Predefined time formation control method based on fuzzy neural network

Abstract

The invention relates to the technical field of formation control of an air-sea heterogeneous cluster system, and discloses a predefined time formation control method based on a fuzzy neural network, and the method comprises the steps: building kinematics and dynamics models of a six-degree-of-freedom four-rotor unmanned plane and a three-degree-of-freedom unmanned ship; establishing a formation tracking error and a speed tracking error of the air-sea heterogeneous cluster system; constructing a preset performance function, and determining an error conversion function; determining a predefined time virtual control law; establishing a predefined time fuzzy neural network, and determining a distributed predefined time formation controller of the air-sea heterogeneous cluster system in combination with a predefined time virtual control law; and performing distributed predefined time error constraint formation control on the air-sea heterogeneous cluster system according to a distributed predefined time formation controller of the air-sea heterogeneous cluster system. According to the method, the robustness of the air-sea heterogeneous cluster is improved, so that the system can converge within the settable predefined time and does not depend on the initial value of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of air-sea heterogeneous cluster system formation control, and in particular to a predefined time formation control method based on fuzzy neural network. Background Art

[0002] The ocean occupies more than 70% of the earth and is rich in minerals and underwater biological resources. In the 21st century, with the increasing importance of marine resources and the growing demand for development, how to efficiently and safely explore and monitor marine resources has become the focus of global attention. The air-sea heterogeneous cluster system, especially the formation tracking control method of quad-rotor drones and unmanned boats, provides innovative solutions in this field. Quad-rotor drones have the characteristics of strong maneuverability and flexible deployment, and can quickly reach the target area for aerial reconnaissance and data collection. The unmanned boat can cruise on the water surface for a long time to conduct underwater detection and environmental monitoring. Through the formation tracking control technology, drones and unmanned boats work together according to the preset paths and mission requirements to achieve comprehensive monitoring of marine resources. This system can realize collaborative operations between the air and the surface, greatly improving the coverage of marine resource exploration and the accuracy of data collection. In the development of marine resources, the formation tracking control method of the air-sea heterogeneous cluster system is particularly important. It can not only improve the efficiency of resource exploration and reduce manpower and material costs, but also provide a safer and more environmentally friendly way of operation in complex marine environments, such as severe weather or sensitive ecological areas. Therefore, researching and developing formation tracking control methods for air-sea heterogeneous swarm systems is of great strategic significance for promoting the sustainable development of marine resources.

[0003] However, the research on formation tracking control methods of quadrotor drones and unmanned boats faces a series of technical challenges. The dynamic and unpredictable nature of the marine environment places high demands on the stability of unmanned systems. Factors such as ocean currents, wind speeds, and weather changes will affect the navigation trajectory of unmanned boats, while drones in the air need to cope with changes in wind and airflow. The formation control algorithm must have a certain degree of adaptive ability to adapt to these external disturbances. At the same time, in many application scenarios such as marine resource exploration, environmental monitoring, or emergency rescue, the rapid response and timely execution of the system are the key to completing marine missions. The settable stable time control algorithm can ensure the immediacy and time certainty of the system. Therefore, it is of great research value and significance to design a reasonable formation controller to ensure that the quadrotor drones and unmanned boats of the air-sea heterogeneous system can smoothly and reliably perform trajectory tracking tasks in a complex marine environment.

[0004] In the field of multi-agent control systems, technologies such as PID control, sliding mode control and adaptive control have been widely adopted. PID control is known for its low requirements on system models and reliable performance in a variety of application scenarios. However, its limitations in anti-interference and rapid response cannot be ignored, and the parameter adjustment process is often cumbersome. Sliding mode control technology is highly praised for its excellent rapid response and robustness to system parameter fluctuations and external disturbances. However, its chattering phenomenon near the sliding surface may have an adverse effect on the stability and performance of the system. The adaptive control strategy updates the control parameters in real time to adapt to the dynamic changes of the system and external disturbances, but this method requires too many parameter updates and increases the computational complexity. In recent years, for the control problems faced by marine unmanned systems due to the uncertainty and complexity of the marine environment, the common method is usually to design a state observer or use various neural networks to approximate, and then design a control law so that the system can achieve trajectory tracking control and ensure the global convergence of the error. However, the observation effect of the observer depends on the observation gain; the complexity of neural network operations and the limited online computing speed lead to some inevitable problems, such as a large number of parameter updates, heavy system computing burden, possible computational explosion problems, and inability to guarantee that the system converges within a limited time.

[0005] At present, in the field of multi-agent tracking control, different types of obstacle functions and preset performance functions are often combined with backstepping methods to design controllers with asymptotic convergence, finite time convergence, and fixed time convergence to limit the system error and achieve better control accuracy. However, the convergence speed of the asymptotically convergent controller is slow, and the convergence speed of the finite time convergent controller is affected by the initial state of the system. It is worth mentioning that the control technology of predefined time convergence has begun to be used in recent years, which can deal with the above problems well. In the field of formation control, the application of error constraints has brought significant advantages to the system. Error constraints allow the formation control algorithm to consider the dynamic characteristics of each agent, so as to design a control strategy with specified performance. Such constraints include but are not limited to position errors and velocity errors, etc., to ensure that the formation members do not exceed their performance boundaries while maintaining the formation, and to ensure the stability of the system. Although there have been many studies on error-constrained control, no corresponding research results have been obtained for the error-constrained air-sea heterogeneous system quadcopter drone and unmanned boat formation control. At the same time, the relevant error-constrained formation control research does not involve the problem of predefined time stability, which makes its control performance have certain limitations.

[0006] In dealing with disturbances in the marine environment, fuzzy neural networks have shown a series of advantages over traditional neural networks, making them an ideal choice for solving such problems. Fuzzy neural networks can handle uncertainty and ambiguity in formation control by integrating fuzzy logic. In the marine environment, unmanned boats may be affected by unpredictable currents and waves, while drones may encounter unpredictable winds. Fuzzy neural networks can process fuzzy data and map them into the network, that is, they can convert these fuzzy environmental variables into precise control instructions, thereby improving the stability and adaptability of the formation. In addition, the advantages of fuzzy neural networks in generalization ability cannot be ignored. In practical applications, the formation may encounter new situations that were not encountered in the training phase. Fuzzy neural networks can use their generalization ability to learn and infer effective control strategies from existing data, thereby maintaining the coordination of the formation in unknown situations. In terms of computational efficiency, fuzzy neural networks usually have lower model complexity, which makes them more advantageous in resource-constrained unmanned systems. Compared with traditional neural networks, fuzzy neural networks require less computing resources, which is crucial to reducing the computational burden of drones and unmanned boats. However, such networks usually lack precise control over convergence time, which can be a significant disadvantage in applications that require fast response. If the fuzzy neural network cannot guarantee to reach a stable state within a specific time, it may reduce the performance and safety of the system. In complex and changing marine environments, this uncertainty may lead to delayed execution of critical tasks and increase the overall risk of the system. Therefore, it is particularly important to explore a technology that can set the convergence time of fuzzy neural networks. Summary of the invention

[0007] In view of the above-mentioned deficiencies in the prior art, the present invention provides a predefined time formation control method based on fuzzy neural network, so that the tracking error of the system is guaranteed to converge within the predefined time and is not affected by the initial state of the system. The combination of fuzzy neural network improves the stability and robustness of the system in complex marine environment, so as to solve the problems of limited control accuracy, external environmental disturbance and defined time stability when applying the current control method of unmanned aerial vehicle and unmanned boat air-sea heterogeneous cluster system.

[0008] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0009] A predefined time formation control method based on fuzzy neural network comprises the following steps:

[0010] Establish kinematic and dynamic models of six-degree-of-freedom quad-rotor UAV and three-degree-of-freedom unmanned boat;

[0011] According to the kinematic and dynamic models of the six-degree-of-freedom quad-rotor UAV and the three-degree-of-freedom unmanned boat, the formation tracking error and velocity tracking error of the air-sea heterogeneous swarm system are established;

[0012] Construct a preset performance function and determine the error conversion function based on the formation tracking error and speed tracking error of the air-sea heterogeneous cluster system;

[0013] Determine a predefined time virtual control law according to a formation tracking error, a speed tracking error and an error conversion function;

[0014] A predefined-time fuzzy neural network is established, and the distributed predefined-time formation controller of the air-sea heterogeneous cluster system is determined in combination with the predefined-time virtual control law;

[0015] The distributed predefined time error constraint formation control of the air-sea heterogeneous cluster system is performed according to the distributed predefined time formation controller of the air-sea heterogeneous cluster system.

[0016] As a preferred embodiment, the kinematic and dynamic models of the six-degree-of-freedom quad-rotor drone are specifically as follows:

[0017]

[0018] Among them, χ ph =[x ph ,y ph ,z ph ] T Represents the position information in the global coordinate system, χ ah =[φ ah ,θ ah ,ψ ah ] T Indicates the roll angle, pitch angle and yaw angle; d xph ,d yph ,d zph ,d φah ,d θah ,d ψah represents the time-varying unknown ocean disturbance function; I x ,I y ,I z represents the moment of inertia; ρ x ,ρ y ,ρ z ,ρ φ ,ρ θ ,ρ ψ Represents the aerodynamic damping coefficient; m a represents mass and inertia parameters; u 1h ,u φh ,u θh ,u ψhrepresents the controller to be designed; g represents the acceleration of gravity; Λ represents the total residual rotor angle; I r Represents the rotor's moment of inertia.

[0019] As a preferred embodiment, the kinematic and dynamic models of the three-degree-of-freedom unmanned boat are specifically as follows:

[0020]

[0021] Among them, χ sk,1 =[χ xsk,1 ,x ysk,1 ,x ψsk,1 ] T represents the position variable and heading angle variable of the kth unmanned boat in the global coordinate system; v k =[v xsk ,v ysk ,v ψsk ] T represents the velocity variable and angular velocity variable of the kth unmanned boat in the fixed coordinates of the body; u sk =[u xsk ,u ysk ,u ψsk ] T represents the control input quantity to be designed for the kth unmanned boat; d sk (t) = [d xk ,d yk ,d ψk ] T represents the external disturbance of the kth unmanned boat; R sk is the rotation matrix; represents the inertia matrix; represents the damping matrix, represents the Coriolis matrix.

[0022] As a preferred embodiment, the formation tracking error of the air-sea heterogeneous cluster system is specifically:

[0023] e i,j =Ξ i (x i,j -β i,j )+E e,i ,i=ph,ah,sk;

[0024]

[0025] Among them, e i,j represents the formation tracking error, Ξ irepresents the formation communication parameters, ph, ah represent the position subsystem and attitude subsystem of the h-th UAV, sk represents the k-th unmanned boat system, h=1,...,N represents the h-th UAV, N represents the number of UAVs in the heterogeneous multi-agent system, k=1,...,M represents the k-th unmanned boat, M represents the number of unmanned boats in the heterogeneous multi-agent system, N+M represents the number of all agents in the heterogeneous multi-agent system, diag represents the diagonal matrix, a h,i represents the communication parameters between the h-th UAV and other intelligent agents, b lh represents the communication parameters between the hth UAV and the leader, a k+N,i represents the communication parameters between the kth unmanned boat and other intelligent agents, I 3×3 represents a 3D unit vector, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ i,j represents the state vector of the agent, β i,j represents the relative state vector that the agent needs to maintain with the leader, E ei,j It represents the state vector that an agent receives from other agents and is defined as follows.

[0026]

[0027] Among them, χ pi,j represents the position state vector of the h-th UAV, β ψai,j represents the relative heading angle state vector that the h-th UAV needs to maintain with the leader, β xpi,j ,β yai,j represents the relative position state vector that the h-th UAV needs to maintain with the leader, E eph,j represents the angle error vector received by the drone from other agents, E esk,j represents the error vector received by the unmanned boat from other intelligent agents, E eph,j represents the position error vector received by the drone from other agents, E eah,j represents the attitude error vector received by the drone from other agents, a k+N,i and a k+N,i+N represents the communication parameters between the kth unmanned boat and other intelligent agents, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ xL,j ,x yL,j ,x zL,j ,x ψL,j represents the state vector of the leader, χ φd,hj and χ θd,hj represents the desired roll angle and pitch angle of the h-th UAV, β pi,j represents the relative position vector that the h-th UAV needs to maintain with the leader, χ xpi,j ,xypi,j represents the position state vector of the hth UAV, χ ψai,j represents the heading angle of the hth UAV, χ xsi,j ,x ysi,j ,x ψsi,j represents the state vector of k unmanned boats, β xsi,j ,β ysi,j ,β ψsi,j Represents the relative state vector that the k unmanned boats need to maintain with the leader.

[0028] As a preferred embodiment, the formation tracking error and speed tracking error of the air-sea heterogeneous cluster system are specifically:

[0029]

[0030] e φah,j = x φah,j -x φd,hj ,e θah,j = x θah,j -x θd,hj ;

[0031]

[0032] Among them, e gph,j and e gsk,j They represent the state errors of the UAV and the unmanned boat on the two-dimensional plane, e ψah,j and e ψsk,j They represent the state errors of the heading angles of the UAV and the unmanned boat, e zh,j Represents the height error, e φah,j and e θah,j Denote the roll angle and pitch angle errors respectively, χ xL,j ,x yL,j ,x ψL,j ,x zL,j represents the trajectory information of the virtual leader, β gph,j ,β gpi,j and β gsk,j ,β gsi,j They represent the relative positions of the UAV and the UAV with respect to the leader, β ψah,j ,β ψai,j and β ψsk,j ,β ψsi,j They represent the relative heading angle state vectors that the UAV and the unmanned boat need to maintain with the leader, β zph,j ,β zpi,j represents the relative altitude state vector that the drone needs to maintain with the leader, a h,i and a h,i+N represents the communication parameters between the h-th UAV and other intelligent agents, a k+N,i and a k+N,i+Nrepresents the communication parameters between the kth unmanned boat and other intelligent agents, b lh represents the communication parameters between the hth UAV and the leader, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ gph,j ,x gpi,j and χ gsk,j ,x gsi,j Respectively represent the position status of the UAV and the unmanned boat, χ ψah,j ,x ψai,j and χ ψsk,j ,x ψsi,j Represent the heading angle state vectors of the UAV and the unmanned boat, respectively, zph,j ,x zpi,j Indicates the altitude status of the drone, χ φah,j and χ θah,j represents the roll angle and pitch angle state vector of the hth unmanned boat, respectively, φd,hj and χ θd,hj represents the desired roll angle and pitch angle of the hth unmanned boat, a ph and a ah represents the virtual control law vector of the UAV, a sk represents the virtual control law vector of the unmanned boat, χ ph,2 and χ ah,2 represents the velocity state vector of the UAV, χ sk,2 represents the velocity state vector of the unmanned boat, E ph and E ah represents the error vector between the UAV speed and the virtual control law, E sk represents the error vector between the speed of the unmanned boat and the virtual control law, E xph ,E yph ,E yph It represents the error between the speed of the UAV in the XYZ direction and the virtual control law, E φah ,E θah ,E ψah represents the error of the virtual control law of the angular velocity and angle of the UAV, E xsk ,E ysk ,E ψsk Represents the error between the speed of the unmanned boat and the virtual control law.

[0033] Preferably, the error conversion function is:

[0034]

[0035] Among them, Z ph,1 ,Z ah,1 and Z ph,2 ,Z ah,2 Represents the UAV formation error conversion function vector, Z sk,1 and Zsk,2 represents the error conversion function vector of the unmanned boat formation, Q ph ,Q ah ,Q sk The performance function index representing the error of the UAV and UAV formation, e xph,1 ,e yph,1 ,e zph,1 represents the formation error of the UAV in the XYZ direction, e φah,1 ,e θah,1 ,e ψah,1 represents the formation error of the UAV’s attitude system, e xsk,1 ,e ysk,1 ,e ψsk,1 represents the formation error of the unmanned boat, E xph ,E yph ,E yph It represents the error between the speed of the UAV in the XYZ direction and the virtual control law, E φah ,E θah ,E ψah represents the error of the virtual control law of the angular velocity and angle of the UAV, E xsk ,E ysk ,E ψsk Represents the error between the speed of the unmanned boat and the virtual control law.

[0036] Preferably, the predefined time virtual control law is specifically:

[0037]

[0038] Among them, a ph and a ah represents the virtual control law of the UAV, a sk represents the virtual control law of the unmanned boat, η represents the exponential parameter, T c represents the time parameter, π represents the circumference, Z ph,1 ,Z ah,1 Represents the UAV formation error conversion function vector, Z sk,1 represents the formation error conversion function vector of the unmanned boat, Ξ ph ,Ξ ah ,Ξ sk represents the communication parameters of the system, E eph,2 ,E eah,2 Indicates the state vector that the drone receives from other agents, E esk,2 Indicates that the unmanned boat receives the state vector of other intelligent agents, represents the bias of the error conversion function, γ ph,1 ,γ ah,1 ,γ sk,1 represents the time-varying coefficient of the error conversion function, β ph,2 ,β ah,2represents the relative state vector that the drone needs to maintain with the leader, β sk,2 Represents the relative state vector that the unmanned boat needs to maintain with the leader.

[0039] Preferably, establishing a predefined time fuzzy neural network includes:

[0040] Establish a fuzzy neural network including an input layer, a radial basis function layer, a normalization layer and an output layer;

[0041] All uncertainties of the air-sea heterogeneous system are estimated using a fuzzy neural network, which is expressed as:

[0042] F i =f i * =W i * S i ,i=ph.ah,sk;

[0043] Among them, F i represents all uncertain functions of the system, f i * represents the ideal value of the neural network fitting the uncertain function, S i Represents the neuron function of the neural network, W i * It represents the ideal weight that can estimate the uncertainty of the system.

[0044] As a preferred embodiment, the weight update law of the predefined time fuzzy neural network is specifically established as follows:

[0045]

[0046] Among them, i = ph, ah, sk represents the serial number of the agent, represents the estimated value of the neural network weight, S i Represents the neuron function of the neural network, Z i,2 represents the error conversion function vector, γ i,2 represents the time-varying coefficient of the error conversion function, η represents the exponential parameter, T c represents the time parameter and π represents the ratio of circumference to circumference.

[0047] As a preferred embodiment, the distributed predefined time formation controller of the air-sea heterogeneous cluster system is specifically:

[0048]

[0049] Among them, u ph and u ah represents the formation control law of UAVs, u sk represents the formation control law of the unmanned boat, A, B and fph ,f ah represents the model parameters of the UAV, C sk ,D sk and M sk represents the model parameters of the unmanned boat, R sk represents the rotation matrix of the unmanned boat, represents the estimated value of the neural network weight, S ph ,S ah ,S sk represents the neuron function of the neural network, Ξ ph ,Ξ ah ,Ξ sk Represents the communication parameters of the system, represents the virtual control law, represents the bias of the error conversion function, γ ph,1 ,γ ah,1 ,γ sk,1 and γ ph,2 ,γ ah,2 ,γ sk,2 represents the time-varying coefficient of the error conversion function, Z ph,1 ,Z ah,1 ,Z sk,1 and Z ph,2 ,Z ah,2 ,Z sk,2 represents the error conversion function vector, χ sk,2 represents the state vector of the unmanned boat, E ph ,E ah ,E sk represents the error vector between the speed and the virtual control law, η represents the exponential parameter, T c represents the time parameter and π represents the ratio of circumference to circumference.

[0050] The present invention has the following beneficial effects:

[0051] 1. The present invention limits the formation tracking error of the air-sea heterogeneous cluster system according to the functional characteristics of the performance function, and constrains the speed tracking error at the same time. Compared with the method of only constraining the position error, the distributed error constraint formation controller designed by the present invention has better control performance.

[0052] 2. Based on the fuzzy neural network, the present invention designs a predefined time fuzzy neural network by improving the weight convergence speed. While ensuring the good generalization ability of the fuzzy neural network, it improves the convergence speed of the neural network, sets and determines the stabilization time of the neural network, and enhances the robustness of the system.

[0053] 3. The present invention targets the air-sea heterogeneous cluster system of unmanned aerial vehicles and unmanned boats with external interference and model uncertainty, and determines a distributed predefined time error constraint formation control method based on fuzzy neural network, which improves the robustness of the air-sea heterogeneous cluster and enables the system to converge within a settable predefined time and is independent of the system initial value. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a predefined time formation control method based on fuzzy neural network;

[0055] Figure 2 This is a schematic diagram of the scenario of the air-sea heterogeneous swarm system of drones and unmanned boats;

[0056] Figure 3 It is a three-dimensional tracking diagram of the air-sea heterogeneous cluster system;

[0057] Figure 4 It is a two-dimensional tracking diagram of the air-sea heterogeneous cluster system;

[0058] Figure 5 It is the error diagram of the air-sea heterogeneous cluster system;

[0059] Figure 6 It is the control input diagram of the air-sea heterogeneous cluster system;

[0060] Figure 7 It is the fuzzy neural network fitting diagram of UAV 1 and UAV 2;

[0061] Figure 8 It is the fuzzy neural network fitting diagram of unmanned boat 1 and unmanned boat 2;

[0062] Fig. 9 This is the fuzzy neural network fitting diagram of unmanned boat 3 and unmanned boat 4. DETAILED DESCRIPTION

[0063] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0064] like Figure 1 As shown, a predefined time formation control method based on a fuzzy neural network provided by an embodiment of the present invention includes the following steps S1 to S6:

[0065] S1. Establish kinematic and dynamic models of six-degree-of-freedom quadrotor drone and three-degree-of-freedom unmanned boat;

[0066] In an optional embodiment of the present invention, the kinematic and dynamic model of the hth six-degree-of-freedom quad-rotor drone is established as follows:

[0067]

[0068] Among them, χ ph =[x ph ,y ph ,z ph ] T Represents the location information in global coordinates, χ ah =[φ ah ,θ ah ,ψ ah ] T Indicates the roll angle, pitch angle and yaw angle; d xph ,d yph ,d zph ,d xph ,d xph ,d xph represents the time-varying unknown ocean disturbance function; I x ,I y ,I z represents the moment of inertia; ρ x ,ρ y ,ρ z ,ρ φ ,ρ θ ,ρ ψ Represents the aerodynamic damping coefficient; m a represents mass and inertia parameters; u 1h ,u φh ,u θh ,u ψh represents the controller to be designed; g represents the acceleration of gravity; Λ represents the total residual rotor angle; I r Represents the rotor's moment of inertia.

[0069] To prevent the drone from experiencing strange phenomena during movement, scroll φ ah and pitch θ ah The following inequality should be satisfied:

[0070] -π / 2<φ ah <π / 2;

[0071] -π / 2<θ ah <π / 2;

[0072] The six-degree-of-freedom model of the h-th UAV can be divided into a position subsystem and an attitude subsystem, which contain model uncertainties and external disturbances:

[0073] Position Subsystem:

[0074]

[0075] Attitude subsystem:

[0076]

[0077] Among them, the three-dimensional position state vector χ is defined ph,1 =[χ xph,1 ,x yph,1 ,x zph,1 ] T = x ph and Define the attitude angle state vector χ ah,1 =[χ φah,1 ,x θah,1 ,x ψah,1 ] T = x ah and

[0078] u ph (t) = R h u 1h =[u xph (t),u yph (t),u zph (t)] T ∈R 3 ;

[0079] u ah (t)=[u φh (t),u θh (t),u ψh (t)] T ∈R 3 ;u xph ,u yph and u zph Represents the virtual control input of the longitudinal, lateral and altitude channels respectively. The control thrust can be calculated according to the formula Calculated; u φh ,u θh and u ψh Denote the ideal control inputs in rolling, pitching and yaw directions respectively; F ph (·)=Δ ph f ph (x ph,2 )+d ph (t)+AΔu ph and F ah (·)=Δ ah f ah (x ah,2 )+d ah (t) is the bounded uncertainty of the h-th UAV; d ph(·)=[d xph ,d yph ,d zph ] T ∈R 3 and Δ ph f ph (·)=Δ ph [f xph ,f yph ,f zph ] T ∈R 3 denote the external disturbance and model uncertainty of the h-th UAV position subsystem; d ah (.)=[d φah ,d θah ,d ψah ] T ∈R 3 and Δ ah f ah (·)=Δ ah [f φah ,f θah ,f ψah ] T ∈R 3 They represent the external disturbance and model uncertainty of the h-th UAV attitude subsystem respectively.

[0080]

[0081] According to the position control input of the UAV, the desired roll angle χ can be obtained φd,h1 and the pitch angle χ θd,h1 Therefore, according to the position of the drone control input and sin 2 (.)+cos 2 (.) = 1; x φd,h1 and χ θd,h1 It can be obtained by the following formula:

[0082]

[0083] This embodiment establishes the k-th three-degree-of-freedom unmanned boat kinematic and dynamic model:

[0084] χ sk,1 =R(χ ψsk,1 )v s,k

[0085]

[0086] Among them, χ sk,1 =[χ xsk,1 ,x ysk,1 ,x ψsk,1 ] Trepresents the position variable and heading angle variable of the kth unmanned boat in the global fixed coordinates; v k =[v xsk ,v ysk ,v ψsk ] T represents the velocity variable and angular velocity variable of the kth unmanned boat in the fixed coordinates of the body; u sk =[u xsk ,u ysk ,u ψsk ] T represents the control input to be designed for the kth unmanned boat; d k (t) = [d xk ,d yk ,d ψk ] T represents the external disturbance of the kth unmanned boat; R sk is a rotation matrix and satisfies ||R sk ||=1、R sk T R sk =I. represents the inertia matrix; represents the damping matrix, represents the Coriolis matrix and the centripetal matrix. The matrix R sk , and The definition is as follows:

[0087]

[0088] Therefore, the three-degree-of-freedom composite dynamic mathematical model of the kth USV can be obtained, which includes model uncertainty and external disturbances, and the formula is as follows:

[0089]

[0090] in, represents the model uncertainty of the kth USV, and M s,k ,C s,k ,D s,k It can be described as follows:

[0091]

[0092] The composite dynamics model of the kth USV is obtained by the following formula:

[0093]

[0094] in, is the total uncertainty of the kth USV, and the speed state information χ sk,2 =[χxsk,2 ,x ysk,2 ,x ψsk,2 ] T , k=1,...,M.

[0095] S2. Based on the kinematic and dynamic models of the six-degree-of-freedom quad-rotor drone and the three-degree-of-freedom unmanned boat, the formation tracking error and velocity tracking error of the air-sea heterogeneous cluster system are established;

[0096] In an optional embodiment of the present invention, the present embodiment firstly represents the communication situation of the air-sea heterogeneous cluster system by graph theory; the communication transmission line between N heterogeneous intelligent agents can be represented by an undirected graph Indicates that, where Γ={v 1 ,v 2 ,…,v N} represents all agents, Represents the communication line, A=[a ij ]∈R N ×N represents a non-negative adjacency matrix, a ij represents the transmission of directed information between agents i and j. Then a ij > 0, indicating that agent i can accept information transmission from agent j; otherwise a kj = 0. Define L = DA as the Laplace matrix L, where D = diag{d v1 ,…,d vN} represents a weighted diagonal matrix, and the elements of the matrix D are expressed as The augmented matrix corresponding to the leader is defined as B, with the diagonal term b i (i=1,…,N), when the agent can obtain the leader information, then b i =1, otherwise b i =0.

[0097] This embodiment then establishes the formation tracking position and velocity error in three-dimensional space:

[0098]

[0099] e φah,j = x φah,j -x φd,hj ,e θah,j = x θah,j -x θd,hj ;

[0100] Where r = 1, ..., N + M, j = 1, 2, e gph,j and e gsk,j They represent the state errors of the UAV and the unmanned boat on the two-dimensional plane, e zh,jRepresents the height error, e φah,j and e θah,j Denote the roll angle and pitch angle errors respectively, χ xL,j ,x yL,j ,x ψL,j ,x zL,j represents the trajectory information of the virtual leader, β gsi,j represents the relative position to the virtual leader. Integrating the formation error, the following formula can be obtained:

[0101] e i,j =Ξ i (x i,j -β i,j )+E e,i ,i=ph,ah,sk

[0102]

[0103]

[0104] Among them, i represents the formation error parameter, E ei,j Indicates that agent i receives information from other agents.

[0105] S3, constructing a preset performance function, and determining an error conversion function based on the formation tracking error and speed tracking error of the air-sea heterogeneous cluster system;

[0106] In an optional embodiment of the present invention, this embodiment establishes a velocity tracking error:

[0107] E i = x i,2 -a i ,i=ph,ah,sk;

[0108] Among them, a i Represents the virtual control law that needs to be designed.

[0109] The following boundary functions are determined as preset performance functions to limit formation tracking error and velocity tracking error:

[0110] Q i (t)=(Q i,0 -Q i,∞ )exp(-λ i t)+Q i,∞ ;

[0111] Where, i=ph,ah,sk;h=1,...,N;k=1,...,M;λ i >0 is the performance function speed parameter. i,0 >Q i,∞>0 is the domain parameter of the performance function. Under the action of the performance function, the formation tracking error and speed error meet the following constraints:

[0112] -Q i (t)<e i,1 <Q i (t)

[0113] -Q i (t)<E i <Q i (t);

[0114] Establish the error constraint error transfer function:

[0115]

[0116] S4. Determine a predefined time virtual control law according to the formation tracking error, the speed tracking error and the error conversion function;

[0117] In an optional embodiment of the present invention, the derivative of the formation error conversion function can be obtained:

[0118]

[0119] Where, i=ph,ah,sk;h=1,...,N;k=1,...,M; and Represents the bias of the error conversion function. γ i,1 represents the time-varying coefficient of the conversion function, which is defined as:

[0120]

[0121] Determine the virtual control law of the hth UAV and the kth unmanned boat:

[0122]

[0123] Where 0<η<1 represents the exponential parameter; T c Represents the time parameter.

[0124] S5, establishing a predefined time fuzzy neural network, and determining a distributed predefined time formation controller of the air-sea heterogeneous cluster system in combination with a predefined time virtual control law;

[0125] In an optional embodiment of the present invention, a fuzzy neural network is defined, which consists of four layers: input layer, radial basis function layer, normalization layer and output layer. The specific description is as follows:

[0126] Input layer: This layer contains q neurons, representing q-dimensional input variables. For the input vector L = [L 1 ,L2 ,…,L q ] T , the output of this layer represents x = [x 1 ,x 2 ,…,x q ] T , where x i Represents the output of the i-th neuron in the input layer. The output of this layer can be expressed as:

[0127] x i =L i ,i=1,...,q;

[0128] Radial basis function layer: This layer contains M neurons. The output calculation formula of this layer is:

[0129]

[0130] in, is the output of the jth neuron in this layer, ξ ij represents the membership of the jth neuron relative to the input of the ith neuron. A Gaussian function is used to represent the membership, μ ij represents the central parameter of the jth neuron’s input to the ith neuron, σ ij Represents the width parameter of the j-th neuron's input to the i-th neuron.

[0131] Normalization layer: This layer contains M neurons. The output calculation formula of this layer is:

[0132]

[0133] Among them, S j is the output value of the jth neuron.

[0134] Output layer: The output layer of the fuzzy neural network has only one neuron, and the network output calculation formula is:

[0135]

[0136] Among them, W j is the weight between the jth neuron in the normalization layer and the output layer, representing the result part of the jth fuzzy rule.

[0137] Using fuzzy neural network to estimate all uncertainties of air-sea heterogeneous system, we can get the following formula:

[0138] F i =f i * =W i * S i ,i=ph.ah,sk;

[0139] Among them, W i * It indicates that the ideal weight of system uncertainty can be estimated, and the weight error is defined as

[0140] Determine the update law for the weights of the fuzzy neural network for a predefined time:

[0141]

[0142] Among them, i=ph,ah,sk.

[0143] This embodiment determines a distributed predefined time formation controller for the air-sea heterogeneous cluster system:

[0144]

[0145] S6. Perform distributed predefined time error constraint formation control on the air-sea heterogeneous cluster system according to the distributed predefined time formation controller of the air-sea heterogeneous cluster system.

[0146] In order to verify the effectiveness of the method of the present invention, a simulation experiment was carried out, as follows:

[0147] The simulation experiment was carried out using the air-sea heterogeneous cluster system of unmanned aerial vehicles and unmanned boats to verify the effectiveness of the designed distributed predefined time error constraint formation control method based on fuzzy neural network. The system parameters are shown in Table 1, and the controller parameters are shown in Table 2:

[0148] Table 1 Parameters of the model of UAV and unmanned boat air-sea heterogeneous swarm system

[0149]

[0150]

[0151] Table 2 Controller parameters

[0152]

[0153] The relative quantity of the UAV attitude subsystem is β ah,1 =[0,0,0] T , the relative position vector β of the UAV and unmanned boat air-sea heterogeneous cluster system ph,1 and β sk,1 The options are as follows:

[0154]

[0155] The simulation results are as follows Figure 2-7As shown in the figure. It shows that the designed distributed predefined time error constraint formation controller can complete the three-dimensional formation tracking control of the air-sea heterogeneous cluster system of UAVs and unmanned boats within the predefined time, with good robustness, fast convergence speed and no dependence on the initial state. The designed predefined time fuzzy neural network improves the convergence speed while ensuring the good generalization ability of the fuzzy neural network.

[0156] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0159] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0160] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A predefined time formation control method based on fuzzy neural network, characterized in that: The following steps are involved: Establish kinematic and dynamic models of six-degree-of-freedom quad-rotor UAV and three-degree-of-freedom unmanned boat; According to the kinematic and dynamic models of the six-degree-of-freedom quad-rotor UAV and the three-degree-of-freedom unmanned boat, the formation tracking error and velocity tracking error of the air-sea heterogeneous swarm system are established; Construct a preset performance function and determine the error conversion function based on the formation tracking error and speed tracking error of the air-sea heterogeneous cluster system; Determine a predefined time virtual control law according to a formation tracking error, a speed tracking error and an error conversion function; A predefined-time fuzzy neural network is established, and the distributed predefined-time formation controller of the air-sea heterogeneous cluster system is determined in combination with the predefined-time virtual control law; The distributed predefined time error constraint formation control of the air-sea heterogeneous cluster system is performed according to the distributed predefined time formation controller of the air-sea heterogeneous cluster system.

2. A predefined time formation control method based on fuzzy neural network according to claim 1, characterized in that: The kinematic and dynamic models of the six-degree-of-freedom quadrotor drone are as follows: Among them, χ ph =[x ph ,y ph ,z ph ] T Represents the position information in the global coordinate system, χ ah =[φ ah ,θ ah ,ψ ah ] T Indicates the roll angle, pitch angle and yaw angle; d xph ,d yph ,d zph ,d φah ,d θah ,d ψah represents the time-varying unknown ocean disturbance function; I x ,I y ,I z represents the moment of inertia; ρ x ,ρ y ,ρ z ,ρ φ ,ρ θ ,ρ ψ Represents the aerodynamic damping coefficient; m a represents mass and inertia parameters; u 1h ,u φh ,u θh ,u ψh represents the controller to be designed; g represents the acceleration of gravity; Λ represents the total residual rotor angle; I r Represents the rotor's moment of inertia.

3. A predefined time formation control method based on fuzzy neural network according to claim 2, characterized in that: The kinematic and dynamic models of the three-degree-of-freedom unmanned boat are as follows: Among them, χ sk,1 =[χ xsk,1 ,x ysk,1 ,x ψsk,1 ] T represents the position variable and heading angle variable of the kth unmanned boat in the global coordinate system; v k =[v xsk ,v ysk ,v ψsk ] T represents the velocity variable and angular velocity variable of the kth unmanned boat in the fixed coordinates of the body; u sk =[u xsk ,u ysk ,u ψsk ] T represents the control input quantity to be designed for the kth unmanned boat; d sk (t) = [d xk ,d yk ,d ψk ] T represents the external disturbance of the kth unmanned boat; R sk is the rotation matrix; represents the inertia matrix; represents the damping matrix, represents the Coriolis matrix.

4. A predefined time formation control method based on fuzzy neural network according to claim 3, characterized in that: The formation tracking error of the air-sea heterogeneous cluster system is specifically: Among them, e i,j represents the formation tracking error, Ξ i represents the formation communication parameters, ph, ah represent the position subsystem and attitude subsystem of the h-th UAV, sk represents the k-th unmanned boat system, h=1,...,N represents the h-th UAV, N represents the number of UAVs in the heterogeneous multi-agent system, k=1,...,M represents the k-th unmanned boat, M represents the number of unmanned boats in the heterogeneous multi-agent system, N+M represents the number of all agents in the heterogeneous multi-agent system, diag represents the diagonal matrix, a h,i represents the communication parameters between the h-th UAV and other intelligent agents, b lh represents the communication parameters between the hth UAV and the leader, a k+N,i represents the communication parameters between the kth unmanned boat and other intelligent agents, I 3×3 represents a 3D unit vector, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ i,j represents the state vector of the agent, β i,j represents the relative state vector that the agent needs to maintain with the leader, E ei,j Indicates that the agent receives the state vector of other agents, which is defined as follows: Among them, χ pi,j represents the position state vector of the h-th UAV, β ψai,j represents the relative heading angle state vector that the h-th UAV needs to maintain with the leader, β xpi,j ,β yai,j represents the relative position state vector that the h-th UAV needs to maintain with the leader, E eph,j represents the angle error vector received by the drone from other agents, E esk,j represents the error vector received by the unmanned boat from other intelligent agents, E eph,j represents the position error vector received by the drone from other agents, E eah,j represents the attitude error vector received by the drone from other agents, a k+N,i and a k+N,i+N represents the communication parameters between the kth unmanned boat and other intelligent agents, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ xL,j ,x yL,j ,x zL,j ,x ψL,j represents the state vector of the leader, χ φd,hj and χ θd,hj represents the desired roll angle and pitch angle of the h-th UAV, β pi,j represents the relative position vector that the h-th UAV needs to maintain with the leader, χ xpi,j ,x ypi,j represents the position state vector of the hth UAV, χ ψai,j represents the heading angle of the hth UAV, χ xsi,j ,x ysi,j ,x ψsi,j represents the state vector of k unmanned boats, β xsi,j ,β ysi,j ,β ψsi,j Represents the relative state vector that the k unmanned boats need to maintain with the leader.

5. A predefined time formation control method based on fuzzy neural network according to claim 4, characterized in that: The formation tracking error and speed tracking error of the air-sea heterogeneous cluster system are specifically: Among them, e gph,j and e gsk,j They represent the state errors of the UAV and the unmanned boat on the two-dimensional plane, e ψah,j and e ψsk,j They represent the state errors of the heading angles of the UAV and the unmanned boat, e zh,j Indicates the height error, e φah,j and e θah,j Denote the roll angle and pitch angle errors respectively, χ xL,j ,x yL,j ,x ψL,j ,x zL,j represents the trajectory information of the virtual leader, β gph,j ,β gpi,j and β gsk,j ,β gsi,j They represent the relative positions of the UAV and the UAV with respect to the leader, β ψah,j ,β ψai,j and β ψsk,j ,β ψsi,j They represent the relative heading angle state vectors that the UAV and the unmanned boat need to maintain with the leader, β zph,j ,β zpi,j represents the relative altitude state vector that the drone needs to maintain with the leader, a h,i and a h,i+N represents the communication parameters between the h-th UAV and other intelligent agents, a k+N,i and a k+N,i+N represents the communication parameters between the kth unmanned boat and other intelligent agents, b lh represents the communication parameters between the hth UAV and the leader, b l(k+N) represents the communication parameters between the kth unmanned boat and the leader, χ gph,j ,x gpi,j and χ gsk,j ,x gsi,j Respectively represent the position status of the UAV and the unmanned boat, χ ψah,j ,x ψai,j and χ ψsk,j ,x ψsi,j Represent the heading angle state vectors of the UAV and the unmanned boat, respectively, zph,j ,x zpi,j Indicates the altitude status of the drone, χ φah,j and χ θah,j represents the roll angle and pitch angle state vector of the hth unmanned boat, respectively, φd,hj and χ θd,hj represents the desired roll angle and pitch angle of the hth unmanned boat, a ph and a ah represents the virtual control law vector of the UAV, a sk represents the virtual control law vector of the unmanned boat, χ ph,2 and χ ah,2 represents the velocity state vector of the UAV, χ sk,2 represents the velocity state vector of the unmanned boat, E ph and E ah represents the error vector between the UAV speed and the virtual control law, E sk represents the error vector between the speed of the unmanned boat and the virtual control law, E xph ,E yph ,E yph It represents the error between the speed of the UAV in the XYZ direction and the virtual control law, E φah ,E θah ,E ψah represents the error of the virtual control law of the angular velocity and angle of the UAV, E xsk ,E ysk ,E ψsk Represents the error between the speed of the unmanned boat and the virtual control law.

6. A predefined time formation control method based on fuzzy neural network according to claim 5, characterized in that: The error conversion function is specifically: Among them, Z ph,1 ,Z ah,1 and Z ph,2 ,Z ah,2 Represents the UAV formation error conversion function vector, Z sk,1 and Z sk,2 represents the error conversion function vector of the unmanned boat formation, Q ph ,Q ah ,Q sk The performance function index representing the error of the UAV and UAV formation, e xph,1 ,e yph,1 ,e zph,1 represents the formation error of the UAV in the XYZ direction, e φah,1 ,e θah,1 ,e ψah,1 represents the formation error of the UAV’s attitude system, e xsk,1 ,e ysk,1 ,e ψsk,1 represents the formation error of the unmanned boat, E xph ,E yph ,E yph It represents the error between the speed of the UAV in the XYZ direction and the virtual control law, E φah ,E θah ,E ψah represents the error of the virtual control law of the angular velocity and angle of the UAV, E xsk ,E ysk ,E ψsk Represents the error between the speed of the unmanned boat and the virtual control law.

7. A predefined time formation control method based on fuzzy neural network according to claim 6, characterized in that: The predefined time virtual control law is as follows: Among them, a ph and a ah represents the virtual control law vector of the UAV, a sk represents the virtual control law vector of the unmanned boat, η represents the exponential parameter, T c represents the time parameter, π represents the circumference, Z ph,1 ,Z ah,1 Represents the UAV formation error conversion function vector, Z sk,1 represents the formation error conversion function vector of the unmanned boat, Ξ ph ,Ξ ah ,Ξ sk represents the communication parameters of the system, E eph,2 ,E eah,2 Indicates the state vector that the drone receives from other agents, E esk,2 Indicates that the unmanned boat receives the state vector of other intelligent agents, represents the bias of the error conversion function, γ ph,1 ,γ ah,1 ,γ sk,1 represents the time-varying coefficient of the error conversion function, β ph,2 ,β ah,2 represents the relative state vector that the drone needs to maintain with the leader, β sk,2 Represents the relative state vector that the unmanned boat needs to maintain with the leader.

8. The predefined time formation control method based on fuzzy neural network according to claim 7, characterized in that: Building a predefined time fuzzy neural network includes: Establish a fuzzy neural network including an input layer, a radial basis function layer, a normalization layer and an output layer; All uncertain functions of the air-sea heterogeneous system are estimated using fuzzy neural networks, which can be expressed as: F i =f i * =W i * S i ,i=ph.ah,sk; Among them, F i represents all uncertain functions of the system, f i * represents the ideal value of the neural network fitting the uncertain function, S i Represents the neuron function of the neural network, W i * It represents the ideal weight that can estimate the uncertainty of the system.

9. A predefined time formation control method based on fuzzy neural network according to claim 8, characterized in that: The weight update law of the established predefined time fuzzy neural network is specifically: Among them, i = ph, ah, sk represents the intelligent agent, represents the estimated value of the neural network weight, S i Represents the neuron function of the neural network, Z i,2 represents the error conversion function vector, γ i,2 represents the time-varying coefficient of the error conversion function, η represents the exponential parameter, T c represents the time parameter and π represents the ratio of circumference to circumference.

10. A predefined time formation control method based on fuzzy neural network according to claim 9, characterized in that: The distributed predefined time formation controller of the air-sea heterogeneous cluster system is as follows: Among them, u ph and u ah represents the formation control law of UAVs, u sk represents the formation control law of the unmanned boats, A, B and f ph ,f ah represents the model parameters of the UAV, C sk ,D sk and M sk represents the model parameters of the unmanned boat, R sk represents the rotation matrix of the unmanned boat, represents the estimated value of the neural network weight, S ph ,S ah ,S sk represents the neuron function of the neural network, Ξ ph ,Ξ ah ,Ξ sk Represents the communication parameters of the system, represents the virtual control law, represents the bias of the error conversion function, γ ph,1 ,γ ah,1 ,γ sk,1 and γ ph,2 ,γ ah,2 ,γ sk,2 represents the time-varying coefficient of the error conversion function, Z ph,1 ,Z ah,1 ,Z sk,1 and Z ph,2 ,Z ah,2 ,Z sk,2 represents the error conversion function vector, χ sk,2 represents the state vector of the unmanned boat, E ph ,E ah ,E sk represents the error vector between the speed and the virtual control law, η represents the exponential parameter, T c represents the time parameter and π represents the ratio of circumference to circumference.

Images