Mechanical property analysis and scheme optimization method for river-crossing ground wire construction process

Abstract

The invention relates to the technical field of bridge and structure engineering construction, and discloses a river-crossing ground wire construction process mechanical property analysis and scheme optimization method, which comprises the following steps of S1, establishing a full-order finite element model of a river-crossing ground wire, extracting effective modal stiffness based on a preset construction working condition, and establishing a full-order finite element model of the river-crossing ground wire; constructing a dynamic order reduction model by using a model order reduction technology; s2, in the construction process, control input and measurement vectors of a construction site are collected in real time; s3, calculating a prior state at a current moment by using the dynamic reduced-order model, and calculating a residual vector between a predicted measurement value corresponding to the prior state and the measurement vector; and S4, based on the residual vector, solving a residual force field vector equivalent to an unmodeled load through an inversion algorithm. According to the method, through residual force field inversion and effective modal stiffness adaptive correction, dual online calibration of the unmeasurable external load and the time-varying stiffness parameter in the dynamic model is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bridge and structure engineering construction, in particular to a method for analyzing mechanical characteristics and optimizing schemes in the construction process of a river-crossing ground wire. BACKGROUND

[0002] The traction and deployment construction of a large-span overhead ground wire across a river or sea is a key link in a power transmission project. Such construction involves a flexible structure several kilometers long, and its mechanical behavior is extremely sensitive to the operation of construction equipment such as a traction machine and a tensioning machine, and is easily affected by complex external environments, which poses extremely high requirements on construction safety and quality control.

[0003] In the prior art, the formulation of a construction scheme is mostly dependent on offline calculation based on statics such as catenary theory, or supplemented by offline simulation of finite elements. However, the ground wire is a complex dynamic system during the construction process. Traditional static analysis cannot capture the dynamic response caused by wind-induced vibration, trolley friction, and changes in traction speed, resulting in uncontrollable deviations of the construction tension from the design value, which easily leads to safety hazards such as over-tensioning or insufficient clearance.

[0004] Although a full-order finite element model can accurately describe the nonlinear mechanical characteristics of the ground wire, it has extremely high degrees of freedom, resulting in huge calculation costs. Such high calculation time costs make the full-order finite element model only suitable for offline checking before construction, and cannot meet the online calculation needs of real-time mechanical analysis and closed-loop control during the construction process.

[0005] Another major challenge faced by the prior art is the handling of uncertainties at the construction site. On the one hand, the ground wire along the line, especially in the river or valley, bears highly random and spatially non-uniform wind loads, which are difficult to accurately measure and model and are the main disturbance source leading to loss of control of the actual tension. On the other hand, the effective stiffness of the ground wire changes with time due to changes in environmental temperature, material creep, and geometric configuration changes as the construction progresses. Existing control models are usually based on fixed initial parameters, and when the model parameters deviate from the actual working conditions, i.e., the model is mismatched, the analysis accuracy and control performance will significantly decrease. SUMMARY

[0006] In view of the deficiencies of the prior art, the present application provides a method for analyzing mechanical characteristics and optimizing schemes in the construction process of a river-crossing ground wire, which solves the problems of inaccurate mechanical analysis and insufficient dynamic control accuracy in the prior art due to low calculation efficiency and the inability to estimate random loads and compensate for time-varying stiffness in real time online.

[0007] To achieve the above purpose, the present application is implemented by the following technical scheme: a method for analyzing mechanical characteristics and optimizing schemes in the construction process of a river-crossing ground wire, comprising the following steps:

[0008] Step S1: a full-order finite element model of the cross-river ground wire is established, effective modal stiffness is extracted based on a preset construction condition, and a dynamic reduced-order model is constructed by using a model reduction technique;

[0009] Step S2: during the construction process, control input and a measurement vector of a construction site are collected in real time;

[0010] Step S3: the dynamic reduced-order model is used to calculate an a priori state at a current time, and a residual vector between a predicted measurement value corresponding to the a priori state and the measurement vector is calculated;

[0011] Step S4: based on the residual vector, a residual force field vector equivalent to an unmodeled load is solved by an inversion algorithm;

[0012] Step S5: a sensitivity gradient of the residual vector or an objective function of an inversion process to the effective modal stiffness is calculated, and the effective modal stiffness is updated online according to the sensitivity gradient, to obtain calibrated effective modal stiffness;

[0013] Step S6: a prediction model is constructed based on the calibrated effective modal stiffness and the residual force field vector, an optimal control sequence in a future prediction time domain is solved by a model predictive control algorithm, and a first control instruction in the optimal control sequence is issued to a construction execution unit.

[0014] Preferably, in the step S1, the full-order finite element model of the cross-river ground wire is established and the effective modal stiffness is extracted, including:

[0015] The ground wire structure is discretized by using a nonlinear string element or a beam element, and a full-order finite element model including a mass matrix, a damping matrix and a nonlinear internal elastic restoring force vector is established;

[0016] The full-order finite element model is subjected to modal analysis at a plurality of preset construction tension levels, and an equivalent stiffness matrix reflecting the stiffness characteristics of the ground wire under different geometric configurations is obtained as the effective modal stiffness;

[0017] The first N-order principal modal shapes of the full-order finite element model are extracted to construct a projection basis matrix, and the dynamic equation of the full-order finite element model is projected into a modal space by using a Galerkin projection method, to generate the dynamic reduced-order model.

[0018] Preferably, the dynamic equation of the dynamic reduced-order model defines a system dynamic characteristic defined by a modal generalized coordinate vector, a reduced-order mass matrix, a reduced-order damping matrix, and an effective modal stiffness matrix dependent on the modal coordinate, and receives a generalized external force input composed of an external load vector projected by the projection basis matrix.

[0019] Preferably, in step S3, calculating the residual vector includes:

[0020] The reduced-order dynamic model is converted into a discrete-time state-space equation.

[0021] The residual vector at the current moment is calculated based on a measurement equation that maps the system state vector to the measurement vector through the observation matrix.

[0022] The residual vector is calculated by subtracting the predicted measurement value obtained by mapping the prior state estimate through the observation matrix from the measurement vector at the current time.

[0023] Preferably, in step S4, the solution is equivalent to the residual force field vector of the unmodeled load, including:

[0024] Construct an inversion objective function, which includes at least one measurement residual term based on the measurement weight matrix and one regularization term based on the regularization weight matrix;

[0025] The measurement residual term is used to measure the deviation between the measurement vector and the predicted measurement value, and the regularization term is used to constrain the solution space of the residual force field vector to be solved.

[0026] The gradient of the inversion objective function with respect to the residual force field vector is calculated using the adjoint method or extended Kalman filter, and the optimal residual force field vector that minimizes the inversion objective function is solved iteratively.

[0027] Preferably, in step S5, calculating the sensitivity gradient includes:

[0028] Within a preset time window, the gradient of the inversion objective function with respect to each element in the effective modal stiffness matrix is ​​calculated using the chain rule;

[0029] The calculation of the gradient involves the partial derivative of the inversion objective function with respect to the state vector of the system matrix in the state-space equation.

[0030] Preferably, in step S5, updating the effective modal stiffness online includes:

[0031] The effective modal stiffness matrix is ​​corrected using the gradient descent method.

[0032] The updated effective modal stiffness matrix is ​​equal to the original effective modal stiffness matrix minus a correction factor.

[0033] The correction amount depends on a preset learning rate, the average sensitivity gradient within the time window, and a projection operator used to maintain the positive definiteness of the stiffness matrix.

[0034] Preferably, in step S6, constructing the prediction model includes:

[0035] The calibrated effective modal stiffness obtained in step S5 is used as the stiffness parameter of the prediction model.

[0036] The residual force field vector obtained in step S4 is used as the external input of the prediction model, and the residual force field vector is extrapolated by maintaining or attenuating it in the prediction time domain.

[0037] Preferably, in step S6, solving for the optimal control sequence using a model predictive control algorithm includes:

[0038] Define a control objective function that includes a state deviation penalty term and a control increment penalty term;

[0039] Set constraints, which include at least the conductor tension safety threshold constraint, the conductor-to-ground clearance geometric constraint, and the execution capability constraint of the construction execution unit;

[0040] The optimal control sequence is obtained by solving a quadratic or nonlinear programming problem that satisfies the constraints and minimizes the control objective function within the prediction time domain.

[0041] A system for analyzing the mechanical properties and optimizing the construction scheme of cross-river conductors includes:

[0042] Offline modeling unit is used to establish a full-order finite element model of the cross-river conductor, extract effective modal stiffness, and generate a reduced-order dynamic model.

[0043] The data acquisition unit is used to collect control inputs and measurement vectors from the construction site in real time.

[0044] The residual force field inversion module is used to calculate the residual vector between the predicted measurement value of the reduced-order dynamic model and the measurement vector, and invert the residual force field vector equivalent to the unmodeled load based on the residual vector.

[0045] The EMS correction module is used to calculate the sensitivity gradient of the residual vector or inverted objective function to the effective modal stiffness, and update the effective modal stiffness online according to the sensitivity gradient;

[0046] The MPC optimization module is used to solve for the optimal control sequence in the future prediction time domain based on the updated effective modal stiffness and the inverted residual force field vector, while satisfying safety constraints.

[0047] The construction execution interface is used to receive the optimal control sequence and issue control commands to the construction equipment.

[0048] This invention provides a method for analyzing the mechanical properties and optimizing the construction scheme of cross-river conductors. It has the following beneficial effects:

[0049] 1. This invention, based on the residual vector, uses an inversion algorithm to solve for the residual force field vector equivalent to unmodeled loads. It can solve for all unmodeled loads, such as non-uniform wind fields and dynamic friction, in real time based on the residual vector between real-time acquired measurement data and model predictions. This design enables the system to accurately quantify unmeasurable disturbances during construction, significantly improving the accuracy and precision of real-time mechanical state analysis of conductors and ground wires, and overcoming the model mismatch problem caused by traditional methods relying on static load assumptions.

[0050] 2. This invention calculates the sensitivity gradient of the residual vector or the objective function of the inversion process to the effective modal stiffness, and updates the effective modal stiffness online accordingly to obtain the calibrated effective modal stiffness. Using the sensitivity gradient of the objective function of the residual force field inversion process to the effective modal stiffness, the stiffness parameters of the reduced-order dynamic model are adaptively calibrated online. This design enables the model to compensate in real time for the time-varying characteristics of structural stiffness caused by changes in ambient temperature, material creep, or geometric shape, ensuring high fidelity and robustness of the reduced-order model during long-term construction.

[0051] 3. This invention employs a predictive model constructed based on calibrated effective modal stiffness and residual force field vectors. A model predictive control algorithm is used to solve for the optimal control sequence in the future prediction time domain. The first control command in this sequence is then issued to the construction execution unit. Furthermore, this control algorithm is a dual-calibrated high-precision predictive model constructed based on the inverted residual force field and calibrated effective modal stiffness. This enables the controller to accurately predict the future dynamic response of the conductor and ground wire under complex loads and time-varying stiffness, thereby achieving smooth, accurate, and efficient control of the construction process while strictly meeting construction safety constraints. Attached Figure Description

[0052] Figure 1 This is a diagram illustrating the method steps of the present invention. Detailed Implementation

[0053] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Please see the appendix Figure 1 This invention provides a method for analyzing the mechanical properties and optimizing the construction scheme of cross-river conductors, comprising the following steps:

[0055] Step S1: Establish a full-order finite element model of the cross-river conductor, extract the effective modal stiffness based on the preset construction conditions, and construct a dynamic reduced-order model using model order reduction technology;

[0056] A full-order finite element model is built offline, effective modal stiffness is extracted, and a reduced-order dynamic model is constructed.

[0057] This step is part of the offline preparation phase, which aims to build a low-dimensional model that can accurately reflect the nonlinear dynamic characteristics of the conductor and ground wire while meeting the requirements of online real-time calculation.

[0058] Establishment of the full-order finite element model: The long-span conductor and ground wire structure is discretized using nonlinear cable elements or beam elements to establish a full-order finite element model (FOM) containing hundreds of thousands of degrees of freedom. This model accurately describes the mass matrix of the conductor and ground wire. Damping matrix And the internal elastic restoring force vector describing geometric nonlinearity and large deformation effects. .

[0059] Extraction of Effective Modal Stiffness (EMS): The stiffness of the conductor and ground wire varies nonlinearly with the tension level. This method performs modal analysis on a full-order finite element model under multiple preset key construction tension levels to obtain the equivalent stiffness matrix reflecting the stiffness characteristics of the conductor and ground wire under different geometric configurations, which serves as the initial value or benchmark of the effective modal stiffness (EMS).

[0060] Construction of the reduced-order dynamic model (ROM): Extracting the pre-FOM First-order principal mode shapes, constructing the projection basis matrix The dynamic equations of FOM are projected onto a graph using the Galerkin projection method. Zhang Cheng's low-dimensional modal space is used to generate the ROM. The dynamic equation of the ROM is:

[0061] ;

[0062] in For modal generalized coordinate vectors; This is a reduced-order quality matrix; The damping matrix is ​​of reduced order. It is the effective modal stiffness matrix; ; represents the external load vector.

[0063] Step S2: During the construction process, control inputs and measurement vectors at the construction site are collected in real time;

[0064] This step is the initial stage of the online loop, and its purpose is to obtain the current system status and the operation information being performed in real time.

[0065] Control input acquisition: Real-time acquisition and recording of construction execution units, such as traction machines and tension machines, at the current moment. The control command applied to the conductor is denoted as the control input. .

[0066] Measurement vector acquisition: Real-time acquisition of sensor data deployed at the construction site to form measurement vectors. Measurement vector It should include at least the tension sensor readings at both ends of the conductor, as well as displacement sensor or strain sensor data at the top of the tower or key points.

[0067] Step S3: Calculate the prior state at the current moment using the reduced-order dynamic model, and calculate the residual vector between the predicted measurement value and the measurement vector corresponding to the prior state;

[0068] This step aims to quantify the discrepancy between the current model predictions and actual measurements, providing input for subsequent inversion and correction. 1. Calculation of Prior States: The reduced-order dynamic model (ROM) is converted into discrete-time state-space equations. Using the previous time step... System state estimate and the control input at the current moment Calculate the current time. Prior state estimation .

[0069] 2. Calculation of predicted measurements: using the observation matrix Prior state estimate Mapping to the measurement space, we obtain the measurement value H predicted by the ROM at the current time. .

[0070] 3. Calculation of residual vector: Calculate the predicted measurement value and the actual measurement vector collected in step S2. The difference between them yields the waste vector.

[0071] ;

[0072] Residual vector It includes the total effect of model mismatch (stiffness deviation) and unmodeled loads, such as environmental disturbances.

[0073] Step S4: Based on the residual vector, solve for the residual force field vector equivalent to the unmodeled load using an inversion algorithm;

[0074] This step aims to extract the residual vector. The components caused by external disturbances are separated to estimate environmental loads in real time. 1. Construction of the inversion objective function: Construct an inversion objective function. It includes the measurement residual term and the residual force field vector. The regularization term. Residual force field vector. This is equivalent to the projection of an unmodeled load into the modal space. The objective function is as follows:

[0075] ;

[0076] in To measure the residual weight matrix, This is the regularization weight matrix.

[0077] 2. Inversion Solution: Using inversion algorithms such as the adjoint method or extended Kalman filter, calculate the residual force field vector of the inversion objective function with respect to the unknown. sensitivity gradient Solving the problem through an iterative optimization process (e.g., the conjugate gradient method) makes... Minimize the optimal residual force field vector .

[0078] Step S5: Calculate the sensitivity gradient of the residual vector or the objective function of the inversion process to the effective modal stiffness, and update the effective modal stiffness online according to the sensitivity gradient to obtain the calibrated effective modal stiffness;

[0079] This step aims to extract the residual vector. The components caused by the mismatch of model stiffness parameters are separated to achieve adaptive calibration of the model.

[0080] 1. Sensitivity gradient calculation: Within a preset rolling time window, the inversion objective function is calculated using the chain rule. Regarding the effective modal stiffness matrix Sensitivity gradient of each element This calculation involves solving for the system state vector. The partial derivatives of .

[0081] 2. Online Update of Effective Modal Stiffness: The effective modal stiffness matrix is ​​corrected in parallel using the gradient descent method. The correction formula should include a projection operator Proj to maintain the positive definiteness (i.e., physical stability) of the stiffness matrix. :

[0082] ;

[0083] in The stiffness before the update, For the updated stiffness, For learning rate, This represents the average sensitivity gradient within the time window. This step completes the adaptive calibration of the ROM to address time-varying stiffness issues during construction.

[0084] Step S6: Construct a prediction model based on the calibrated effective modal stiffness and residual force field vector, solve the optimal control sequence in the future prediction time domain through the model predictive control algorithm, and issue the first control command in the optimal control sequence to the construction execution unit.

[0085] This step is the final control decision stage, where the optimal control command is solved and issued using the dual-night-accurate local precision medical model.

[0086] 1. Construction of the prediction model: The effective modal stiffness obtained in step S5 after calibration is used as the model. Used as a stiffness parameter in the prediction model.

[0087] The residual force field vector obtained in step S4 It serves as an external input to the prediction model and is used to maintain or decay the residual force field vector extrapolated in the future prediction time domain.

[0088] 2. Solving for the optimal control sequence: The Model Predictive Control (MPC) algorithm is used, and a control objective function is defined that includes state deviation penalty terms and control increment penalty terms. At the current moment Solving the Future The optimal control sequence within a time step

[0089] ;

[0090] in To predict the state, For reference only. and This is the weight matrix.

[0091] 3. Constraint Satisfaction: The solution process must satisfy a series of physical and safety constraints, including at least the conductor / ground wire tension safety national value constraint, the conductor / ground wire clearance geometry constraint, and the execution capability constraint of the construction execution unit. 4. Command Issuance: The optimal control sequence... The first control command in The document was issued to the construction execution unit, and then the time was stepped to... Re-execute the online loops from S2 to S6.

[0092] This invention also provides a system for analyzing the mechanical properties and optimizing the construction scheme of cross-river conductors, including:

[0093] Offline modeling unit is used to establish a full-order finite element model of the cross-river conductor, extract effective modal stiffness, and generate a reduced-order dynamic model.

[0094] This unit is mainly executed before construction begins to build a high-precision, computationally efficient structural dynamics model, laying the foundation for online real-time analysis and control.

[0095] Full-order finite element modeling: Responsible for accurately discretizing the long-span conductor and ground wire structure using nonlinear cable elements or beam elements, and establishing a full-order finite element model (FOM) that includes geometric nonlinear characteristics.

[0096] Effective Modal Stiffness Extraction: This function is responsible for performing modal analysis on the FOM (Form of Mechanism) under multiple preset key construction tension levels and extracting the equivalent stiffness matrix that reflects the stiffness characteristics of the structure under multiple working conditions, which serves as the benchmark parameter for the effective modal stiffness (EMS).

[0097] Dynamics Reduction Model Generation: Based on the feature vectors of the FOM, this function uses model reduction techniques such as Galerkin projection to project the dynamic equations of the FOM to a low-dimensional modal space, thereby generating a dynamics reduction model (ROM) with significantly reduced degrees of freedom to meet the needs of online real-time computation.

[0098] The data acquisition unit is used to collect control inputs and measurement vectors from the construction site in real time.

[0099] This unit serves as the physical interface between the system and the construction site, responsible for acquiring status information and operation instructions during the construction process in real time and synchronously.

[0100] Control Input Acquisition: This function receives control commands in real-time from construction equipment such as traction machines and tension machines via communication interfaces like Modbus TCP / IP and Ethernet. These commands, such as traction / layout speed and tension setpoints, serve as the system's control inputs. .

[0101] Measurement Vector Acquisition: Responsible for synchronously acquiring data from distributed sensors via multi-channel data acquisition cards and sensor interfaces, including tension sensors at both ends of the conductor, displacement gauges, inclinometers, environmental anemometers, and thermometers at key tower tops or mid-span locations, and integrating this data into the system's current measurement vector. This unit must ensure that the timestamp synchronization and accuracy of the data meet the control cycle requirements.

[0102] The residual force field inversion module is used to calculate the residual vector between the predicted measurement value and the measurement vector of the dynamic reduced-order model, and invert the residual force field vector equivalent to the unmodeled load based on the residual vector.

[0103] Residual force field inversion module

[0104] Function Description: This module is responsible for online estimation of the combined effects of all external random disturbances not accurately described by the model, such as non-uniform wind loads and dynamic friction, on the system.

[0105] Residual vector calculation: Responsible for calculating the measurement vector acquired at the current moment. Compared with the reduced-order dynamic model (ROM) based on prior states Predicted measurement value The two are compared, the deviation between them is calculated, and a residual vector is generated. .

[0106] Residual Force Field Inversion (RFI): Responsible for executing inversion optimization algorithms, such as the adjoint method, variational assimilation, or extended Kalman filter, to obtain the residual vector. As input, the real-time inverse solution is equivalent to the residual force field vector projected onto the modal space of all unmodeled loads. This process significantly improves the system's ability to perceive uncertainties in the construction site environment.

[0107] The EMS correction module is used to calculate the sensitivity gradient of the residual vector or inverted objective function to the effective modal stiffness, and update the effective modal stiffness online based on the sensitivity gradient.

[0108] This module is responsible for online adaptive calibration of stiffness parameters in the dynamic model that may change over time in order to address model mismatch issues.

[0109] Sensitivity gradient calculation: Responsible for using the chain rule to calculate the objective function or equivalent loss function in the residual force field inversion process with respect to the effective modal stiffness (EMS) matrix. The sensitivity gradient of each element. This gradient characterizes the degree to which the stiffness parameter deviation contributes to the measurement error.

[0110] Online Update: Responsible for updating the EMS matrix based on sensitivity gradients, using gradient descent with projection constraints or similar adaptive algorithms. Perform iterative corrections. Updated stiffness parameters. It can compensate for the time-varying characteristics of structural stiffness caused by temperature changes, material creep, or changes in geometric configuration in real time, ensuring the high fidelity of ROM under long-term and multi-condition conditions.

[0111] The MPC optimization module is used to solve for the optimal control sequence in the future prediction time domain based on the updated effective modal stiffness and the inverted residual force field vector, while satisfying safety constraints.

[0112] This module is the core decision-making unit of the system, which uses a high-precision model after double calibration to predict and optimize future control commands.

[0113] Predictive model construction: Responsible for integrating the calibrated effective modal stiffness output from the EMS correction module. The residual force field vector output by the residual force field inversion module We will construct a prediction model that combines high computational efficiency and high fidelity.

[0114] Constraint optimization solution: Responsible for executing the Model Predictive Control (MPC) algorithm. Within each control cycle, with the objective function of minimizing the deviation between the predicted state and the target state and minimizing the control increment, and while satisfying all construction safety constraints, including maximum tension limits, minimum ground clearance limits, and equipment capability limits, it continuously solves for future predictions in the time domain. Optimal control sequence within .

[0115] The construction execution interface is used to receive the optimal control sequence and issue control commands to the construction equipment.

[0116] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for analyzing the mechanical properties and optimizing the construction scheme of cross-river conductors, characterized in that, Includes the following steps: Step S1: Establish a full-order finite element model of the cross-river conductor, extract the effective modal stiffness based on the preset construction conditions, and construct a dynamic reduced-order model using model order reduction technology; Step S2: During the construction process, control inputs and measurement vectors at the construction site are collected in real time; Step S3: Calculate the prior state at the current moment using the reduced-order dynamic model, and calculate the residual vector between the predicted measurement value corresponding to the prior state and the measurement vector; Step S4: Based on the residual vector, solve for the residual force field vector equivalent to the unmodeled load using an inversion algorithm; Step S5: Calculate the sensitivity gradient of the residual vector or the objective function of the inversion process to the effective modal stiffness, and update the effective modal stiffness online according to the sensitivity gradient to obtain the calibrated effective modal stiffness; Step S6: Construct a prediction model based on the calibrated effective modal stiffness and the residual force field vector, solve the optimal control sequence in the future prediction time domain through the model predictive control algorithm, and issue the first control command in the optimal control sequence to the construction execution unit.

2. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 1, characterized in that, In step S1, establishing a full-order finite element model of the cross-river conductor and extracting the effective modal stiffness includes: A full-order finite element model containing a mass matrix, a damping matrix, and a nonlinear internal elastic restoring force vector is established by using nonlinear cable elements or beam elements to discretize the conductor ground wire structure. Modal analysis was performed on the full-order finite element model under multiple preset construction tension levels to obtain the equivalent stiffness matrix that reflects the stiffness characteristics of the conductor and ground wire under different geometric configurations, which is used as the effective modal stiffness. The first N principal modes of the full-order finite element model are extracted to construct the projection basis matrix. The dynamic equations of the full-order finite element model are then projected onto the modal space using the Galerkin projection method to generate the reduced-order dynamic model.

3. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 2, characterized in that, The dynamic equations of the reduced-order dynamic model define the dynamic characteristics of the system as defined by the modal generalized coordinate vector, the reduced-order mass matrix, the reduced-order damping matrix, and the effective modal stiffness matrix dependent on the modal coordinates, and receive a generalized external force input consisting of an external load vector projected from the projection basis matrix.

4. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 1, characterized in that, In step S3, calculating the residual vector includes: The reduced-order dynamic model is converted into a discrete-time state-space equation. The residual vector at the current moment is calculated based on a measurement equation that maps the system state vector to the measurement vector through the observation matrix. The residual vector is calculated by subtracting the predicted measurement value obtained by mapping the prior state estimate through the observation matrix from the measurement vector at the current time.

5. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 1, characterized in that, In step S4, the solution is equivalent to the residual force field vector of the unmodeled load, including: Construct an inversion objective function, which includes at least one measurement residual term based on the measurement weight matrix and one regularization term based on the regularization weight matrix; The measurement residual term is used to measure the deviation between the measurement vector and the predicted measurement value, and the regularization term is used to constrain the solution space of the residual force field vector to be solved. The gradient of the inversion objective function with respect to the residual force field vector is calculated using the adjoint method or extended Kalman filter, and the optimal residual force field vector that minimizes the inversion objective function is solved iteratively.

6. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 1, characterized in that, In step S5, calculating the sensitivity gradient includes: Within a preset time window, the gradient of the inversion objective function with respect to each element in the effective modal stiffness matrix is ​​calculated using the chain rule; The calculation of the gradient involves the partial derivative of the inversion objective function with respect to the state vector of the system matrix in the state-space equation.

7. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor as described in claim 1, characterized in that, In step S5, updating the effective modal stiffness online includes: The effective modal stiffness matrix is ​​corrected using the gradient descent method. The updated effective modal stiffness matrix is ​​equal to the original effective modal stiffness matrix minus a correction factor. The correction amount depends on a preset learning rate, the average sensitivity gradient within the time window, and a projection operator used to maintain the positive definiteness of the stiffness matrix.

8. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor according to claim 1, characterized in that, In step S6, constructing the prediction model includes: The calibrated effective modal stiffness obtained in step S5 is used as the stiffness parameter of the prediction model. The residual force field vector obtained in step S4 is used as the external input of the prediction model, and the residual force field vector is extrapolated by maintaining or attenuating it in the prediction time domain.

9. The method for analyzing the mechanical properties and optimizing the scheme during the construction of a cross-river conductor according to claim 1, characterized in that, In step S6, solving for the optimal control sequence using the model predictive control algorithm includes: Define a control objective function that includes a state deviation penalty term and a control increment penalty term; Set constraints, which include at least the conductor tension safety threshold constraint, the conductor-to-ground clearance geometric constraint, and the execution capability constraint of the construction execution unit; The optimal control sequence is obtained by solving a quadratic or nonlinear programming problem that satisfies the constraints and minimizes the control objective function within the prediction time domain.

10. A system for analyzing the mechanical properties and optimizing the construction scheme of a cross-river conductor / ground wire, used in accordance with any one of claims 1-9, characterized in that, include: Offline modeling unit is used to establish a full-order finite element model of the cross-river conductor, extract effective modal stiffness, and generate a reduced-order dynamic model. The data acquisition unit is used to collect control inputs and measurement vectors from the construction site in real time. The residual force field inversion module is used to calculate the residual vector between the predicted measurement value of the reduced-order dynamic model and the measurement vector, and invert the residual force field vector equivalent to the unmodeled load based on the residual vector. The EMS correction module is used to calculate the sensitivity gradient of the residual vector or inverted objective function to the effective modal stiffness, and update the effective modal stiffness online according to the sensitivity gradient; The MPC optimization module is used to solve for the optimal control sequence in the future prediction time domain based on the updated effective modal stiffness and the inverted residual force field vector, while satisfying safety constraints. The construction execution interface is used to receive the optimal control sequence and issue control commands to the construction equipment.

Images