Error constraint cluster synchronization control method for seabed flight nodes
By adopting the combination of restraint control and adaptive neural network in the subsea flight node system, the accuracy and stability problems of cluster synchronization control are solved, and high-precision multi-AUV system control is realized, reducing computing complexity and communication costs. .
Patent Information
- Application Number
- CN202411228094.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-03
AI Technical Summary
The cluster synchronization control of subsea flight node clusters has problems such as limited control accuracy and poor stability. Especially in the face of external perturbations and nonlinear model uncertainty, it is difficult for the prior art to achieve high-precision multi-AUV system control.
Adaptive neural network controller based on restraint control and preset performance is adopted to approximate generalized interference terms through the neural network, design cluster synchronization control law, and combine preset performance functions and smooth functions to realize error constraint cluster clusters for subsea flight nodes. Synchronous control.
It effectively improves the control accuracy and stability of multi-AUV systems, and can achieve accurate tracking of position and attitude when there are external perturbations and model uncertainties, reducing computing complexity and communication costs.
Smart Images

Figure CN119087813A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of autonomous underwater vehicle control, and in particular relates to a cluster synchronization control method of seabed flight nodes. Background Art
[0002] Autonomous Underwater Vehicle (AUV) has become a hot topic of research due to its excellent performance. With the deepening of research on AUV, the application scope of AUV has also become wider. AUV not only plays an important role in the sampling and collection of seabed resources, but also can complete the maintenance of underwater facilities, underwater topography mapping and other tasks. Among them, the Ocean bottom flying node (OBFN) belongs to a type of extended model AUV. Its model is as follows Figure 1 shown.
[0003] There are complexities and diversity in maritime missions. In order to perform multiple tasks simultaneously and improve the efficiency of mission execution, it is necessary to divide the multi-AUV system into multiple subgroups, with each group of AUVs performing different tracking tasks. For example, when searching and surveying a large area of sea is required, dividing the AUV into multiple subgroups can cover different areas at the same time and improve search efficiency. However, due to high nonlinearity, cross-coupling and external ocean current disturbances, these will cause random deviations in the motion trajectory of the AUV. Therefore, it is very important to control the position of the AUV within the expected range, which puts forward further requirements on the control accuracy of the AUV.
[0004] There are external disturbances and nonlinear model uncertainties in the AUV system. The common treatment method is to assume that the factors that affect the stability of the system have a constant upper bound and estimate this constant. This method has low accuracy and poor stability; or to consider the above factors separately and design algorithms to control them, but it will increase the complexity of the calculation. In recent years, there has been a method that integrates factors such as external disturbances and nonlinear model uncertainties, regards them as general uncertainties of the system, and uses neural networks for estimation and approximation. However, this method has limited control accuracy for AUVs. With the gradual increase in requirements for AUV operations, simple multi-AUV formation control can no longer meet people's control needs. More and more research has begun to involve how to improve the control accuracy of multi-AUV systems. Summary of the invention
[0005] The present invention aims to solve the problems of limited control accuracy and poor stability in cluster synchronization control of OBFN.
[0006] An error-constrained cluster synchronization control method for submarine flight nodes, comprising:
[0007] S1. Based on the dynamic model of the seabed flight node OBFN, the model of the OBFN with interference is obtained:
[0008]
[0009]
[0010] Wherein, the subscript i represents the parameter corresponding to the i-th OBFN in the multi-OBFN system, M ηi , C RBηi , C Aηi , D ηi , g ηi Indicates the M corresponding to the i-th OBFN η =MJ -1 , C Aη =C A (v r )J -1 , D η =D(v r )J -1 , g η =g(η),v r =vv c , v r is the velocity of OBFN relative to the ocean current, v c is the velocity of the ocean current in the moving coordinate system; M is the mass inertia matrix, J is the transformation matrix between the earth coordinate system and the moving coordinate system, C RB is the Coriolis force and centripetal force matrix of the rigid body, C A is the Coriolis force and centripetal force matrix of the added mass, D is the hydrodynamic damping matrix, g η are the force and torque vectors generated by gravity and buoyancy, and the parameter subscript η represents the parameters in the geodetic coordinate system; v ηi =J i (η i )v i represents the velocity in the geodetic coordinate system, v i represents the velocity of the i-th OBFN in the body coordinate system, v = [u, v, w, p, q, r] T is the velocity and angular velocity in the motion coordinate system; is η i First-order derivative, the · above the parameter represents the first-order derivative of the corresponding parameter, η i is the η of the i-th OBFN, η=[x,y,z,φ,θ,ψ] T F represents the vector composed of the position coordinates and the pitch angle of the USV in the geodetic coordinate system; i is the generalized disturbance composed of model uncertainty and external disturbance; τ iis the control force and torque generated by the propulsion system of the i-th OBFN; the generalized disturbance F i Using neural network φ i Approximate approach to processing;
[0011] Based on the OBFN model, a cluster synchronization controller is used to synchronize the OBFN clusters; the cluster synchronization controller is as follows:
[0012] Cluster synchronization control law:
[0013] τ i =τ i1 +τ i2 +τ i3
[0014]
[0015]
[0016]
[0017] in, is the σth i Cluster V σi A subset of The OBFN system in D receives information from the OBFN of other clusters, and the containment control needs to weaken the influence of the OBFN from other clusters; i = 0 means that the i-th OBFN cannot receive information from other OBFNs; these OBFNs are selected as pinning nodes; ω i The estimated value of The adaptive law is for The estimated value of Δ i is a positive constant; definition is the position error and velocity error, η σi represents the expected position and attitude of the i-th OBFN; v ησi represents the expected velocity and angular velocity of the i-th OBFN; the subscript σ i Used to represent the σth i The expected value of the cluster; Υ=diag[ρ i (t)], R = diag[r i ], i = 1, ..., N, N represents the number of OBFNs; ρ i (t) is the performance function, δi is the adjustment parameter, δ i , Denotes δ i The upper and lower bounds of S(·) is a smooth function.
[0018] Furthermore, the performance function Among them, ρ i0 , i∞ and is a positive constant and t is time.
[0019] Furthermore, the performance function ρ i (t) satisfies the following time-varying constraints:
[0020]
[0021] Where t ≥ 0 and δ i is the adjustment parameter, δ i , Denotes δ i The upper and lower bounds of , i=1,...,N.
[0022] Furthermore, the smooth function
[0023] Furthermore, the dynamic model of the seabed flying node OBFN is as follows:
[0024]
[0025] Where τ is the control force and torque generated by the propulsion system.
[0026] Furthermore, the generalized interference F i Using neural network φ i The approach to processing is as follows:
[0027]
[0028] in, F i The estimated value, that is, F approximated by neural network i , is the estimated value of the weight matrix, is the neural network input, e i =η i -η σi Represents tracking error.
[0029] Furthermore, the φ i (z i ) uses the Gaussian basis function.
[0030] Furthermore, generalized interference where F i is the generalized disturbance composed of model uncertainty and external disturbance, d i Indicates ocean current disturbances, represents the disturbance caused by uncertainty in the OBFN model, ΔM ηi , ΔC RBηi , ΔD ηi , Δg ηi They represent the M generated due to system uncertainty. ηi , C RBηi , D ηi , g ηi The corresponding uncertainty values.
[0031] Beneficial effects:
[0032] Aiming at the cluster synchronization control problem of OBFN, the present invention comprehensively considers the influence of external disturbance and model uncertainty, and proposes an adaptive neural network controller based on pinning control and preset performance to realize error-constrained cluster synchronization control of OBFN, which can effectively solve the problem of poor stability of cluster synchronization control of OBFN and improve control accuracy. The results obtained through simulation are as follows Figure 3 and Figure 4 The control law designed in the present invention shows the cluster consistency control effect of the multi-OBFN system. The curves in the figure represent the posture states of the virtual leader OBFN and the follower OBFN in six degrees of freedom. Figure 5 and Figure 6 The figure shows the constraint effect of the control law designed in the present invention on the position error and attitude error. The curves in the figure respectively represent the difference between the posture state of each follower OBFN in six degrees of freedom and the expected state. The two curves that always constrain the error are the set constraint boundary curves. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 This is a model diagram of the submarine flight node.
[0034] Figure 2 Schematic diagram of the communication topology between the virtual leader and the follower OBFN;
[0035] Figure 3 Linear displacement tracking curves of the follower and virtual navigator OBFN;
[0036] Figure 4 Angular displacement tracking curves of the follower and virtual navigator OBFN;
[0037] Figure 5 It is the position error curve and error boundary diagram following OBFN;
[0038] Figure 6 The posture error curve and error boundary diagram following OBFN. DETAILED DESCRIPTION Specific implementation method one:
[0040] This embodiment is an error-constrained cluster synchronization control method for submarine flight nodes. Before the specific description, the coordinate system and parameter definition involved in the present invention are first described:
[0041] The geodetic coordinate system is represented by E-ξηζ, where E is the origin, which can be selected as any point in the OBFN working sea area. The selected reference plane is tangent to the surface of the OBFN operating area. Eζ is the axis with the positive direction pointing to the center of the earth. The Eξ axis points to the geographic north, and the Eη axis points to the geographic east.
[0042] O-xyz represents the body coordinate system, which is also called the motion coordinate system because it moves with the OBFN. It should be noted that as long as the body coordinate system can be guaranteed to move with the OBFN, in this embodiment, for the convenience of explanation, the origin O of the motion coordinate system is selected as the center of gravity of the OBFN, the x-axis points to the bow of the OBFN along the middle longitudinal axis of the OBFN, the y-axis points from the center of gravity of the OBFN to the starboard side of the OBFN, and the z-axis is perpendicular to the xOy plane and points from the center of gravity of the OBFN to the bottom of the OBFN, with downward being positive.
[0043] Parameter definition: M——mass inertia matrix; η=[x,y,z,φ,θ,ψ] T ——OBFN six-degree-of-freedom position and attitude value in the geodetic coordinate system; η σ =[x σ ,y σ ,z σ ,φ σ ,θ σ ,ψ σ ] T ——Expected values of the six-degree-of-freedom position and attitude of the OBFN cluster in the geodetic coordinate system; e = η-η σ ——Tracking error; v = [u, v, w, p, q, r] T ——Velocity and angular velocity in the motion coordinate system; J——Transformation matrix between the earth coordinate system and the motion coordinate system; C RB ——Coriolis force and centripetal force matrix of rigid body; C A ——Coriolis force and centripetal force matrix of added mass; D——hydrodynamic damping matrix; g η ——the force and torque vectors generated by gravity and buoyancy; τ——the control forces and torques generated by the propulsion system; v r ——the speed of OBFN relative to the ocean current; vc ——the speed of the ocean current in the moving coordinate system; F——the generalized uncertainty of the system;
[0044] The present invention selects OBFN for control, thereby achieving the effect of dividing groups. Compared with the full equipment control method, the containment control method effectively reduces the communication cost and the complexity of calculation; at the same time, the preset performance control method is adopted to make the designed error included in the set set from the beginning, and gradually converge to a predefined small set, and the convergence speed and overshoot value also meet the requirements. The preset performance function has been used to provide a smooth tracking response of the control signal with a sufficient range, and to ensure transient and tracking performance with specified characteristics.
[0045] In order to achieve cluster synchronization control of OBFN using an adaptive neural network controller based on pinning control and preset performance method, the object of the present invention is to design a controller τ so that the position and attitude η of the OBFN system can still track the expected value η in the presence of external disturbances and model uncertainty. σ And let the tracking error e = η-η σ There are predetermined dynamic performance and steady-state response conditions.
[0046] This embodiment is an error-constrained cluster synchronization control method for submarine flight nodes, which includes the following steps:
[0047] S1. Transformation of the dynamic model of the seabed flight node:
[0048] The dynamic model of the seabed flight node OBFN is established. The dynamic equation of OBFN is expressed by a six-degree-of-freedom nonlinear model based on the Fossen outline.
[0049]
[0050] Where M η =MJ -1 , C Aη =C A (v r )J -1 , D η =D(v r )J -1 , g η =g(η), v r =vv c ; M is the mass inertia matrix, η=[x,y,z,φ,θ,ψ] T The vector represents the position and attitude of the OBFN in the geodetic coordinate system, v = [u, v, w, p, q, r] Trepresents the velocity and angular velocity in the motion coordinate system, J is the transformation matrix between the earth coordinate system and the motion coordinate system; C RB is the Coriolis force and centripetal force matrix of the rigid body, C A is the Coriolis force and centripetal force matrix of the added mass; D is the hydrodynamic damping matrix, g η is the force and torque vector generated by gravity and buoyancy, τ is the control force and torque generated by the propulsion system, v r is the velocity of OBFN relative to the ocean current, v c is the velocity of the ocean current in the moving coordinate system.
[0051] Considering the model uncertainty and external interference of OBFN, it can be rewritten as follows
[0052]
[0053] The parameter subscript i represents the parameter corresponding to the i-th OBFN in the multi-OBFN system, v ηi =J i (η i )v i Represents the velocity in the geodetic coordinate system (the OBFN system directly obtains v i Body coordinate system), is the generalized disturbance composed of model uncertainty and external disturbance, where d i Indicates ocean current disturbances, represents the disturbance caused by uncertainty in the OBFN model, ΔM ηi , ΔC RBηi , ΔD ηi , Δg ηi They represent the M generated due to system uncertainty. ηi , C RBηi , D ηi , g ηi The corresponding uncertainty value. i The neural network approach is used for processing.
[0054] Determining position error and velocity error
[0055]
[0056]
[0057] Among them, η σi Indicates the i-th OBFN (located at the σth i The expected position and attitude of the cluster; v ησi represents the expected velocity and angular velocity of the i-th OBFN; the subscript σ i Used to represent the σth i The expected value of the cluster.
[0058] S2. Designing performance functions based on the preset performance control method:
[0059] Performance function ρ(t) and error components Related, and ρ(t) should be a smooth function so that ρ(t)∈R + is a positive decreasing function, The default performance function can be written as
[0060]
[0061] Among them, R + represents a positive real number, ρ i0 , i∞ and is a positive constant; t is time.
[0062] In order to achieve the desired preset performance, the following time-varying constraints are proposed:
[0063]
[0064] Where t ≥ 0 and δ i is the adjustment parameter, δ i , Denotes δ i The upper and lower bounds of , i=1,...,N.
[0065] In order to transform the error of the nonlinear system from the constrained to unconstrained state in (4), the following transformation is performed:
[0066]
[0067] Among them, ψ(·) is the intermediate variable function, representing the nonlinear time-varying dynamic constraint;
[0068] (5) can be converted into the following form
[0069]
[0070] where S(·) is a smooth function and the inverse of ψ(·).
[0071] The general form of S(·) is as follows
[0072]
[0073] The conversion error
[0074]
[0075] Because i(t)>0, then (8) can be rewritten as
[0076]
[0077] The conversion error in (5) is as follows:
[0078]
[0079] To reduce the impact of chattering, a transformation error of the following form is used:
[0080]
[0081] in, ξ>0 is a design parameter.
[0082] The derivative of (11) is expressed as follows:
[0083]
[0084] set up
[0085]
[0086] Combining (12) and (13), we have
[0087]
[0088] where Ξ=[Ξ1,Ξ2,...,Ξ N ] T , Υ=diag[ρ i (t)], R = diag[r i ], i = 1, ..., N represents the number of OBFNs;
[0089] S3. Select radial basis function neural network parameters:
[0090] For the generalized interference term in the system, an adaptive neural network is used to approximate it, and the neural network input is taken as Then the estimate of the generalized interference term can be written as
[0091]
[0092] in, is the estimated value of the weight matrix; φ i (z i ) can be in the form of a Gaussian basis function, φ i (z i )=[φ i1 (z i ),φ i2 (zi ),...,φ ij (z i ),...,φ im (z i )] T ∈R m , m is the number of hidden layer nodes in the network; e i =η i -η σi Represents the tracking error.
[0093] S4. Design cluster synchronization controller:
[0094] The following cluster synchronization control law is proposed:
[0095] τ i =τ i1 +τ i2 +τ i3 (16)
[0096]
[0097]
[0098] in, is the σth i Cluster V σi A subset of . The OBFN system in the cluster can receive information from the OBFN of other clusters, and the containment control needs to reduce the influence of the OBFN from other clusters. i = 0 means that the i-th OBFN cannot receive information from other OBFNs. These OBFNs can be selected as pinning nodes. Definition
[0099]
[0100]
[0101]
[0102] in, ω i The estimated value of
[0103] Design the adaptive law as
[0104]
[0105] in for The estimated value of Δ i is a positive constant. And define
[0106] The controller design and verification process is as follows:
[0107] 1. Dynamic model of OBFN:
[0108] The dynamic equation of OBFN is expressed by a six-degree-of-freedom nonlinear model based on the Fossen outline.
[0109]
[0110] Where M η =MJ -1 , C Aη =C A (v r )J -1 , D η =D(v r )J -1 , g η =g(η), v r =vv c ; M is the mass inertia matrix, η=[x,y,z,φ,θ,ψ] T The vector represents the motion and posture of OBFN in the body coordinate system, v = [u,v,w,p,q,r] T represents the velocity and angular velocity in the motion coordinate system, J is the transformation matrix between the earth coordinate system and the motion coordinate system; C RB is the Coriolis force and centripetal force matrix of the rigid body, C A is the Coriolis force and centripetal force matrix of the added mass; D is the hydrodynamic damping matrix, g η is the force and torque vector generated by gravity and buoyancy, τ is the control force and torque generated by the propulsion system, v r is the velocity of OBFN relative to the ocean current, v c is the velocity of the ocean current in the moving coordinate system.
[0111] Considering the model uncertainty and external interference of OBFN, it can be rewritten as follows
[0112]
[0113] in The generalized interference composed of model uncertainty and external interference is processed by neural network approximation.
[0114] Defining position error and velocity error
[0115]
[0116] If in any initial state, OBFN satisfies the following conditions
[0117]
[0118]
[0119] Then the position and velocity information of OBFNs in different clusters are bounded, and the state error converges to an arbitrary small neighborhood of zero in a finite time, and the multi-OBFN system can achieve cluster synchronization.
[0120] Assumption 1: i The position and speed information of the cluster's virtual leader OBFN is
[0121]
[0122] Assumption 2 For a multi-cluster OBFN system, there exists at least one directed spanning tree.
[0123] Assumption 3 By choosing appropriate parameters, we can get the following inequality
[0124]
[0125]
[0126] in L is the Laplacian matrix of the directed graph, is the submatrix of L minus l rows and columns. i is the pinning coefficient, for M∈R N×N and D = diag(d1, d2, ... d l ...0,...0) N×N , where d l > 0.
[0127]
[0128] Where M is a symmetric matrix; M l is the submatrix of the matrix M with the first l rows and l columns removed; Both B and C are matrices of corresponding dimensions.
[0129] For the matrix (29), if all d l Satisfy l >λ max (B-CM l -1 C T ), then M l <0 is equivalent to MD<0.
[0130] 2. Preset performance control method:
[0131] Performance function ρ(t) and error components Related, and ρ(t) should be a smooth function so that ρ(t)∈R + is a positive decreasing function The default performance function can be written as
[0132]
[0133] where ρ i0 , ρ i∞ and is a positive constant. In order to achieve the desired preset performance, the following time-varying constraints are proposed:
[0134]
[0135] where t ≥ 0 and 0 ≤ δ i ≤1, i=1,...,N.
[0136] In order to transform the error of the nonlinear system from the constrained to unconstrained state in (4), the following transformation is performed:
[0137]
[0138] (5) can be converted into the following form
[0139]
[0140] where S(·) is a smooth function and the inverse of ψ(·).
[0141] The general form of S(·) is as follows
[0142]
[0143] The conversion error
[0144]
[0145] Because i (t)>0, then (8) can be rewritten as
[0146]
[0147] The conversion error in (5) is as follows:
[0148]
[0149] In order to reduce the impact of chattering, the following form of transformation error is used:
[0150]
[0151] in ξ>0 is a design parameter.
[0152] The derivative of (11) is expressed as follows:
[0153]
[0154] set up
[0155]
[0156] Combining (12) and (13), we have
[0157]
[0158] where Ξ=[Ξ1,Ξ2,...,Ξ N ] T , Υ=diag[ρ i (t)], R = diag[r i ], i = 1, ..., N. If the controller τ is designed to make Ξ bounded, then according to equation (41) we can get and Bounded.
[0159] 3. Cluster synchronization controller design:
[0160] For the generalized interference term in the system, an adaptive neural network is used to approximate it, that is,
[0161]
[0162] where ω i is the weight matrix, φ i (z i )=[φ1(z i ),φ2(z i ),...,φ j (z i ),…,φ m (z i )] T ∈R m , m is the number of hidden layer nodes in the network, φ i (z i ) can be in the form of Gaussian basis function,
[0163]
[0164] where c ij and b ij are the center and width of the Gaussian function respectively.
[0165] The neural network input is taken as Then the estimate of the generalized interference term can be written as
[0166]
[0167] in is the estimated value of the weight matrix.
[0168] The following cluster synchronization control law is proposed:
[0169] τ i =τ i1 +τ i2 +τ i3 (45)
[0170]
[0171]
[0172] in is the σth i Cluster V σi A subset of . The OBFN system in the cluster can receive information from the OBFNs of other clusters, and the containment control needs to reduce the influence of the OBFNs from other clusters. i = 0 means that the i-th OBFN cannot receive information from other OBFNs. These OBFNs can be selected as pinning nodes. Definition
[0173]
[0174]
[0175]
[0176] in ω i The estimated value of
[0177] Design the adaptive law as
[0178]
[0179] in for The estimated value of Δ i is a positive constant. And define
[0180] By using the control algorithm (45) and the adaptive laws (50) and (51), the signal of the multi-OBFN system will be semi-globally uniformly bounded. The follower OBFN can follow the trajectory of the virtual leader OBFN in the same subgroup, and the cluster synchronization of multiple OBFNs can be achieved.
[0181] Proof: Design the following Lyapunov function
[0182]
[0183] in By properly designing φ i accomplish.
[0184] make Then, taking the derivative of (52), substituting (2), (26) and (25) into (52), we can obtain
[0185]
[0186] Selecting an appropriate Lyapunov function
[0187]
[0188] The derivative is
[0189]
[0190] in χ i =I6-κ i .
[0191] Position error is bounded, then
[0192] According to the properties of the matrix, we can get
[0193]
[0194] Similarly, we can get
[0195]
[0196] We can further obtain
[0197]
[0198] Where N, α1 and β1 satisfy the following conditions
[0199]
[0200]
[0201]
[0202] in Σ=diag[ε1,ε2,...,ε N], Λ=diag[||Λ1|| / 2,||Λ2|| / 2,...,||Λ N || / 2].
[0203] Simplifying (58) we can get
[0204]
[0205] The expressions of α2 and β2 are as follows
[0206]
[0207]
[0208] Then we can conclude that the system is uniformly ultimately bounded.
[0209]
[0210] Design the following parameters
[0211]
[0212] Through (60) and (61), we can get
[0213]
[0214] in
[0215] because Then we can get
[0216]
[0217] From (62) and (63), it can be seen that all error variables of the system can converge to a finite range, and it can be judged that the multi-OBFN system can eventually achieve cluster synchronization.
[0218] Example
[0219] In order to verify the effectiveness of the control method designed by the present invention, it is combined with the OBFN model for simulation, taking into account the influence of external disturbances and model uncertainty.
[0220] Since the communication between multiple AUVs (i.e. OBFN) will be greatly affected underwater, the communication between multiple AUVs is carried out using a directed communication topology. The information transmission relationship between the leader and follower AUV is as follows: Figure 2 As shown. Figure 2 In the above equation, we know that AUV3 in group G2 cannot obtain information from other AUVs in groups G1 and G2. Therefore, in the containment control law (47), AUV2 and AUV4 in group G1 can obtain information from AUV3 in group G2, that is, they can obtain information from AUVs in other groups, so and Therefore, AUV2, AUV3 and AUV4 were selected as the pinning nodes.
[0221] The hydrodynamic parameters, inertia coefficients, initial values of position and attitude, initial values of velocity and expected trajectory of the corresponding models are shown in Table 1.
[0222] Table 1 Inertial parameters of AUVs in group G1
[0223]
[0224] Table 2 Inertial parameters of AUVs in group G2
[0225]
[0226] Table 3 Hydrodynamic coefficients of AUVs in group G1
[0227]
[0228] Table 4 Hydrodynamic coefficients of AUVs in group G2
[0229]
[0230] Table 5 Initial position of the following AUV
[0231]
[0232]
[0233] Table 6 Initial speed of following AUV
[0234]
[0235] Table 7 Movement trajectories of the virtual pilot AUV in subgroups G1 and G2
[0236]
[0237] Model uncertainty: In view of the uncertainty of the AUV model itself, a 30% model error was selected in the simulation experiment, i.e. Δ in =0.3.
[0238] Ocean current disturbance: Assume that the external disturbance is described as
[0239] d i (t) = [d i1 (t),d i2 (t),d i3 (t),di4 (t),d i5 (t),d i6 (t)] T (64)
[0240]
[0241] Controller parameters: The control parameter is selected as ρ i0 =10,ρ i∞ =0.6, l i =1, i = 1, ..., 6, the default performance form is ρ i (t) = 9.4exp(-t) + 0.6. Error conversion function δ 1= δ 2= δ 4=0.5, δ 3 = 0.1, δ 5 = 0.3, δ 6=0.6. The control gain parameters in the control method are d2=d4=15, d3=20. The parameters Δ in the adaptive law i =0.03, i=1,...,6.
[0242] The results obtained through simulation are as follows Figure 3 and Figure 4 The control law designed in the present invention shows the cluster consistency control effect of the multi-AUV system. The curves in the figure respectively represent the posture states of the virtual leader AUV and the follower AUV in six degrees of freedom.
[0243] Figure 5 and Figure 6 The figure shows the constraint effect of the control law designed in the present invention on the position error and attitude error. The curves in the figure respectively represent the difference between the posture state of each follower AUV in six degrees of freedom and the expected state. The two curves that always constrain the error are the set constraint boundary curves.
[0244] from Figure 3-Figure 6 It can be seen that the method proposed in the present invention can enable the system to achieve grouping, and the system error can converge to a specific range, achieving error-constrained cluster consistent control.
[0245] Comparison with existing technical solutions: If external disturbances and model uncertainties are taken into account to achieve the requirements for synchronous control of a cluster of submarine flight nodes, in addition to the algorithm of the present invention, there are also solutions based on full-equipment control methods, adaptive neural networks, etc. The following briefly introduces these two solutions and compares them with the algorithm of the present invention.
[0246] A. Solution based on full equipment control
[0247] "Scaled group consensus in multiagent systems with first / second-order continuous dynamics" studies the problem of scaled cluster synchronization of multi-agents with first / second-order linear continuous dynamics based on the full-device control method. "Adaptive Output Consensus Design in Clustered Networks of Heterogeneous Linear Multi-Agent Systems" mainly studies the problem of cluster synchronization in networks divided into heterogeneous agent clusters, and achieves consistency control by controlling each node. Both are based on the full-device control method to achieve cluster consistency control. However, compared with the control of the entire agent system in the full-device control method, the containment control method in the present invention only needs to control a part of the agent system, which will effectively reduce the communication cost and computational complexity.
[0248] Therefore, the present invention adopts the pinning control method to realize the synchronous control of multi-AUV system clusters, regards the external disturbance and model uncertainty as the generalized interference of the system, and uses an adaptive neural network to estimate and approximate it, which reduces the complexity of the calculation, saves computing resources, and is closer to actual engineering needs.
[0249] B. Neural network-based solutions
[0250] Radial basis function neural networks and adaptive methods are often used to deal with unknown external disturbances. Sliding mode based neural adaptive formation control of underactuated AUVs with leader-followers strategy combines backstepping, adaptive neural network and dynamic surface control technology to propose a robust sliding mode formation control strategy to solve the master-slave formation control problem of underactuated underwater robots under uncertain ocean disturbances. Three-dimensional neural network tracking control of a moving target by underactuated autonomous underwater vehicles uses a multi-layer neural network to approximate an unknown nonlinear dynamic system, and uses adaptive robust control technology to compensate for the disturbances caused by waves, wind and ocean currents to ensure that the tracking error of the AUV is semi-globally consistent and ultimately bounded. Robust adaptive sliding mode control of underactuated autonomous underwater vehicles with uncertain dynamics studies the trajectory tracking control problem of a class of underactuated five-degree-of-freedom underwater robots under uncertain dynamics and unknown environmental disturbances. Considering the actual situation of system parameter uncertainty and external disturbances, a direct adaptive neural network controller combined with a conditional integrator is proposed to improve the robustness and adaptability of the system.
[0251] All three of them use adaptive neural networks to estimate and approximate the factors that affect the stability of the system, but compared with the algorithm of the present invention, the above schemes do not take into account the problem of system control accuracy. Therefore, the algorithm of the present invention is improved on this basis, by introducing a preset performance method to control the set system error within a specific range, thereby improving the stability of the system.
[0252] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.
Claims
1. An error-constrained cluster synchronization control method for submarine flight nodes, characterized in that: include: S1. Based on the dynamic model of the seabed flight node OBFN, the model of the OBFN with interference is obtained: Wherein, the subscript i represents the parameter corresponding to the i-th OBFN in the multi-OBFN system, M ηi , C RBηi , C Aηi , D ηi , g ηi Indicates the M corresponding to the i-th OBFN η =MJ -1 , C Aη =C A (v r )J -1 , D η =D(v r )J -1 , g η =g(η),v r =vv c , v r is the velocity of OBFN relative to the ocean current, v c is the velocity of the ocean current in the moving coordinate system; M is the mass inertia matrix, J is the transformation matrix between the earth coordinate system and the moving coordinate system, C RB is the Coriolis force and centripetal force matrix of the rigid body, C A is the Coriolis force and centripetal force matrix of the added mass, D is the hydrodynamic damping matrix, g η are the force and torque vectors generated by gravity and buoyancy, and the parameter subscript η represents the parameter in the geodetic coordinate system; v ηi =J i (η i )v i represents the velocity in the geodetic coordinate system, v i represents the velocity of the i-th OBFN in the body coordinate system, v = [u, v, w, p, q, r] T is the velocity and angular velocity in the motion coordinate system; is η i First-order derivative, the · above the parameter represents the first-order derivative of the corresponding parameter, η i is the η of the i-th OBFN, η=[x,y,z,φ,θ,ψ] T F represents the vector composed of the position coordinates and the pitch angle of the USV in the geodetic coordinate system; i is the generalized disturbance composed of model uncertainty and external disturbance; τ i is the control force and torque generated by the propulsion system of the i-th OBFN; the generalized disturbance F i Using neural network φ i Approximate approach to processing; Based on the OBFN model, a cluster synchronization controller is used to synchronize the OBFN clusters; the cluster synchronization controller is as follows: Cluster synchronization control law: t i =t i1 +t i2 +t i3 in, is the σth i Cluster V σi A subset of The OBFN system in D receives information from the OBFN of other clusters, and the containment control needs to weaken the influence of the OBFN from other clusters; i = 0 means that the i-th OBFN cannot receive information from other OBFNs; these OBFNs are selected as pinning nodes; ω i The estimated value of The adaptive law is for The estimated value of Δ i is a positive constant; definition is the position error and velocity error, η σi represents the expected position and attitude of the i-th OBFN; v ησi represents the expected velocity and angular velocity of the i-th OBFN; the subscript σ i Used to represent the σth i The expected value of the cluster; Υ=diag[ρ i (t)], R = diag[r i ], i = 1, ..., N, N represents the number of OBFNs; ρ i (t) is the performance function, δ i is the adjustment parameter, δ i , Denotes δ i The upper and lower bounds of S(·) is a smooth function.
2. The error-constrained cluster synchronization control method of a submarine flight node according to claim 1, characterized in that: Performance functions Among them, ρ i0 , i∞ and is a positive constant and t is time.
3. The error-constrained cluster synchronization control method of a submarine flight node according to claim 2, characterized in that: Performance function ρ i (t) satisfies the following time-varying constraints: Where t ≥ 0 and δ i is the adjustment parameter, δ i , Denotes δ i The upper and lower bounds of , i=1,...,N.
4. The error-constrained cluster synchronization control method for submarine flight nodes according to claim 3 is characterized in that: Smooth function 5. The error-constrained cluster synchronization control method for submarine flight nodes according to any one of claims 1 to 4, characterized in that: The dynamic model of the seabed flying node OBFN is as follows: Where τ is the control force and torque generated by the propulsion system.
6. According to any one of claims 1 to 4, the error-constrained cluster synchronization control method for submarine flight nodes, the generalized interference F i Using neural network φ i The approach to processing is as follows: in, F i The estimated value, that is, F approximated by neural network i , is the estimated value of the weight matrix, is the neural network input, e i =η i -η σi Represents the tracking error.
7. The error-constrained cluster synchronization control method of a submarine flight node according to claim 6, characterized in that: The φ i (z i ) uses the Gaussian basis function.
8. An error-constrained cluster synchronization control method for submarine flight nodes according to claim 6, characterized in that: Generalized interference where F i is the generalized disturbance composed of model uncertainty and external disturbance, d i Indicates ocean current disturbances, represents the disturbance caused by uncertainty in the OBFN model, ΔM ηi , ΔC RBηi , ΔD ηi , Δg ηi They represent the M generated due to system uncertainty. ηi , C RBηi , D ηi , g ηi The corresponding uncertainty values.
Citation Information
Patent Citations
Track tracking control method for seabed flight nodes
CN109343347A
Method for rapidly tracking and controlling trajectory of unmanned surface vehicle based on fixed time observer
CN112965371A
Intelligent hierarchical control selection method for unmanned aerial vehicle autonomous control system
CN113341712A
Model-free control method for unmanned ship based on dynamic event triggering
CN114756029A
Dynamic network bounded cluster synchronization method and device, and storage medium
CN117675594A
