A prefabricated component installation and positioning method under complex hydrological conditions

Abstract

The application relates to the field of water engineering and discloses a prefabricated component installation and positioning method under complex hydrological conditions, which comprises the following steps: a sensor network is arranged on the surface of a component; an initial reconstruction model of strain-water dynamic load mapping is established based on finite element analysis; strain data of the sensor is collected in real time; component pose information of an external positioning system is synchronously acquired; real-time water dynamic disturbance is calculated through the initial model; sensor health diagnosis is carried out in combination with a dynamic threshold method; a Kalman filter is expanded to fuse the pose and disturbance data, model parameters are updated online to generate a calibration model; based on calibration model output and state estimation, a feedforward-feedback composite control instruction is generated to drive an actuator to adjust the pose. Through the online calibration model and the composite control architecture, the redundant actuator is optimally distributed, water dynamic disturbance is counteracted in real time, and millimeter-level pose regulation and control are realized, so that the technical bottleneck of low precision and poor adaptability of the traditional method is broken through.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water engineering, in particular to a prefabricated component installation positioning method under complex hydrological conditions. BACKGROUND

[0002] With the development of marine resources to the deep sea, underwater engineering equipment is facing the severe challenge of complex hydrodynamic environment. Disturbance such as strong ocean current and vortex leads to instability of component pose, which directly affects the operation accuracy and equipment safety. Traditional pose control methods rely on feedback regulation and empirical model, which is difficult to meet the high-precision control demand in dynamic flow field, and new intelligent control technology is needed to improve the anti-interference ability of the system.

[0003] In the prior art, a typical scheme adopts feedforward compensation based on a fixed parameter model combined with PID feedback control. A pressure sensor array arranged on the surface of the component is used to obtain flow field information, and a static mapping relationship between hydrodynamic disturbance and actuator thrust is established. Some improved schemes introduce adaptive filtering algorithm to process noise reduction of sensor data, or update model parameters through periodic offline calibration.

[0004] Although the prior art improves the stability of pose control to some extent, there are still some deficiencies: first, the fixed parameter model cannot adapt to the time-varying characteristics caused by material performance degradation and sensor drift, and the control accuracy gradually decreases in long-term operation; second, the single feedback control architecture is difficult to cope with sudden flow field changes due to disturbance compensation lag; third, the sensor network lacks online health monitoring mechanism, and abnormal data directly leads to model misalignment and control command oscillation; finally, the actuator thrust distribution strategy does not consider energy optimization under redundant configuration, which restricts the continuous operation ability of the system. These defects are due to the lack of model dynamic calibration, multi-source data fusion and abnormal diagnosis in the prior art, which leads to insufficient robustness and adaptability in complex marine environment. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a prefabricated component installation positioning method under complex hydrological conditions, which solves the problems of low underwater component pose control accuracy and poor environmental adaptability caused by model parameter solidification, disturbance compensation lag and insufficient sensor reliability in the prior art.

[0006] To achieve the above purpose, the present application is realized by the following technical scheme: a prefabricated component installation positioning method under complex hydrological conditions, comprising the following steps:

[0007] S1, a sensor network is arranged on the surface of the prefabricated component, and an initial reconstruction model is established by finite element analysis, which maps the sensor strain response to equivalent hydrodynamic load;

[0008] S2. Real-time acquisition of strain data from the sensor network, and simultaneous acquisition of component pose data from the external positioning system;

[0009] S3. Based on the strain data, calculate the real-time hydrodynamic disturbance through the initial reconstruction model, and at the same time perform a health diagnosis on the sensor network to remove abnormal data.

[0010] S4. Input the pose data and the real-time hydrodynamic disturbance into the extended Kalman filter, update the parameters of the initial reconstruction model online, and generate the calibrated reconstruction model;

[0011] S5. Based on the disturbance estimation output by the calibrated reconstruction model and the state estimation result of the extended Kalman filter, a control command is generated to counteract the hydrodynamic disturbance and drive the actuator to adjust the component pose.

[0012] Preferably, step S1 includes the following steps:

[0013] A distributed fiber optic grating sensor array is pre-embedded below the outer surface of the underwater part of the prefabricated component, with the sensor spacing not exceeding 5% of the characteristic length of the component, forming a fully covered strain sensing network.

[0014] A three-dimensional mechanical model of the component is established using finite element analysis software, and the elastic modulus of the material is defined. Poisson's ratio and boundary conditions;

[0015] Apply six-degree-of-freedom unit loads sequentially to the model:

[0016] Calculate the strain response of all sensors under each load. ;

[0017] Constructing the initial reconstruction matrix Establish strain-load mapping relationship .

[0018] Preferably, step S2 includes:

[0019] Through a high-speed fiber optic demodulator. The sampling frequency reads the wavelength offset of the FBG sensor in real time. And through calibration formula Convert to strain value, where, For the first The strain sensitivity coefficient of each sensor; For the first A fiber grating sensor at time Wavelength offset; For the first Each sensor at time The strain measurement value;

[0020] Through GNSS / INS combined system Frequency acquisition of global pose data of the water surface portion of the component. In the formula, The eastward coordinates of the centroid of the water surface portion of the component in the global coordinate system; The north coordinates of the centroid of the water surface portion of the component in the global coordinate system; The vertical coordinates of the centroid of the water surface portion of the component in the global coordinate system; The component's roll angle; The component's pitch angle; The heading angle of the component;

[0021] Using underwater acoustic positioning systems Frequency acquisition determines the relative position of the underwater portion of the component. And use a timestamp alignment algorithm to and Fusion into complete pose data In the formula, The eastward offset of the reference point of the underwater part of the component relative to the acoustic beacon; This represents the northward offset of the reference point of the underwater part of the component relative to the acoustic beacon; This represents the vertical offset of the underwater reference point of the component relative to the acoustic beacon.

[0022] Preferably, the timestamp alignment algorithm is as follows:

[0023] For each strain data point pose data points ,when hour Synchronization data is generated through linear interpolation. In the formula, These are strain measurement values; The fused pose data; This is the maximum permissible time deviation; This is the aligned unified timestamp.

[0024] Preferably, step S3 includes:

[0025] For the first in the sensor network Each sensor defines a set of sensors in spatial proximity. Calculate the local strain residual of the sensor. The formula is:

[0026] ;

[0027] wherein, is the strain measurement value of the i-th sensor at time t; is the number of valid sensors in the neighborhood set; is the strain measurement value of the i-th sensor at time t; is the spatio-temporal residual of the i-th FBG sensor; is the spatial neighborhood set of the i-th sensor; is the strain measurement value of the i-th neighbor sensor at time t;

[0028] set a dynamic fault threshold , wherein, is the standard deviation of the historical strain data within a sliding time window , if and lasts more than a time window , the i-th sensor is determined to be faulty;

[0029] remove all elements corresponding to faulty sensors from the strain data vector , and generate a valid strain vector , wherein, is the original strain measurement vector of all sensors; is the strain value of the i-th sensor at time t; is the total number of sensors; is the set of faulty sensor indices; is the valid strain vector after removing faulty sensors; is the number of faulty sensors;

[0030] delete the row corresponding to from the initial reconstruction matrix , and obtain a reduced-dimension reconstruction matrix , and calculate the real-time hydrodynamic disturbance by least squares method:

[0031] ;

[0032] wherein, is the six-degree-of-freedom hydrodynamic disturbance estimate; is the reduced-dimension reconstruction matrix after removing faulty sensors; is the set of faulty sensor indices; is the valid strain vector after removing faulty sensors; is the number of faulty sensors; is the total number of sensors.​​​​​​​​

[0033] Preferably, the step S4 comprises:

[0034] The motion state of the component is combined with the reconstructed model parameters into an augmented state vector, which is given by:

[0035]

[0036] where, is the reconstructed model parameters vector; is the component motion state, which includes:

[0037]

[0038] is the position of the component's center of mass in the global coordinate system; is the component's attitude quaternion; is the component's linear velocity; is the component's angular velocity;

[0039] Based on the component's dynamics equation and the water dynamic disturbance input, a nonlinear state transition function is constructed, which is given by:

[0040]

[0041] where, is the state vector at the current time, which includes the motion state and the model parameters; is the nonlinear state transition function, which describes the dynamics evolution; is the state vector at the previous time; is the control input at the previous time; is the estimated water dynamic disturbance at the previous time; is the process noise, which is subject to Gaussian distribution;

[0042] The pose data measured by the external positioning system is mapped into an observation function of the state vector, which is given by:

[0043]

[0044] where, is the observation vector at time , which includes the pose data measured by the external positioning system; is the nonlinear observation function, which maps the state vector into the pose observation space; is the augmented state vector; is the observation noise;

[0045] In the prediction stage, the prior state estimation and covariance are calculated, which is given by: ​​​​

[0046] ;

[0047] ;

[0048] where, is the prior state estimate at time ; is a nonlinear state transition function describing the system dynamics; is the posterior state estimate at time ; is the control input at the previous time step; is the estimated hydrodynamic disturbance at the previous time step; is the prior estimate covariance matrix; is the Jacobian matrix of the state transition function; is the posterior covariance matrix at time ; is the process noise covariance matrix.

[0049] The update stage of the extended Kalman filter preferably fuses the pose observation data to update the state estimate, with the formula:

[0050] ;

[0051] where, is the Kalman gain matrix; is the Jacobian matrix of the observation function; is the identity matrix; is the posterior state estimate at time ; is the updated posterior covariance matrix; is the observation noise covariance matrix; is the prior state estimate; is the actual observation value; is the predicted observation value;

[0052] Model parameter online calibration: extract the parameter vector from the updated state estimate , update the initial reconstruction matrix according to the preset mapping relationship, generate the calibrated reconstruction model , satisfying:

[0053] ;

[0054] where, is the effective strain vector at time ; is the calibrated reconstruction matrix; is the equivalent six-degree-of-freedom hydrodynamic force vector.

[0055] Preferably, the step S5 comprises:

[0056] Based on the disturbance estimation output by the calibrated reconstruction model, a feedforward compensation term is calculated as:

[0057] ;

[0058] wherein, is the feedforward control vector, which directly counteracts the hydrodynamic disturbance; is the control allocation matrix; is the pseudo-inverse of ; is the reconstructed equivalent six-degree-of-freedom hydrodynamic force;

[0059] Based on the state estimation result of the extended Kalman filter, a closed-loop feedback term is calculated as:

[0060] ;

[0061] wherein, is the feedback control vector, which eliminates the deviation between the target and the estimated state; is the proportional gain matrix, a preset weight parameter; is the differential gain matrix, a preset damping parameter; is the target state vector; is the estimated motion state vector; is the target state rate of change; is the estimated state rate of change;

[0062] The feedforward and feedback control amounts are superimposed to generate a total control instruction and drive the actuator, which is calculated as:

[0063] ;

[0064] wherein, is the total control instruction vector, which is sent to the actuator; is the feedforward control term; is the feedback control term;

[0065] is sent to the following actuators through a real-time communication protocol:

[0066] The power positioning thruster array: according to the thrust instruction adjusts the speed of each thruster to generate a six-degree-of-freedom control force;

[0067] The anchor winch system: through the cable tension control, it assists in suppressing low-frequency drift; ​

[0068] Hydraulic leveling mechanism: fine-tuning component pitch angle and roll angle Ensure the installation surface is level.

[0069] Preferably, the control allocation matrix The construction method is as follows:

[0070] For each thruster Define the control allocation matrix The Listed as the number The thrust contribution of each thruster is calculated using the following formula:

[0071] ;

[0072] in: For the first The contribution vector of each actuator to the six degrees of freedom control force; For the first A unit vector representing the thrust direction of each thruster; For the first The position vector of the thruster installation position relative to the centroid of the component; This is the vector cross product operator.

[0073] Preferably, the actuator includes:

[0074] Dynamically positioned thruster array, whose thrust direction covers six degrees of freedom;

[0075] Anchor cable winch system, which uses tension control to adjust the position and orientation of auxiliary components;

[0076] The hydraulic leveling mechanism is used to fine-tune the pitch and roll attitude of the components.

[0077] This invention provides a method for installing and positioning prefabricated components under complex hydrological conditions. It has the following beneficial effects:

[0078] 1. This invention, through the synergistic effect of feedforward control and online calibration reconstruction model, directly generates compensation commands based on real-time observed disturbance values, effectively overcoming the hysteresis problem inherent in traditional feedback control. Combined with the dynamic parameter calibration capability of the extended Kalman filter, it improves the accuracy of disturbance estimation in complex flow field environments, ensuring the rapid response characteristics of component pose control.

[0079] 2. This invention introduces a sensor health diagnosis mechanism and a redundant actuator optimization allocation strategy. In the event of sensor malfunction or partial actuator failure, it maintains stable system operation through dynamic reconfiguration and thrust reallocation. This design significantly reduces the risk of loss of control due to single-point failures and enhances the continuous operation capability of marine equipment under harsh conditions.

[0080] 3、The present application compensates for time-varying factors such as material performance degradation and sensor drift through multi-source pose data fusion and model parameter online updating mechanism. The combination of space-time synchronization algorithm and closed-loop calibration technology enables the system to maintain high-precision control performance for a long time, avoiding the precision decay problem caused by environmental changes in traditional offline calibration methods.

[0081] 4、The present application decouples feedforward disturbance cancellation and feedback error correction through a composite control architecture, reducing the oscillation amplitude of control commands. Combined with the weighted optimization distribution strategy of the actuator, high energy-efficient execution units are preferentially called under the premise of meeting control requirements, prolonging the service life of key equipment. BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 The present application is a method flowchart. DETAILED DESCRIPTION

[0083] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0084] Please refer to the accompanying Figure 1 The present application provides a prefabricated component installation and positioning method under complex hydrological conditions, comprising the following steps:

[0085] S1, a sensor network is laid out on the surface of the prefabricated component, and an initial reconstruction model is established through finite element analysis, which maps the sensor strain response to equivalent hydrodynamic load;

[0086] In this embodiment, the layout of the sensor network on the surface of the prefabricated component and the establishment of the initial reconstruction model are realized through the following technical solutions:

[0087] A distributed Fiber Bragg Grating (FBG) sensor array is embedded in a grid layout under the outer surface of the underwater part of the prefabricated component. The sensor array covers the flow surface, corner area and key stress points determined by finite element analysis of the component, and the optimal sensor spacing is not more than 5% of the component characteristic length (such as the maximum side length or diameter), to ensure the spatial resolution capability for complex hydrodynamic load.

[0088] The fiber optic grating sensor is made entirely of quartz glass, encapsulated in a titanium alloy sheath, and internally filled with silicone gel cushioning material, achieving an IP68 protection rating. It can withstand water pressure up to 50 meters deep and long-term seawater corrosion. The sensor leads are connected to the surface signal processing unit via waterproof conduits embedded within the component, preventing damage to the wiring from external water flow.

[0089] Based on the three-dimensional geometric model of the component, a mechanical simulation model is established using finite element analysis software (such as ANSYS Mechanical). The material properties of the component, including the elastic modulus, are explicitly defined in the model. Poisson's ratio and density And set fixed constraint boundary conditions that match the mounting base.

[0090] A six-degree-of-freedom unit load, including unit forces in three orthogonal directions, is sequentially applied to the mechanical model. and unit torque Forming a unit load set , of which Each load is represented as:

[0091] ;

[0092] In the formula, For the first A six-degree-of-freedom unit load vector; Let Kronecker function be used when = The value is 1 if it is true, and 0 otherwise.

[0093] Calculate each unit load Strain response of all sensors under action In the formula, This represents the total number of sensors; Indicates the first The sensor at the first Strain values ​​under each load.

[0094] Combine all unit load responses column-wise to construct the initial reconstruction matrix. The formula is:

[0095] ;

[0096] In the formula, This is the initial reconstruction matrix; These are the strain response vectors under unit loads of the first to sixth six degrees of freedom, respectively. The total number of sensors deployed on the surface of precast components; The mathematical space representing a matrix.

[0097] This matrix establishes a linear mapping relationship from strain response to equivalent hydrodynamic load. ,in Let be the equivalent hydrodynamic load vector to be determined.

[0098] Using the aforementioned techniques, the sensor network can sensitively capture changes in the strain distribution on the surface of components caused by water flow impact. The application of a unit load in the finite element model simulates the actual action of hydrodynamic loads, while the initial reconstruction matrix... The construction of the system is essentially based on the principle of linear superposition, which maps discrete strain measurements to equivalent load distributions in continuous space.

[0099] S2. Real-time acquisition of strain data from the sensor network and simultaneous acquisition of component pose data from the external positioning system;

[0100] In this embodiment, the real-time acquisition and synchronization of sensor network strain data and component pose data are achieved through the following technical solution:

[0101] The wavelength offset signals of each sensor on the surface of the precast component are read in real time using a fiber optic grating demodulator. With strain value The conversion relationship is as follows:

[0102] ;

[0103] In the formula, For the first The strain sensitivity coefficient of each sensor; For the first A fiber grating sensor at time Wavelength offset; For the first Each sensor at time The strain measurement value.

[0104] Preferably, the demodulator adopts wavelength scanning demodulation technology, supports multi-channel parallel acquisition, and ensures that the strain data of all sensors are output with a unified time reference.

[0105] Simultaneously acquire pose data from the following heterogeneous positioning systems:

[0106] GNSS / INS integrated navigation system: outputs the global pose of the water surface portion of the component. It includes three-dimensional position and Euler angle information, where, The eastward coordinates of the centroid of the water surface portion of the component in the global coordinate system; The north coordinates of the centroid of the water surface portion of the component in the global coordinate system; The vertical coordinates of the centroid of the water surface portion of the component in the global coordinate system; is the roll angle of the component; is the pitch angle of the component; is the yaw angle of the component;

[0107] Underwater acoustic positioning system: through the base array to emit acoustic signals, to solve the local position of the underwater part of the component relative to the installation base , preferably the acoustic signal frequency is between 30 kHz-50 kHz to balance the precision and anti-interference ability, in which, is the eastward offset of the reference point of the underwater part of the component relative to the acoustic beacon; is the northward offset of the reference point of the underwater part of the component relative to the acoustic beacon; is the vertical offset of the reference point of the underwater part of the component relative to the acoustic beacon.

[0108] For the time reference difference of multi-source heterogeneous data, a sliding window interpolation algorithm is used to realize timestamp alignment:

[0109] A high-precision hardware clock mark is added to each data source, and the time synchronization error is not more than ;

[0110] Within the time window , cubic spline interpolation is performed on the asynchronously arrived pose data to generate a unified time sequence , in which, is the fused position vector; is the normalized attitude quaternion.

[0111] Strain data acquisition is based on the wavelength modulation characteristics of fiber Bragg grating, and the dynamic strain distribution on the surface of the component is inversely calculated by monitoring the Bragg wavelength shift. The multi-source fusion of pose data solves the problems of signal attenuation, multipath effect and other problems existing in underwater environment of single positioning system, and the space-time synchronization algorithm ensures the causal consistency of strain and pose data.

[0112] S3, based on the strain data, the real-time hydrodynamic disturbance is calculated through the initial reconstruction model, and the sensor network is simultaneously diagnosed for health to eliminate abnormal data;

[0113] In this embodiment, the real-time hydrodynamic disturbance calculation and sensor health diagnosis are realized through the following technical solutions:

[0114] Sensor health state dynamic evaluation: based on the real-time strain data obtained in step S2, local consistency test is performed on each sensor node. The first sensor node is defined as the spatially adjacent sensor set , the local strain residual of the sensor is calculated , and the formula is:

[0115] ;

[0116] wherein, is the strain measurement value of the th sensor at time ; is the number of active sensors in the neighborhood set; is the spatio-temporal residual of the th fiber Bragg grating sensor; is the spatial neighborhood set of the th sensor; is the strain measurement value of the th neighbor sensor at time . Preferably, the neighborhood relationship is predefined according to the sensor deployment topology, ensuring that physically adjacent sensors are included in the same inspection group.

[0117] Dynamic threshold setting and fault determination: the standard deviation of the historical data within the sliding time window is calculated , and the dynamic fault threshold is set as:

[0118] ;

[0119] wherein, is the standard deviation of the historical strain data within the sliding time window , if and lasts more than the time window , the th sensor is determined to be faulty.

[0120] All elements corresponding to faulty sensors are removed from the strain data vector , generating the effective strain vector , wherein, is the original strain measurement vector of all sensors; is the strain value of the th sensor at time ; is the total number of sensors; is the set of faulty sensor indices; is the effective strain vector after removing faulty sensors; is the number of faulty sensors.

[0121] Dynamic dimension reduction of the reconstruction model: the row vector corresponding to the faulty sensors is removed from the initial reconstruction matrix , obtaining the reduced dimension reconstruction matrix .

[0122] Real-time hydrodynamic disturbance reconstruction: the reduced dimension reconstruction matrix is used to reconstruct the effective strain data The real-time hydrodynamic disturbance estimates are obtained using the least squares method:

[0123] ;

[0124] In the formula, This is an estimate of the six-degree-of-freedom hydrodynamic disturbance. The dimension-reduced reconstruction matrix after removing faulty sensors; For the set of fault sensor indices; The effective strain vector after removing faulty sensors; The number of faulty sensors; This represents the total number of sensors.

[0125] Health diagnosis is based on the role of spatially adjacent sensor strain response:

[0126] Under normal operating conditions, adjacent sensors exhibit a consistent trend of change due to similar loads. The residual difference usually indicates sensor failure or local structural damage.

[0127] The dynamic threshold mechanism can adapt to changes in environmental noise levels, avoiding misjudgments or missed detections caused by fixed thresholds.

[0128] The dimensionality reduction and reconstruction matrix maintains the effective rank of the model by removing the row vectors corresponding to fault data, thus ensuring the numerical stability of the perturbation estimation.

[0129] S4. Input the pose data and real-time hydrodynamic disturbance into the extended Kalman filter, update the parameters of the initial reconstruction model online, and generate the calibrated reconstruction model.

[0130] In this embodiment, the online calibration of the extended Kalman filter parameters is achieved through the following technical solution:

[0131] The dynamic motion characteristics of the components are jointly modeled with the time-varying parameters of the reconstructed model to construct an augmented vector containing multi-dimensional state information. Specifically, the augmented state vector is defined as follows:

[0132] ;

[0133] In the formula, To reconstruct the model parameter vector; The motion state of the component includes:

[0134] ;

[0135] This represents the position of the component's centroid in the global coordinate system. The quaternion represents the component's attitude. The linear velocity of the component; is the angular velocity of the component. By embedding the parameters into the state vector, the model error and the motion state are estimated synchronously.

[0136] Based on the component dynamics equation and the water dynamic disturbance input, the nonlinear state transition function is constructed, which is:

[0137] ;

[0138] where, is the state vector at the current time, which contains the motion state and the model parameters; is the nonlinear state transition function, which describes the dynamics evolution; is the state vector at the last time; is the control input at the last time; is the estimated water dynamic disturbance at the last time; is the process noise, which is subject to Gaussian distribution.

[0139] The pose data measured by the external positioning system is mapped to the observation function of the state vector, which is:

[0140] ;

[0141] where, is the observation vector at time , which contains the pose data measured by the external positioning system; is the nonlinear observation function, which maps the state vector to the pose observation space; is the augmented state vector; is the observation noise.

[0142] In the prediction stage, the prior state estimation and covariance are calculated, which is:

[0143] ;

[0144] ;

[0145] where, is the prior state estimation at time ; is the nonlinear state transition function, which describes the system dynamics; is the posterior state estimation at time ; is the control input at the last time; is the estimated water dynamic disturbance at the last time; is the prior estimation covariance matrix; is the Jacobian matrix of the state transition function; is the time a posteriori covariance matrix of the state estimate; a process noise covariance matrix.

[0146] fusing the pose observation data, updating the state estimate, formula is:

[0147]

[0148] wherein, a Kalman gain matrix; a Jacobian matrix of the observation function; an identity matrix; a posteriori state estimate at time ; an updated a posteriori covariance matrix; an observation noise covariance matrix; a priori state estimate; an actual observation value; a predicted observation value.

[0149] extracting a parameter vector from the updated state estimate , updating the initial reconstruction matrix according to a preset mapping relationship, generating a calibrated reconstruction model , satisfying:

[0150]

[0151] wherein, an effective strain vector at time ; a calibrated reconstruction matrix; an equivalent six-degree-of-freedom hydrodynamic load vector.

[0152] Preferably, in order to solve the numerical stability problem of quaternion update, the attitude quaternion is normalized after each filtering iteration, formula is: ; at the same time, a physical constraint is applied to the parameter vector , to ensure the physical meaning of the reconstruction matrix is reasonable, wherein, an updated attitude quaternion estimate value of the extended Kalman filter; denotes the Euclidean norm of the quaternion; denotes an assignment operation, the normalized quaternion is re-assigned to to ensure that it satisfies the unit constraint condition.

[0153] S5, according to the disturbance estimate output by the calibrated reconstruction model and the state estimate result of the extended Kalman filter, a control command is generated to offset the hydrodynamic disturbance, to drive the actuator to adjust the member pose; ​​

[0154] In this embodiment, the water dynamic disturbance offset and the component pose adjustment are realized by the following technical solutions:

[0155] The real-time water dynamic disturbance estimation value output based on the calibrated reconstruction model A feedforward control term is designed to directly offset the influence of the disturbance on the component pose. The calculation of the feedforward control quantity is based on the pseudo-inverse operation of the control allocation matrix, and the formula is:

[0156] ;

[0157] In the formula, is the feedforward control vector, which is used to directly offset the water dynamic disturbance; is the control allocation matrix; is the pseudo-inverse matrix of is the reconstructed equivalent six-degree-of-freedom water dynamic load. The pseudo-inverse of the matrix is calculated by singular value decomposition (SVD), and the specific form is:

[0158] ;

[0159] In the formula, is the control allocation matrix; is the left singular vector matrix; is the singular value matrix; is the right singular vector matrix; is the pseudo-inverse matrix; this method can effectively handle the thrust allocation problem under redundant configuration of actuators, and ensure the controllability of the system when some actuators fail.

[0160] Combined with the component motion state estimation value output by the extended Kalman filter , a proportional-derivative (PD) control strategy is used to generate a feedback control quantity to eliminate the cumulative deviation of the target pose. The calculation formula of the feedback control quantity is:

[0161] ;

[0162] In the formula, is the feedback control vector, which is used to eliminate the deviation between the target and the estimated state; is the proportional gain matrix, a preset weight parameter; is the derivative gain matrix, a preset damping parameter; is the target state vector; is the estimated motion state vector; is the target state change rate;​ To estimate the rate of state change. By adjusting the weight coefficients of the gain matrix, the response speed and stability of the system can be balanced.

[0163] The feedforward and feedback control amounts are superimposed to generate total control commands and drive the pose adjustment through the actuator system:

[0164] ;

[0165] In the formula, is the total control command vector, sent to the actuator; is the feedforward control term; is the feedback control term.

[0166] The control commands are transmitted to the following actuators through real-time communication protocols:

[0167] Full-rotation thruster array: adjust the rotation speed and vector direction of each thruster according to the thrust command to generate six-degree-of-freedom control forces and moments. The installation position and thrust direction of the thruster are pre-calibrated through three-dimensional modeling to ensure the accuracy of the control distribution matrix.

[0168] Anchor winch system: suppress the low-frequency drift motion of the member by dynamically adjusting the cable tension. Preferably, a tension-position double closed-loop control strategy is adopted to maintain the preset tension while tracking the target position.

[0169] Hydraulic leveling mechanism: drive multiple sets of telescopic oil cylinders to adjust the pitch angle and roll angle of the member to ensure that the installation plane level meets the millimeter-level precision requirement. The pressure-flow closed-loop control of the hydraulic system can effectively suppress the vibration caused by water flow impact.

[0170] Control distribution matrix is constructed based on the geometric layout and mechanical properties of the actuators. For the th thruster, its thrust contribution vector to the six-degree-of-freedom control force is: ;

[0171] Wherein: is the contribution vector of the th actuator to the six-degree-of-freedom control force; is the unit vector of the thrust direction of the th thruster; is the position vector of the installation position of the th thruster relative to the center of mass of the member; is the vector cross operator.

[0172] Preferably, for redundant actuator configurations, the weighted least squares method is used to optimize thrust distribution, with the formula being:

[0173] ;

[0174] wherein, is the optimal control command vector to be solved; is the weight matrix; is the ideal thrust setpoint vector; is the control allocation matrix; is the desired six-degree-of-freedom control force vector.

[0175] Step S5 achieves efficient cancellation of hydrodynamic disturbance and precise pose regulation through a feedforward-feedback composite control architecture. The feedforward control directly generates compensatory thrust based on real-time disturbance observations, prospectively eliminating the main disturbance effects; the feedback control eliminates residual errors and unmodeled disturbances through closed-loop regulation. The control allocation algorithm maps abstract control forces to the coordinated action of multiple actuators, balancing system energy consumption and actuator life in combination with optimization strategies.

[0176] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for installing and positioning prefabricated components under complex hydrological conditions, characterized in that, Includes the following steps: S1. A sensor network is deployed on the surface of the precast component, and an initial reconstruction model is established through finite element analysis. The initial reconstruction model maps the sensor strain response to an equivalent hydrodynamic load. S2. Real-time acquisition of strain data from the sensor network, and simultaneous acquisition of component pose data from the external positioning system; S3. Based on the strain data, calculate the real-time hydrodynamic disturbance through the initial reconstruction model, and at the same time perform a health diagnosis on the sensor network to remove abnormal data. S4. Input the pose data and the real-time hydrodynamic disturbance into the extended Kalman filter, update the parameters of the initial reconstruction model online, and generate the calibrated reconstruction model; S5. Based on the disturbance estimation output by the calibrated reconstruction model and the state estimation result of the extended Kalman filter, a control command is generated to counteract the hydrodynamic disturbance and drive the actuator to adjust the component pose.

2. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 1, characterized in that, Step S1 includes the following steps: A distributed fiber optic grating sensor array is pre-embedded below the outer surface of the underwater part of the prefabricated component, with the sensor spacing not exceeding 5% of the characteristic length of the component, forming a fully covered strain sensing network. A three-dimensional mechanical model of the component is established using finite element analysis software, and the elastic modulus of the material is defined. Poisson's ratio and density And set fixed constraint boundary conditions that match the mounting base; Apply six-degree-of-freedom unit loads sequentially to the model: Calculate the strain response of all sensors under each load. In the formula, For the first A six-degree-of-freedom unit load vector; The Kronecker function; Constructing the initial reconstruction matrix Establish strain-load mapping relationship In the formula, This is the initial reconstruction matrix; These are the strain response vectors under unit loads of the first to sixth six degrees of freedom, respectively. The total number of sensors deployed on the surface of precast components; The mathematical space representing a matrix.

3. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 2, characterized in that, Step S2 includes: Through a high-speed fiber optic demodulator. The sampling frequency reads the wavelength offset of the FBG sensor in real time. And through calibration formula Convert to strain value, where, For the first The strain sensitivity coefficient of each sensor; For the first A fiber grating sensor at time Wavelength offset; For the first Each sensor at time The strain measurement value; Through GNSS / INS combined system Frequency acquisition of global pose data of the water surface portion of the component. In the formula, The eastward coordinates of the centroid of the water surface portion of the component in the global coordinate system; The north coordinates of the centroid of the water surface portion of the component in the global coordinate system; The vertical coordinates of the centroid of the water surface portion of the component in the global coordinate system; The component's roll angle; The pitch angle of the component; The heading angle of the component; Using underwater acoustic positioning systems Frequency acquisition determines the relative position of the underwater portion of the component. And use a timestamp alignment algorithm to and Fusion into complete pose data In the formula, The eastward offset of the reference point of the underwater part of the component relative to the acoustic beacon; This represents the northward offset of the reference point of the underwater part of the component relative to the acoustic beacon; This represents the vertical offset of the underwater reference point of the component relative to the acoustic beacon.

4. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 3, characterized in that, The timestamp alignment algorithm is as follows: For each strain data point pose data points ,when hour Synchronization data is generated through linear interpolation. In the formula, These are strain measurement values; The fused pose data; This is the maximum permissible time deviation; This is the aligned unified timestamp.

5. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 4, characterized in that, Step S3 includes: For the first in the sensor network Each sensor defines a set of sensors in spatial proximity. Calculate the local strain residual of the sensor. The formula is: ； In the formula, For the first Each sensor at time The strain measurement value; The number of valid sensors within the neighborhood set; For the first Spatiotemporal residuals of a fiber Bragg grating sensor; For the first The spatial neighborhood set of a sensor; For the first Each neighbor sensor at time The strain measurement value; Set dynamic fault threshold In the formula, Historical strain data in a sliding time window The standard deviation within, if it satisfies And it continues beyond the time window Then determine the first One sensor malfunctioned; From strain data vector Remove all elements corresponding to faulty sensors to generate an effective strain vector. In the formula, This represents the original strain measurement vector for all sensors; For the first Each sensor at time The strain value; This represents the total number of sensors; For the set of fault sensor indices; The effective strain vector after removing faulty sensors; The number of faulty sensors; From the initial reconstruction matrix Delete with The corresponding rows yield the dimensionality reduction reconstruction matrix. Real-time hydrodynamic disturbances were calculated using the least squares method. : ； In the formula, This is an estimate of the six-degree-of-freedom hydrodynamic disturbance. The dimension-reduced reconstruction matrix after removing faulty sensors; For the set of fault sensor indices; The effective strain vector after removing faulty sensors; The number of faulty sensors; This represents the total number of sensors.

6. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 5, characterized in that, Step S4 includes: The motion state of the component and the parameters of the reconstructed model are combined into an augmented state vector, as shown in the formula: ； In the formula, To reconstruct the model parameter vector; The motion state of the component includes: ； This represents the position of the component's centroid in the global coordinate system. The quaternion represents the component's attitude. The linear velocity of the component; The component's angular velocity; Based on the component dynamics equations and hydrodynamic disturbance input, a nonlinear state transition function is constructed, the formula of which is: ； In the formula, This is the current state vector, containing the motion state and model parameters; This is a nonlinear state transition function that describes the dynamic evolution. This is the state vector from the previous time step; This is the control input from the previous moment; This refers to the hydrodynamic disturbance estimated at the previous moment; This is process noise, and it follows a Gaussian distribution. The observation function that maps pose data measured by an external positioning system to a state vector is given by the following formula: ； In the formula, For a moment The observation vector contains pose data measured by an external positioning system; For nonlinear observation functions, the state vector is... Map to the pose observation space; For augmented state vectors; To observe noise; In the prediction phase, the prior state estimate and covariance are calculated using the following formula: ； ； In the formula, For a moment Prior state estimation; It is a nonlinear state transition function that describes the system dynamics; For a moment Posterior state estimation; This is the control input from the previous moment; This refers to the hydrodynamic disturbance estimated at the previous moment; To estimate the covariance matrix a priori; Let be the Jacobian matrix of the state transition function; For a moment The posterior covariance matrix; Let be the process noise covariance matrix.

7. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 6, characterized in that, In the update phase of the extended Kalman filter, pose observation data is fused to update the state estimate, and the formula is: ； In the formula, The Kalman gain matrix; Let be the Jacobian matrix of the observation function; It is the identity matrix; For a moment Posterior state estimation; This is the updated posterior covariance matrix; To observe the noise covariance matrix; For prior state estimation; These are actual observed values; Online calibration of model parameters: from updated state estimates Extracting parameter vectors Update the initial reconstruction matrix according to the preset mapping relationship. Generate a calibrated reconstructed model ,satisfy: ； In the formula, For a moment The effective strain vector; The reconstructed matrix after calibration; This is the equivalent six-degree-of-freedom hydrodynamic load vector.

8. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 7, characterized in that, Step S5 includes: Based on the perturbation estimate of the reconstructed model output after calibration, the feedforward compensation term is calculated using the following formula: ； In the formula, It is a feedforward control vector used to directly counteract hydrodynamic disturbances; To control the allocation matrix; for The pseudo-inverse matrix; For the reconstructed equivalent six-degree-of-freedom hydrodynamic load; Based on the state estimation results of the extended Kalman filter, the closed-loop feedback term is calculated using the following formula: ； In the formula, This is a feedback control vector used to eliminate the deviation between the target and the estimated state; This is the proportional gain matrix, with preset weight parameters; The differential gain matrix contains preset damping parameters. The target state vector; This is the estimated motion state vector; The target state change rate; To estimate the rate of change of state; The feedforward and feedback control values ​​are superimposed to generate a total control command that drives the actuator. The formula is as follows: ； In the formula, This is the overall control command vector, sent to the actuator. For feedforward control terms; For feedback control items; Through real-time communication protocols Send to the following implementing agencies: Dynamically positioned thruster array: based on thrust command Adjusting the rotational speed of each thruster generates a six-degree-of-freedom control force; Anchor winch system: Assists in suppressing low-frequency drift through cable tension control; Hydraulic leveling mechanism: fine-tuning component pitch angle and roll angle Ensure the installation surface is level.

9. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 8, characterized in that, The control allocation matrix The construction method is as follows: For each thruster Define the control allocation matrix The Listed as the number The thrust contribution of each thruster is calculated using the following formula: ； in: For the first The contribution vector of each actuator to the six degrees of freedom control force; For the first A unit vector representing the thrust direction of each thruster; For the first The position vector of the thruster installation position relative to the centroid of the component; This is the vector cross product operator.

10. The method for installing and positioning prefabricated components under complex hydrological conditions according to claim 9, characterized in that, The implementing mechanism includes: Dynamically positioned thruster array, whose thrust directions cover six degrees of freedom; Anchor cable winch system, which uses tension control to adjust the position and orientation of auxiliary components; The hydraulic leveling mechanism is used to fine-tune the pitch and roll attitude of the components.

Images