. Detailed Implementation

[0054] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0055] The present invention discloses a multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft, the overall flowchart of which is shown below. Figure 1 As shown, the specific steps are as follows:

[0056] Step 1: Dynamic Model Building

[0057] During flight, a STOVL aircraft experiences aerodynamic forces from the airflow due to the deflection of control surfaces and changes in attitude. These forces include longitudinal drag. and lift The aerodynamic torque generated in the body coordinate system is only the pitching torque in the longitudinal direction. These all vary depending on the aircraft's flight state. The specific expressions for the aircraft's drag, lift, and pitch moment are as follows:

[0058] (11)

[0059] In the formula, The longitudinal dimensionless forward force coefficient; This is the dimensionless vertical force coefficient in the longitudinal direction; This is the longitudinal dimensionless pitching moment coefficient; This refers to the elevator deflection angle; For dynamic pressure; This refers to the wing reference area. The average aerodynamic chord length; For aircraft speed; This refers to the aircraft's altitude. and These are the aircraft's angle of attack and sideslip angle, respectively.

[0060] When establishing the STOVL aircraft dynamics equations, due to the different considerations of aircraft motion in the inertial frame and the kinematic frame, the equations of motion for the center of mass translation of the STOVL aircraft in the body coordinate system are as follows:

[0061] (12)

[0062] In the formula, The velocity vector in the body coordinate system The absolute derivative, The velocity vector in the velocity coordinate system The relative derivative, The angular velocity vector represents the rotational angular velocity, which is a three-axis angular velocity between the body coordinate system and the velocity coordinate system. , T = [ T xt T yt T zt ] T The forces generated by the tail nozzle and lift fan along the three axes of the fuselage coordinate system. This is the transformation matrix from the velocity coordinate system to the body coordinate system. The aerodynamic vector in the velocity coordinate system. L BA F aero = F = [ F x q F y q F z q ] T The three-axis aerodynamic forces in the body coordinate system, This is the transformation matrix from the ground coordinate system to the body coordinate system. Let g be the gravity vector in the ground coordinate system. Substituting the values ​​into the equation, we can obtain the expression for the velocity components in the body coordinate system:

[0063] (13)

[0064] In the formula, Let be the acceleration about the three axes in the body coordinate system. The pitch angle, For roll angle, For aircraft quality, It is the acceleration due to gravity; V B = [ V x V y V z ] T The velocities of the STOVL aircraft in the three axes are given in the body coordinate system. ω B = [ ω x ω y ω z ] T Let be the angular velocity of the STOVL aircraft rotating about its three axes. Similarly, establish the dynamic equations for the STOVL aircraft rotating about its center of mass:

[0065] (14)

[0066] In the formula, angular momentum in body coordinate system The absolute derivative, Angular momentum in the velocity coordinate system The relative derivative, It is the angular velocity vector about the three axes. M aero = [ L ¯ xq M ¯ yq N ¯ zq ] T For the aerodynamic torque on the three axes, M T = [ L ¯ xt M ¯ yt N ¯ zt ] T The torque generated by the tail nozzle and lift fan along the three axes. Angular momentum For the inertial tensor, the expression is:

[0067] (15)

[0068] In the formula, , , These are the roll, yaw, and pitch moments of inertia of a STOVL aircraft. This is the product of rotational inertia. Substituting the values ​​into the equation, we can obtain the expressions for the three-axis angular velocity components of the STOVL aircraft rotating about its center of mass:

[0069] (16)

[0070] In the formula, Angular acceleration about the three axes in the body coordinate system.

[0071] Furthermore, equations for linear and angular displacements are established in both the ground coordinate system and the body coordinate system. The equation for linear displacement in the ground coordinate system is:

[0072] (17)

[0073] In the formula, , , These represent the displacements of the STOVL aircraft along the three axes in the ground coordinate system. , , Velocity in the three axes in the ground coordinate system This is the yaw angle.

[0074] The equation of motion for angular displacement in the body coordinate system is:

[0075] (18)

[0076] In the formula, For roll angle Pitch angle Yaw angle The rate of change.

[0077] This invention focuses on the longitudinal motion of a STOVL aircraft, therefore... , The aerodynamic control surfaces only provide pitch control torque through the deflection of the all-moving horizontal stabilizer, therefore there are Considering only the change in the longitudinal attitude of the aircraft, there is Substituting the above conditions into the established equations of motion, we can obtain the STOVL aircraft longitudinal dynamics model as follows:

[0078] (19)

[0079] In the formula, For aerodynamic drag, The thrust generated by the tail nozzle, The thrust generated by the lift fan, The deflection angle of the tail nozzle. This refers to the deflection angle of the lift fan. Let be the lever arm from the point of application of the tail nozzle thrust to the center of mass. The lever arm is the distance from the point of application of the thrust of the lift fan to the center of mass.

[0080] Step 2: Design of the reference control law

[0081] Step 2.1: Design of Transient Mode Control Law

[0082] The transition flight phase of a STOVL aircraft is a crucial process for seamlessly transitioning from cruise to vertical landing mode. During this phase, the propulsion system shifts from horizontal thrust output to vertical lift generation, leading to significant changes in aerodynamic characteristics and mechanical performance, placing stringent demands on the dynamic response and precision of the control system. Therefore, the transition phase is divided into two stages for control design: power deflection and deceleration transition. In the power deflection stage, the tail nozzle and lift fan deflect linearly from horizontal to vertical, gradually changing the thrust direction and ensuring stable attitude. In the deceleration transition stage, by adjusting the nozzle deflection pattern and coordinating thrust distribution, the forward velocity is continuously reduced while maintaining horizontal velocity and altitude, creating stable conditions for transitioning to vertical landing mode. The transition mode control block diagram is shown below. Figure 2 As shown.

[0083] Figure 2 The control design for the STOVL aircraft's transition modes was demonstrated, with a focus on forward flight speed. ,high and pitch angle The closed-loop control mechanism is used. When the aircraft transitions from level flight cruise mode to transition mode, the control logic changes significantly due to the introduction of the power system. At this time, the focus of the power deflection phase is to complete the fixed deflection of the tail nozzle and lift fan, and to ensure the stability of the aircraft's attitude. Therefore, the original pitch angle... As the inner loop of altitude control, now the pitch angle command... The forward speed of the aircraft is given directly from the outside. No control is exercised.

[0084] During the deceleration transition phase, it is necessary to adjust the forward flight speed. Control is performed. The outer loop of the forward speed control uses a PID controller, which is activated by the forward speed command. Compared with actual speed The deviation is calculated to generate the pitch angle command, as shown in the following expression:

[0085] (20)

[0086] In the formula, during the transition mode dynamic deflection stage, The proportional term coefficient of the forward speed control PID controller. The integral term coefficient of the forward speed control PID controller is... The coefficients of the derivative term in the forward speed control PID controller are... This refers to the forward velocity deviation during the transition mode.

[0087] The pitch control loop is designed using the cascaded NDI method. This cascaded control design enhances the rapid pitch tracking capability while effectively improving the system's robustness and dynamic adaptability. In this case, the elevator not only needs to control the pitch angle command generated by the aircraft's forward speed but also needs to compensate for the additional disturbance torque generated by the power system deflection.

[0088] The altitude control is implemented using a PID controller. The altitude controller calculates the throttle opening of the exhaust nozzle and lift fan based on the deviation between the altitude command and the actual altitude, thereby adjusting the thrust of the exhaust nozzle and lift fan to achieve precise control of the flight altitude. Its expression is as follows:

[0089] (twenty one)

[0090] In the formula, during the transition mode deceleration transition phase, This is the throttle opening command for the tail nozzle. This is the throttle opening command for the lift fan. To control the proportional term coefficient of the PID controller, To highly control the integral term coefficient of the PID controller, To highly control the derivative coefficients of the PID controller, This is for height deviation.

[0091] During the powered deflection phase, the deflection of the tail nozzle and lift fan is given by external commands and deflects linearly over time. The tail nozzle is commanded to deflect linearly from 0 degrees to 100 degrees within 6 seconds, while the lift fan starts after the powered deflection phase begins, with a command to deflect linearly from 0 degrees to -10 degrees within 6 seconds, in order to balance the disturbance torque generated by the deflection of the tail nozzle and lift fan.

[0092] When the aircraft is in the deceleration transition phase, the deflection of the tail nozzle and lift fan is given by the same external command, but at this time the deflection is linear according to the forward speed of the aircraft. As the flight speed drops from 50m / s to 0m / s, the tail nozzle deflects from the deflection limit of 100 degrees to 90 degrees, and the lift fan deflects from -10 degrees to 0 degrees.

[0093] Step 2.2: Design of Control Laws for Vertical Descent Phase

[0094] Traditional cascaded inner and outer loop control methods have two main limitations: first, indirectly controlling forward velocity through pitch angle results in a slow dynamic response, making it difficult to adapt to rapid changes and strong disturbances during vertical landing; second, pitch angle simultaneously affects both velocity and attitude, and this coupling can easily reduce system robustness under disturbances, even leading to instability. Therefore, by directly generating the resultant force command required for forward velocity and coordinating attitude and altitude control, combined with precise thrust distribution, efficient coordinated control of velocity, attitude, and altitude is achieved. This method significantly improves dynamic response speed and anti-disturbance robustness, laying the foundation for stable vertical landing of STOVL aircraft.

[0095] To address the complex disturbances and dynamic characteristics of STOVL aircraft during vertical landing on carriers, this study focuses on the aircraft's altitude. Forward speed and pitch angle State control. To achieve precise control during the landing process, PID and cascaded nonlinear NDI methods were used to decouple the aircraft's dynamics and kinematics, respectively, to obtain the aircraft's control law commands. For the resultant force and torque calculated by the control law, a comprehensive nonlinear optimization dynamic control allocation was performed to solve for the throttle and yaw angles of the tail nozzle and lift fan that satisfy the current commands. Specific design details are as follows... Figure 3 As shown:

[0096] Kinematics primarily studies the longitudinal trajectory motion of an aircraft, starting with the aircraft's altitude as an external factor. and forward speed command Then, the forward resultant force command of the kinematics is obtained through the PID controller. and vertical resultant force command As shown in the following formula:

[0097] (twenty two)

[0098] In the formula, in the vertical descent mode, Forward speed deviation, The proportional term coefficient of the forward speed control PID controller. The integral term coefficient of the forward speed control PID controller is... The derivative coefficients of the forward speed control PID are given.

[0099] Forward resultant force command and vertical resultant force command These outputs, respectively for forward speed control and altitude control, provide target values ​​for the next thrust and attitude allocation, enabling the aircraft to fly according to the set altitude and forward speed. In the pitch angle control section, the system employs a cascaded NDI control structure. The first layer, through the NDI controller, generates a pitch rate command based on the error between the pitch angle command and the actual pitch angle; the second-layer NDI controller further processes the pitch rate command and outputs a torque command to adjust the aircraft's attitude.

[0100] The core component of the vertical landing mode is the nonlinear dynamic control allocation module. This module integrates vertical resultant force commands, forward resultant force commands, and pitch moment commands to rationally allocate the thrust and yaw angle of the exhaust nozzle and lift fan. Through nonlinear optimization methods, this module can achieve optimal thrust and moment allocation under complex dynamic conditions, improving the system's control accuracy and stability.

[0101] Step 3: Multimodal control allocation and simulation

[0102] Step 3.1: Transition Mode Control Allocation Strategy

[0103] In the transient mode, the control assignment objective of an STOVL aircraft is to assign the desired torque command calculated from the control law. In addition to the pitch torque command provided by the power system The obtained aerodynamic torque command Elevator deflection angle Provides the main pitch moment and aileron deflection angle Provides some pitching moment. When the aircraft is in transition mode, the tail nozzle and lift fan participate in attitude and altitude control, working in conjunction with the aerodynamic control surfaces to provide the necessary forces and moments.

[0104] During the control allocation process, it is necessary to consider the aircraft's current flight speed. Flight altitude and attack angle The transient process control allocation strategy block diagram is shown below, after the input information is used to solve the problem. Figure 4 As shown:

[0105] Design a transition mode control allocation algorithm:

[0106] With minimizing the allocation error as the primary optimization objective, the direct lift mode carrier-based aircraft control allocation problem is described as the following optimal control problem:

[0107] (twenty three)

[0108] In the formula, u 1 = [ ele , ail ] Optimize the output rudder deflection for the transition mode. For elevators, For aileron; For the transition mode manipulation performance matrix, For the desired control command in the transition mode, The proportion of dynamic optimization terms in the total optimization objective is considered to be 0.1, ensuring that the accuracy of pitch moment distribution error is the dominant factor; The objective function for the transition mode consists of two parts: a static optimization term and a dynamic optimization term. The static optimization term is... The main goal is to ensure the minimum accuracy of pitch moment command allocation error; the dynamic optimization term is... This primarily ensures that the deflection of the secondary control surfaces is minimized, thereby reducing the energy consumption of the secondary control surfaces. This is the transition mode weighting coefficient matrix. Given a 2×2 real matrix space, in the transition mode, considering manipulation effectiveness, the weighting coefficient matrix is ​​designed as follows:

[0109] (twenty four)

[0110] Ensure that the elevator, as the dominant control surface, provides the main pitch moment, while the ailerons provide a portion of the pitch moment.

[0111] Based on the elevator deflection angle range and deflection rate in the transition mode, the constraints for SQP optimization during cruise are designed as follows:

[0112] (25)

[0113] Therefore, the Sequential Quadratic Programming (SQP) algorithm was used to solve the dynamic optimal control allocation problem, and the transient mode objective function satisfying the constraints was obtained. Optimal control surface deflection This allows for optimal control assignment in a STOVL (Simultaneous Transitional Mode) aircraft. Control assignment simulation using the above method yields the transitional mode image as shown below. Figure 5-12 As shown.

[0114] Step 3.2, Vertical Landing Mode Control Allocation Strategy

[0115] In the vertical landing mode, the desired forward force command in the inertial frame calculated by the control law needs to be applied. Vertical expected force command and desired torque command Deflection angles allocated to the tail nozzle and lift fan and Throttle of the tail nozzle, lift fan and roll nozzle , and To reduce the computational burden on the control allocation algorithm, the gravity and desired force commands experienced by the aircraft in the inertial frame are considered. and Transform to the airframe coordinate system, and then subtract the aerodynamic forces and moments generated by the aerodynamic components (elevator) (i.e., consider the aerodynamic components as interference during takeoff and landing) to obtain the expected force of the power system components (tail nozzle and lift fan) in the airframe coordinate system. , and torque command :

[0116] (26)

[0117] In the formula, The aerodynamic torque generated by the elevator is calculated as follows:

[0118] (27)

[0119] In the formula, This is the pitch moment coefficient (dimensionless coefficient) of the elevator.

[0120] and The aerodynamic force generated by the elevator is calculated according to the following formula:

[0121] (28)

[0122] In the formula, The drag generated by the elevator, The lift generated by the elevator:

[0123] (29)

[0124] In the formula, This is the drag coefficient (dimensionless coefficient) of the elevator. This is the lift coefficient (dimensionless coefficient) of the elevator.

[0125] The transformation matrix from the body coordinate system to the airflow coordinate system is as follows:

[0126] (30)

[0127] In this way, the objective of the vertical landing mode control allocation strategy can be transformed into: to express the desired force command of the STOVL aircraft power system in the body coordinate system. and and desired torque command Deflection angles allocated to the tail nozzle and lift fan and And the throttle of the tail nozzle, lift fan and roll nozzle. , , The design strategy for the vertical landing mode control assignment of the STOVL aircraft is as follows: Figure 13 As shown:

[0128] With minimizing the allocation error as the primary optimization objective, and considering both the constraints of control surface effectiveness loss and the difference in control surface deflection rates, the direct lift mode carrier-based aircraft control allocation problem is described as the following optimal control problem:

[0129] (31)

[0130] In the formula, u 2 = [ δ n , δ s , η w , η s , η n ] T Optimize the output rudder deflection for vertical landing mode. The vertical landing mode control performance matrix. The desired control command for the vertical landing mode. The proportion of dynamic optimization terms in the total optimization objective is considered to be 0.1, ensuring that the accuracy of pitch moment distribution error is the dominant factor; The objective function for the vertical landing mode consists of two parts: static optimization terms and dynamic optimization terms. The static optimization terms are... The main goal is to ensure the minimum accuracy of pitch moment command allocation error; the dynamic optimization term is... This primarily ensures that the deflection of the secondary control surfaces is minimized, thereby reducing the energy consumption of the secondary control surfaces. As a weighted coefficient matrix, in the vertical descent mode, considering the control effectiveness, the weighted coefficient matrix is ​​designed as follows: This ensures that the tail nozzle and lift fan deflect significantly while the throttle change is relatively small, with the tail nozzle and lift fan serving as the primary power sources and the rolling nozzle playing a secondary role.

[0131] The constraints for SQP optimization during vertical descent are as follows:

[0132] (32)

[0133] Therefore, the Sequential Quadratic Programming (SQP) algorithm was used to solve the above dynamic optimal control allocation problem, and the objective function satisfying the constraints was obtained. Optimal change of actuator This allows for optimal control allocation for a STOVL (Short-Terminal-Landing) aircraft. The control allocation simulation using the above method yields the altitude simulation diagram shown below. Figure 14 As shown, the simulation diagram of the tail nozzle and lift fan throttle is as follows: Figure 15 As shown in the simulation diagram of the tail nozzle and lift fan deflection angle, see below. Figure 16 As shown in the simulation diagram of forward flight speed. Figure 17 As shown, the pitch angle simulation diagram is as follows: Figure 18 As shown, the angle of attack simulation diagram is as follows: Figure 19 As shown.

[0134] Step 4: Design and Simulation of Multimodal Switching Method

[0135] During the multi-mode switching process of the STOVL aircraft, the flight control system acquires the pitch angle in real time. angular velocity These status parameters provide a basis for smooth switching. Among them, the pitch angle error change rate... This directly reflects the stability and transition quality of the switching process. For this highly nonlinear, time-varying switching control problem, fuzzy control does not require a precise model. It can dynamically adjust command weights based on real-time state, optimize the transition curve, effectively suppress oscillations caused by sudden command changes, and improve control quality. Therefore, a fuzzy controller suitable for two-mode switching can be established. Its input is the switching time. and pitch angle error rate of change The output is the fusion weight of the two-modal commands. and By designing reasonable fuzzy rules and initializing and verifying them using flight test data or simulation platforms, the onboard computer can calculate and output smoothed control commands in real time. A two-mode switching fuzzy controller can be established, such as... Figure 20 As shown.

[0136] The input variables are fuzzified, and membership functions for the input and output variables are established. The switching time is then... Normalization is performed, and its fuzzy set is defined as {VS, S, MS, M, ML, L, VL}. The specific steps are as follows:

[0137] 1) Switch time Map the normalized time interval to [0,1] and calculate the normalized time. :

[0138] (33)

[0139] In the formula, This is the start time of the switching phase. The duration of the switching phase.

[0140] 2) Based on normalized time Define 7 Gaussian membership functions, corresponding to very short (VS), short (S), medium short (MS), medium (M), medium long (ML), long (L), and very long (VL), respectively.

[0141] 3) Select the Gaussian function as the membership function, and set the center position and width parameters of each membership function to ensure that they are evenly distributed and appropriately overlapped on the time axis.

[0142] Rate of change of pitch angle error Normalization is performed, and its fuzzy set is defined as {NB, NM, NS, ZE, PS, PM, PB}. After completing the fuzzification of the switching time as described above, the specific steps for fuzzifying the error change rate are as follows:

[0143] 1) The rate of change of pitch angle error Limit the range to [-10, 10] degrees / second and normalize it;

[0144] 2) Define 7 Gaussian membership functions, corresponding to negative large (NB), negative medium (NM), negative small (NS), zero (ZE), positive small (PS), positive medium (PM), and positive large (PB), and select Gaussian functions as membership functions;

[0145] 3) Based on the dynamic response characteristics of the system, adjust the parameters of each membership function so that the membership functions of the ZE region are narrower, the NS and PS regions are moderate, and the NB and PB regions are wider.

[0146] Define output weight coefficients and The fuzzy sets of both inputs are defined as {VS, S, MS, M, ML, L, VL}, with values ​​ranging from [0,1]. Gaussian membership functions are used, and parameters are appropriately set to ensure that each function is evenly distributed and smoothly transitions within the [0,1] interval. Finally, the completeness of this set of membership functions needs to be verified to ensure that effective weight outputs are generated for all input combinations.

[0147] A fuzzy rule base is designed based on the dynamic characteristics of the system, establishing a time-based rule system. and error change rate To weighting coefficients and The mapping relationship. Considering that the time characteristics and dynamic response characteristics of the flight control system during multi-modal transitions have clear physical laws, the following fuzzy inference rule base is designed:

[0148] 1) When the switching time is in the very short (VS) stage, the current instruction remains dominant regardless of the error rate of change. At this time, the weight of the current instruction is determined. It should be set to Very Large (VL), target instruction weight. It should be set to very small (VS);

[0149] 2) When the switching time is in the short (S) phase, regardless of the error change rate, the current instruction remains dominant, but the dominant weight is smaller than in the VS phase, and the instruction weight is lower. It should be set to large (L), the target instruction weight. It should be set to small (S);

[0150] 3) When the switching time is in the medium-short range (MS) and the error rate of change is positive (PB), the system response is too slow, and the weight of the current instruction needs to be reduced to ( ). To expedite response, the target instruction weight is increased by ( );

[0151] 4) When the switching time is in the medium (M) stage and the error change rate is zero (ZE), a balanced transition strategy is adopted, and the current instruction weight ( ) and target instruction weight ( Keep them equal;

[0152] 5) When the switching time is in the medium-long (ML) stage and the error change rate is positive (PB), the system response is too fast, and the weight of the current instruction needs to be reduced to ( ). To expedite response, the target instruction weight is increased by ( );

[0153] 6) When the switching time is in the long (L) stage and the error change rate is positive (PB), the system response is too fast, and the weight of the current instruction needs to be reduced to ( To prevent overshoot, the target instruction weight is reduced by ( When the error rate of change is zero (ZE), the system response is too fast, and the weight of the current instruction needs to be reduced to ( ); To prevent overshoot, the target instruction weight is reduced by ( ).

[0154] 7) When the switching time is in the very long (VL) phase, regardless of the error rate of change, the current instruction remains dominant, and the weight of the current instruction is determined. The target instruction weight should be set to very small (VS). It should be set to Very Large (VL);

[0155] A complete set of 49 fuzzy rules is constructed, covering all combinations of time {VS,S,MS,M,ML,L,VL} and error change rate {NB,NM,NS,ZE,PS,PM,PB}.

[0156]

[0157] Fuzzy inference is performed using the Mamdani inference method to calculate the activation strength of each rule. The calculation process of Mamdani inference is as follows:

[0158] 1) The fuzzy rules are designed as follows:

[0159]

[0160] 2) Calculate the activation strength of the rule antecedent. :

[0161] (34)

[0162] in, and These are the membership functions of the corresponding fuzzy sets;

[0163] 3) Based on activation intensity The membership function of the cut-off part is used to obtain the output fuzzy set of each rule. and ,in:

[0164] (35)

[0165] The fuzzy output set is obtained by synthesizing the rule outputs using a weighted average method. The specific steps for synthesizing the fuzzy output set after completing the reasoning calculations for all rules are as follows:

[0166] The output fuzzy sets of all rules are superimposed, and the largest value is used to synthesize the total output fuzzy set:

[0167] (36)

[0168] Verify the continuity and rationality of the synthesized fuzzy output set to ensure a smooth transition of the output when the input space changes, and normalize the synthesized fuzzy output set to prepare for the subsequent defuzzification step.

[0169] The centroid method is used to defuzzify the fuzzy output set. Considering the distribution characteristics of the membership function of the fuzzy inference output, the centroid method is used to calculate the precise values ​​of each output variable. The calculation formula is as follows:

[0170] (37)

[0171] In the formula, Indicates the first Membership degree of each output fuzzy set; This represents the center value of the corresponding fuzzy set; To output the total number of fuzzy sets; This is an estimated value for the defuzzified weight coefficients.

[0172] The deblurred weight coefficients are normalized to ensure that the sum of the weight coefficients for the two modes is 1. Considering the integrity requirements of the control commands, the weight coefficients for the two modes are calculated using normalization:

[0173] (38)

[0174] In the formula, and These represent the estimated defuzzification weights for mode 1 and mode 2, respectively. and These are the final weighting coefficients after normalization.

[0175] The weighting coefficients are limited according to system stability requirements to prevent sudden command changes. Considering the dynamic response characteristics of the actual control system, a threshold for the rate of change of the weights is set. Limit the weight changes between adjacent sampling periods:

[0176] (39)

[0177] In the formula, and These represent the weighting coefficients for the current time step and the previous time step, respectively. To ensure a smooth change in the weight coefficients and avoid drastic changes in control commands, this process is designed to maximize the allowable rate of change in weights.

[0178] Weighting coefficients and (in ), and transition mode instructions and vertical takeoff and landing mode target commands Using the linear weighted sum formula:

[0179] (40)

[0180] Calculate the final control command at each moment. As time goes by and weights A smooth transition from 0 to 1, with the final instruction also changing from... Smooth transition to .

[0181] A comparative simulation was performed on the transition pitch angle command switching and direct hard switching based on the output weights, and the results were as follows: Figure 21 The simulation diagram comparing the pitch angle commands shown is as follows: Figure 22 The simulation images showing the pitch angle tracking error are shown. The simulation images showing the nozzle deflection angle and the difference in deflection angle at each step are also shown. Figure 23 , 24As shown in the simulation diagram, the deflection angle and the difference in deflection angle at each step of the lift fan are compared. Figure 25 , 26 As shown.

[0182] During handover, a hard handover causes a sudden change in pitch angle command, leading to significant overshoot and error fluctuations, requiring approximately 3 seconds to stabilize. A smooth handover, through a gentle transition of commands, avoids instantaneous jumps, significantly reducing both the peak error and its duration. To quantify the improvement, maximum and minimum pitch angle errors are introduced. Perform a comparative analysis before and after the smooth transition.

[0183] Pitch angle error statistics:

[0184]

[0185] like Figure 23 and Figure 25 As shown, during the 5-second mode switching, the hard switching strategy causes a significant change in the deflection angle of the tail nozzle and lift fan, which then oscillates continuously; while the smooth switching strategy makes the deflection process stable and the rate of change significantly moderate. Figure 24 and Figure 26 Further analysis shows that the deflection difference at each step under hard switching is often close to the physical rate limit, while the deflection difference under smooth switching fluctuates less, with most step changes remaining around 1°, far from the saturation rate. To further quantify this effect, the sum of the absolute values ​​of the deflection differences per second of the exhaust nozzle and lift fan is introduced as an evaluation index. The formula is as follows:

[0186] (41)

[0187] In the formula, This is the sum of the absolute values ​​of the tail nozzle deviation per second; The sum of the absolute values ​​of the lift fan deviation per second; For each step of the tail nozzle deviation; Deviation of lift fan in each step.

[0188] Tail nozzle deflection difference statistics:

[0189]

[0190] Smoothing effect: Average deflection difference reduced by 34.99%.

[0191] Lift fan deflection difference statistics:

[0192]

[0193] Smoothing effect: Average deflection difference reduced by 43.92%.

[0194] Smooth switching reduces the sum of the absolute values ​​of the deflection difference per second of the nozzle and lift fan by 34.99% and 43.92% respectively compared to hard switching, and the amplitude of the single-step deflection difference is significantly reduced. This indicates a significant reduction in overall deflection demand and actuator stress. In summary, smooth control at the moment of switching effectively suppresses the peak and abrupt changes in the deflection rates of the nozzle and lift fan, not only reducing the instantaneous impact and wear on the actuators, but also improving the stability and safety of pitch control to a certain extent.

Claims

1. A multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft, characterized in that, Includes the following steps: Step 1: Dynamic Model Building Step 2: Design of the reference control law In the specific design of the control law, the mathematical model of the inner loop of pitch angular velocity must first be established: (1) ； In the formula, For pitch angular velocity command, For pitch rate of change command, For pitch moment command, The pitching moment command generated by the aerodynamic control surfaces. The pitch torque command generated by the power system. The moment of inertia is the pitching moment. Based on the NDI control method and combined with the state-space representation, equation (1) is reconstructed into the following form: (2) ； In the formula, This indicates the current pitch rate command status. To correspond to the pitch rate of change command state, For inner loop control input, and Describe the nonlinear dynamic characteristics of the inner loop of the system; When designing the inner-loop control law, the desired pitch rate of change command is... Pitch angular velocity command This is obtained through slow loop calculation, i.e.: (3) ； In the formula, The bandwidth of the pitch angular velocity channel. The current pitch angular velocity; According to the NDI control method, in order to obtain the desired form of equation (3), the fast-loop NDI control law expression is: (4) ； In the cascaded control architecture, the slow loop is the attitude angle loop, which is the outer loop control loop. It receives attitude angle commands from the command generation module and outputs angular rate commands to the inner loop control loop as input setpoints for the inner loop controller. The following equations give the equation set of the slow loop: (5) ； In the formula, For pitch angle command, For pitch angular velocity command, This indicates the current pitch angle command status. The pitch angle change rate command status is... For outer loop control input, and Describe the nonlinear dynamic characteristics of the outer loop of the system; When designing the outer-loop control law, the pitch angular velocity command External pitch angle command The calculation yields the following expression: (6) ； In the formula, The bandwidth of the pitch angle channel. The pitch angle; According to the NDI control method, in order to obtain the desired form of equation (6), the fast-loop NDI control law expression is: (7) ； By implementing time-scale separation and NDI cascaded control design for the pitch angle channel, inner and outer loop control laws were constructed respectively, and the reference pitch moment command of the aircraft was generated by combining them. Step 3: Multimodal control allocation The intelligent control allocation algorithm based on Sequential Quadratic Programming (SQP) decomposes the torque command generated by the reference control law obtained in step 2 into each heterogeneous control surface. Step 4: Design of Multimodal Switching Method For the switching between transition mode and vertical descent mode, a fuzzy controller is constructed to fuzzify the input variables. Then, fuzzy inference is performed using the designed fuzzy rules to obtain the output variables. The output variables are then defuzzified to obtain the final output value. The fuzzy logic control system consists of three parts: the input part, the fuzzy controller, and the output part. The input and output parts need to define their respective membership functions. By defining the membership functions of each influencing factor, the values ​​of the influencing factors are normalized and fuzzified through grey relational analysis and used as the input of the system. Then, fuzzy inference is performed using the defined fuzzy rules, and finally, the fuzziness is defuzzified using the membership function of the defined optimization level to output the accurate value.

2. The multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft according to claim 1, characterized in that, Step 4 is as follows: Receive input variable switching time and pitch angle error rate of change Data normalization is completed; each input variable corresponds to multiple Gaussian membership functions, and the degree to which the input variable belongs to each fuzzy set is calculated using the following formula: (8) ； In the formula, The membership function used for fuzzification. It is an input variable. For the first The first input variable A fuzzy set, The center point of the fuzzy set, The standard deviation of the width of the fuzzy set. Indicates the number of input variables. This represents the number of membership functions for each input; The input variables are fuzzified, and membership functions for the input and output variables are established; the switching time is... Normalization is performed, and its fuzzy set is defined as {VS, S, MS, M, ML, L, VL}, where VS is very short, S is short, MS is medium short, M is medium, ML is medium long, L is long, and VL is very long. Rate of change of pitch angle error Normalization is performed, and its fuzzy set is defined as {NB, NM, NS, ZE, PS, PM, PB}, where NB is negative large, NM is negative medium, NS is negative small, ZE is zero, PS is positive small, PM is positive medium, and PB is positive large. A Gaussian function is selected as the membership function. Define the weight coefficients of the output variables. and The fuzzy sets correspond to the current instruction weight and the target instruction weight, respectively; The input variable is determined to be the switching time. and pitch angle error rate of change The output is the weighting coefficients for both modes. and Furthermore, the relationship between different input states and outputs is analyzed, and a fuzzy rule base is designed. Based on the designed fuzzy rule base, the corresponding input conditions will output the corresponding variables. The centroid method is used to defuzzify the fuzzy output set. The weight coefficients after defuzzification are normalized. According to the system stability requirements, the weight coefficients are limited to prevent sudden changes in instructions. The system performs weighted fusion calculations; the weighting coefficients are... and and transition mode instructions and vertical landing mode target command Perform linear weighting; (9)。 3. The multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft according to claim 1, characterized in that, In step 1, the STOVL aircraft longitudinal dynamics model is as follows: (19) ； In the formula, For aerodynamic drag, The thrust generated by the tail nozzle, The thrust generated by the lift fan, The deflection angle of the tail nozzle. This refers to the deflection angle of the lift fan. Let be the lever arm from the point of application of the tail nozzle thrust to the center of mass. The lever arm is the distance from the point of application of the lift fan thrust to the center of mass.

4. The multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft according to claim 1, characterized in that, Step 2 is as follows: Step 2.1: Design of Transient Mode Control Law During the deceleration transition phase, it is necessary to adjust the forward flight speed. Control is performed; the outer loop of the forward speed control uses a PID controller, which is activated by the forward speed command. Compared with actual speed The deviation is calculated to generate the pitch angle command, as shown in the following expression: (20) ； In the formula, during the transition mode dynamic deflection stage, The proportional term coefficient of the forward speed control PID controller. The integral term coefficient of the forward speed control PID controller is... The coefficients of the derivative term in the forward speed control PID controller are... For the forward velocity deviation of the transition mode; Pitch control loop design using cascaded NDI method; The altitude control section is implemented using a PID controller. The altitude controller calculates the throttle opening of the exhaust nozzle and lift fan based on the deviation between the altitude command and the actual altitude, thereby adjusting the thrust of the exhaust nozzle and lift fan to achieve precise control of the flight altitude. Its expression is as follows: (21) ； In the formula, during the transition mode deceleration transition phase, This is the throttle opening command for the tail nozzle. This is the throttle opening command for the lift fan. To control the proportional term coefficient of the PID controller, To highly control the integral term coefficient of the PID, To highly control the derivative coefficients of the PID controller, For height deviation; Step 2.2: Design of Control Laws for Vertical Descent Phase The dynamics and kinematics of the aircraft are decoupled using PID and cascaded nonlinear NDI methods to obtain the control law commands. Based on the resultant force and torque calculated by the control law, nonlinear optimization dynamic control allocation is performed to solve for the throttle and yaw angle of the tail nozzle and lift fan that satisfy the current commands. Kinematics studies the longitudinal trajectory motion of an aircraft; it begins by externally determining the aircraft's altitude. and forward speed command Then, the forward resultant force command of the kinematics is obtained through the PID controller. and vertical resultant force command As shown in the following formula: (22) ； In the formula, in the vertical descent mode, Forward speed deviation, The proportional term coefficient of the forward speed control PID controller. The integral term coefficient of the forward speed control PID controller is... The differential coefficients of the forward speed control PID controller; Forward resultant force command and vertical resultant force command These outputs, respectively for forward speed control and altitude control, provide target values ​​for the next thrust and attitude allocation, enabling the aircraft to fly according to the set altitude and forward speed. In the pitch angle control section, the system adopts a cascaded NDI control structure. The first layer, through the NDI controller, generates a pitch rate command based on the error between the pitch angle command and the actual pitch angle. The second layer NDI controller further processes the pitch rate command and outputs a torque command to adjust the aircraft's attitude.

5. The multi-modal control design method for the landing phase of a short takeoff and vertical landing aircraft according to claim 1, characterized in that, Step 3 is as follows: Step 3.1: Transition Mode Control Allocation Strategy In the transient mode, the control assignment objective of an STOVL aircraft is to assign the desired torque command calculated from the control law. In addition to the pitch torque command provided by the power system The obtained aerodynamic torque command Elevator deflection angle Provides pitch moment and aileron deflection angle Provides partial pitch moment; Design a transition mode control allocation algorithm: With the goal of minimizing the allocation error, the control allocation problem for carrier-based aircraft in direct lift mode can be described as the following optimal control problem: (23) ； In the formula, u 1 = [ ele , ail ] To optimize the output rudder deflection for the transition mode, For elevators, For aileron; For the transition mode manipulation performance matrix, For the desired control command in the transition mode, The proportion of dynamic optimization terms in the total optimization objective is considered to be 0.1, ensuring that the pitch moment distribution error accuracy is the dominant factor; The objective function for the transition mode consists of two parts: a static optimization term and a dynamic optimization term. The static optimization term is... To ensure the pitch moment command allocation error is minimized; the dynamic optimization term is... This ensures that the deflection of the secondary control surfaces is minimized, thereby reducing the energy consumption of the secondary control surfaces; This is the transition mode weighting coefficient matrix. Given a 2×2 real matrix space, in the transition mode, considering the manipulation effectiveness, the weighting coefficient matrix is ​​designed as follows: (24) ； Ensure that the elevator provides pitch moment as the dominant control surface, and the ailerons provide part of the pitch moment; Based on the elevator deflection angle range and deflection rate in the transition mode, the constraints for SQP optimization during cruise are designed as follows: (25) ； Therefore, the Sequential Quadratic Programming (SQP) algorithm was used to solve the dynamic optimal control allocation problem, and the transient mode objective function was obtained under the constraints. Optimal control surface deflection This enables optimal control allocation for STOVL aircraft in transition mode; Step 3.2, Vertical Landing Mode Control Allocation Strategy In the vertical landing mode, the desired forward force command in the inertial frame calculated by the control law needs to be applied. Vertical expected force command and desired torque command Deflection angles allocated to the tail nozzle and lift fan and Throttle of the tail nozzle, lift fan and roll nozzle , and The commands for gravity and desired force acting on the aircraft in an inertial frame. and Transform to the body coordinate system, and then subtract the aerodynamic forces / torques generated by the aerodynamic components to obtain the desired force of the thrust vector component in the body coordinate system. , and torque command : (26) ； In the formula, The aerodynamic torque generated by the elevator is calculated as follows: (27) ； In the formula, This is the pitch moment coefficient of the elevator; and The aerodynamic force generated by the elevator is calculated according to the following formula: (28) ； In the formula, The drag generated by the elevator, The lift generated by the elevator: (29) ； In the formula, The drag coefficient of the elevator. This is the lift coefficient of the elevator; The transformation matrix from the body coordinate system to the airflow coordinate system is as follows: (30) ； Therefore, the objective of the vertical landing mode control allocation strategy is transformed into: to assign the desired force command under the thrust vectoring system of the STOVL aircraft. and Desired torque command Deflection angles allocated to the tail nozzle and lift fan and Throttle of tail nozzle, lift fan and roll nozzle , , ; With the goal of minimizing the allocation error, and considering both the constraints of control surface effectiveness loss and the difference in control surface deflection rates, the direct lift mode carrier-based aircraft control allocation problem is described as the following optimal control problem: (31) ； In the formula, u 2 = [ δ n , δ s , η w , η s , η n ] T Optimize the output rudder deflection for vertical landing mode. The vertical landing mode control performance matrix. The desired control command for the vertical landing mode. The proportion of dynamic optimization terms in the total optimization objective is considered to be 0.1, ensuring that the pitch moment distribution error accuracy is the dominant factor; The objective function for the vertical landing mode consists of two parts: static optimization terms and dynamic optimization terms. The static optimization terms are... To ensure the pitch moment command allocation error is minimized; the dynamic optimization term is... This ensures that the deflection of the secondary control surfaces is minimized, thereby reducing the energy consumption of the secondary control surfaces; As a weighted coefficient matrix, in the vertical descent mode, considering the handling effectiveness, the weighted coefficient matrix is ​​designed as follows: ; The constraints for SQP optimization during vertical descent are as follows: (32) ； Thus, the Sequential Quadratic Programming (SQP) algorithm was used to solve the dynamic optimal control allocation problem, yielding the objective function that satisfies the constraints. Optimal change of actuator This enables optimal control allocation for STOVL aircraft in vertical landing mode.

Images