A fan hub blade butt joint posture real-time correction control method

Abstract

The application discloses a fan hub blade butt joint posture real-time correction control method and belongs to the technical field of wind power generation equipment installation. The method establishes an interface relative constraint graph with the hub blade installation end face as a local reference, extracts continuous drift, periodic swing and transient disturbance from a continuous control cycle to form a dynamic butt joint state; generates a plurality of candidate correction control quantities after correlating the current constraint graph with the dynamic state, and predicts the influence of each candidate quantity on the future interface deviation evolution in a limited time domain; selects the candidate quantity that continuously converges the deviation according to the prediction result, determines the correction direction and generates a posture correction flow, and controls the execution mechanism to implement limited-amplitude translation and rotation correction; updates the state after correction and cycles until the butt joint deviation continuously satisfies the threshold value, and then axial advancement is completed to complete butt joint. The application can effectively suppress common motion interference, over-regulation and oscillation, and improve the stability and safety of fine butt joint.

Description

Technical Field

[0001] This invention relates to the field of wind power equipment installation technology, specifically to a method for real-time correction and control of the docking attitude of wind turbine hub blades. Background Technology

[0002] During the installation and docking of large wind turbine blades, factors such as gusts, sling sway, lifting tool movement errors, and actuator lag cause continuous radial, axial, and angular deviations between the blade root mounting face and the hub blade mounting face. Existing methods typically compensate directly based on the current attitude error in a fixed coordinate system. However, when the hub, lifting tool, or measuring device move together, the measured absolute attitude change cannot accurately reflect the true docking deviation. Furthermore, blades have significant inertia and periodic oscillation characteristics; the current correction action will affect multiple subsequent control cycles. Relying solely on instantaneous errors can easily lead to overshoot, reverse oscillations, and frequent micro-motions. Especially in the precise docking stage near the hub, if continuous drift, periodic oscillation, and transient disturbances cannot be distinguished, and the subsequent effects of correction actions cannot be predicted in advance, misalignment or even collision at the connection point can easily occur.

[0003] Therefore, a control method is needed that can perform real-time correction based on the actual relative relationship between the blade root mounting end face and the hub blade mounting end face, and in conjunction with the future deviation evolution. Summary of the Invention

[0004] This invention addresses the problems of unstable relative docking state characterization, overlapping blade oscillations and control actions, and near-term overshooting during wind turbine blade hoisting and docking. It provides a real-time correction control method. This method establishes an interface relative constraint diagram using the hub blade mounting end face as a local reference. It extracts continuous drift, periodic oscillation, and transient disturbances from the interface state changes in continuous control cycles to form a dynamic docking state. Based on this, it performs finite-time-domain propagation prediction on multiple candidate correction control quantities. Based on the prediction results, it determines the correction direction that continuously converges subsequent docking deviations, achieving real-time, progressive attitude correction.

[0005] This invention provides a real-time correction and control method for the docking attitude of wind turbine hub blades, comprising the following steps:

[0006] Step S1: Interface Relative State Construction

[0007] The system collects real-time position, attitude, and positioning feature information of the blade root mounting end face and the hub blade mounting end face. A local relative docking coordinate system is established with the hub blade mounting end face as the reference. The position, attitude, and positioning features of the blade root mounting end face are transformed into the local relative docking coordinate system. Based on the relative relationship between the center, normal, and circumferential references and corresponding positioning features of the two mounting end faces, the interface relative constraint diagram for the current control cycle is generated.

[0008] Step S2: Dynamic docking status extraction

[0009] The interface relative constraint diagrams formed by multiple consecutive control cycles are arranged into a docking state sequence according to time order. Continuous drift, periodic oscillation, and transient disturbances at different time scales are extracted from the docking state sequence. The translational and rotational motion states relative to the local relative docking coordinate system are determined by the position and attitude changes of the blade root mounting end face. At the same time, the translational and rotational adjustment states implemented by the attitude adjustment actuator in the previous control cycle are obtained. The continuous drift, periodic oscillation, and transient disturbances are associated with the translational motion state, rotational motion state, and adjustment states implemented in the previous control cycle to form a dynamic docking state. This dynamic state characterizes the current relative motion trend and existing control effects of the blade root mounting end face and reduces the impact of transient disturbances on subsequent attitude correction control.

[0010] Step S3: Finite-time domain docking deviation propagation

[0011] The interface relative constraint diagram of the current control cycle is associated with the dynamic docking state to construct the joint docking state of the current control cycle. Multiple candidate correction control quantities are generated for the joint docking state. The docking state change of the blade root mounting end face relative to the hub blade mounting end face in the future preset control cycle after the action of each candidate correction control quantity is predicted. The predicted docking states of each future control cycle corresponding to the same candidate correction control quantity are combined in chronological order to form the finite time domain docking deviation propagation state corresponding to the candidate correction control quantity.

[0012] Step S4: Attitude correction flow generation and execution

[0013] Based on the finite-time domain docking deviation propagation state corresponding to each candidate correction control quantity, and combined with the interface geometric constraints between the two mounting end faces, the motion range of the attitude adjustment actuator, and the safety distance, the future docking state corresponding to each candidate correction control quantity is screened and evaluated to determine the candidate correction control quantity that makes the docking deviation tend to converge in the finite-time domain, and the local correction direction corresponding to the current joint docking state is determined accordingly. Based on the local correction direction, the attitude correction flow of the current control cycle is formed, and the attitude adjustment actuator is controlled according to the attitude correction flow to perform finite-amplitude translation and rotation attitude correction on the blade root mounting end face.

[0014] Step S5: Feedback Update and Integration Determination

[0015] After the attitude adjustment actuator completes the attitude correction for the current control cycle, it re-acquires the real-time position, attitude, and positioning feature information of the blade root mounting end face and the hub blade mounting end face, and returns to step S1 to update the interface relative constraint diagram, dynamic docking status, finite-time domain docking deviation propagation status, and attitude correction flow in sequence. It then determines whether the various docking deviations between the two mounting end faces continuously meet the preset docking thresholds. If not, it continues to execute the attitude correction for the next control cycle. If they are continuously met, it controls the blade root mounting end face to approach the hub blade mounting end face at a low speed along the axial direction until docking is completed.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] This invention establishes an interface relative constraint diagram using the hub blade mounting end face as a local reference, and extracts continuous drift, periodic oscillation, and transient disturbances based on the interface state changes during continuous control cycles. This effectively eliminates the interference of the joint movement of the hoisting platform, hub, or measuring device on absolute posture measurement, allowing the characterization of docking deviation to directly reflect the true relative relationship between the blade root mounting end face and the hub blade mounting end face, thereby improving the pertinence and accuracy of attitude correction.

[0018] This invention combines dynamic docking status with finite time-domain propagation prediction to uniformly evaluate the interface deviation evolution of multiple candidate correction control quantities over multiple control cycles in the future. Based on the prediction results, it selects the correction direction that makes the docking deviation continuously converge. This can effectively suppress overshoot, reverse oscillation and frequent micro-motion caused by instantaneous error compensation, reduce the superposition of large inertia blade oscillation and control actions, and improve the control stability during the fine docking stage.

[0019] This invention incorporates interface geometric constraints, actuator movement range, and safety distance as hard constraints into the candidate control quantity screening process. As the axial distance gradually decreases, it adaptively tightens the single-cycle correction amplitude. When approaching the hub, it can actively limit the amplitude of the correction action to avoid misalignment or collision of the connection part due to excessive correction. At the same time, by allowing continuous determination of the docking domain and conditional control of axial advancement, it provides a closed-loop guarantee for safe and gradual docking. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall process of the real-time correction and control method for the docking attitude of wind turbine hub blades according to the present invention;

[0021] Figure 2 A schematic diagram of the coordinate system for the partial relative alignment between the blade root mounting end face and the hub blade mounting end face. Detailed Implementation

[0022] The present invention will be further described below with reference to embodiments. The following embodiments take the dynamic docking of the blade root mounting end face and the hub blade mounting end face as the object, using the relative rigid body pose between the hub interface coordinate system and the blade root interface coordinate system as the basic measurement object, and establishing a closed-loop control process around the interface relative constraint diagram, dynamic docking state, and finite-time domain docking deviation propagation state. In the embodiments, the structural dimension thresholds are determined by the specific wind turbine interface structure, the measured noise parameters are obtained by the measurement system calibration, the actuator response parameters are obtained by offline identification by the lifting device, and the state identification parameters are updated online by continuous control cycle data, without replacing the actual calibration results with fixed empirical constants without engineering basis. The overall process of the present invention is as follows: Figure 1 As shown.

[0023] Example 1: Correcting the relative constraint diagram of the interface and constructing the dynamic docking status

[0024] The blade spreader has translational adjustment capabilities in two local radial directions, as well as pitch, yaw, and circumferential rotation adjustment capabilities. A visual measurement device obtains the three-dimensional positioning features of the hub blade mounting face and the blade root mounting face, respectively; an inertial measurement unit obtains the blade angular motion information; and the attitude adjustment actuator provides feedback on the actual displacement and rotation completed in the previous control cycle.

[0025] Interface relative pose establishment: Establish a fixed measurement coordinate system, a hub interface coordinate system, and a blade root interface coordinate system. For example... Figure 2 As shown, the hub interface coordinate system takes the center of the hub blade mounting end face as its origin, and its... Along the axis normal to the end face, The axis is determined by the projection direction of the circumferential positioning datum onto the end face. The axes are determined using the right-hand rule. The blade root interface coordinate system has its origin at the center of the blade root mounting end face, and the coordinate axes are defined in the same way.

[0026] No. The relative pose transformation matrix of the blade root interface relative to the hub interface for each control cycle is given by the following equation:

[0027] ;

[0028] in, This is the homogeneous transformation matrix of the hub interface coordinate system relative to the fixed measurement coordinate system; This represents the inverse operation of a matrix; This is the homogeneous transformation matrix of the leaf root interface coordinate system relative to the fixed measurement coordinate system; The relative pose transformation matrix of the blade root interface relative to the hub interface directly describes the relative transformation between the two interface coordinate systems. When the fixed measurement coordinate system, the hoisting platform, or the hub undergoes common rigid body motion, the common motion components are naturally eliminated in the relative transformation.

[0029] Relative pose transformation matrix It consists of a relative rotation matrix and a relative translation vector:

[0030] ;

[0031] in, It is a relative rotation matrix, belonging to the three-dimensional rotation group. This describes the attitude relationship between the blade root interface and the hub interface. The relative translation vector describes the three-dimensional position offset of the origin of the leaf root interface coordinate system relative to the origin of the hub interface coordinate system. for The zero vector, is the normalized scaling factor for homogeneous coordinates.

[0032] The relative rotation matrix and relative translation vector are calculated from the attitude and position of the two end interfaces, respectively:

[0033] ;

[0034] ;

[0035] in, This is the attitude matrix of the hub interface coordinate system relative to the fixed measurement coordinate system; This is the attitude matrix of the leaf root interface coordinate system relative to the fixed measurement coordinate system; The three-dimensional position vector of the center of the hub mounting end face in a fixed measurement coordinate system; The three-dimensional position vector of the blade root end face center in a fixed measurement coordinate system is used; This represents the transpose of the matrix.

[0036] To avoid describing the 3D attitude error using only a single normal angle, a rotation group is used for the relative rotation matrix. The logarithmic mapping on the x-axis yields the three-dimensional rotation error vector:

[0037] ;

[0038] in, This is a three-dimensional rotation error vector; and These respectively characterize the tilting misalignment in two orthogonal radial directions; Characterizes the circumferential rotational misalignment of the blade root interface relative to the hub interface; For a three-dimensional rotation group The matrix logarithmic mapping on the matrix maps the rotation matrix to an antisymmetric matrix; symbol This indicates that the antisymmetric matrix is ​​mapped to a three-dimensional vector.

[0039] Interface relative constraint graph generation: Based on the current relative pose, map the positioning features of each blade root to the hub interface coordinate system to obtain the local closure error of each set of corresponding positioning features:

[0040] ;

[0041] in, For the first The local closure error vector corresponding to the localization features of the group; For the first The calibration coordinates of each leaf root positioning feature in the leaf root interface coordinate system; For the first The calibration coordinates of the target positioning features of the wheel hub in the wheel hub interface coordinate system; subscript This is the location feature number. , This represents the total number of feature groups used for localization.

[0042] For those with The interface corresponding to the positioning features is grouped, with the interface center, end face direction reference, and positioning features as nodes, and the geometric correspondence between the blade root positioning features and the corresponding wheel hub target positioning features as constraint edges, forming the interface relative constraint diagram for the current control cycle:

[0043] ;

[0044] in, For the first Interface relative constraint diagram for each control cycle; It is a set of nodes, including the interface center, end face direction reference, and various positioning features; The corresponding constraint edge set represents the geometric correspondence between the leaf root positioning features and the wheel hub target positioning features; This is a set of constraint attributes; the set of constraint attributes must contain at least the relative translation vector, the rotation error vector, and the closure error of each positioning feature.

[0045] ;

[0046] This graph structure preserves the complete data association between "interface geometric objects - correspondence - constraint residuals". When the interface centers are nearly coincident but local positioning features are still misaligned, the closure error of the corresponding constraint edges can still preserve the local geometric inconsistency.

[0047] Multi-timescale dynamic docking state extraction: The interface relative constraint diagrams of multiple consecutive control cycles are arranged into a docking state sequence in chronological order. For any deviation component extracted from the constraint attribute set, long-term drift extraction, periodic oscillation extraction, and transient disturbance determination are performed.

[0048] Among them: long-term drift extraction uses window least squares slope estimation to obtain the low-frequency drift baseline and its rate of change; periodic oscillation extraction identifies the oscillation period and angular frequency by searching for extreme point intervals; transient disturbance determination is based on the timeliness and persistence of residual analysis.

[0049] The resulting dynamic docking state vector is:

[0050] ;

[0051] in, For the first Dynamic docking state vector for each control cycle; As a low-frequency drift baseline; This represents the rate of change of the drift baseline. This refers to the amplitude of the periodic oscillation. The current phase is oscillating periodically; For transient residuals; The relative motion spin of the blade root interface with respect to the hub interface (including the three-axis velocities and the three-axis angular velocities). This represents the translational and rotational adjustment amounts actually output by the attitude adjustment actuator in the previous control cycle. This state vector simultaneously describes the deviation center, deviation trend, cyclic oscillation stage, short-term disturbance, current relative motion, and existing control action, providing a unified state input for subsequent finite-time-domain propagation.

[0052] Example 2: Finite-Time Domain Docking Deviation Propagation Based on Online Local State Model

[0053] Based on the interface relative constraint diagram and dynamic docking state, this embodiment performs finite time-domain propagation of the future interface state after the application of multiple candidate correction control quantities.

[0054] Actuator Local Response Model Calibration: Limited amplitude translational and rotational calibration inputs are applied to the attitude adjustment actuator, and the discrete response between the input command and the interface relative motion is obtained based on actual feedback. For the finite prediction time domain near the current operating point, it is written as a local discrete state model:

[0055] ;

[0056] in, The propagation state vector includes the interface relative pose, relative velocity, and dynamic trend terms; This is the propagation state vector for the next control cycle; This is the local state transition matrix for the current operating point; Provide the local input response matrix for the actuator; The actual action of the actuator (translational and rotational adjustment); Input matrix for external trends; The external trend vector is reconstructed from the drift baseline, the rate of change of the drift baseline, the amplitude of the periodic oscillation, and the current phase of the periodic oscillation in the dynamic docking state. and The propagation model is determined by a combination of recent stable control data and offline calibration results of the actuator, and is corrected online using a recursive least squares method, so that the propagation model is updated as the sling length, blade attitude and actuator operating point change.

[0057] Candidate correction control quantity construction: Based on the current axial distance, interface constraint residuals, and the allowable range of the actuator, finite amplitude candidate quantities are constructed for both translational and rotational degrees of freedom, and combined into a candidate set:

[0058] ;

[0059] in, For the first A set of candidate corrective control variables for each control cycle; For the first One candidate correction control command, superscript Candidate number; This represents the total number of candidate control variables. The upper and lower limits of the candidate control variables are determined by the single-cycle allowable displacement, allowable rotation angle, velocity, and acceleration boundaries obtained from the actuator calibration. Considering actuator response hysteresis, the candidate commands are first converted into actual action predictions through the calibrated actuator discrete response model before being written into the interface state propagation model.

[0060] Finite-time propagation and constrained property prediction: For the th Each candidate control variable is used as a common initial value to recursively predict the state variables for multiple future control cycles, and these variables are mapped to predictive interface constraint attributes. The finite-time-domain docking deviation propagation state corresponding to each candidate control variable is defined as follows:

[0061] ;

[0062] in, For the first The finite-time-domain docking deviation propagation state corresponding to each candidate control variable; In order to be in Time prediction Interface constraint attributes for each control cycle (including predicted relative translation and rotation errors, and closure errors of each positioning feature); subscript Indicates in Always The predicted value at any given time; For the prediction step number, ; This is to predict the total number of control cycles contained in the time domain. The propagation state preserves the complete interface geometry constraint evolution of the same candidate action over multiple future control cycles.

[0063] Convergence evaluation and safety hard constraints: The predicted constraint attributes are dimensionless to obtain a dimensionless interface constraint vector. For the ... A candidate control variable is used, and the finite-time domain evaluation index is constructed by predicting the interface deviation and the control increment:

[0064] ;

[0065] in, For the first Evaluation index values ​​for each candidate control quantity; A dimensionless vector for predicting interface constraint attributes; This is the interface constraint weight matrix, used to multiply the dimensionless interface constraint vector on the left, and set in stages according to the current axial distance. In the far-distance stage, the weight of the interface center position and overall attitude constraint is increased, and in the near-interface stage, the weight of the positioning feature closure error is increased. The Euclidean norm of a vector; The control increment is normalized to the maximum allowable action per single period for each degree of freedom; The penalty coefficient is used to control changes and suppress frequent actions of the actuator.

[0066] Collision safety, actuator travel, and single-cycle motion capability are treated as hard constraints and are not included in the above-mentioned penalty items:

[0067] ;

[0068] ;

[0069] ;

[0070] in, The minimum geometric distance between the outer edge of the blade root, the positioning feature or connector and the hub structure in the predicted state; The safety distance threshold is determined by structural dimensional errors, measurement uncertainties, and safety margins. Forecast output for the implementing agency; For the implementing agency, predict the single-cycle increment; , These are the lower and upper bounds of the actuator output, respectively. , These are the lower and upper bounds of the single-cycle increment, respectively.

[0071] Any candidate control variable that violates the above hard constraints at any prediction step is directly determined to be infeasible. For the remaining candidate variables, the candidate variable with the smallest evaluation index and whose prediction end interface constraint norm decreases relative to the current state is selected as the current local correction direction.

[0072] Example 3: Adaptive Correction under Periodic Oscillation, Gust Disturbance, and Proximity to Interface Conditions

[0073] This embodiment illustrates how periodic oscillations and transient disturbances in a dynamic docking state change the prediction results of candidate actions, and explains the adaptive tightening method of the correction amplitude when approaching the hub.

[0074] Control strategy under periodic oscillation conditions: When a radial deviation is identified as being formed by the superposition of the drift center and the periodic oscillation, if the current deviation is on the positive side but the oscillation velocity points towards zero deviation, a large negative correction will be superimposed on the blade's own motion. Finite-time propagation will manifest as the error continuously increasing after crossing zero in subsequent prediction steps, leading to an increase in the evaluation index of the candidate quantity or its rejection due to failure to meet the terminal convergence condition. The controller thus automatically selects to maintain or perform a smaller correction, rather than performing full reverse compensation based on the current instantaneous error. When the blade approaches the extreme value of the oscillation and the relative radial velocity decreases, if the drift center is still significantly deviating from the allowable docking region, the correction effect against the drift center is allowed to be increased within the hard constraint range, changing the control objective from point-by-point following of the periodic oscillation to gradually moving the oscillation center.

[0075] Suppression mechanism of transient disturbances caused by gusts: Calculation of normalized disturbance intensity based on the measurement noise covariance matrix of transient residuals.

[0076] ;

[0077] in, For the first Normalized disturbance intensity for each control cycle; For transient residuals; The noise covariance matrix was obtained from static and dynamic calibration data of the vision measurement system and the inertial measurement unit. When If the disturbance threshold is exceeded within a short period of time but the drift baseline does not move accordingly, the change is retained in the transient residual term, and its contribution to the prediction of external trends is limited. When the residual exists continuously and the drift baseline moves synchronously, it is then transferred to the drift state. This mechanism effectively avoids the high-frequency, short-term ineffective response caused by large-inertia actuators following a single gust of wind.

[0078] Axial distance-related correction capability tightened: Based on the allowable control limits for both long-distance and near-distance interface conditions, and the distance to the fine docking zone, a distance adjustment factor and the current allowable single-cycle control limit are defined.

[0079] ;

[0080] ;

[0081] in, For the first The axial distance between the two mounting faces in one control cycle; This is a distance adjustment factor that varies continuously with axial distance, taking a value between 0 and 1. The distance threshold for entering the fine docking zone; This represents the upper limit of single-cycle control allowed under long-distance operating conditions. This represents the upper limit of single-cycle control allowed under near-interface operating conditions; This represents the single-cycle control upper limit corresponding to the current axial distance. This expression allows the control upper limit to change continuously with the axial distance, avoiding the use of multiple discontinuous artificial piecewise constants. and All of these are obtained by calibrating the actuator's capabilities, the interface structure's permissible motion, and safety margins.

[0082] Allowable docking region determination and axial advance condition: After normalizing the interface constraint vector according to the allowable scale, the allowable docking region is defined as follows:

[0083] ;

[0084] in, To allow docking of the set of domains; This is a dimensionless interface constraint vector; The infinity norm of a vector (i.e., the maximum absolute value of its components). This is defined as the infinity norm of a vector when the current interface constraint vector satisfies this norm for multiple consecutive control cycles. (i.e., falling into the permissible docking zone), and when all candidate safety distances in the predicted time domain meet the structural safety requirements, it is determined that the blade root mounting end face has stably entered the permissible docking zone.

[0085] After entering the permissible docking region, the radial and angular actuators maintain low-amplitude closed-loop correction, while the axial actuator operates along the hub interface coordinate system. The shaft advances at low speed. If the following occurs during the advancement process:

[0086] ;

[0087] in, The exit coefficient is set based on the measurement uncertainty, and If the axial advance is interrupted, attitude correction will resume until the system re-stabilizes and enters the permissible docking zone.

[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for real-time correction and control of the docking attitude of wind turbine hub blades, characterized in that, include: Step S1: Collect the real-time status of the blade root mounting end face and the hub blade mounting end face, establish a local relative docking coordinate system with the hub blade mounting end face as the reference, convert the real-time status of the blade root mounting end face to the local relative docking coordinate system, and generate the interface relative constraint diagram of the current control cycle based on the relative geometric relationship of the two end faces. Step S2: Arrange the interface relative constraint diagrams of multiple consecutive control cycles into a docking state sequence in chronological order, extract continuous drift, periodic oscillation and transient disturbance, and combine them with the motion state of the blade root mounting end face and the actual adjustment state of the attitude adjustment actuator in the previous control cycle to form a dynamic docking state. Step S3: Associate the interface relative constraint diagram of the current control cycle with the dynamic docking state to construct a joint docking state, generate multiple candidate correction control quantities, predict the docking state evolution of each candidate correction control quantity in the future preset control cycle, and combine the predicted docking states of each future control cycle corresponding to the same candidate correction control quantity to form a finite time domain docking deviation propagation state. Step S4: Based on the propagation state of the docking deviation in the finite time domain, combined with the interface geometric constraints between the two end faces, the motion range and safety distance of the attitude adjustment actuator, the candidate correction control quantities are screened and evaluated to determine the target candidate correction control quantity and local correction direction that will make the docking deviation converge; based on the local correction direction, an attitude correction flow is formed to control the attitude adjustment actuator to perform attitude correction on the blade root mounting end face. Step S5: After completing the attitude correction for the current control cycle, re-acquire the real-time status and return to step S1 to update the status. Determine whether the various docking deviations between the two end faces continuously meet the preset docking thresholds. If not, continue to execute the attitude correction for the next control cycle. If continuously met, control the blade root mounting end face to approach at a low speed along the axial direction until docking is completed.

2. The method according to claim 1, characterized in that, In step S1, the real-time status includes real-time position, attitude and positioning feature information; the interface relative constraint diagram takes the interface center of both ends, the end face direction reference and positioning features as nodes, the geometric correspondence between the blade root positioning features and the hub target positioning features as constraint edges, and the relative translation vector, rotation error vector and local closure error as constraint attributes.

3. The method according to claim 1, characterized in that, In step S2, the specific methods for extracting continuous drift, periodic oscillation, and transient disturbance are as follows: Window least squares slope estimation is performed on each deviation component in the docking state sequence to obtain the low-frequency drift baseline and rate of change; the periodic oscillation amplitude and current phase are identified by searching for extreme point intervals; transient disturbances are determined based on the timeliness and persistence of the residuals, and transient residuals are extracted; the dynamic docking state includes the low-frequency drift baseline and rate of change, the periodic oscillation amplitude and current phase, transient residuals, the relative motion spin of the blade root mounting end face relative to the hub blade mounting end face, and the actual adjustment state of the attitude adjustment actuator in the previous control cycle.

4. The method according to claim 1, characterized in that, In step S3, the docking state evolution is predicted based on the pre-calibrated local response model of the actuator. The local response model of the actuator uses a local discrete state model to describe the propagation of the interface relative pose, relative velocity and dynamic trend term. The state transition matrix and input response matrix in the model are jointly determined by the control data of the most recent time period and the offline calibration results of the attitude adjustment actuator, and are corrected online using recursive least squares.

5. The method according to claim 4, characterized in that, In step S3, the method for generating multiple candidate correction control quantities is as follows: based on the current axial distance, interface constraint residual, and allowable range of the attitude adjustment actuator, finite amplitude candidate quantities are constructed and combined in each degree of freedom; each candidate correction control quantity instruction is first converted into actual action prediction quantity by the local response model of the actuator, and then the docking state evolution is predicted based on the actual action prediction quantity.

6. The method according to claim 1, characterized in that, In step S4, the specific method for screening and evaluating each candidate correction control quantity is as follows: the predicted interface constraint attributes in the finite time domain docking deviation propagation state are evaluated by weighted comprehensive evaluation, and the collision safety distance, actuator travel boundary and single-cycle incremental boundary are used as hard constraints. Candidate correction control quantities that violate any hard constraint are directly eliminated. Among the remaining candidate correction control quantities, the candidate correction control quantity with the smallest evaluation index and the predicted end interface constraint norm that decreases relative to the current state is selected as the target candidate correction control quantity.

7. The method according to claim 6, characterized in that, In step S4, when determining the local correction direction: the radial deviation is determined based on the interface relative constraint diagram of the current control cycle, and the oscillation speed is determined based on the adjacent state changes of the periodic oscillation component in the docking state sequence; when the radial deviation is formed by the superposition of the low-frequency drift baseline and the periodic oscillation, and the oscillation speed points in the direction of decreasing radial deviation, the correction amount in the same direction as the oscillation speed is suppressed; when the low-frequency drift baseline does not change accordingly, the contribution of transient residuals to future predicted evolution is limited; when transient residuals exist continuously and the low-frequency drift baseline changes synchronously, the corresponding change of transient residuals is included in the low-frequency drift baseline.

8. The method according to claim 1, characterized in that, In step S5, the constraint attributes in the interface relative constraint diagram of the current control cycle are normalized and combined to form the current interface constraint vector. When the current interface constraint vector is within the preset allowable docking domain for multiple consecutive control cycles, and the safety distance of each prediction step in the prediction time domain meets the structural safety requirements, the root mounting end face of the control blade approaches at a low speed along the axial direction. If the current interface constraint vector exceeds the preset exit range during the low-speed axial approach, the low-speed axial approach is paused and attitude correction is resumed.

9. The method according to claim 1, characterized in that, In each control cycle, the upper limit of the single-cycle correction amplitude is dynamically determined based on the current axial distance between the two end faces; a distance adjustment factor that changes continuously with the current axial distance is set, and the upper limit of the current single-cycle correction amplitude is determined based on the distance adjustment factor between the upper limit of the far-distance allowable control and the upper limit of the adjacent interface allowable control; wherein, the upper limit of the adjacent interface allowable control is less than the upper limit of the far-distance allowable control, so that the upper limit of the single-cycle correction amplitude decreases continuously as the current axial distance decreases.

Images