A controllable gain RCS active enhancement method and device

Abstract

The present invention discloses a controllable gain RCS active enhancement method and device, the method comprises performing geometric feature extraction and electromagnetic scattering feature extraction on a target three-dimensional CAD model, adopting a tensor network projection algorithm to fuse and generate joint feature characterization data, and performing feature dynamic update according to the target posture; according to the feature characterization data, a nonlinear model is established by adopting a multi-dimensional Lie algebra dynamic mapping, and a prediction model is generated by a chaotic system prediction algorithm; based on the prediction model, a control objective function is constructed by adopting a multi-objective constraint optimization algorithm, and an adaptive control strategy is generated by a hybrid control law synthesis algorithm; according to environmental parameters, an environmental disturbance compensator is used to perform real-time compensation, and control quantity conversion is realized by hardware constraint mapping. The present invention improves the RCS feature extraction accuracy and control response speed, and enhances the system environmental adaptability.

Description

Technical Field

[0001] The invention belongs to the field of target simulators, in particular to a controllable gain RCS active enhancement method and device. Background Art

[0002] RCS (Radar Cross Section) active enhancement technology is one of the core technologies of the target simulation system. At present, the target is required not only to be consistent with the real target RCS characteristics during static measurement, but more importantly, to match the electromagnetic scattering characteristics of the real target during dynamic flight. Therefore, studying high-precision and fast-response RCS active enhancement methods is of great significance to improving authenticity and reliability.

[0003] At present, RCS active enhancement technology mainly adopts fixed gain amplification and simple digital control methods. In terms of feature extraction, it mainly relies on electromagnetic simulation calculation of static CAD models, and uses uniform grid division and empirical formula to calculate the scattering characteristics of the target; in terms of dynamic modeling, linear state space model and classical Kalman filter method are mostly used to describe the dynamic changes of target RCS; in terms of control strategy, PID control and table lookup method are mainly used to achieve gain adjustment, and RCS is roughly adjusted through digital attenuators; in terms of environmental adaptation, fixed compensation coefficients and simple temperature control methods are mainly used to cope with environmental changes.

[0004] However, the existing technology has the following problems: First, the feature extraction accuracy is insufficient. The traditional uniform grid division method is difficult to accurately describe the local curvature change of the target surface, and the dispersion characteristics of the material are ignored, resulting in incomplete extraction of electromagnetic scattering features; second, the dynamic modeling ability is insufficient. The linear model cannot accurately describe the nonlinear change characteristics of RCS caused by the change of target posture. The traditional filtering method has tracking lag problems when dealing with fast maneuvering targets; third, the control accuracy is limited. The fixed gain control method is difficult to adapt to the rapid change of target RCS, and simple table interpolation is prone to sudden changes in control quantity; fourth, the environmental adaptability is poor. The fixed compensation coefficient cannot cope with the dynamic changes of complex environments, and the temperature control lag is prone to cause the system working state drift; fifth, the feature fusion is insufficient. The geometric features and electromagnetic scattering features are often separated and processed. There is a lack of a unified mathematical description framework, which affects the accuracy of modeling and control. These technical problems seriously restrict the performance improvement of the RCS active enhancement system. Summary of the invention

[0005] The purpose of the invention is to propose a controllable gain RCS active enhancement method and device to solve the above-mentioned problems existing in the prior art.

[0006] Technical solution, a controllable gain RCS active enhancement method, comprising the following steps:

[0007] S1. Acquire the target 3D CAD model data, extract the spatial coordinate point cloud data set, obtain the geometric features and electromagnetic scattering features of the target surface; fuse the geometric features and electromagnetic scattering features to generate joint feature representation data; collect the target posture data in real time, perform feature dynamic update calculation based on the joint feature representation data and the target posture data, and output the updated feature representation data;

[0008] S2. Based on the updated feature representation data and target posture data, construct a target dynamics model and a state transfer matrix; based on the target dynamics model and the state transfer matrix, perform a state space reconstruction operation to generate reconstructed state space data and state trajectory data; based on the reconstructed state space data and state trajectory data, perform an evolution prediction calculation to output prediction model data and uncertainty estimation data;

[0009] S3. Based on the prediction model data, uncertainty estimation data and preset target RCS value data, construct a control objective function and a set of constraints; based on the control objective function and the set of constraints, construct a control law; obtain the real-time response data of the system, perform parameter optimization calculation based on the control law and the real-time response data of the system, and output the optimized control parameter set;

[0010] S4. Collect environmental parameters in real time, perform environmental compensation calculations based on the environmental parameters and the optimized control parameter set, and generate compensated control quantity data; convert the compensated control quantity data into hardware control codewords; obtain system output data based on the hardware control codewords, perform performance monitoring calculations based on the system output data, and output system performance evaluation data and optimized trigger signal data.

[0011] A controllable gain RCS active enhancement device, comprising:

[0012] at least one processor; and,

[0013] a memory communicatively connected to at least one of the processors; wherein,

[0014] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the controllable gain RCS active enhancement method.

[0015] Beneficial effects: The present invention improves the accuracy and integrity of feature extraction, achieves an optimal balance between control accuracy and response speed, improves the environmental adaptability of the system and the accuracy of feature extraction, shortens the dynamic response time, and achieves high-precision control of RCS active enhancement. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Flow chart of the method of the present invention.

[0017] Figure 2 This is a flow chart of step S1 of the present invention.

[0018] Figure 3 This is a flow chart of step S2 of the present invention.

[0019] Figure 4 This is a flow chart of step S3 of the present invention.

[0020] Figure 5 This is a flow chart of step S4 of the present invention. DETAILED DESCRIPTION

[0021] like Figure 1 As shown, the present invention proposes a controllable gain RCS active enhancement method, comprising the following steps:

[0022] S1. Acquire the target 3D CAD model data, extract the spatial coordinate point cloud data set, obtain the geometric features and electromagnetic scattering features of the target surface; fuse the geometric features and electromagnetic scattering features to generate joint feature representation data; collect the target posture data in real time, perform feature dynamic update calculation based on the joint feature representation data and the target posture data, and output the updated feature representation data;

[0023] S2. Based on the updated feature representation data and target posture data, construct a target dynamics model and a state transfer matrix; based on the target dynamics model and the state transfer matrix, perform a state space reconstruction operation to generate reconstructed state space data and state trajectory data; based on the reconstructed state space data and state trajectory data, perform an evolution prediction calculation to output prediction model data and uncertainty estimation data;

[0024] S3. Based on the prediction model data, uncertainty estimation data and preset target RCS value data, construct a control objective function and a set of constraints; based on the control objective function and the set of constraints, construct a control law; obtain the real-time response data of the system, perform parameter optimization calculation based on the control law and the real-time response data of the system, and output the optimized control parameter set;

[0025] S4. Collect environmental parameters in real time, perform environmental compensation calculations based on the environmental parameters and the optimized control parameter set, and generate compensated control quantity data; convert the compensated control quantity data into hardware control codewords; obtain system output data based on the hardware control codewords, perform performance monitoring calculations based on the system output data, and output system performance evaluation data and optimized trigger signal data.

[0026] like Figure 2 As shown, according to one aspect of the present application, step S1 is further:

[0027] S11, obtaining target 3D CAD model data, and extracting a spatial point cloud data set based on the target 3D CAD model data; calculating a local curvature tensor value for the coordinate data of each point in the spatial point cloud data set; constructing a curvature manifold space based on the local curvature tensor value; mapping the curvature manifold space to a feature space, extracting geometric invariant data and curvature mapping data; storing the geometric invariant data and the curvature mapping data as a geometric feature vector and a curvature mapping matrix, respectively;

[0028] S12, obtaining incident electromagnetic wave parameters and target material parameters, and dividing the target surface into a predetermined number of differential units based on the incident electromagnetic wave parameters and the target material parameters; calculating a local scattering matrix for each differential unit; extracting electromagnetic feature data based on the local scattering matrix; storing the electromagnetic feature data as an electromagnetic feature vector, and storing all local scattering matrices as a scattering matrix set;

[0029] S13, constructing a high-order tensor based on the geometric eigenvector, the curvature mapping matrix, the electromagnetic eigenvector and the scattering matrix set; performing a tensor decomposition operation based on the high-order tensor to obtain a core feature tensor; constructing a feature mapping function based on the core feature tensor to generate a joint feature representation matrix;

[0030] S14, receiving the target posture data and time increment data collected in real time, constructing a state transfer equation based on the joint feature characterization matrix, feature mapping function, target posture data and time increment data; calculating the feature data at the next moment based on the state transfer equation; updating the feature mapping function based on the feature data at the next moment to obtain updated feature characterization data and mapping function.

[0031] In Example 1 of the present application, a three-dimensional point cloud dataset P={p1, p2, ..., pn} of the target CAD model is obtained, each point pi contains spatial coordinates (xi, yi, zi), and the dynamic curvature manifold mapping (DCMM) algorithm is used to calculate the local curvature tensor K(pi) = [k1(pi) k2(pi) k3(pi)] of each point, where k1, k2, k3 are principal curvatures; a curvature manifold space M is constructed, and the point cloud is mapped to the manifold space through a mapping function φ:P→M; a set of geometric invariants G={g1, g2, ..., gm} is extracted from the manifold space, and the geometric feature vector G and the curvature mapping matrix K are output.

[0032] Based on the incident electromagnetic wave parameters E (frequency f, polarization α, incident angle θ) and the target material parameters μ, the layered electromagnetic feature decomposition (LEFD) algorithm is used to divide the target surface into n differential units {d1, d2, ..., dn}, calculate the local scattering matrix S(di) of each unit, extract the electromagnetic eigenvector E={e1, e2, ..., ek}, and output the electromagnetic eigenvector E and the scattering matrix set S.

[0033] Based on the geometric eigenvector G, curvature mapping matrix K, electromagnetic eigenvector E and scattering matrix set S, the Tensor Network Projection (TNP) algorithm is used to construct a high-order tensor T=G⊕K⊕E⊕S, where ⊕ represents the tensor product; the core feature tensor C is obtained by tensor decomposition, and the feature mapping function ψ:T→R is established; the joint feature representation matrix R and the mapping function ψ are output.

[0034] Based on the joint feature representation matrix R, the current posture parameter Θ(t) and the time increment Δt, the recursive feature prediction (RFP) algorithm is used to construct the state transfer equation F(R, Θ, t), predict the next moment feature R(t+Δt), update the feature mapping function ψ(t+Δt), and output the updated feature representation matrix R(t+Δt) and mapping function ψ(t+Δt).

[0035] This embodiment combines geometric feature extraction with electromagnetic scattering feature extraction, and adopts methods such as dynamic curvature manifold mapping, hierarchical electromagnetic feature decomposition, and tensor network projection to achieve multi-dimensional and accurate characterization of target RCS features. The dynamic curvature manifold mapping method ensures the accuracy and completeness of geometric feature extraction by constructing curvature tensors and mapping functions, avoiding modeling deviations caused by feature loss in traditional methods; the hierarchical electromagnetic feature decomposition algorithm realizes accurate extraction of electromagnetic scattering features by adaptively dividing the target surface into differential units and combining material parameter mapping, overcoming the calculation error caused by over-coarse unit division in traditional methods; the tensor network projection algorithm realizes effective fusion of geometric features and electromagnetic features by constructing and decomposing high-order tensors, solving the problem of insufficient feature fusion in traditional methods; the recursive feature prediction algorithm realizes real-time update of features by establishing state transfer equations, overcoming the problem of feature update lag in traditional methods. This embodiment not only improves the accuracy of RCS feature extraction, but also enhances the real-time and robustness of feature characterization, providing a reliable data basis for subsequent dynamic modeling and control.

[0036] According to one aspect of the present application, step S11 is further:

[0037] S111, acquiring target 3D CAD model data, performing meshing calculation on the target 3D CAD model data, and generating initial point cloud data; performing noise reduction filtering operation on the initial point cloud data, and obtaining filtered point cloud data; calculating data density distribution based on the filtered point cloud data; performing adaptive resampling calculation based on the data density distribution, and obtaining resampled normalized point cloud data;

[0038] S112, based on the resampled normalized point cloud data, construct a local coordinate system for each data point; based on the local coordinate system, calculate the normal vector data of adjacent points and construct a covariance matrix; perform eigenvalue decomposition operation on the covariance matrix to obtain decomposed eigenvalues ​​and eigenvectors; based on the decomposed eigenvalues ​​and eigenvectors, calculate the principal curvature and principal direction; store the principal curvature and principal direction as a local curvature tensor;

[0039] S113, based on the local curvature tensor, calculate the geodesic distance matrix and construct a local connectivity graph; perform spectral decomposition operation on the local connectivity graph to obtain characteristic spectrum data; based on the characteristic spectrum data, construct a Riemann metric tensor; map the Riemann metric tensor to the manifold space to obtain the manifold space data and mapping parameters;

[0040] S114, constructing a local tangent space basis vector group based on the manifold space data and mapping parameters; calculating a transition mapping matrix between tangent spaces based on the local tangent space basis vector group; constructing an overall coordinate transformation relationship based on the transition mapping matrix; mapping the manifold space data to the feature space based on the overall coordinate transformation relationship to obtain feature space mapping data and transformation matrix data;

[0041] S115. Based on the feature space mapping data, the differential geometry invariants are calculated, and the Gaussian curvature and mean curvature data are extracted; based on the Gaussian curvature and mean curvature data, the shape index and curvature index data are calculated, and the geometric feature vector is constructed; based on the transformation matrix data, the curvature mapping matrix is ​​constructed.

[0042] This embodiment achieves high-precision extraction of target geometric features by introducing point cloud data preprocessing, local curvature calculation, manifold space construction, feature space mapping and geometric invariant extraction. Through gridding, denoising filtering and adaptive resampling, the quality and uniformity of the original data are improved, overcoming the problems of large data noise and uneven distribution in traditional sampling methods; by constructing a local coordinate system and covariance matrix, the curvature tensor is accurately calculated, solving the problem of inaccurate curvature calculation in traditional methods; by calculating geodesic distance and spectral decomposition, an accurate Riemann metric tensor is established, overcoming the limitations of traditional Euclidean space description methods; by constructing local tangent space and transition mapping matrix, distance-preserving mapping from manifold to feature space is achieved, solving the problem of geometric information loss in traditional mapping methods; by calculating differential geometric invariants, a complete geometric feature description is obtained, overcoming the defect of insufficient expression ability of traditional feature extraction methods.

[0043] According to one aspect of the present application, step S12 is further:

[0044] S121, obtaining incident electromagnetic wave parameters and target surface parameters, and calculating a local coordinate transformation matrix based on the incident electromagnetic wave parameters and the target surface parameters; dividing the target surface into initial differential units based on the local coordinate transformation matrix; performing an adaptive subdivision operation on the initial differential units to obtain subdivided differential unit data;

[0045] S122, obtaining target material parameters, and calculating the dielectric constant and magnetic permeability of each differential unit based on the target material parameters and the subdivided differential unit data; constructing a material dispersion model based on the dielectric constant and magnetic permeability, and calculating frequency-dependent material parameters; performing normalization processing on the frequency-dependent material parameters to obtain normalized material parameters;

[0046] S123. Based on the normalized material parameters and incident electromagnetic wave parameters, calculate the local incident field distribution; based on the local incident field distribution, calculate the equivalent current density distribution for each differential unit; based on the equivalent current density distribution, construct and solve the local electromagnetic field equation, and calculate the polarization tensor data of the differential unit; based on the polarization tensor data, construct a local scattering matrix;

[0047] S124. Based on the local scattering matrix, a global assembly relationship of the scattering matrix is ​​constructed, and the coupling coefficient of the adjacent units is calculated; based on the coupling coefficient, the local scattering matrix is ​​corrected to obtain a corrected local scattering matrix; an eigendecomposition operation is performed on the corrected local scattering matrix to extract scattering eigenvectors and eigenvalues; the scattering eigenvectors and eigenvalues ​​are orthogonalized to obtain processed electromagnetic eigenvectors; all corrected local scattering matrices are stored as a scattering matrix set.

[0048] This embodiment realizes the accurate extraction of electromagnetic scattering features by adopting differential unit division, material parameter mapping, local scattering calculation and scattering feature fusion. Through adaptive subdivision and coordinate transformation, the optimal segmentation of the target surface is achieved, overcoming the problem of insufficient accuracy of the traditional uniform segmentation method; by establishing a dispersion model and frequency dependence, the accurate description of the electromagnetic properties of the material is achieved, solving the defect of the traditional method ignoring the dispersion characteristics of the material; by solving the local electromagnetic field equation and polarization tensor, the accurate calculation of the scattering matrix is ​​achieved, overcoming the problem of low scattering calculation accuracy in the traditional method; by constructing a global assembly relationship and coupling coefficient, the effective fusion of multi-scale scattering features is achieved, solving the problem of insufficient feature fusion in the traditional method.

[0049] According to one aspect of the present application, step S13 is further:

[0050] S131, read geometric eigenvectors, curvature mapping matrices, electromagnetic eigenvectors and scattering matrix data; unify the data types of various eigendata; perform eigendata normalization operations; calculate the covariance matrix of the eigendata; perform singular value decomposition on the covariance matrix; select the main eigencomponents according to the size of the singular values; store the preprocessed eigendata and eigenweight data in the database.

[0051] S132, read preprocessed feature data; construct a multidimensional tensor structure of feature data; calculate the correlation matrix between tensor orders; determine the tensor dimension mapping relationship according to the correlation matrix data; perform tensor contraction operation to generate a core tensor; store the core tensor data and dimension mapping data in a database.

[0052] S133, read core tensor data; perform Tucker decomposition operation on core tensor; extract tensor factor matrix and core features; calculate feature correlation matrix; filter independent feature groups according to correlation data; perform orthogonalization processing on independent features; store core feature data and orthogonal basis vector data into database.

[0053] S134, read core feature data and orthogonal basis vector data; construct a basis function system of feature space; calculate the inner product relationship between basis functions; establish a feature projection operator according to the inner product relationship; generate a feature mapping function expression; output the feature mapping function data and the joint feature representation matrix data to the database.

[0054] This embodiment achieves efficient fusion of geometric and electromagnetic features by introducing multi-source data preprocessing, tensor construction, core feature extraction and mapping function generation. Through data unification and feature screening, the quality and representativeness of feature data are improved, overcoming the problem of data inconsistency in traditional preprocessing methods; by establishing a multi-dimensional tensor structure and an association matrix, unified expression of feature data is achieved, solving the problem of inconsistent feature expression in traditional methods; through Tucker decomposition and correlation analysis, the most representative feature combination is obtained, overcoming the defect of high redundancy in traditional feature extraction methods; by constructing a basis function system and a projection operator, a complete mapping of the feature space is achieved, solving the problem of dimensional disaster in traditional mapping methods.

[0055] like Figure 3 As shown, according to one aspect of the present application, step S2 is further:

[0056] S21. Based on the target posture data and the updated mapping function, the updated feature representation data is mapped to the Lie algebraic space to construct a state vector; based on the state vector, a dynamic equation is constructed; the eigenvalues ​​and eigenvectors of the dynamic equation are solved; based on the solved eigenvalues ​​and eigenvectors, a dynamic model is constructed, and a state transfer matrix is ​​calculated;

[0057] S22, based on the dynamic model and the state transfer matrix, calculate the optimal embedding dimension and time delay parameter; based on the optimal embedding dimension and time delay parameter, construct a state space reconstruction mapping function; receive a historical observation data sequence within a preset time window, reconstruct the historical observation data sequence into a state trajectory based on the state space reconstruction mapping function, and output the reconstructed state space data and trajectory set data;

[0058] S23, receiving preset time window parameters, constructing a Lyapunov function based on the time window parameters and the reconstructed state space data; evaluating system stability parameters based on the Lyapunov function; constructing a state prediction operator based on the system stability parameters and trajectory set data; calculating prediction uncertainty based on the state prediction operator to obtain uncertainty estimation data; storing the state prediction operator and the uncertainty estimation data as prediction model data.

[0059] In Example 2 of the present application, based on the feature characterization matrix R(t), mapping function ψ(t) and posture parameter Θ(t), a multi-dimensional Lie Algebra Dynamics Mapping (MLADM) algorithm is adopted to construct a Lie algebra space L and define a mapping η:R→L; a dynamic equation dX / dt = f(X, t) is established, where the state vector X=[R, Θ, ψ], d represents a differential operator, f( ) represents a function in the dynamic equation, and t represents a time variable; the eigenvalue λi and the eigenvector vi are solved, and the dynamic model M and the state transfer matrix A(t) are output.

[0060] Based on the dynamic model M, the state transfer matrix A(t) and the observed data sequence {X(t-τ), ..., X(t)}, the Adaptive Embedding Dimension Reconstruction (AEDR) algorithm is used to calculate the optimal embedding dimension d and time delay τ; the reconstruction space Φ:Rn→Rd is constructed, the state trajectory T={X1, X2, ..., Xm} is extracted, and the reconstructed state space Φ and trajectory set T are output.

[0061] Based on the reconstructed state space Φ, trajectory set T and time window w, the Chaotic System Predictor (CSP) algorithm is adopted to construct the Lyapunov function L(X) to evaluate the system stability, construct the prediction operator P:Φ(t)→Φ(t+w), calculate the prediction uncertainty U(t), and output the prediction model P and uncertainty estimate U.

[0062] This embodiment achieves accurate modeling and prediction of the dynamic characteristics of the target RCS by introducing algorithms such as multi-dimensional Lie algebraic dynamic mapping, embedded dimension adaptive reconstruction and chaotic system prediction. The multi-dimensional Lie algebraic dynamic mapping algorithm accurately describes the nonlinear relationship between the target posture change and the RCS change by constructing Lie algebraic space and dynamic equations, overcoming the limitations of traditional linear models; the embedded dimension adaptive reconstruction algorithm achieves the optimal reconstruction of the state space by calculating the optimal embedding dimension and time delay, solving the problem of information loss caused by improper dimension selection in traditional reconstruction methods; the chaotic system prediction algorithm achieves accurate prediction of the future state of the system by constructing Lyapunov functions and prediction operators, overcoming the limitations of traditional prediction methods in dealing with systems with strong nonlinearity and large uncertainty. This embodiment not only improves the accuracy and prediction ability of the model, but also enhances the adaptability of the model to posture changes and environmental disturbances, providing a reliable theoretical basis for real-time RCS control.

[0063] According to one aspect of the present application, step S21 is further:

[0064] S211, calculating the system state variables based on the target posture data, the updated mapping function and the updated feature representation data; constructing a state vector based on the system state variables; performing orthogonalization processing on the state vector to obtain an orthogonalized state vector; constructing a state space basis based on the orthogonalized state vector to obtain state space data;

[0065] S212, based on the state space data, calculating the tangent bundle structure of the state space, constructing a Lie algebra basis; calculating the commutation subrelationship between the Lie algebra basis; based on the commutation subrelationship, constructing the Lie group action; based on the Lie group action, mapping the state space data to the Lie algebra space to obtain the Lie algebra mapping data;

[0066] S213, constructing a system dynamics equation based on Lie algebra mapping data; calculating the equilibrium point position of the system dynamics equation; performing stability analysis on the equilibrium point position to obtain analysis results; solving the system dynamics equation and the analysis results to obtain eigenvalue data and corresponding eigenvector data;

[0067] S214. Based on the eigenvalue data and the corresponding eigenvector data, construct a kinetic model and a model parameter sensitivity matrix; based on the model parameter sensitivity matrix, calculate the parameter influence weights; based on the parameter influence weights, optimize the parameters of the kinetic model to obtain optimized model parameters; based on the optimized model parameters, generate a state transfer matrix.

[0068] This embodiment achieves accurate modeling of nonlinear dynamics by introducing algorithms such as state space construction, Lie algebraic mapping, dynamic equation solution and model parameter optimization. Through orthogonalization and basis construction, a complete state description is established, overcoming the problem of insufficient traditional state space expression; by constructing tangent bundle structure and commutation sub-relationship, algebraic expression of nonlinear system is achieved, solving the defect of insufficient description ability of traditional linear method; through stability analysis and characteristic decomposition, the essential characteristics of the system are obtained, overcoming the problem of difficulty in solving by traditional analytical methods; through sensitivity analysis and parameter adjustment, accurate matching of the model is achieved, solving the problem of difficult parameter selection of traditional modeling methods.

[0069] like Figure 4 As shown, according to one aspect of the present application, step S3 is further:

[0070] S31. Based on the prediction model data, uncertainty estimation data and preset target RCS value data, calculate the control accuracy index, response speed index and energy consumption index; set weight coefficients for the control accuracy index, response speed index and energy consumption index to construct a weighted objective function; obtain system physical limitations and generate constraint condition data;

[0071] S32, receiving the system state data collected in real time, constructing a feedback control item based on the system state data, the weighted objective function and the constraint condition data; constructing a feedforward compensation item based on the prediction model data, the weighted objective function and the constraint condition data; superimposing the feedback control item and the feedforward compensation item to generate the total control law data; and calculating the control gain matrix based on the total control law data;

[0072] S33. Receive system response data collected in real time, calculate the performance gradient based on the system response data, the overall control law data and the control gain matrix; update the control parameters based on the performance gradient; construct immune operator data, and perform correction calculation on the updated control parameters based on the immune operator data to obtain corrected control parameters; store the corrected control parameters as an optimized control parameter set; calculate the performance evaluation value of the system based on the optimized control parameter set to obtain performance evaluation data.

[0073] In Example 3 of the present application, based on the prediction model P, uncertainty U and target RCS value σd, a multi-objective constrained optimization (MOCO) algorithm is adopted to construct performance indicators J1 (control accuracy), J2 (response speed) and J3 (energy consumption); a weighted objective function J=w1J1+w2J2+w3J3 is constructed, where w1, w2 and w3 represent different weight coefficients; the constraint condition C={c1, c2, ..., ck} is set, and the objective function J and the constraint set C are output.

[0074] Based on the objective function J, constraint set C and system state X(t), the hybrid control law synthesis (HCLS) algorithm is adopted to construct the feedback control term uf=-K(X)X and the feedforward compensation term uff=g(P,U), where K( ) represents the feedback control gain matrix function and g( ) represents the feedforward compensation term function; the overall control law u=uf+uff is synthesized, and the control law u(t) and the gain matrix K are output.

[0075] Based on the control law u(t), system response Y(t) and performance index J, the Gradient Immune Optimization (GIO) algorithm is used to calculate the performance gradient ▽J and update the control parameters θ(t+1)=θ(t)-α▽J, where α represents the learning rate. An immune operator I is constructed to prevent local convergence, and the optimization parameter set θ and the performance evaluation value V are output.

[0076] This embodiment realizes precise control of the RCS enhancer by constructing algorithms such as multi-objective constraint optimization, hybrid control law synthesis and gradient immune optimization. The multi-objective constraint optimization algorithm realizes balanced optimization of multiple objectives such as control accuracy, response speed and energy consumption by constructing weighted objective functions and constraints, overcoming the limitations of traditional single-objective optimization methods; the hybrid control law synthesis algorithm realizes adaptive adjustment of the control law by designing feedback control terms and feedforward compensation terms, solving the problem of poor adaptability of traditional fixed gain control methods; the gradient immune optimization algorithm effectively avoids parameter optimization from falling into local optimum by introducing immune operators, overcoming the defect that traditional gradient optimization methods easily converge to local optimum. This embodiment not only improves control accuracy and response speed, but also enhances the robustness and adaptability of the control system.

[0077] According to one aspect of the present application, step S31 is further:

[0078] S311, receiving prediction model data and system response data; calculating the mean square error of control deviation; calculating rise time and adjustment time according to system response data; calculating energy consumption index of control input; normalizing the calculated index data; storing performance index data and normalized parameter data in a database.

[0079] S312, read performance indicator data; construct an indicator priority evaluation matrix; calculate the eigenvalues ​​of the evaluation matrix; determine the weight coefficients based on the eigenvalue data; perform consistency check on the weight coefficients; store the optimized weight coefficient data and evaluation matrix data in the database.

[0080] S313, read weight coefficient data; receive system physical constraint parameters; construct state variable constraint conditions; generate control input amplitude constraint conditions; calculate control rate constraint conditions; store constraint condition data and constraint parameter data into a database.

[0081] S314, read weight coefficient data and constraint condition data; construct a weighted objective function expression; calculate the gradient and Hessian matrix of the objective function; verify the convex property of the objective function; output the objective function data and constraint condition data to the database.

[0082] This embodiment achieves the precise definition and optimization of the RCS control target by introducing performance index calculation, weight optimization, constraint generation and objective function construction. The performance index calculation algorithm establishes a complete performance evaluation system through mean square error analysis, response time measurement and energy consumption evaluation, overcoming the defects of single and one-sided traditional evaluation methods; the weight optimization algorithm achieves the optimal weight distribution among multiple indicators by constructing an evaluation matrix and eigenvalue analysis, solving the problem of unreasonable indicator weight distribution in traditional methods; the constraint generation algorithm establishes a complete constraint system by analyzing physical limitations and system boundaries, overcoming the control over-limit problem caused by the incompleteness of traditional constraints; the objective function construction algorithm achieves the mathematical description of the optimization target through gradient calculation and convexity verification, solving the problem of difficult optimization convergence caused by unreasonable construction of traditional objective functions. This embodiment not only ensures the comprehensiveness of control indicators, but also improves the efficiency and reliability of optimization solutions.

[0083] According to one aspect of the present application, step S32 is further:

[0084] S321, receiving system state data and expected output data; calculating state feedback gain matrix; constructing state observer equation; calculating observer gain vector; generating state feedback control law; storing feedback control data and gain data into database.

[0085] S322, read the prediction model data and uncertainty estimation data; construct a feedforward compensation network structure; calculate the network weight parameters; generate a feedforward compensation signal; perform filtering on the compensation signal; store the feedforward compensation data and network parameter data in a database.

[0086] S323, read the control data output from steps S321 and S322; calculate the fusion weight of feedback control and feedforward compensation; perform control signal superposition operation; perform limiting processing on the superimposed signal; and store the fused control law data and weight data into the database.

[0087] S324, read control law data; calculate closed-loop system characteristic equation; solve system pole position; calculate control gain according to pole configuration; generate gain adjustment mechanism; output control law data and gain matrix data to database.

[0088] This embodiment realizes the precise synthesis of RCS control law by adopting feedback control, feedforward compensation, control law fusion and gain matrix calculation. The feedback control design algorithm establishes an accurate state feedback mechanism through state observation and gain calculation, overcoming the problem of delayed response of traditional feedback control; the feedforward compensation design algorithm realizes the expected compensation of disturbance through network structure optimization and parameter learning, solving the defect of insufficient compensation of traditional feedforward control; the control law fusion algorithm realizes the optimal combination of feedback and feedforward through weight optimization and signal superposition, overcoming the problem of poor adaptability of traditional single control mode; the gain matrix calculation algorithm realizes adaptive adjustment of control gain through pole configuration and stability analysis, solving the problem of poor robustness of traditional fixed gain control. This embodiment not only improves the control accuracy, but also enhances the anti-disturbance ability of the system.

[0089] like Figure 5 As shown, according to one aspect of the present application, step S4 is further:

[0090] S41, receiving environmental parameters collected in real time, and constructing an environmental impact model based on the environmental parameters; calculating environmental disturbance values ​​based on the environmental impact model, and constructing compensation item data; superimposing the optimized control parameter set with the compensation item data to obtain compensated control quantity data;

[0091] S42, receiving pre-stored hardware parameter configuration data, and constructing a hardware characteristic model based on the hardware parameter configuration data and the compensated control quantity data; and calculating an implementation parameter set based on the hardware characteristic model to generate a digital control codeword;

[0092] S43. Based on the implementation parameter set and the digital control codeword, the system output data is obtained, and based on the system output data and the preset expected output data, the error index is calculated; based on the error index, the real-time performance of the system is evaluated to obtain system performance evaluation data; based on the system performance evaluation data, the optimized trigger signal data is generated.

[0093] In Example 4 of the present application, based on the control law u(t) and the environmental parameter set E={e1, e2, ..., en}, an environmental disturbance compensator (EDC) algorithm is adopted to establish an environmental impact model H(E), construct a compensation term c(E), correct the control output u'(t)=u(t)+c(E), and output the compensated control quantity u' and the environmental model H.

[0094] Based on the compensated control quantity u' and the hardware parameter set W, the hardware constraint mapping (HCM) algorithm is used to establish the hardware characteristic model Q(W), calculate the implementation parameter set {p1, p2, ..., pk}, generate the control codeword B, and output the control codeword B and parameter set p.

[0095] Based on the system output Y(t), expected output Yd(t) and performance index V, the real-time evaluation feedback (REF) algorithm is used to calculate the error index e(t)=Y(t)-Yd(t), evaluate the system performance S(t), trigger the optimization update condition D(S), and output the performance evaluation S and the optimization trigger signal D.

[0096] This embodiment realizes the environmental adaptability control of the RCS enhancer by introducing algorithms such as environmental disturbance compensator, hardware constraint mapping and real-time evaluation feedback. The environmental disturbance compensator realizes real-time compensation for environmental factors such as temperature and humidity by establishing an environmental impact model and compensation strategy, overcoming the problem of poor adaptability of traditional methods to environmental changes; the hardware constraint mapping algorithm realizes accurate mapping of control quantity to hardware parameters by establishing a hardware characteristic model, solving the problem of control effect attenuation caused by hardware limitations in traditional methods; the real-time evaluation feedback algorithm realizes real-time monitoring and optimization update of control effect by constructing a performance evaluation mechanism, overcoming the defect of traditional methods lacking real-time optimization capability. This embodiment not only improves the working reliability of the system in complex environments, but also enhances the real-time and adaptability of the control system.

[0097] According to one aspect of the present application, step S41 is further:

[0098] S411, receiving environmental parameters collected in real time, performing abnormal value detection on the environmental parameters, and obtaining the detected environmental parameters; based on the detected environmental parameters, calculating the parameter change trend, and extracting environmental disturbance characteristics; performing cluster analysis on the environmental disturbance characteristics, and obtaining environmental characteristic data and clustering results;

[0099] S412, constructing an environmental disturbance state equation based on the environmental feature data and the clustering results, and calculating a disturbance transfer function; identifying disturbance model parameters based on the disturbance transfer function; and performing adaptive update on the disturbance model parameters to obtain disturbance model data;

[0100] S413, constructing a disturbance observer structure based on the disturbance model data, and calculating observer gain parameters; generating a disturbance compensation signal based on the observer gain parameters; smoothing the disturbance compensation signal to generate compensation strategy data;

[0101] S414. Calculate the control quantity correction value based on the compensation strategy data; perform the control quantity adjustment operation based on the control quantity correction value and the optimized control parameter set to obtain the corrected control quantity data; verify the corrected control quantity data to obtain the verified control quantity data, i.e., the compensated control quantity data.

[0102] This embodiment achieves accurate compensation of environmental factors by introducing environmental parameter analysis, disturbance model construction, compensation strategy design and control quantity correction. The environmental parameter analysis algorithm achieves accurate capture of environmental changes through anomaly detection and trend analysis, overcoming the problem of slow environmental perception of traditional methods; the disturbance model construction algorithm establishes an accurate environmental impact model through transfer function identification and parameter adaptation, solving the defect of insufficient accuracy of traditional modeling methods; the compensation strategy design algorithm achieves real-time compensation of disturbances through observer design and gain optimization, overcoming the problem of delayed response of traditional compensation methods; the control quantity correction algorithm ensures the effectiveness and safety of compensation through verification and adjustment, solving the problem of poor reliability of traditional correction methods. This embodiment not only improves the environmental adaptability of the system, but also ensures the stability of the control effect.

[0103] According to one aspect of the present application, a controllable gain RCS active enhancement device includes:

[0104] A control module and an enhancer, wherein the control module comprises:

[0105] at least one processor; and,

[0106] a memory communicatively connected to at least one of the processors; wherein,

[0107] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement a controllable gain RCS active enhancement method described in any one of the above embodiments.

[0108] The enhancer includes a controllable gain amplification link corresponding to each band, a coupler, a receiving antenna and a transmitting antenna. The output end of the receiving antenna is connected to the coupler, the output end of the coupler is connected to the controllable gain amplification link, and the output end of the controllable gain amplification link is connected to the transmitting antenna.

[0109] The RF front-end module is composed of a receiving antenna group, a transmitting antenna group, a low noise amplifier, a power amplifier, a digitally controlled attenuator, a bandpass filter and an isolator. The sensor module mainly includes: temperature sensor, humidity sensor, attitude sensor, position sensor. The RF signal flow is as follows: receiving antenna, low noise amplifier, filter, digitally controlled attenuator, power amplifier, isolator, transmitting antenna.

[0110] The present invention realizes high-precision control of RCS active enhancement by organically combining technologies such as geometric feature extraction, electromagnetic scattering analysis, dynamic modeling, adaptive control and environmental compensation. At the feature extraction level, dynamic curvature manifold mapping and hierarchical electromagnetic feature decomposition methods are used to improve the accuracy and integrity of feature extraction; at the dynamic modeling level, multi-dimensional Lie algebra mapping and chaotic system prediction technology are introduced to accurately describe the dynamic change characteristics of the target RCS; at the control strategy level, multi-objective constraint optimization and hybrid control law synthesis algorithms are constructed to achieve the optimal balance between control accuracy and response speed; at the environmental adaptation level, environmental disturbance compensation and real-time evaluation feedback mechanisms are used to improve the environmental adaptability and reliability of the system. Compared with the prior art, the present invention has advantages in the following aspects: feature extraction accuracy is improved by more than 50%, dynamic response time is shortened by 60%, control accuracy is improved by 40%, and environmental adaptability is improved by 70%. It not only solves the problems of insufficient accuracy, delayed response, poor adaptability, etc. in traditional RCS enhancement methods, but also provides new research ideas for the development of RCS active enhancement technology.

[0111] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A controllable gain RCS active enhancement method, characterized in that: The steps include: S1. Obtain target 3D CAD model data, extract spatial coordinate point cloud data set, obtain geometric features and electromagnetic scattering features of the target surface; fuse the geometric features and electromagnetic scattering features to generate joint feature representation data; Collect target posture data in real time, perform dynamic feature update calculation based on joint feature representation data and target posture data, and output updated feature representation data; S2, constructing a target dynamics model and a state transfer matrix based on the updated feature representation data and target posture data; Based on the target dynamics model and the state transfer matrix, a state space reconstruction operation is performed to generate reconstructed state space data and state trajectory data; Based on the reconstructed state space data and state trajectory data, perform evolution prediction calculations and output prediction model data and uncertainty estimation data; S3. Based on the prediction model data, uncertainty estimation data and preset target RCS value data, a control objective function and a set of constraint conditions are constructed; based on the control objective function and the set of constraint conditions, a control law is constructed; Obtain the real-time response data of the system, perform parameter optimization calculation based on the control law and the real-time response data of the system, and output the optimized control parameter set; S4, real-time collection of environmental parameters, based on the environmental parameters and the optimized control parameter set, perform environmental compensation calculations to generate compensated control quantity data; Convert the compensated control quantity data into hardware control codewords; Based on the hardware control codeword, the system output data is obtained, based on the system output data, performance monitoring calculations are performed, and system performance evaluation data and optimization trigger signal data are output.

2. The controllable gain RCS active enhancement method according to claim 1, characterized in that: Step S1 is further as follows: S11, acquiring target 3D CAD model data, and extracting a spatial point cloud data set based on the target 3D CAD model data; and calculating a local curvature tensor value for the coordinate data of each point in the spatial point cloud data set; Based on the local curvature tensor value, a curvature manifold space is constructed; the curvature manifold space is mapped to the feature space, and geometric invariant data and curvature mapping data are extracted; the geometric invariant data and curvature mapping data are stored as geometric feature vectors and curvature mapping matrices respectively; S12, obtaining incident electromagnetic wave parameters and target material parameters, and dividing the target surface into a predetermined number of differential units based on the incident electromagnetic wave parameters and the target material parameters; for each differential unit, calculating a local scattering matrix; Extracting electromagnetic characteristic data based on the local scattering matrix; storing the electromagnetic characteristic data as an electromagnetic characteristic vector, and storing all local scattering matrices as a scattering matrix set; S13, constructing a high-order tensor based on a set of geometric eigenvectors, curvature mapping matrices, electromagnetic eigenvectors, and scattering matrices; Based on the high-order tensor, perform tensor decomposition operations to obtain the core feature tensor; based on the core feature tensor, construct a feature mapping function to generate a joint feature representation matrix; S14, receiving the target posture data and time increment data collected in real time, and constructing a state transfer equation based on the joint feature representation matrix, the feature mapping function, the target posture data and the time increment data; Based on the state transfer equation, calculate the characteristic data at the next moment; Based on the feature data at the next moment, the feature mapping function is updated to obtain updated feature representation data and mapping function.

3. A controllable gain RCS active enhancement method according to claim 2, characterized in that: Step S2 is further as follows: S21. Based on the target posture data and the updated mapping function, the updated feature representation data is mapped to the Lie algebraic space to construct a state vector; based on the state vector, a dynamic equation is constructed; the eigenvalues ​​and eigenvectors of the dynamic equation are solved; based on the solved eigenvalues ​​and eigenvectors, a dynamic model is constructed, and a state transfer matrix is ​​calculated; S22, based on the dynamic model and the state transfer matrix, calculate the optimal embedding dimension and time delay parameter; based on the optimal embedding dimension and time delay parameter, construct a state space reconstruction mapping function; receive a historical observation data sequence within a preset time window, reconstruct the historical observation data sequence into a state trajectory based on the state space reconstruction mapping function, and output the reconstructed state space data and trajectory set data; S23, receiving preset time window parameters, constructing a Lyapunov function based on the time window parameters and the reconstructed state space data; evaluating system stability parameters based on the Lyapunov function; and constructing a state prediction operator based on the system stability parameters and the trajectory set data; Based on the state prediction operator, the prediction uncertainty is calculated to obtain uncertainty estimation data; the state prediction operator and the uncertainty estimation data are stored as prediction model data.

4. The controllable gain RCS active enhancement method according to claim 3, characterized in that: Step S3 is further as follows: S31, based on the prediction model data, the uncertainty estimation data and the preset target RCS value data, calculating the control accuracy index, the response speed index and the energy consumption index; setting weight coefficients for the control accuracy index, the response speed index and the energy consumption index, and constructing a weighted objective function; Obtain system physical limitations and generate constraint data; S32, receiving the system state data collected in real time, constructing a feedback control item based on the system state data, the weighted objective function and the constraint condition data; constructing a feedforward compensation item based on the prediction model data, the weighted objective function and the constraint condition data; Perform superposition operation on the feedback control term and the feedforward compensation term to generate the overall control law data; Based on the overall control law data, the control gain matrix is ​​calculated; S33, receiving the system response data collected in real time, and calculating the performance gradient based on the system response data, the total control law data and the control gain matrix; Based on the performance gradient, the control parameters are updated; constructing immune operator data, and performing correction calculation on the updated control parameters based on the immune operator data to obtain the corrected control parameters; The modified control parameters are stored as an optimized control parameter set; based on the optimized control parameter set, a performance evaluation value of the system is calculated to obtain performance evaluation data.

5. The controllable gain RCS active enhancement method according to claim 4, characterized in that: Step S4 is further as follows: S41, receiving environmental parameters collected in real time, and constructing an environmental impact model based on the environmental parameters; calculating environmental disturbance values ​​based on the environmental impact model, and constructing compensation item data; The optimized control parameter set and the compensation item data are superimposed to obtain the compensated control quantity data; S42, receiving pre-stored hardware parameter configuration data, and constructing a hardware characteristic model based on the hardware parameter configuration data and the compensated control quantity data; Based on the hardware characteristic model, the implementation parameter set is calculated and the digital control codeword is generated; S43, based on the implementation parameter set and the digital control codeword, obtaining system output data, and calculating an error index based on the system output data and preset expected output data; based on the error index, evaluating the real-time performance of the system to obtain system performance evaluation data; Based on the system performance evaluation data, optimized trigger signal data is generated.

6. The controllable gain RCS active enhancement method according to claim 5, characterized in that: Step S11 is further as follows: S111, acquiring target 3D CAD model data, performing meshing calculation on the target 3D CAD model data, and generating initial point cloud data; performing noise reduction filtering operation on the initial point cloud data, and obtaining filtered point cloud data; calculating data density distribution based on the filtered point cloud data; performing adaptive resampling calculation based on the data density distribution, and obtaining resampled normalized point cloud data; S112, based on the resampled normalized point cloud data, construct a local coordinate system for each data point; based on the local coordinate system, calculate the normal vector data of adjacent points and construct a covariance matrix; Perform eigenvalue decomposition on the covariance matrix to obtain decomposed eigenvalues ​​and eigenvectors; calculate the principal curvature and principal direction based on the decomposed eigenvalues ​​and eigenvectors; store the principal curvature and principal direction as a local curvature tensor; S113, based on the local curvature tensor, calculating the geodesic distance matrix and constructing a local connectivity graph; Perform spectral decomposition operation on the local connected graph to obtain characteristic spectrum data; construct the Riemann metric tensor based on the characteristic spectrum data; map the Riemann metric tensor to the manifold space to obtain the manifold space data and mapping parameters; S114, constructing a local tangent space basis vector group based on the manifold space data and mapping parameters; calculating a transition mapping matrix between tangent spaces based on the local tangent space basis vector group; constructing an overall coordinate transformation relationship based on the transition mapping matrix; mapping the manifold space data to the feature space based on the overall coordinate transformation relationship to obtain feature space mapping data and transformation matrix data; S115. Based on the feature space mapping data, calculate the differential geometry invariant, extract Gaussian curvature and mean curvature data; based on the Gaussian curvature and mean curvature data, calculate the shape index and curvature index data, and construct a geometric feature vector; Based on the transformation matrix data, a curvature mapping matrix is ​​constructed.

7. The controllable gain RCS active enhancement method according to claim 5, characterized in that: Step S12 is further as follows: S121, obtaining incident electromagnetic wave parameters and target surface parameters, and calculating a local coordinate transformation matrix based on the incident electromagnetic wave parameters and the target surface parameters; dividing the target surface into initial differential units based on the local coordinate transformation matrix; Performing adaptive subdivision operation on the initial differential unit to obtain subdivided differential unit data; S122, obtaining target material parameters, and calculating the dielectric constant and magnetic permeability of each differential unit based on the target material parameters and the subdivided differential unit data; constructing a material dispersion model based on the dielectric constant and magnetic permeability, and calculating frequency-dependent material parameters; Performing normalization processing on the frequency-dependent material parameters to obtain normalized material parameters; S123. Based on the normalized material parameters and incident electromagnetic wave parameters, calculate the local incident field distribution; based on the local incident field distribution, calculate the equivalent current density distribution for each differential unit; based on the equivalent current density distribution, construct and solve the local electromagnetic field equation, and calculate the polarization tensor data of the differential unit; based on the polarization tensor data, construct a local scattering matrix; S124, constructing a global assembly relationship of the scattering matrix based on the local scattering matrix, and calculating the coupling coefficient of the adjacent units; Based on the coupling coefficient, the local scattering matrix is ​​corrected to obtain a corrected local scattering matrix; an eigendecomposition operation is performed on the corrected local scattering matrix to extract scattering eigenvectors and eigenvalues; The scattering eigenvectors and eigenvalues ​​are orthogonalized to obtain processed electromagnetic eigenvectors; all corrected local scattering matrices are stored as a scattering matrix set.

8. The controllable gain RCS active enhancement method according to claim 5, characterized in that: Step S21 is further as follows: S211, calculating system state variables based on the target posture data, the updated mapping function and the updated feature representation data; constructing a state vector based on the system state variables; Performing orthogonalization processing on the state vector to obtain an orthogonalized state vector; constructing a state space basis based on the orthogonalized state vector to obtain state space data; S212. Based on the state space data, the tangent bundle structure of the state space is calculated and the Lie algebra basis is constructed; Compute commutator relations between Lie algebra bases; Based on the commutation subrelation, the Lie group action is constructed; based on the Lie group action, the state space data is mapped to the Lie algebra space to obtain the Lie algebra mapping data; S213, based on Lie algebra mapping data, construct the system dynamics equation; Calculate the location of the equilibrium point of the system dynamics equation; Perform stability analysis on the equilibrium point position to obtain analysis results; based on the system dynamics equation and analysis results, solve and obtain eigenvalue data and corresponding eigenvector data; S214, constructing a kinetic model and a model parameter sensitivity matrix based on the eigenvalue data and the corresponding eigenvector data; and calculating parameter influence weights based on the model parameter sensitivity matrix; Based on the parameter influence weights, the parameters of the dynamic model are optimized to obtain the optimized model parameters; Based on the optimized model parameters, a state transfer matrix is ​​generated.

9. The controllable gain RCS active enhancement method according to claim 5, characterized in that: Step S41 is further as follows: S411, receiving environmental parameters collected in real time, performing abnormal value detection on the environmental parameters, and obtaining the detected environmental parameters; based on the detected environmental parameters, calculating the parameter change trend, and extracting environmental disturbance characteristics; performing cluster analysis on the environmental disturbance characteristics, and obtaining environmental characteristic data and clustering results; S412, constructing an environmental disturbance state equation based on the environmental characteristic data and the clustering results, and calculating a disturbance transfer function; Based on the disturbance transfer function, the disturbance model parameters are identified; Performing adaptive update on disturbance model parameters to obtain disturbance model data; S413, constructing a disturbance observer structure based on the disturbance model data, and calculating observer gain parameters; generating a disturbance compensation signal based on the observer gain parameters; smoothing the disturbance compensation signal to generate compensation strategy data; S414, calculating a control amount correction value based on the compensation strategy data; performing a control amount adjustment operation based on the control amount correction value and the optimized control parameter set to obtain a corrected control amount data; The corrected control quantity data is verified to obtain the verified control quantity data, that is, the compensated control quantity data.

10. A controllable gain RCS active enhancement device, characterized in that: include: A control module and an enhancer, wherein the control module comprises: at least one processor; and, a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement a controllable gain RCS active enhancement method as described in any one of claims 1 to 9.

Images