Bionic learning-based multi-variable aircraft attitude cooperative control system and method

Abstract

The application discloses a multi-configuration aircraft attitude cooperative control system and method based on bionic learning, belongs to the technical field of multi-configuration aircrafts, and comprises a time-delay interference decoupling module, a multi-configuration aircraft bionic variable configuration module and a multi-configuration aircraft attitude cooperative module. The time-delay interference decoupling module learns and separates the time-delay and interference coupling mechanism of the multi-configuration aircraft, and realizes compensation of the time delay and the interference. The multi-configuration aircraft bionic variable configuration module realizes configuration control of the wing span of the multi-configuration aircraft through bionic learning. The multi-configuration aircraft attitude cooperative module realizes integrated modeling and attitude cooperative control of the multi-configuration aircraft. The time-delay interference decoupling module, the multi-configuration aircraft bionic variable configuration module and the multi-configuration aircraft attitude cooperative module all adopt balanced counterwork networks. The application can make the multi-configuration aircraft have strong autonomous, adaptive and survival bionic intelligent behavior capabilities, and can improve the intelligentization, safety and reliability of the multi-configuration aircraft.

Description

Technical Field

[0001] This invention belongs to the field of variable-structure aircraft technology, specifically relating to a biomimetic learning-based attitude cooperative control system and method for variable-structure aircraft. Background Technology

[0002] In recent years, with the rapid development of logistics, environmental protection, emergency rescue, urban air traffic, and military operations, the demand for unmanned aerial vehicles (UAVs) with "multi-performance capabilities including high and low altitude / high and low speed, long-duration cruise / short takeoff and landing, and multi-mission capability" has become increasingly urgent. Previous fixed-wing monoplane and multiplane UAVs have struggled to meet these requirements, making multi-configuration UAVs the most promising breakthrough approach. However, the wing-changing and attitude coordination process faces challenges such as strong nonlinearity, strong uncertainty, time delay, and interference. Furthermore, the time delay of the deformation actuator leads to uncertainty in the change of the UAV's center of mass, resulting in uncertainty in aerodynamic torque and moment of inertia—a type of model uncertainty—indicating a coupling relationship between the deformation actuator time delay and model uncertainty. A single-configuration UAV is affected by the external environment, causing wing vibration during deformation. This vibration, transmitted via communication, is then transmitted to other communicating UAVs, and due to time delays, the corresponding communicating UAVs experience delayed wing vibration during deformation. These coupling relationships exacerbate the difficulty of attitude coordination control for multi-configuration UAVs. The above factors make the coordinated control of variable wing and attitude a bottleneck problem restricting the research and development and application of multi-configuration aircraft.

[0003] Current technologies typically focus on the attitude cooperative control of multiple aircraft or the control problem of a single variable-configuration aircraft separately, without simultaneously considering the control problem of multiple variable-configuration aircraft. Furthermore, there is a lack of research on the multi-source time delay problems caused by communication and actuator time delays during the configuration process, as well as the cross-coupling problems caused by time delays and interference. Therefore, developing a biomimetic learning-based attitude cooperative control system and method for multiple variable-configuration aircraft is particularly important. Summary of the Invention

[0004] This invention addresses the aforementioned problems and overcomes the shortcomings of existing technologies by providing a biomimetic learning-based attitude cooperative control system and method for multi-configuration aircraft. This invention can effectively improve the reliability, safety, and environmental adaptability of multi-configuration aircraft operation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] The present invention provides a biomimetic learning-based attitude cooperative control system for a multi-configuration aircraft, comprising a time-delay disturbance decoupling module, a biomimetic variable configuration module for the multi-configuration aircraft, and a multi-configuration aircraft attitude cooperative module.

[0007] The time-delay interference decoupling module is used to learn and separate the time-delay and interference coupling mechanism of the multi-configuration aircraft, and to compensate for the time delay and interference.

[0008] The biomimetic variable structure module of the variable structure aircraft is used to achieve configuration control of the wingspan of the variable structure aircraft through biomimetic learning.

[0009] The variable-structure aircraft attitude coordination module is used for integrated modeling and attitude coordination control of the variable-structure aircraft.

[0010] The time-delay interference decoupling module, the biomimetic variant module of the multi-variable aircraft, and the attitude coordination module of the multi-variable aircraft are all implemented using a balanced adversarial network.

[0011] As a preferred embodiment of the present invention, the time-delay interference decoupling module includes a coupling mechanism learning and separation unit and a time-delay interference compensation unit;

[0012] The coupling mechanism learning and separation unit is used to learn the coupling characteristics between the deformation actuator delay and the model uncertainty, as well as the coupling characteristics between external interference and communication delay, through the balanced adversarial network, and output the interference amount and the delay amount.

[0013] The time delay interference compensation unit is used to synchronously compensate for and suppress time delay and interference through a multi-source compensation and suppression framework.

[0014] As another preferred embodiment of the present invention, the multi-source compensation and suppression framework includes a time delay observer, a time delay compensator, and an interference observer; the time delay observer is used to learn the multi-source time delays online through an equalized adversarial network to identify flight actuator time delays, deformable actuator time delays, and communication time delays; the time delay compensator is used to compensate for the multi-source time delays by recursively calculating k steps through the equalized adversarial network; the interference observer is used to learn the interference online through the equalized adversarial network and output the interference value.

[0015] As another preferred embodiment of the present invention, the biomimetic variable structure module of the multi-variable aircraft learns the control decision mechanism of the goose when changing its wingspan through an equal adversarial network. The input is environmental information and goose configuration parameters, and the output is the deformation decision mechanism of the multi-variable aircraft, so as to achieve precise control of the wingspan of the multi-variable aircraft.

[0016] As another preferred embodiment of the present invention, the attitude coordination module of the multi-configuration aircraft adopts a hybrid driving modeling method that combines mechanism modeling and data-driven modeling to perform integrated modeling of the multi-configuration aircraft; wherein, the data-driven modeling describes the uncertainty of the model through an equilibrium adversarial network, the input of the data-driven model is the angular velocity state value, and the output is the inertia matrix uncertainty value.

[0017] As another preferred embodiment of the present invention, the mechanism modeling adopts the following state vector representation of the first... i The status of each aircraft:

[0018] ,

[0019] ;

[0020] in, V i For flight speed, c i The inclination angle of the flight path. oh i The vector of the body's angular velocity. l i These are configuration parameters; T i Powered by the engine, Let α be the rate of change of velocity. i For the angle of attack, D i For aerodynamic drag, m ( l ( ) represents the mass of the variable-configuration aircraft, and g represents the acceleration due to gravity. The rate of change of the track angle. L i For aerodynamic lift, J i ( l i ) is the variable configuration inertia matrix. t aero，i For aerodynamic torque, t morph，i Additional torque for the variable configuration, Δ J i Due to the uncertainty of the inertia matrix, d ω，i To control surface jitter interference, m λ,i For deformation control input, k λ The natural decay coefficient, t λ,i For the time delay of the deformation actuator, t α,i For the attitude control actuator time delay, t c,ij For from the first j Machine to the i Communication delay of the machine.

[0021] As another preferred embodiment of the present invention, the multi-configuration aircraft attitude coordination module establishes an equilibrium-adversarial composite controller through an equilibrium-adversarial network, and adopts a composite anti-interference control strategy to achieve attitude coordination control; the specific expression of the equilibrium-adversarial composite controller is as follows:

[0022] ,

[0023] ,

[0024] ,

[0025] ,

[0026] ;

[0027] This includes two constraints, the first constraint... For: Deformation rate constraint, the second constraint condition For: Time-delay safety constraints; For the first k The predicted angular velocity of the step. oh ref For reference angular velocity, l opt For optimal configuration parameters, For the maximum deformation rate, N p To predict the time-domain step size, Q This is the angular velocity error weighting matrix; For the first k The predicted configuration parameters are used to establish an equal adversarial network. k The output of the step prediction model; w This is the configuration error weight matrix. t λ,i For the time delay of the deformable actuator, △ l safe For safety deformation margin; The control surface jitter interference observation is represented by the interference observer output value.

[0028] As another preferred embodiment of the present invention, the balanced adversarial network includes an input layer, an intermediate layer, and an output layer;

[0029] The intermediate layer includes:

[0030] Feature extractors are used to extract shared and private features;

[0031] A competitive adversarial device is used to select the winning neuron through a competitive mechanism;

[0032] A policy generator is used to generate a final policy based on shared and private features.

[0033] As another preferred embodiment of the present invention, the learning algorithm of the balanced adversarial network adopts any one of the following: gradient optimization method, meta-learning algorithm, recursive least squares algorithm, or wake-up-sleep algorithm.

[0034] Furthermore, the biomimetic learning-based attitude cooperative control method for multi-configuration aircraft provided by this invention, implemented using the aforementioned biomimetic learning-based attitude cooperative control system for multi-configuration aircraft, includes the following steps:

[0035] Step 1: Learn and separate the coupling mechanism of time delay and interference through a balanced adversarial network, and realize compensation for time delay and interference;

[0036] Step 2: Learn the control decision-making mechanism of the goose's variable wingspan through an equilibrium adversarial network to achieve biomimetic control of the wingspan of a variable-structure aircraft;

[0037] Step 3: Use a hybrid-driven modeling method that combines mechanistic modeling and data-driven modeling to perform integrated modeling of the multi-configuration aircraft;

[0038] Step 4: Construct a composite controller through an equal adversarial network to achieve attitude cooperative control of the multi-configuration aircraft.

[0039] Beneficial effects of this invention:

[0040] This invention provides a biomimetic learning-based attitude cooperative control system and method for multi-configuration aircraft. By combining the aforementioned time-delay interference decoupling module, the biomimetic variant module, and the attitude cooperative module, the multi-configuration aircraft possesses strong autonomy, adaptability, and survivability—bionic intelligent behavioral capabilities—effectively improving the intelligence, safety, and reliability of the multi-configuration aircraft. This invention provides a robust modeling and control scheme for attitude cooperative control of multi-configuration aircraft, enhancing their mission adaptability, flight safety, and energy economy, thus contributing to the development of important planning fields such as ecological environment, modern agriculture, and smart cities. Attached Figure Description

[0041] Figure 1 This diagram shows the mapping relationship between the input layer, intermediate layer, and output layer of the balanced adversarial network of the biomimetic learning-based multi-configuration aircraft attitude cooperative control system and method of the present invention.

[0042] In the diagram, 1 represents the input layer, 2 the intermediate layer, and 3 the output layer; 201 is the feature extractor, 202 the adversarial processor, and 203 the policy generator. Detailed Implementation

[0043] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0044] Inspired by flying creatures in nature, such as geese, which can fly in formation with alternating wings, flexibly changing their wingspan and other parts according to the mission, environment, and their own state to achieve optimized flight capabilities at high and low altitudes and speeds, this invention abstracts the goose into a multi-configuration aircraft for research, thereby overcoming the application limitations of fixed-wing monoplane and multiplane unmanned aerial vehicles (UAVs). Therefore, by implementing the biomimetic learning-based multi-configuration aircraft attitude cooperative control system and method involved in this invention, cooperative control of the multi-configuration aircraft's attitude can be achieved, improving the reliability, safety, and environmental adaptability of the multi-configuration aircraft's operation.

[0045] Combination Figure 1 As shown in the figure, an embodiment of the present invention provides a biomimetic learning-based attitude cooperative control system for a multi-configuration aircraft, including a time-delay interference decoupling module, a biomimetic configuration module for the multi-configuration aircraft, and a attitude cooperative module for the multi-configuration aircraft. The time-delay interference decoupling module is used to learn and separate the time-delay and interference coupling mechanism of the multi-configuration aircraft and to compensate for the time-delay and interference. The biomimetic configuration module for the multi-configuration aircraft is used to achieve configuration control of the wingspan of the multi-configuration aircraft through biomimetic learning. The attitude cooperative module for the multi-configuration aircraft is used to perform integrated modeling and attitude cooperative control of the multi-configuration aircraft. The time-delay interference decoupling module, the biomimetic configuration module, and the attitude cooperative module for the multi-configuration aircraft are all implemented using a balanced adversarial network. This endows the multi-configuration aircraft with strong autonomy, strong adaptability, and strong survivability, effectively improving the intelligence, safety, and reliability of the multi-configuration aircraft.

[0046] Specifically, the time-delay interference decoupling module has functions for learning and separating time-delay interference coupling mechanisms and compensating for time-delay interference. The module includes a coupling mechanism learning and separation unit and a time-delay interference compensation unit. The coupling mechanism learning and separation unit is used to learn the coupling characteristics between the deformation actuator time delay and model uncertainty, and the coupling characteristics between external interference and communication time delay, respectively, through an equalized adversarial network, and outputs the interference amount and time delay amount. At this time, the input to the equalized adversarial network is the time-delay interference coupling information. The time-delay interference compensation unit is used to synchronously compensate for and suppress time delay and interference through a multi-source compensation and suppression framework.

[0047] The multi-source compensation and suppression framework includes a time-delay observer, a time-delay compensator, and an interference observer. The time-delay observer is used to learn multi-source time delays online through an equalized adversarial network, identifying flight actuator time delays, deformable actuator time delays, and communication time delays. The time-delay compensator is used to compensate for multi-source time delays by recursively calculating k steps through the equalized adversarial network. The interference observer is used to learn interference online through the equalized adversarial network and output interference values. Specifically, firstly, a time-delay observer is established using an equalized adversarial network to learn time delays online. Based on the time delay attributes, the sources of time delays in the multi-variable aircraft are analyzed. The multi-source time delays include flight actuator time delays, deformable actuator time delays, and communication time delays. At this point, the input to the time-delay observer is the multi-source time delay, and the output is the flight actuator time delay, deformable actuator time delay, and communication time delay. Then, a recursive k-step time delay compensator is established through a balanced adversarial network to perform refined compensation for multi-source time delays. The input to the time delay compensator is the state value of the multi-configuration aircraft and the time delays of the first k-1 steps, with the output being the time delay at the k-th step. Finally, an interference observer is established using the balanced adversarial network to learn the interference online. The input to the interference observer is the state value of the multi-configuration aircraft, and the output is the interference value.

[0048] Specifically, the biomimetic morphological module of the multi-morphic aircraft can realize the function of learning and controlling the wingspan of a goose; the biomimetic morphological module of the multi-morphic aircraft learns the control decision mechanism of a goose when changing its wingspan through an equilibrium adversarial network, with environmental information and goose configuration parameters as inputs, and the deformation decision mechanism of the multi-morphic aircraft as output. This is to achieve precise control over the wingspan of variable-structure aircraft.

[0049] Specifically, the attitude coordination module for the multi-configuration aircraft has integrated modeling and attitude coordination control functions for the multi-configuration aircraft. The integrated modeling function of the attitude coordination module for the multi-configuration aircraft employs a hybrid driving modeling method that combines mechanistic modeling and data-driven modeling to perform integrated modeling of the multi-configuration aircraft. The data-driven modeling describes the uncertainty of the model through an equalized adversarial network. The input of the data-driven model is the angular velocity state value, and the output is the uncertainty value of the inertia matrix.

[0050] Specifically, there are N variable-structure aircraft, and the mechanism modeling is represented by the following state vector. i The status of a variant aircraft:

[0051] ,

[0052] ;

[0053] in, V i For flight speed, ci The inclination angle of the flight path. oh i The vector of the body's angular velocity. l i These are configuration parameters; T i Powered by the engine, Let α be the rate of change of velocity. i For the angle of attack, D i For aerodynamic drag, m ( l ( ) represents the mass of the variable-configuration aircraft, and g represents the acceleration due to gravity. The rate of change of the track angle. L i For aerodynamic lift, J i ( l i ) is the variable configuration inertia matrix. t aero，i For aerodynamic torque, t morph，i Additional torque for the variable configuration, Δ J i Due to the uncertainty of the inertia matrix, d ω，i To control surface jitter interference, m λ,i For deformation control input, k λ The natural decay coefficient, t λ,i For the time delay of the deformation actuator, t α,i For the attitude control actuator time delay, t c,ij For from the first j Machine to the i Communication delay of the machine;

[0054] The additional torque caused by the configuration change is: ;

[0055] The variable mass inertia matrix is: ;

[0056] Variational parameter vector: ;

[0057] in, l 1 The wingspan scaling factor, l 2 This is the sweep angle variation coefficient. l 3 This is the coefficient for wing area variation.

[0058] Specifically, the attitude coordination module of the multi-configuration aircraft establishes a balanced adversarial composite controller through a balanced adversarial network, and adopts a composite anti-interference control strategy to achieve attitude coordination control; the specific expression of the balanced adversarial composite controller is as follows:

[0059] ,

[0060] ,

[0061] ,

[0062] ,

[0063] ;

[0064] This includes two constraints, the first constraint... For: Deformation rate constraint, the second constraint condition For: Time-delay safety constraints; For the first k The predicted angular velocity of the step. oh ref For reference angular velocity, l opt For optimal configuration parameters, For the maximum deformation rate, N p To predict the time-domain step size, Q This is the angular velocity error weighting matrix; For the first k The predicted configuration parameters are used to establish an equal adversarial network. k The output of the step prediction model; w This is the configuration error weight matrix. t λ,i For the time delay of the deformable actuator, △ l safe For safety deformation margin; The control surface jitter interference observation is represented by the interference observer output value.

[0065] Specifically, the equilibrium adversarial network mimics the brain's decision-making mechanism, completing the final feature classification and description through game-like competition between neurons; such as Figure 1As shown, the balanced adversarial network includes an input layer 1, an intermediate layer 2, and an output layer 3. The intermediate layer 2 includes: a feature extractor 201 for extracting shared and private features; a competitive adversarial network 202 for selecting the winning neuron through a competitive mechanism; and a policy generator 203 for generating the final policy based on the shared and private features. The output of the feature extractor 201 serves as the input to the competitive adversarial network, and the output of the competitive adversarial network 202 serves as the input to the policy generator 203. The policy generator 203 is fully connected, as detailed below:

[0066] (1) The feature extractor 201 includes shared features f s With private characteristics The specific expression is as follows:

[0067] ;

[0068] in, s In a state of confrontation, W s This is the weight matrix. b s ReLU is the bias, and ReLU is the activation function.

[0069] ;

[0070] in, s This is the current state. This represents the action at the previous moment. LSTM stands for Long Short-Term Memory Network.

[0071] (2) The competitive adversarial device 202: When an input signal is received, the neurons in the competitive layer compare the similarity between the input vector and their own weights, and finally only one neuron wins (with the highest activation value), representing the feature or category of the input; specifically as follows:

[0072] The similarity between a neuron and its input is typically measured using Euclidean distance. x With neuron weights w j The similarity is calculated using the following formula: Select the winning neuron c , c The expression is: .

[0073] (3) The policy generator 203 is represented as follows:

[0074] ;

[0075] in, Shared features fs (Global state information) and private features (Competitors' personalized information) is concatenated into a joint feature vector; For the competitor's unique weight matrix, is the corresponding bias term; softmax is the normalization.

[0076] Specifically, the learning algorithm of the balanced adversarial network adopts any one of the following: gradient optimization method, meta-learning algorithm, recursive least squares algorithm, or wake-up-sleep algorithm; the learning algorithm can also adopt other algorithms that can satisfy the optimization function, and is not limited to the above learning algorithms.

[0077] Furthermore, the biomimetic learning-based attitude cooperative control method for multi-configuration aircraft provided by this invention, implemented using the aforementioned biomimetic learning-based attitude cooperative control system for multi-configuration aircraft, includes the following steps:

[0078] Step 1: Learn and separate the coupling mechanism of time delay and interference through a balanced adversarial network, and realize compensation for time delay and interference;

[0079] Step 2: Learn the control decision-making mechanism of the goose's variable wingspan through an equilibrium adversarial network to achieve biomimetic control of the wingspan of a variable-structure aircraft;

[0080] Step 3: Use a hybrid-driven modeling method that combines mechanistic modeling and data-driven modeling to perform integrated modeling of the multi-configuration aircraft;

[0081] Step 4: Construct a composite controller through an equal adversarial network to achieve attitude cooperative control of the multi-configuration aircraft.

[0082] In summary, the following key advantages of this invention can be identified through the embodiments provided by this invention:

[0083] (1) This invention integrates bionic technology, anti-interference technology, automatic control technology and artificial intelligence technology to design a bionic learning-based attitude cooperative control system and method for multi-configuration aircraft. Through time-delay interference coupling mechanism learning and separation, time-delay interference compensation function, and goose-wingspan learning and control function; integrated modeling of multi-configuration aircraft and attitude cooperative control function of multi-configuration aircraft, it endows multi-configuration aircraft with strong autonomy, strong adaptability and strong survival and other bionic intelligent behavior capabilities, effectively improving the intelligence, safety and reliability of multi-configuration aircraft.

[0084] (2) This invention integrates bionic technology, automatic control technology and artificial intelligence technology, and uses an equal adversarial network as a coupling mechanism learning model to learn and separate the time delay interference coupling mechanism of multi-configuration aircraft, which can effectively and accurately separate time delay and interference.

[0085] (3) This invention integrates bionic technology, automatic control technology and artificial intelligence technology, and uses a balanced adversarial network to construct a time delay observer, a time delay compensator and an interference observer, which can estimate and compensate for the time delay and interference of multi-configuration aircraft.

[0086] (4) This invention integrates bionic technology, automatic control technology and artificial intelligence technology. By learning the configuration parameters of geese changing their wingspan under different environments through balanced adversarial networks, it can accurately obtain their control decision mechanism and thus effectively control the wingspan of multi-configuration aircraft.

[0087] (5) This invention integrates bionic technology, automatic control technology and artificial intelligence technology, and adopts a combination of mechanism modeling and data-driven modeling to construct an integrated model of multi-configuration aircraft. The data-driven modeling uses an equilibrium adversarial network to characterize the uncertainty of multi-configuration aircraft, which can effectively improve the modeling accuracy.

[0088] (6) This invention integrates bionic technology, automatic control technology and artificial intelligence technology, establishes a balanced adversarial composite controller through a balanced adversarial network, and adopts a composite anti-interference control strategy to control the multi-configuration aircraft, effectively improving control accuracy and reliability.

[0089] (7) This invention integrates artificial intelligence technology to establish a balanced adversarial network, including an input layer, an intermediate layer and an output layer. The intermediate layer includes a feature extractor, a competitive adversarial unit and a policy generator. It has strong nonlinear approximation, fast convergence and strong generalization ability. It has "self-learning ability", robustness and anti-interference performance. It can be well used as an approximation model and a classification model.

[0090] In summary, this invention combines biomimetic technology, anti-interference technology, automatic control technology, and artificial intelligence technology to provide a biomimetic learning-based attitude collaborative control system and method for multi-configuration aircraft. This system enables multi-configuration aircraft to possess biomimetic intelligent behaviors such as strong autonomy, strong adaptability, and strong survivability, thus improving the intelligence, safety, and reliability of multi-configuration aircraft.

[0091] It is understood that the above specific description of the present invention is only for illustrating the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention to achieve the same technical effect; as long as the use needs are met, they are all within the protection scope of the present invention.

Claims

1. A biomimetic learning-based attitude cooperative control system for a variable-structure aircraft, characterized in that: This includes a time-delay interference decoupling module, a biomimetic variant module for multi-variable aircraft, and an attitude coordination module for multi-variable aircraft. The time-delay interference decoupling module is used to learn and separate the time-delay and interference coupling mechanism of the multi-configuration aircraft, and to compensate for the time delay and interference. The time-delay interference decoupling module includes a coupling mechanism learning and separation unit and a time-delay interference compensation unit. The coupling mechanism learning and separation unit is used to learn the coupling characteristics between the deformation actuator delay and model uncertainty, and the coupling characteristics between external interference and communication delay, respectively, through the balanced adversarial network, and output the interference amount and the delay amount. The time delay interference compensation unit is used to synchronously compensate and suppress time delay and interference through a multi-source compensation and suppression framework. The biomimetic variable structure module of the variable structure aircraft is used to achieve configuration control of the wingspan of the variable structure aircraft through biomimetic learning. The variable-structure aircraft attitude coordination module is used for integrated modeling and attitude coordination control of the variable-structure aircraft. The time-delay interference decoupling module, the biomimetic variant module of the multi-variable aircraft, and the attitude coordination module of the multi-variable aircraft are all implemented using a balanced adversarial network.

2. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 1, characterized in that: The multi-source compensation and suppression framework includes a time-delay observer, a time-delay compensator, and an interference observer; The time delay observer is used to learn multi-source time delays online through a balanced adversarial network to identify flight actuator time delays, deformable actuator time delays, and communication time delays. The time delay compensator is used to compensate for multi-source time delays by recursively calculating k steps through a balanced adversarial network. The interference observer is used to learn the interference online through a balanced adversarial network and output the interference value.

3. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 1, characterized in that: The biomimetic variant module of the multi-variable aircraft learns the control decision mechanism of a goose when changing its wingspan through an equal adversarial network. The input is environmental information and goose configuration parameters, and the output is the deformation decision mechanism of the multi-variable aircraft, so as to achieve precise control of the wingspan of the multi-variable aircraft.

4. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 2, characterized in that: The attitude coordination module of the multi-modal aircraft adopts a hybrid driving modeling method that combines mechanism modeling and data-driven modeling to perform integrated modeling of the multi-modal aircraft. The data-driven modeling describes the uncertainty of the model through an equalized adversarial network. The input of the data-driven model is the angular velocity state value, and the output is the uncertainty value of the inertia matrix.

5. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 4, characterized in that: The mechanism modeling uses the following state vector to represent the first... i The status of a variant aircraft: ， ； in, V i For flight speed, γ i The inclination angle of the flight path. ω i The vector of the body's angular velocity. λ i These are configuration parameters; T i Powered by the engine, Let α be the rate of change of velocity. i For the angle of attack, D i For aerodynamic drag, m ( λ ( ) represents the mass of the variable-configuration aircraft, and g represents the acceleration due to gravity. The rate of change of the track angle. L i For aerodynamic lift, J i ( λ i ) is the variable configuration inertia matrix. τ aero，i For aerodynamic torque, τ morph，i Additional torque for the variable configuration, Δ J i Due to the uncertainty of the inertia matrix, d ω，i To control surface jitter interference, μ λ,i For deformation control input, k λ The natural decay coefficient, τ λ,i For the time delay of the deformation actuator, τ α,i For the attitude control actuator time delay, τ c,ij For from the first j Machine to the i Communication delay of the machine.

6. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 2, characterized in that: The attitude coordination module of the multi-configuration aircraft establishes a balanced adversarial composite controller through a balanced adversarial network, and adopts a composite anti-interference control strategy to achieve attitude coordination control; the specific expression of the balanced adversarial composite controller is as follows. ， ， ， ， ； This includes two constraints, the first constraint... For: Deformation rate constraint, the second constraint condition For: Time-delay safety constraints; For the first k The predicted angular velocity of the step. ω ref For reference angular velocity, λ opt For optimal configuration parameters, For the maximum deformation rate, N p To predict the time-domain step size, Q This is the angular velocity error weighting matrix; For the first k The predicted configuration parameters are used to establish an equal adversarial network. k The output of the step prediction model; w This is the configuration error weight matrix. τ λ,i For the time delay of the deformable actuator, △ λ safe For safety deformation margin; The control surface jitter interference observation is represented by the interference observer output value.

7. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 1, characterized in that: The balanced adversarial network includes an input layer, an intermediate layer, and an output layer; The intermediate layer includes: Feature extractors are used to extract shared and private features; A competitive adversarial device is used to select the winning neuron through a competitive mechanism; A policy generator is used to generate a final policy based on shared and private features.

8. The biomimetic learning-based multi-configuration aircraft attitude cooperative control system according to claim 1, characterized in that: The learning algorithm of the balanced adversarial network adopts any one of the following: gradient optimization method, meta-learning algorithm, recursive least squares algorithm, or wake-up-sleep algorithm.

9. A biomimetic learning-based attitude cooperative control method for multi-configuration aircraft, characterized by: The attitude cooperative control system for a multi-configuration aircraft based on biomimetic learning, as described in any one of claims 1 to 8, includes the following steps: Step 1: Learn and separate the coupling mechanism of time delay and interference through a balanced adversarial network, and realize compensation for time delay and interference; Step 2: Learn the control decision-making mechanism of the goose's variable wingspan through an equilibrium adversarial network to achieve biomimetic control of the wingspan of a variable-structure aircraft; Step 3: Use a hybrid-driven modeling method that combines mechanistic modeling and data-driven modeling to perform integrated modeling of the multi-configuration aircraft; Step 4: Construct a composite controller through an equal adversarial network to achieve attitude cooperative control of the multi-configuration aircraft.

Images