A tiltrotor aircraft transition section double-layer safety control method, device, equipment, medium and product

Abstract

The application discloses a tilt-rotor aircraft transition section double-layer safety control method, device, equipment, medium and product, and relates to the field of tilt-rotor aircraft safety control. The method comprises the following steps: acquiring information data of a target tilt-rotor aircraft at a current time; in an upper-layer control architecture, based on a current-time corresponding tight format dynamic linearization model and a corresponding safety corridor constraint, rolling optimization is performed through a particle swarm optimization algorithm with a penalty function; in a lower-layer control architecture, according to optimization data at the current time, control distribution is performed through a multi-objective weighted quadratic programming method; and the real-time dynamic state of the target tilt-rotor aircraft is determined according to a control instruction at the current time, and is fed back to the tight format dynamic linearization model in the upper-layer control architecture through a multi-source on-board sensor to perform closed-loop control. The application aims to realize stable and safe transition control of a tilt-rotor aircraft that can land.

Description

Technical Field

[0001] This application relates to the field of tiltrotor aircraft safety control, and in particular to a method, device, equipment, medium, and product for a dual-layer safety control of the transition section of a tiltrotor aircraft. Background Technology

[0002] Tiltrotor aircraft combine the advantages of helicopter vertical takeoff and landing with the high-speed cruise of fixed-wing aircraft, but their core control challenge lies in the tilt transition process (helicopter mode to fixed-wing mode). Current control schemes (such as traditional PID (Proportional-Integral-Derivative), conventional MPC (Model Predictive Control), or fixed control assignment strategies) suffer from three major pain points in this stage: 1. Rapidly changing dynamics and difficult model building: Within the tilt corridor, rotor wake interference and wing aerodynamic interference are extremely severe, and the tilt angle... Changes in aerodynamic characteristics can lead to drastic, nonlinear abrupt changes. Using a fixed model makes control mismatch highly likely.

[0003] 2. Significant safety risks: During the transition phase, if the left and right rotors experience significant thrust asymmetry due to environmental interference or control calculations (such as excessive speed difference), it will generate extremely large yaw and roll moments, making tiltrotor aircraft highly susceptible to irreversible rollover and crash.

[0004] 3. Redundant Countermeasures of Multiple Control Variables: Tiltrotors not only possess the collective pitch and cyclic pitch control of helicopters, but also the aileron and elevator control of fixed wings. During the transition phase, both systems operate simultaneously. Due to the greater number of actuators than control degrees of freedom and their different physical constraints, conventional control can easily lead to the rotor and control surfaces issuing opposite commands, resulting in control countermeasures and energy consumption. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, equipment, medium, and product for dual-layer safety control of the transition section of a tiltrotor aircraft, aiming to achieve stable and safe transition control for the tiltrotor aircraft to land.

[0006] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a dual-layer safety control method for the transition section of a tiltrotor aircraft, wherein the dual-layer safety control method for the transition section of a tiltrotor aircraft employs a dual-layer safety control architecture; the dual-layer safety control architecture includes: an upper-layer control architecture and a lower-layer control architecture; The dual-layer safety control method for the transition section of the tiltrotor aircraft includes: Acquire information data of the target tiltrotor aircraft at the current moment; the information data includes: tilt angle, tilt rate and status data fed back by multiple airborne sensors; the status data includes: the difference in rotational speed between the left and right rotors, the airframe roll rate and roll acceleration measured by the IMU, and the strain data of the rotor root structure. Within the upper-level control architecture, based on the current time-bound compact dynamic linearization model and the corresponding safety corridor constraints, rolling optimization is performed using a particle swarm optimization algorithm with a penalty function to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity. The compact dynamic linearization model is obtained by adaptive adjustment based on the state data and the optimization data corresponding to the previous time-bound moment using a tilt angle gating update mechanism. The tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by the rapid change of aerodynamic characteristics in the transition section. Within the lower-level control architecture, based on the optimized data at the current moment, control allocation is performed through a multi-objective weighted quadratic programming method to obtain the control command at the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled with the tilt angle to dynamically penalize the usage cost of different actuators, thereby achieving a smooth transfer of the control weight between the rotor channel and the aerodynamic control surface during the transition period. The real-time dynamic state of the target tiltrotor is determined based on the control commands at the current moment, and fed back to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

[0007] Secondly, this application provides a dual-layer safety control device for the transition section of a tiltrotor aircraft, comprising: The information data acquisition module is used to acquire information data of the target tiltrotor aircraft at the current moment; the information data includes: tilt angle, tilt rate and status data fed back by multi-source airborne sensors; the status data includes: the difference in rotational speed between the left and right rotors, the roll rate and roll acceleration of the airframe measured by the IMU, and the strain data of the rotor root structure. The rolling optimization module is used within the upper-level control architecture to perform rolling optimization using a particle swarm optimization algorithm with a penalty function, based on the current time-bound compact dynamic linearization model and the corresponding safety corridor constraints, to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity. The compact dynamic linearization model is obtained by adaptive adjustment based on the state data and the optimization data from the previous time-bound moment using a tilt angle gating update mechanism. The tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by rapid changes in aerodynamic characteristics during the transition section. The control allocation module is used to perform control allocation based on the current optimized data within the lower control architecture, using a multi-objective weighted quadratic programming method to obtain the control command for the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled according to the tilt angle, so as to dynamically penalize the usage cost of different actuators and realize the smooth transfer of the control weight of the rotor channel and aerodynamic control surface during the transition period. The closed-loop control module is used to determine the real-time dynamic state of the target tiltrotor aircraft based on the control command at the current moment, and to feed back the data to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

[0008] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described tiltrotor transition section dual-layer safety control method.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned tiltrotor transition section dual-layer safety control method.

[0010] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned tiltrotor aircraft transition section dual-layer safety control method.

[0011] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a two-layer safety control method, device, equipment, medium, and product for the transition phase of a tiltrotor aircraft. It acquires information data of the target tiltrotor aircraft at the current moment. Within the upper-layer control architecture, rolling optimization is performed using a particle swarm optimization algorithm with a penalty function, based on the current moment's compact-format dynamic linearization model and corresponding safety corridor constraints. Within the lower-layer control architecture, control allocation is performed using a multi-objective weighted quadratic programming method based on the current moment's optimized data. The real-time dynamic state of the target tiltrotor aircraft is determined based on the current moment's control commands and fed back to the compact-format dynamic linearization model within the upper-layer control architecture via multi-source airborne sensors for closed-loop control. This application introduces a tilt angle gating update mechanism within the upper-layer control architecture, updating the pseudo-partial derivative matrix online and dynamically adjusting the forgetting factor through the tilt angle rate, enabling rapid adaptation to nonlinear aerodynamic changes during the transition phase. Furthermore, a mode weighting matrix scheduled according to the tilt angle is introduced within the lower-layer control architecture, allowing for seamless and smooth dynamic transfer of control weights between the rotor and control surfaces. This achieves stable and safe transition control for the tiltrotor aircraft, enabling it to land safely. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart of a dual-layer safety control method for the transition section of a tiltrotor aircraft; Figure 2 This is a schematic diagram of a two-layer security control architecture; Figure 3 This outlines the overall process flow of the entire technical solution. Figure 4 Optimize the PSO rolling flowchart; Figure 5 This is a framework diagram of all steps in the entire technical solution; Figure 6 This is a structural diagram of the dual-layer safety control device for the transition section of a tiltrotor aircraft. Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] In one exemplary embodiment, a dual-layer safety control method for the transition section of a tiltrotor aircraft is provided. The dual-layer safety control method employs a dual-layer safety control architecture, which includes an upper-layer control architecture and a lower-layer control architecture.

[0017] like Figure 1 As shown, the dual-layer safety control method for the transition section of the tiltrotor aircraft includes: Step 100: Acquire information data of the target tiltrotor aircraft at the current moment. Information data includes: tilt angle, tilt rate, and status data fed back by multiple airborne sensors; status data includes: the difference in rotational speed between the left and right rotors, the airframe roll rate and roll acceleration measured by the IMU, and the strain data of the rotor root structure.

[0018] Step 200: Within the upper-level control architecture, based on the current time-bound compact dynamic linearization model and the corresponding safety corridor constraints, rolling optimization is performed using a particle swarm optimization algorithm with a penalty function to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity; the compact dynamic linearization model is obtained by adaptive adjustment based on the state data and the optimization data corresponding to the previous time-bound moment using a tilt angle gating update mechanism; wherein, the tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by the rapid change of aerodynamic characteristics in the transition section.

[0019] Rolling optimization is performed using a particle swarm optimization algorithm with a penalty function. The corresponding optimization objective function is: .

[0020] .

[0021] in, To optimize the objective function; For prediction in the time domain; This refers to the sequence number of future discrete time steps within the prediction or control time domain. For the first The expected state trajectory at each moment; The desired state trajectory; The predicted state is calculated recursively using a compact scheme dynamic linearization model; The first, calculated recursively using the compact scheme dynamic linearization model, is... The predicted state at each moment; It is the square of the second norm of the vector (representing the sum of squares of error or energy); To control the time domain; To control the incremental weighting coefficient; For the first The increment of the virtual control quantity at each moment; For penalty weighting; For proxy functions; , , All are weighting coefficients; The difference in rotational speed between the left and right rotors; For the present The difference in rotational speed between the left and right rotors at any given moment; , , All of these are coefficients that determine the steepness of the penalty; The roll velocity of the aircraft as measured by the IMU; The maximum permissible safe limit for the aircraft's roll rate; For the present Strain data of the rotor blade root structure at any given time; This represents the maximum permissible structural strain safety limit value at the wing root.

[0022] The safety corridor constraints include: safe airspeed and tilt angle matching corridor constraints, actuator asymmetric limit constraints, and virtual torque increment constraints.

[0023] The corridor constraint for matching safe airspeed and tilt angle is: .

[0024] The asymmetric limit constraint of the actuator is: .

[0025] The virtual torque increment constraint is: .

[0026] in, The tilt angle; This is the critical tilt angle for stall. Airspeed; The difference in rotational speed between the left and right rotors; This is the absolute value of the difference in rotational speed between the left and right rotors; This represents the maximum difference in rotational speed between the left and right rotors. For the increment of virtual control quantity; The lower limit of the rate of change (increment) of the three-axis virtual torque that the tiltrotor aircraft as a whole can provide; The upper limit of the rate of change (increment) of the three-axis virtual torque that can be provided for the tiltrotor aircraft as a whole.

[0027] Step 300: Within the lower-level control architecture, based on the optimized data at the current moment, control allocation is performed using a multi-objective weighted quadratic programming method to obtain the control command for the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled according to the tilt angle to dynamically penalize the usage cost of different actuators, thereby achieving a smooth transfer of control weights between the rotor channel and aerodynamic control surfaces during the transition period.

[0028] Within the lower-level control architecture, based on the current optimization data, control allocation is performed using a multi-objective weighted quadratic programming method to obtain the current control commands, specifically including: Within the lower-level control architecture, based on the current time-of-flight optimization data, the optimization objective function of the multi-objective QP control allocation is solved using a multi-objective weighted quadratic programming method, with physical limit constraints, to obtain the control command for the current time-of-flight.

[0029] The objective function for multi-objective QP control allocation is: .

[0030] Physical limit constraints include position limiting constraints and velocity limiting constraints.

[0031] The position limit constraint is: .

[0032] The rate limiting constraint is: .

[0033] in, An optimization objective function is assigned to multi-objective QP control; The control command vector of the specific executor to be solved in the process of solving a multi-objective quadratic programming problem; For tracking weight matrix of virtual control variables; For the control effectiveness matrix; for Timing control commands; To optimize the data; The tilt mode weighting matrix; The square of the second norm of the vector; The lower limit of the physical position of the actuator (such as the maximum negative physical deflection angle limit of the control surface); The upper limit of the physical position of the actuator (such as the maximum positive physical deflection angle limit of the control surface); for Timing control commands; The physical rate limit of the actuator (such as the maximum positive acceleration of the motor or the maximum positive deflection angular velocity limit of the servo motor). The lower limit of the physical rate of the actuator (such as the maximum negative acceleration of the motor or the maximum negative deflection angular velocity limit of the servo motor).

[0034] tilt mode weighted matrix It is a diagonal matrix with the cost weights of each executor as its diagonal elements; the mathematical expression for the tilt mode weighting matrix is: .

[0035] in, Let be the cost function for the first rotor channel; The tilt angle; For the first The cost function of each rotor channel; The cost function for the first aerodynamic control surface channel; For the first The cost function of an aerodynamic control surface channel; This is a function for constructing diagonal matrices, representing a diagonal matrix constructed using the elements within the parentheses as the main diagonal elements.

[0036] Step 400: Determine the real-time dynamic state of the target tiltrotor aircraft based on the control command at the current moment, and feed it back to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

[0037] Specifically, the real-time dynamic state of the target tiltrotor aircraft is determined based on the control commands at the current moment, and fed back to the compact-format dynamic linearized model within the upper-level control architecture via multi-source airborne sensors for closed-loop control. This includes: The aerodynamics and torque of the target tiltrotor aircraft are determined based on the control commands at the current moment to obtain the real-time dynamic state; the real-time dynamic state is the flight state response.

[0038] The real-time dynamic state, three-axis attitude angles and three-axis angular velocities of the airframe are synchronously collected by multiple airborne sensors to form a state vector corresponding to the state data. Additionally, the tilt angle and tilt angular velocity are collected in real time to obtain the corresponding information data, which are used as new multi-source sensing data.

[0039] The new multi-source sensing data is distributed to the compact-format dynamic linearized model in the upper-level control architecture of the two-layer control architecture for closed-loop control. The closed-loop process corresponding to the closed-loop control is continuously executed at a fixed operation frequency until it smoothly transitions to the target fixed-wing cruise mode.

[0040] This application proposes a two-layer security control architecture, such as Figure 2 As shown: The upper layer uses an algorithm that combines Model-Free Adaptive Predictive Control (MFAPC) and Particle Swarm Optimization (PSO) to generate the desired three-axis virtual torque, while the lower layer uses a Quadratic Programming (QP) allocation algorithm with mode weights to generate specific actuator instructions.

[0041] This application introduces a tilt angle-gated model update mechanism: a forgetting factor that dynamically changes with the tilt angle and tilt angle rate is designed to adaptively adjust the pseudo-partial derivative matrix of the Compact Form Dynamic Linearization (CFDL) model, thereby solving the mismatch problem caused by the rapid change of aerodynamic characteristics in the transition section.

[0042] This application constructs a PSO objective function that includes a roll risk penalty: safety boundaries such as the difference in rotational speed between the left and right rotors and the roll state of the airframe are explicitly added to the upper-level optimization objective to actively avoid the roll risk caused by thrust asymmetry.

[0043] This application designs a control allocation strategy with tilt mode weights: by dynamically scheduling the control mode weight matrix with tilt angle, the smooth transfer of control weights between the rotor channel and aerodynamic control surfaces during the transition period is achieved, thus solving the problem of redundancy and counteraction among multiple control mechanisms.

[0044] Upper-level control architecture (MFAPC-PSO): It does not rely on a precise mechanistic model, but learns a local model online using input-output data, and uses PSO to continuously optimize the output data, i.e., the desired three-axis torque / virtual control quantity, expressed as... .

[0045] Lower-level control architecture (QP control allocation): ... Under the constraints of actuator limits, speed, and power, specific control commands are assigned. (collective pitch / cyclic pitch / speed / rudder surface / tilt mechanism, etc.).

[0046] Specifically, such as Figure 3 and Figure 5 As shown, the method mentioned in this application includes the following steps: Step 1: 1.1 Constructing the CFDL model: The actual output of the system using sensor feedback (Three-axis attitude angles, angular rates, etc.) and the virtual control quantities input by the actuator at the previous moment. (Desired three-axis torque / virtual control quantity) and perform data-driven modeling.

[0047] A compact-format dynamic linearized data model is established, and its discrete mathematical expression is as follows: .

[0048] in, The pseudo-partial derivative (PPD) matrix is ​​estimated online, which is equivalent to "the sensitivity of the input to the output at the current operating point"; This represents the increment of the virtual control quantity. for Real-time flight status response.

[0049] Tilting angle gating mechanism introduces online update of projection estimation algorithm And utilize tilt angle and tilt rate Dynamically adjust the forgetting factor : When the tilt is violent (i.e., the tilt angular rate) (The value of the forgetting factor is relatively large), and the value of the forgetting factor is adaptively reduced to give greater weight to the latest sampled data, thereby speeding up the model update speed to cope with rapid changes in aerodynamic properties.

[0050] When in a steady state (i.e., the tilt rate is small or zero), the forgetting factor is adaptively increased (to make it approach 1), thereby enhancing the system's noise resistance and maintaining a stable update pace.

[0051] 1.2 Based on the CFDL model constructed above and the safety corridor constraint, PSO rolling optimization is performed using the Penalty-based PSO algorithm to obtain the desired three-axis torque / virtual control quantity. Figure 4 Flowchart for PSO rolling optimization.

[0052] The direct output of the upper-level optimization controller (i.e., the upper-level control architecture) is not the lower-level aerodynamic control surface or rotor speed command, but rather the desired three-axis virtual control torque (desired three-axis torque / virtual control quantity). This design makes the upper-level control architecture more in line with the "physical consistency" of aircraft rigid body dynamics, while significantly reducing the dimensionality of the optimization solution and improving computational efficiency.

[0053] The optimization objective of the PSO algorithm is to find a set of future control increment sequences that makes the predicted output as close as possible to the desired trajectory, while ensuring smooth control actions and strictly avoiding the risk of rollover. Its optimization objective function is... The specific expression is constructed as follows: .

[0054] Tracking item: , The predicted state is obtained through recursive calculation using the CFDL model.

[0055] Control smoothing term: , < , The incremental weighting coefficient is used to penalize drastic changes in virtual torque and prevent high-frequency oscillations.

[0056] Side tilt risk penalty items (core safety indicators): ,in, This constitutes a very high penalty weight. It is due to the difference in rotational speed between the left and right rotors IMU-measured roll velocity of the aircraft and roll acceleration And a surrogate function constructed by fusing strain data from the root structure of the rotor wing.

[0057] Specifically, the surrogate function adopts a normalized weighted model based on an exponential barrier function, and its expression is as follows: .

[0058] Each denominator represents the corresponding physical safety limit value.

[0059] When the predicted control input leads to severe thrust asymmetry and approaches the instability critical point... The fitness of the particle increases exponentially, forcing the PSO optimizer to rate the particle as extremely poor, thus actively avoiding dangerous control commands with large asymmetry.

[0060] Specific hard constraints for PSO rolling optimization: In the optimization process, in addition to the soft penalty in the objective function mentioned above, the position of each particle in the search space must also strictly satisfy the following physical hard constraints (i.e., safe corridor constraints), otherwise it will be forced to be pulled back to the boundary or eliminated: 1. Corridor constraints for matching safe airspeed and tilt angle: That is, the current real airspeed The tilt angle must be greater than or equal to the stall critical tilt angle to prevent the aircraft from losing altitude and stalling.

[0061] 2. Asymmetric limit constraints of actuators: Limiting the absolute value of the maximum speed difference between the left and right rotors locks the source of the maximum tilting moment that causes rollover from the physical boundary.

[0062] 3. Virtual torque increment constraint: This represents the maximum rate of change of the three-axis torque that the tiltrotor aircraft as a whole can provide.

[0063] PSO computing power guarantee mechanism and optimization process for airborne real-time requirements: To meet the extremely high real-time requirements of airborne flight control, the PSO algorithm in this application has made two core improvements to the standard process: 1. Hot Start Mechanism: At time... k When initializing the particle swarm, instead of randomly generating all particles, the optimal control increment sequence obtained at the previous time step k-1 is shifted one step and used as part of the initial population value at the current time step. This makes the population extremely close to the global optimum in the first generation, significantly reducing the number of iterations required for convergence.

[0064] 2. Truncation Mechanism: Within a single interruption cycle of the flight control system, a maximum computation time limit T_{max} or a maximum number of iterations limit is set. Once the time limit is reached, regardless of whether the current particle swarm has fully converged, the optimization program is forcibly truncated, and the currently found global optimal solution (gbest) is directly output, ensuring that the system outputs control input in every cycle and never experiences computational bottleneck.

[0065] Step 2: Based on the desired three-axis torque / virtual control quantity obtained above, control allocation is performed using a multi-objective weighted quadratic programming method to obtain specific control commands.

[0066] Because tiltrotor aircraft exhibit severe redundancy in control variables during the transition phase (i.e., rotor collective pitch, cyclic pitch, ailerons, elevators, etc., all function simultaneously, resulting in a number of actuators far exceeding the required degrees of control freedom), a multi-objective weighted quadratic programming algorithm is introduced into the lower-level controller. This algorithm combines a control performance matrix dynamically scheduled according to the tilt angle. With pattern weighted matrix The desired torque sent down from the upper level, i.e. Real-time allocation of the optimal specific execution mechanism instructions, i.e., control instructions. .

[0067] Construct the optimization objective function for multi-objective QP control allocation: To ensure accurate tracking of the desired torque while minimizing energy consumption and wear of each actuator, a multi-objective QP optimization function is constructed, consisting of minimizing allocation error and minimizing control energy consumption: .

[0068] It includes rotor channel commands (such as collective pitch of left and right rotors, longitudinal and lateral cyclic pitch) and aerodynamic control surface commands (such as aileron and elevator deflection).

[0069] : Control effectiveness matrix. It represents the control effectiveness at the current tilt angle. The physical efficiency of each actuator in generating triaxial torque is obtained in real time through wind tunnel calibration or high-fidelity aerodynamic data lookup table interpolation.

[0070] The virtual control quantity tracking weight matrix (a constant diagonal matrix) is used to determine the penalty priority for allocating errors to the roll, pitch, and yaw channels.

[0071] : Tilting mode weighted matrix, used to dynamically penalize the usage costs of different actuators.

[0072] Setting physical hard constraints: In solving the above During the process of finding the minimum value, it is essential to ensure that the solution is found. Satisfy the physical limit constraints of all current actuators: Position limiting constraints: (For example, the maximum physical deflection limit of the control surface).

[0073] Rate limiting constraint: (For example, the maximum acceleration of the motor or the maximum angular velocity limit of the servo motor).

[0074] tilt mode weighted matrix Mechanism of action and calculation relationship: (1) Function of parameters: During transitional flight, if the system is allowed to allocate resources freely, it is highly likely that the rotor and aerodynamic control surfaces will output opposite commands for the same attitude target (i.e., "controllability conflict" or "internal friction"). Mode weighting matrix This is equivalent to attaching a "price tag" to each executor. When the weight (price) of an executor is set extremely high, the QP optimization solver, in order to... If the lowest weight is reached, the executor will be automatically discarded and replaced with an executor with a lower weight (price).

[0075] (2) Computational relationship with the actuator and dynamic scheduling logic: It is a diagonal matrix with the cost weights of each executor as its diagonal elements, and its mathematical structure is as follows: .

[0076] With tilt angle The computational scheduling relationship is as follows: In the initial transition phase (helicopter mode), (Approaching 90°): At this point, the airspeed is low, and the aerodynamic control surface efficiency is extremely poor. The scheduling function assigns weights to the aerodynamic control surfaces, that is, it... Setting it to the maximum value will increase the rotor channel weight, which means... Set to a minimum value. The QP solver is penalized with extremely high control surface costs, which suppresses aerodynamic control surface deflection during allocation, making the system rely entirely on the rotor channel for control torque.

[0077] In the later stages of the transition (airspeed established, approaching fixed-wing mode), (Approaching 0°): As the tilt angle decreases and airspeed increases, the efficiency of aerodynamic control surfaces gradually becomes apparent, while the cyclic pitch capability of the rotor gradually becomes limited. The scheduling function makes... Follow The decrease in speed leads to a smoother decay, while simultaneously making... Follow It gradually increases as it decreases.

[0078] Through this diagonal matrix element random With a computational relationship that exhibits reverse cross-changes, the QP allocator naturally achieves a seamless and smooth transfer of control authority from rotor-dominated to control surface-dominated at the mathematical solution level, completely resolving the control conflict problem between multiple actuator systems.

[0079] Step 3: Obtain the real-time dynamic state of the tiltrotor aircraft based on specific control commands, and feed it back to the upper-level CFDL prediction model for closed-loop control through multi-source airborne sensors.

[0080] The tiltrotor aircraft receives the specific actuator commands calculated by the underlying distributor, that is... Subsequently, the aircraft will generate actual dynamic responses at the physical level and complete state perception and multi-dimensional closed-loop feedback through an onboard sensor network. The specific closed-loop control process is as follows: Physical command execution and state evolution: The various physical actuators of the tiltrotor aircraft (including the rotor collective pitch servo mechanism, cyclic pitch swashplate, aileron, elevator, and tilt servo motor, etc.) strictly follow... These actions generate actual aerodynamic forces and torques in real, complex flow field environments, changing the system's dynamic state from its current state. The moment evolves to the next control beat At all times, output the latest real flight status response. .

[0081] Data acquisition and state estimation of multi-source heterogeneous sensors: In At any given moment, the airborne multi-source sensor network and state estimation module synchronously collect the latest flight state and physical boundary data: High-precision integrated navigation system (IMU / AHRS): measures and outputs the latest three-axis attitude angles and three-axis angular rates of the airframe. These, etc., constitute a new actual system output state vector.

[0082] Atmospheric Data Computer / Pittograph: Collects the current real forward airspeed of the aircraft. .

[0083] High-precision encoder and tachometer: Real-time reading of the absolute tilt angle of the current tilt nacelle. (k+1) and calculate the tilt rate using difference. Simultaneously, the actual rotational speeds of the left and right rotors are obtained and the speed difference is calculated. .

[0084] Global multidimensional closed-loop feedback mechanism (core coordination process): The latest multi-source sensing data collected above will be specifically distributed to different modules of the two-layer control architecture to achieve closed-loop updating of the algorithm's core parameters: (1) Feedback to the upper-level CFDL prediction model: The latest actual output state y(k+1) is fed back to the predictor, and the true prediction error is calculated by comparing it with the prediction state from the previous cycle. This error is then combined with the virtual control increment from the previous cycle. A novel pseudo-partial derivative (PPD) matrix is ​​calculated and adaptively updated using a projection estimation algorithm. .

[0085] At the same time, the latest tilt rate A tilt angle gating mechanism is implemented to dynamically refresh the forgetting factor. Thus, the data-driven model completes online learning and closed-loop correction of the latest flow field aerodynamic characteristics.

[0086] (2) Feedback to PSO optimizer (dynamic perception of tilt risk): The latest collected speed difference Asymmetric physical quantities such as roll rate p(k+1) are re-introduced into the risk penalty model to update the roll risk penalty term for the next control cycle. If external gusts exacerbate the aircraft's actual asymmetry, the updated penalty term will increase significantly, forcing the next PSO algorithm to proactively output a more conservative virtual torque, thus achieving a dynamic closed loop at the safety boundary.

[0087] (3) Feedback to the lower-level QP allocator (dynamic scheduling of allocation weights): The latest tilt angle (k+1) and airspeed V are fed into the lower-level controller, and the system dynamically updates the control performance matrix. Simultaneously, the diagonal weighted elements are recalculated, and the tilt mode weighted matrix is ​​refreshed. This allows for a smoother transfer of control between the rotor and aerodynamic control surfaces in the next calculation cycle.

[0088] Timing and Rolling Optimization: After updating all the models and parameters, the system control timing is advanced from k to k+1. The flight control system then uses the updated timing... , and With the new initial environmental conditions, a new round of PSO rolling optimization and QP allocation calculation is restarted. This closed-loop process is continuously executed at a fixed calculation frequency until a smooth transition to the target fixed-wing cruise mode is achieved.

[0089] This application overcomes the bottleneck of the extreme difficulty in accurately establishing aerodynamic models for the tilt transition section (high robustness). It relies on CFDL data-driven modeling based on tilt angle gating in step one. Traditional mechanistic modeling or fixed MPC models often suffer severe mismatch due to abrupt aerodynamic changes during the tilt transition. This application directly uses input and output data to update the pseudo-partial derivative matrix online and dynamically adjusts the forgetting factor through tilt angle rate, enabling the model to adapt rapidly to the nonlinear aerodynamic changes in the transition section.

[0090] This application significantly reduces the risk of tilting and flipping during the transition period from the algorithm's source (high safety), namely the tilting risk index introduced by the PSO objective function in step one. Traditional control systems often passively correct dangerous attitudes only after they occur. In contrast, this application treats safety hazards such as the difference in thrust between the left and right rotors as predictive penalties, proactively "bypassing" control quantities that could lead to catastrophic rollovers during the command generation stage (within the future time domain), thus internalizing safety.

[0091] This application solves the control resistance problem (high smoothness) caused by redundancy of multiple control mechanisms. Specifically, it addresses the QP control allocation with tilt mode weights in step two. Because the rotor and aerodynamic control surfaces operate simultaneously during the transition phase, a "struggle" between the control surfaces and the rotor is highly likely. This application introduces a mode weighting matrix that is adjusted according to the tilt angle. This enables a seamless and smooth dynamic transfer of control authority between the rotor and the control surfaces.

[0092] This application meets the hard real-time requirements (high solution efficiency) of airborne flight control systems for computing resources, based on the hot start and time-limited truncation mechanism of the PSO algorithm in step one. Conventional nonlinear PSO optimization has a huge computational load, making it difficult to deploy on airborne microcontrollers. This application significantly reduces the number of iterations by shifting the solution of the previous cycle for hot start, and ensures that instructions are issued within the control cycle through the truncation mechanism, thus completely solving the engineering pain point of airborne computing lag.

[0093] In addition, in the rolling optimization algorithm of the upper control architecture, besides using the particle swarm optimization algorithm (PSO) with a penalty function, genetic algorithm (GA), differential evolution algorithm (DE), or sequential quadratic programming (SQP) can also be used to solve the problem. However, PSO has a better balance between optimization efficiency and engineering implementation difficulty when dealing with discontinuous penalty functions and hot start.

[0094] In the data-driven prediction stage, besides using CFDL (Compact Scheme Dynamic Linearization), BP neural networks (Back Propagation Neural Networks) or radial basis function (RBF) neural networks can also be used for online identification. However, neural networks require long training times and large sample databases, have weak interpretability, and are prone to gradient explosion or overfitting in rapidly changing tilt transitions. The CFDL proposed in this application is more conducive to explicit mathematical gating intervention combined with the tilt angle.

[0095] In one exemplary embodiment, such as Figure 6 As shown, a dual-layer safety control device for the transition section of a tiltrotor aircraft is provided, comprising: The information data acquisition module is used to acquire information data of the target tiltrotor aircraft at the current moment; the information data includes: tilt angle, tilt rate, and status data fed back by multiple airborne sensors; the status data includes: the difference in rotational speed between the left and right rotors, the airframe roll rate and roll acceleration measured by the IMU, and the strain data of the rotor root structure.

[0096] The rolling optimization module is used within the upper-level control architecture to perform rolling optimization using a particle swarm optimization algorithm with a penalty function, based on the current time-bound compact dynamic linearized model and the corresponding safety corridor constraints, to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity. The compact dynamic linearized model is obtained by adaptive adjustment based on the state data and the optimization data from the previous time-bound moment using a tilt angle gating update mechanism. The tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by rapid changes in aerodynamic characteristics during the transition section.

[0097] The control allocation module is used to allocate control based on the current optimized data within the lower control architecture using a multi-objective weighted quadratic programming method to obtain the control command for the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled according to the tilt angle to dynamically penalize the usage cost of different actuators, thereby achieving a smooth transfer of the control weights of the rotor channel and aerodynamic control surfaces during the transition period.

[0098] The closed-loop control module is used to determine the real-time dynamic state of the target tiltrotor aircraft based on the control command at the current moment, and to feed back the data to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

[0099] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 7 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores dual-layer safety control data for the tiltrotor transition section. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the dual-layer safety control method for the tiltrotor transition section.

[0100] Those skilled in the art will understand that Figure 7The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0101] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0102] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0103] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0104] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0105] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0106] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logic devices, etc., and are not limited to these.

[0107] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0108] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A dual-layer safety control method for the transition section of a tiltrotor aircraft, characterized in that, The tiltrotor aircraft transition section dual-layer safety control method adopts a dual-layer safety control architecture for control. The dual-layer security control architecture includes: an upper-layer control architecture and a lower-layer control architecture; The dual-layer safety control method for the transition section of the tiltrotor aircraft includes: Acquire information data of the target tiltrotor aircraft at the current moment; the information data includes: tilt angle, tilt rate and status data fed back by multiple airborne sensors; the status data includes: the difference in rotational speed between the left and right rotors, the airframe roll rate and roll acceleration measured by the IMU, and the strain data of the rotor root structure. Within the upper-level control architecture, based on the current time-bound compact dynamic linearization model and the corresponding safety corridor constraints, rolling optimization is performed using a particle swarm optimization algorithm with a penalty function to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity. The compact dynamic linearization model is obtained by adaptive adjustment based on the state data and the optimization data corresponding to the previous time-bound moment using a tilt angle gating update mechanism. The tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by the rapid change of aerodynamic characteristics in the transition section. Within the lower-level control architecture, based on the optimized data at the current moment, control allocation is performed through a multi-objective weighted quadratic programming method to obtain the control command at the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled with the tilt angle to dynamically penalize the usage cost of different actuators, thereby achieving a smooth transfer of the control weight between the rotor channel and the aerodynamic control surface during the transition period. The real-time dynamic state of the target tiltrotor is determined based on the control commands at the current moment, and fed back to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

2. The dual-layer safety control method for the transition section of a tiltrotor aircraft according to claim 1, characterized in that, Rolling optimization is performed using a particle swarm optimization algorithm with a penalty function. The corresponding optimization objective function is: ； ； in, To optimize the objective function; For prediction in the time domain; This refers to the sequence number of future discrete time steps within the prediction or control time domain. For the first The expected state trajectory at each moment; The desired state trajectory; The predicted state is calculated recursively using a compact scheme dynamic linearization model; The first, calculated recursively using the compact scheme dynamic linearization model, is... The predicted state at each moment; Let be the square of the second norm of the vector; To control the time domain; To control the incremental weighting coefficient; For the first The increment of the virtual control quantity at each moment; For penalty weighting; For proxy functions; , , All are weighting coefficients; The difference in rotational speed between the left and right rotors; For the present The difference in rotational speed between the left and right rotors at any given moment; This represents the maximum difference in rotational speed between the left and right rotors. , , All of these are coefficients that determine the steepness of the penalty; The roll velocity of the aircraft as measured by the IMU; The maximum permissible safe limit for the aircraft's roll rate; For the present Strain data of the rotor blade root structure at any given time; This represents the maximum permissible safety limit value for the structural strain at the wing root. For the present The roll velocity of the aircraft measured by the IMU at a given time.

3. The dual-layer safety control method for the transition section of a tiltrotor aircraft according to claim 1, characterized in that, The safety corridor constraints include: safe airspeed and tilt angle matching corridor constraints, actuator asymmetric limit constraints, and virtual torque increment constraints; The safe airspeed and tilt angle matching corridor constraint is as follows: ； The asymmetric limit constraint of the actuator is: ； The virtual torque increment constraint is: ； in, The tilt angle; This is the critical tilt angle for stall. Airspeed; The difference in rotational speed between the left and right rotors; This is the absolute value of the difference in rotational speed between the left and right rotors; This represents the maximum difference in rotational speed between the left and right rotors. For the increment of virtual control quantity; The lower limit of the three-axis virtual torque variation rate that can be provided for the tiltrotor aircraft as a whole; The upper limit of the three-axis virtual torque variation rate that can be provided for the tiltrotor aircraft as a whole.

4. The dual-layer safety control method for the transition section of a tiltrotor aircraft according to claim 1, characterized in that, Within the lower-level control architecture, based on the current optimization data, control allocation is performed using a multi-objective weighted quadratic programming method to obtain the current control commands, specifically including: Within the lower-level control architecture, based on the current optimization data, the optimization objective function of the multi-objective QP control allocation is solved using a multi-objective weighted quadratic programming method, based on physical limit constraints, to obtain the control command at the current moment. The objective function for multi-objective QP control allocation is: ； The physical limit constraints include: position limiting constraints and rate limiting constraints; The position limiting constraint is as follows: ； The rate limiting constraint is: ； in, An optimization objective function is assigned to multi-objective QP control; The control command vector of the specific executor to be solved in the process of solving a multi-objective quadratic programming problem; For tracking weight matrix of virtual control variables; For the control effectiveness matrix; for Timing control commands; To optimize the data; The tilt mode weighting matrix; Let be the square of the second norm of the vector; The lower limit of the physical location of the actuator; The upper limit of the physical location of the actuator; for Timing control commands; The upper limit of the physical rate of the actuator; This is the lower limit of the physical rate limit for the actuator.

5. The dual-layer safety control method for the transition section of a tiltrotor aircraft according to claim 1, characterized in that, tilt mode weighted matrix It is a diagonal matrix with the cost weights of each executor as its diagonal elements; The mathematical expression corresponding to the tilt mode weighting matrix is: ； in, Let be the cost function for the first rotor channel; The tilt angle; For the first The cost function of each rotor channel; The cost function for the first aerodynamic control surface channel; For the first The cost function of an aerodynamic control surface channel; This is a function for constructing diagonal matrices, representing a diagonal matrix constructed using the elements within the parentheses as the main diagonal elements.

6. The dual-layer safety control method for the transition section of a tiltrotor aircraft according to claim 1, characterized in that, The real-time dynamic state of the target tiltrotor aircraft is determined based on the control commands at the current moment, and fed back to the compact-format dynamic linearized model within the upper-level control architecture via multi-source airborne sensors for closed-loop control. Specifically, this includes: The aerodynamics and torque of the target tiltrotor aircraft are determined based on the control commands at the current moment to obtain the real-time dynamic state; the real-time dynamic state is the flight state response. The real-time dynamic state, three-axis attitude angles and three-axis angular rates of the airframe are synchronously collected by multiple airborne sensors to form a state vector corresponding to the state data. Additionally, the tilt angle and tilt angular rate are collected in real time to obtain the corresponding information data, which are used as new multi-source sensing data. The new multi-source sensing data is distributed to the compact-format dynamic linearized model in the upper-level control architecture of the two-layer control architecture for closed-loop control. The closed-loop process corresponding to the closed-loop control is continuously executed at a fixed operation frequency until it smoothly transitions to the target fixed-wing cruise mode.

7. A dual-layer safety control device for the transition section of a tiltrotor aircraft, characterized in that, include: The information data acquisition module is used to acquire information data of the target tiltrotor aircraft at the current moment; The information data includes: tilt angle, tilt rate, and status data fed back by multiple airborne sensors; the status data includes: the difference in rotational speed between the left and right rotors, the airframe roll rate and roll acceleration measured by the IMU, and the strain data of the rotor root structure. The rolling optimization module is used within the upper-level control architecture to perform rolling optimization using a particle swarm optimization algorithm with a penalty function, based on the current time-bound compact dynamic linearization model and the corresponding safety corridor constraints, to determine the current time-bound optimization data. The optimization data is the desired three-axis torque / virtual control quantity. The compact dynamic linearization model is obtained by adaptive adjustment based on the state data and the optimization data from the previous time-bound moment using a tilt angle gating update mechanism. The tilt angle gating update mechanism uses a projection estimation algorithm to update the pseudo-partial derivative matrix and dynamically adjusts the forgetting factor using the tilt angle and tilt angle rate to solve the mismatch problem caused by rapid changes in aerodynamic characteristics during the transition section. The control allocation module is used to perform control allocation based on the current optimized data within the lower control architecture, using a multi-objective weighted quadratic programming method to obtain the control command for the current moment. The multi-objective weighted quadratic programming method introduces a tilt mode weighting matrix that is dynamically scheduled according to the tilt angle, so as to dynamically penalize the usage cost of different actuators and realize the smooth transfer of the control weight of the rotor channel and aerodynamic control surface during the transition period. The closed-loop control module is used to determine the real-time dynamic state of the target tiltrotor aircraft based on the control command at the current moment, and to feed back the data to the compact dynamic linearization model in the upper control architecture through multi-source airborne sensors for closed-loop control.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the tiltrotor transition section dual-layer safety control method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the dual-layer safety control method for the transition section of a tiltrotor aircraft as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the dual-layer safety control method for the transition section of a tiltrotor aircraft as described in any one of claims 1-6.

Images