Landing control method and system under fault condition of single-side aileron of airplane
Through a pre-trained neural network, the aircraft's unilateral aileron failure is identified and trimmed and adaptively controlled, the problem of aircraft landing control is solved under the unilateral aileron failure, and the flight safety and maneuverability efficiency are improved.
Patent Information
- Application Number
- CN202411742276.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-07-18
AI Technical Summary
In the case of a single-side aileron failure of the aircraft, the difficulty of flight landing control is increased, especially on mobile platforms, affected by landing area restrictions, platform movement and airflow disturbances, and the existing technology is difficult to effectively solve.
A pre-trained neural network is used to identify aileron fault categories, determine the climb rate and turn rate of steady-state climbing turns to match, and solve the flight control parameters through adaptive control, and combine rolling angle and side-slip angle instructions for adaptive control.
Assist pilots to achieve double-action platform drop and fall trajectory control, improve flight quality and safety, and reduce the handling burden.
Smart Images
Figure CN120335463A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the technical field of flight control, and particularly relates to a landing control method and system for an aircraft in the case of a single-side aileron failure. Background Art
[0002] The aileron flap is one of the important actuators for aircraft flight control. When a single-side aileron fails and is damaged, the remaining flight ability of the aircraft can still meet the requirements for basic return and landing. The flight control system can collect the aileron flap deflection for trimming, so as to calculate the flight control law. However, in some fault states, the effective reference cannot be formed by the feedback value of the flap position, which increases the difficulty of flight landing control. For landing on a moving platform, affected by various factors such as landing area limitation, platform movement, and airflow disturbance, the landing control in the case of a single-side aileron failure becomes more difficult. Summary of the Invention
[0003] In order to solve the above problems, this application provides a landing control method and system for an aircraft in the case of a single-side aileron failure, so as to solve the problem of aircraft landing control when the parameters of the faulty aileron cannot be obtained.
[0004] The first aspect of this application provides a landing control method for an aircraft in the case of a single-side aileron failure, which mainly includes:
[0005] Step S1: Determine the single-side aileron failure category of the aircraft based on a pre-trained neural network;
[0006] Step S2: Determine the climb rate and turn rate that enable the aircraft to climb and turn steadily in a state corresponding to the aileron failure category, trim the aircraft, and obtain the trim control parameter state parameters;
[0007] Step S3: Calculate the flight control parameters according to the real-time positions of the current position of the aircraft and the landing point;
[0008] Step S4: Perform adaptive control on the roll angle command and the sideslip angle command.
[0009] Preferably, further before step S1, training the neural network is included, and the training process includes:
[0010] Step S11: Construct training samples based on the flight data sets of the aircraft in normal and fault states. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the labeled aileron failure category;
[0011] Step S12: Input the input data of the training samples into a neural network model including two hidden layers to obtain the model output data;
[0012] Step S13: Compare the model output data with the output data of the training samples, and optimize the parameters of the neural network model based on the comparison results.
[0013] Preferably, in step S11, the flight states and control variables include pitch angle, angle of attack, speed roll angle, roll rate, pitch rate, yaw rate, aileron deflection, elevator deflection, rudder deflection, and throttle control amount; the aileron failure categories include jamming, drift, saturation, and partial damage.
[0014] Preferably, in step S3, the flight control parameters include heading angle command, lateral guidance command, climb angle command, and longitudinal guidance command.
[0015] The second aspect of the present application provides a landing control system for an aircraft in the case of a single-sided aileron failure, mainly including:
[0016] An aileron failure category recognition module for determining the aileron failure category of the aircraft on one side based on a pre-trained neural network;
[0017] A flight trim module for determining the climb rate and turn rate for the aircraft to perform a steady-state climb and turn corresponding to the aileron failure category, trimming the aircraft, and obtaining the trim control parameter state parameters;
[0018] A flight control parameter calculation module for calculating flight control parameters according to the real-time positions of the current position of the aircraft and the landing point;
[0019] An adaptive control module for adaptively controlling the roll angle command and the sideslip angle command.
[0020] Preferably, the aileron failure category recognition module includes a neural network training unit for training the neural network. The neural network training unit includes:
[0021] A training sample acquisition sub-unit for constructing training samples based on the flight data sets of the aircraft in normal and failure states. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the labeled aileron failure category;
[0022] A neural network model output data acquisition sub-unit for inputting the input data of the training samples into a neural network model with two hidden layers to obtain the model output data;
[0023] A neural network model parameter optimization sub-unit for comparing the model output data with the output data of the training samples and optimizing the parameters of the neural network model based on the comparison results.
[0024] Preferably, the flight state and control variables include pitch angle, angle of attack, speed roll angle, roll rate, pitch rate, yaw rate, aileron deflection, elevator deflection, rudder deflection, and throttle control amount; the aileron failure categories include jamming, drift, saturation, and partial damage.
[0025] Preferably, the flight control parameters include heading angle command, lateral guidance command, climb angle command, and longitudinal guidance command.
[0026] This application can assist the pilot in controlling the landing motion trajectory of the dual-actuator platform, improving flight quality and safety, and reducing the pilot's operating burden under fault conditions. Description of the Drawings
[0027] Figure 1 is a flowchart of a preferred embodiment of the landing control method for an aircraft with a single-sided aileron failure in this application.
[0028] Figure 2 is a schematic diagram of the lateral and directional adaptive landing stability augmentation control architecture. Detailed Embodiments
[0029] To make the purpose, technical solutions, and advantages of the implementation of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the drawings in the embodiments of this application. In the drawings, the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The described embodiments are some, but not all, of the embodiments of this application. The embodiments described below by referring to the drawings are exemplary and are intended to explain this application and should not be construed as limiting this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of this application without creative efforts shall fall within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the drawings.
[0030] The first aspect of this application provides a landing control method for an aircraft with a single-sided aileron failure, as Figure 1 and Figure 2 shown, mainly including:
[0031] Step S1: Determine the failure category of the single-sided aileron of the aircraft based on a pre-trained neural network.
[0032] In step S1 of this application, the inference and identification of the effectiveness of the failed aileron are directly implemented based on the neural network to detect the aileron failure type and damage degree, providing a basis for subsequent trimming and adaptive compensation.
[0033] In some alternative embodiments, further training of the neural network is included before step S1, and the training process includes:
[0034] Step S11: Construct training samples based on the flight data sets of the aircraft in the normal state and the fault state. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the labeled aileron fault category; the flight states and control variables include pitch angle, angle of attack, speed roll angle, roll rate, pitch rate, yaw rate, aileron deflection, elevator deflection, rudder deflection, and throttle control amount; the aileron fault category includes jamming, drift, saturation, and partial damage.
[0035] Step S12: Input the input data of the training samples into a neural network model with two hidden layers to obtain the model output data.
[0036] Step S13: Compare the model output data with the output data of the training samples, and optimize the parameters of the neural network model based on the comparison results.
[0037] In this embodiment, the flight data sets of the aircraft in the normal state and the fault state are constructed. The aircraft simulates the injection of actuator system faults on the simulation platform. The present application selects the data of 10 relevant variables, including the following: pitch angle (theta), angle of attack (alpha), speed roll angle (miu), roll rate (p), pitch rate (q), yaw rate (r), aileron deflection (delta_a), elevator deflection (delta_e), rudder deflection (delta_r), and throttle control amount (delta_p). The constructed neural network model is a four-layer network structure, where: the number of data sample nodes in the input layer is set to 10n, where n is the sequence within the specified time range, that is, a single input data includes n thetas, n alphas... n delta_ps.
[0038] To improve the fitting efficiency, the present application sets 2 hidden layers. The aileron fault state is divided into 4 categories, so the output layer contains 4 neurons. Calculate their respective probabilities, and select the data label corresponding to the maximum probability as the diagnosis result of the neural network model. Through the simulation results, a diagnosis success rate of more than 90% can be achieved.
[0039] Step S2: Determine the climb rate and turn rate that enable the aircraft to perform a steady-state climbing turn corresponding to the aileron fault category, trim the aircraft, and obtain the trim control parameter state parameters.
[0040] The aileron fault breaks the original balance of the aircraft, and the aircraft needs to be trimmed according to the new kinematic characteristics of the aircraft. Select the flight condition after aileron damage or fault as the steady-state climbing turn flight. Perform trim calculations at each flight state according to the nonlinear optimization, and then a trim database under different faults can be obtained. The six-degree-of-freedom nonlinear equation of the rigid aircraft can be abbreviated as The 12 - dimensional state vector x T =[ρ T , ν T can be divided into two vectors. ρ T =[x, y, h, φ, θ, ψ] that describes the kinematic state of the aircraft and ν T =[V t , α, β, p, q, r]. The trimmed state is a stable flight state. At this time, the steady - state climbing and turning flight conditions are determined by the following state equations:
[0041]
[0042] where, respectively represent the climbing rate and turning rate of the desired trimmed state. A stable climbing and turning flight state can be completely defined by eight state variables of the aircraft, written as a set of simplified state vectors as:
[0043]
[0044] At a specific solving can obtain the trimmed control input and the state According to the climbing and turning flight conditions, and are related to the four parameters , then the solution of the equilibrium point can be expressed as and
[0045] In order to find the above - mentioned series of trimmed states, according to the conditions of steady - state climbing and turning flight, applying constrained non - linear constrained optimization gives the mathematical model of the above problem as follows:
[0046]
[0047] obj trim (x, u) is the optimization objective function, Q is an arbitrary stabilizing matrix. Among the constraint conditions,
[0048]
[0049] where, a = cosαcosβ, b = sinφsinβ + cosφsinαcosβ, the trimmed flight path angle γ * satisfies the relationship The first two constraints directly limit the altitude and airspeed of the aircraft, the third constraint indirectly specifies the required climbing rate, and the last three constraints indirectly specify the required turning rate. At this time, the climbing rate and turning rate that reflect the flight performance under the climbing and turning flight conditions can be obtained:
[0050]
[0051] s(u) represents the set of control input limits, including engine thrust, the extreme deflection positions of control surfaces, etc. When obj trim (x,u) reaches the minimum, the aircraft is trimmed under the flight conditions and at this time the objective function is expressed as According to the above relationship, the equilibrium point can be solved based on the relevant algorithms for solving constrained non-linear optimization problems, and finally the preliminary trim state under the condition of single-side aileron failure is obtained.
[0052] Step S3: Calculate the flight control parameters according to the real-time positions of the current position of the aircraft and the landing point.
[0053] In some alternative embodiments, in step S3, the flight control parameters include heading angle command, lateral guidance command, climb angle command, and longitudinal guidance command.
[0054] For the landing process on a mobile platform, it is required that the aircraft fly precisely along the reference glide path attached to the mobile platform until it reaches the desired landing point on the mobile platform. First, a double-moving platform coordinate system of the aircraft and the mobile platform is established.
[0055] The inertial coordinate system is where the three coordinate axes point north, east, and local directions respectively. The path transformation coordinate system is fixedly connected to the moving target, and its origin coordinates in the inertial system are is the desired geometric path defined in the path transformation coordinate system, where l represents the path length. The linear velocity and angular velocity in the {P} system are expressed in the {I} system as The coordinate system is fixedly connected to the desired path p d and the origin is located at the virtual target point. In the coordinate system {F}, is along the tangent direction of the desired path p d , is perpendicular to and points to the right, and the direction follows the right-hand rule and points downward. The flight trajectory coordinate system is fixedly connected to the aircraft, points to the ground speed direction, points to the right, points downward.
[0056] The flight trajectory angles, χ and γ, in the moving path tracking are the heading angle and climb angle of the aircraft respectively. is the angle between the coordinate system {I} and the coordinate system {F}, χF , γ F is the angle between the coordinate system {F} and the coordinate system {W}, and there is a geometric relationship
[0057] Furthermore, design the aircraft landing guidance law after the control surface fails. According to the aircraft model, the translational dynamics equation of trajectory tracking can be rewritten as:
[0058]
[0059] where
[0060] Δ = -cosγ F cosχ F ( I v dx + I ω dy Δz - I ω dz Δy) - cosγ F sinχ F ( I v dy + I ω dz Δx - I ω dx Δz) + sinγ F ( I v dz + I ω dx Δy - I ω dy Δx);
[0061] Σ = sinχ F ( I v dx + I ω dy Δz - I ω dz Δy) - cosχ F ( I v dy + I ω dz Δx - I ω dx Δz);
[0062] Γ = -sinγ F cosχ F ( I v dx + I ω dy Δz - I ω dz Δy) - sinγ Fsinχ F ( I v dy + I ω dz Δx - I ω dx Δz) - cosγ F ( I v dz + I ω dx Δy - I ω dy Δx).
[0063] In three - dimensional moving path tracking, it is required that the aircraft track the virtual target point on the desired path at a speed , so the desired x, y, z coordinates are respectively
[0064] Furthermore, the path update law is designed as:
[0065]
[0066] Furthermore, the heading angle command is designed as:
[0067]
[0068] Considering the lateral simplified model It can be obtained that The lateral guidance command is:
[0069]
[0070] Furthermore, the climb angle command is designed:
[0071]
[0072] Considering that the aircraft is in a coordinated flight state during the landing process, that is, the desired sideslip angle β * is zero. When β is very small, there is a relationship between the climb angle γ, pitch angle θ and angle of attack α, expressed as θ = α + γ.
[0073] Furthermore, the longitudinal guidance command is obtained as:
[0074] Step S4: Perform adaptive control on the roll angle command and sideslip angle command.
[0075] Figure 2 The lateral - directional adaptive landing stability - augmentation control law framework is given. Referring to Figure 2 , the lateral - directional angle loop control during the aircraft landing process consists of two parts.
[0076] (1) The roll angle control law is designed as follows.
[0077] The roll angle dynamic equation is as follows:
[0078]
[0079] Where f φ = qsinφtanθ + rcosφtanθ, g φ = 1.
[0080] Design the transfer function of the command filter:
[0081] Controller
[0082] Thus, the roll angle control law is obtained:
[0083]
[0084] (2) The sideslip angle control law is designed as follows.
[0085] The sideslip angle dynamic equation is as follows:
[0086]
[0087] Where g β = -cosα, where a y is the lateral acceleration.
[0088] Controller Where β ref = 0.
[0089] Thus, the sideslip angle control law is obtained:
[0090]
[0091] Furthermore, the lateral and yaw angular velocity control law is designed as follows. The lateral and yaw angular velocity loop control law consists of a reference control law and an adaptive control law.
[0092] The lateral angle loop model is obtained as:
[0093] Design the transfer function of the command filter:
[0094]
[0095] Design the controller:
[0096]
[0097] The control law obtained is:
[0098]
[0099] Furthermore, the adaptive controller consists of three parts: a state predictor, an adaptation law, and a control law. The partial closed-loop system model generated by the reference control law for the state predictor is:
[0100]
[0101] where σ p , σ r are unknown time-varying inputs.
[0102]
[0103] Let σ p = B mp σ mp + B ump σ ump , σ r = B mr σ mr + B umr σ umr ;
[0104]
[0105] The state predictor is obtained as:
[0106]
[0107] Furthermore, design the adaptation law:
[0108]
[0109] where T s is the sampling time and k is the sampling index.
[0110]
[0111]
[0112] Combining the above formulas, we get:
[0113]
[0114] where is a low-pass filter.
[0115] The final control law is obtained as:
[0116]
[0117] This application infers the failure condition of the unilateral aileron through a neural network, without relying on the control surface sensor, which has high reliability and the accuracy can meet the usage requirements. Based on the diagnosis result, this application uses a landing control strategy that combines the initial trim of the normal control surface and the lateral adaptive compensation, which has better robustness. This application adopts a landing guidance method based on the motion trajectory, giving full play to the strong perception and high-precision advantages of the automatic control system, and assisting the pilot in landing control after the failure and damage.
[0118] The second aspect of this application provides a landing control system in the case of an aircraft unilateral aileron failure corresponding to the above method, mainly including:
[0119] An aileron failure category identification module, which is used to determine the aircraft unilateral aileron failure category based on a pre-trained neural network;
[0120] A flight trim module, which is used to determine the climb rate and turn rate that make the aircraft climb and turn steadily corresponding to the aileron failure category, trim the aircraft, and obtain the trim control parameter state parameters;
[0121] A flight control parameter calculation module, which is used to calculate the flight control parameters according to the real-time positions of the aircraft's current position and the landing point;
[0122] An adaptive control module, which is used to adaptively control the roll angle command and the sideslip angle command.
[0123] In some alternative embodiments, the aileron failure category identification module includes a neural network training unit, which is used to train the neural network. The neural network training unit includes:
[0124] A training sample acquisition sub-unit, which is used to construct training samples based on the flight data sets of the aircraft in the normal state and the failure state. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the marked aileron failure category;
[0125] A neural network model output data acquisition sub-unit, which is used to input the input data of the training samples into a neural network model with two hidden layers to obtain the model output data;
[0126] A neural network model parameter optimization sub-unit, which is used to compare the model output data with the output data of the training samples and optimize the parameters of the neural network model based on the comparison result.
[0127] In some alternative embodiments, the flight states and control variables include pitch angle, angle of attack, speed roll angle, roll angle rate, pitch angle rate, yaw angle rate, aileron deflection angle, elevator deflection angle, rudder deflection angle, throttle control amount; the aileron failure categories include jamming, drift, saturation, and partial damage.
[0128] In some alternative embodiments, the flight control parameters include a heading angle command, a lateral guidance command, a climb angle command, and a longitudinal guidance command.
[0129] As described above, only the specific embodiments of the present application are provided, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims
1. A landing control method in the case of a single-side aileron failure of an aircraft, characterized in that Including: Step S1: Determine the failure category of the unilateral aileron of the aircraft based on a pre-trained neural network; Step S2: Determine the climb rate and turn rate for the aircraft to perform a steady-state climbing turn corresponding to the aileron failure category, trim the aircraft, and obtain the trim control parameter state parameters; Step S3: Calculate the flight control parameters according to the real-time positions of the current position of the aircraft and the landing point; Step S4: Perform adaptive control on the roll angle command and the sideslip angle command.
2. The landing control method in the case of a single-sided aileron failure of an aircraft according to claim 1, characterized in that, Before step S1, it further includes training the neural network. The training process includes: Step S11: Construct training samples based on the flight data sets of the aircraft in normal and failure states. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the labeled aileron failure category; Step S12: Input the input data of the training samples into a neural network model with two hidden layers to obtain the model output data; Step S13: Compare the model output data with the output data of the training samples, and optimize the parameters of the neural network model based on the comparison results.
3. The landing control method in the case of a single-sided aileron failure of an aircraft according to claim 2, wherein In step S11, the flight states and control variables include pitch angle, angle of attack, speed roll angle, roll angle rate, pitch angle rate, yaw angle rate, aileron deflection, elevator deflection, rudder deflection, and throttle control amount; the aileron failure categories include jamming, drift, saturation, and partial damage.
4. The landing control method in the case of a single-side aileron failure of an aircraft according to claim 1, characterized in that, In step S3, the flight control parameters include heading angle command, lateral guidance command, climb angle command, and longitudinal guidance command.
5. A landing control system in the case of a failure of one-sided ailerons of an aircraft, characterized in that, Including: An aileron failure category recognition module, which is used to determine the failure category of the unilateral aileron of the aircraft based on a pre-trained neural network; A flight trim module, which is used to determine the climb rate and turn rate for the aircraft to perform a steady-state climbing turn corresponding to the aileron failure category, trim the aircraft, and obtain the trim control parameter state parameters; A flight control parameter calculation module, which is used to calculate the flight control parameters according to the real-time positions of the current position of the aircraft and the landing point; An adaptive control module, which is used to perform adaptive control on the roll angle command and the sideslip angle command.
6. The landing control system in the case of a single-sided aileron failure of an aircraft according to claim 5, characterized in that, The aileron failure category recognition module includes a neural network training unit for training the neural network. The neural network training unit includes: A training sample acquisition sub-unit, which is used to construct training samples based on the flight data sets of the aircraft in normal and failure states. The input data of the training samples includes the sequence values of multiple flight states and control variables within a specified time range, and the output data of the training samples is the labeled aileron failure category; A neural network model output data acquisition sub-unit, which is used to input the input data of the training samples into a neural network model with two hidden layers to obtain the model output data; A neural network model parameter optimization sub-unit, which is used to compare the model output data with the output data of the training samples, and optimize the parameters of the neural network model based on the comparison results.
7. The landing control system in the case of a single-sided aileron failure of an aircraft according to claim 6, characterized in that The flight states and control variables include pitch angle, angle of attack, speed roll angle, roll angle rate, pitch angle rate, yaw angle rate, aileron deflection, elevator deflection, rudder deflection, and throttle control amount; the aileron failure categories include jamming, drift, saturation, and partial damage.
8. The landing control system in the case of a single-sided aileron failure of an aircraft according to claim 1, characterized in that, The flight control parameters include heading angle command, lateral guidance command, climb angle command and longitudinal guidance command.