An adaptive magneto-rheological fluid inertial damper array system based on digital twin driving and modal space decoupling and a control method thereof

Abstract

The application discloses a kind of based on digital twin driving and modal space decoupling adaptive magnetorheological liquid force inertia damper array system and control method.System includes sensor module, control module and execution module.Sensor module gathers tower cylinder multiple-point vibration response signal and prospective wind field data;Execution module is composed of magnetorheological liquid force inertia damper arranged at different heights of tower cylinder, and each damper realizes independent continuous regulation of damping coefficient and inertia coefficient by exciting coil and flow regulating valve;Control module is based on intrinsic orthogonal decomposition algorithm and decouples multimode generalized coordinate online, combined with digital twin calculation engine and enhanced fuzzy neural network to execute double time scale collaborative decision, and the generalized modal control force is mapped into each damper independent execution instruction by independent modal space control strategy, and the array actuation phase is controlled in cooperation, to eliminate control force internal loss.System has model online evolution and hierarchical redundant fault-tolerant mechanism.

Description

Technical Field

[0001] This invention relates to the field of structural vibration control technology, specifically to an adaptive magnetorheological fluid-inertial capacitive damper array system and control method based on digital twin drive and modal space decoupling. Background Technology

[0002] As wind turbines become larger, the tower height increases while the structural stiffness decreases. Under the influence of atmospheric boundary layer turbulence, wind shear, and wake effects, the tower is prone to multimodal coupled vibrations, including first-order bending, second-order bending, and torsion. Long-term multimodal vibrations accelerate fatigue damage at the tower flanges and welds, affecting the service life and operational safety of the wind turbine.

[0003] Tuned mass dampers are widely used vibration suppression devices in engineering. Their basic principle is to add a mass-spring system to the main structure and tune the natural frequency of this system to near the target vibration frequency, dissipating vibration energy through reverse-phase work. However, tuned mass dampers have two inherent drawbacks in engineering practice: First, to achieve effective vibration suppression, tuned mass dampers typically require a solid mass block equivalent to a certain proportion of the main structure's mass. For tower structures with significant height and high sensitivity to top loads, this can result in a large additional static load and increased construction costs. Second, during long-term service, the natural frequency of a tower can drift due to factors such as changes in ambient temperature, material fatigue, and foundation stiffness degradation. Since the parameters of the tuned mass damper are fixed after manufacturing, if the main structure's frequency drifts beyond the originally set tuning range, the vibration suppression performance will significantly decrease, potentially even causing adverse effects under certain conditions.

[0004] To overcome the problem of excessive weight of the physical mass block, passive inertial capacitive dampers amplify the equivalent mass through internal mechanical or hydraulic mechanisms, reducing dependence on the physical mass to some extent. However, the inertial mass coefficient and damping coefficient of traditional inertial capacitive devices are fixed values, which also cannot solve the frequency detuning problem. The semi-active inertial capacitive damper, which combines the characteristics of magnetorheological fluid, has made an improvement on this. Under the action of an external magnetic field, the magnetorheological fluid can change the yield stress within milliseconds, thereby realizing continuous adjustment of the damping force; and when the system is powered off, the magnetorheological fluid returns to a viscous state, and the device can still operate as a passive energy dissipator, providing a certain degree of safety redundancy.

[0005] However, existing magnetorheological vibration control systems still have the following shortcomings when facing multimodal vibration control of large towers: First, existing systems generally adopt pure feedback closed-loop control logic, and the generation of control commands depends on the vibration response that has already occurred. The physical action lags behind the external excitation in time, and it is impossible to achieve forward intervention in the structural response. Second, for higher-order vibration modes, the suppression capability of single-point dampers is limited, and multi-point damper arrays need to be arranged at different heights of the tower. However, existing technologies lack effective spatial mode decoupling methods, and dampers at different elevations lack phase coordination, which can easily lead to the problem of the control forces of each damper canceling each other in space, resulting in a loss of control efficiency. Third, existing control algorithms rely on fixed rule parameters and lack the ability to identify and adaptively update the time-varying characteristics of the tower structure online, making it difficult to maintain effective control performance throughout the entire life cycle of the tower. Summary of the Invention

[0006] To overcome the existing problems and shortcomings, this invention proposes an adaptive magnetorheological hydraulic inertial capacitive damper array system based on digital twin drive and modal space decoupling for multimodal vibration control of wind turbine towers, including:

[0007] The sensor module includes: a state sensing unit for acquiring real-time vibration response signals at multiple points on the tower, and an environmental detection unit for proactively acquiring external environmental excitation data;

[0008] The execution module includes at least two damper groups, each damper group comprising at least two magnetorheological hydraulic inertial-capacitive dampers; each damper group is arranged at different heights in the tower; each magnetorheological hydraulic inertial-capacitive damper includes: a cylinder filled with magnetorheological fluid, a piston dividing the cylinder cavity into an upper chamber and a lower chamber, an excitation coil disposed on the piston, and an inertial-capacitive unit connecting the upper chamber and the lower chamber; the inertial-capacitive unit is provided with a flow regulating valve; the excitation coil and the flow regulating valve respectively receive independent control signals to achieve independent and continuous adjustment of the damping coefficient and the inertial-capacitive coefficient;

[0009] The control module has its input terminal connected to the sensor module and its output terminal connected to the execution module. The control module is configured as follows:

[0010] Based on the modal decoupling algorithm, the physical vibration signal collected by the state sensing unit is reduced in order online, and the generalized coordinates of each dominant mode of the tower are extracted.

[0011] Based on the digital twin computing engine, the feedforward generalized modal reference command is generated by combining the generalized coordinates with the look-ahead data of the environmental detection unit.

[0012] Based on the intelligent feedback algorithm, feedback generalized mode correction instructions are generated according to the real-time modal residuals;

[0013] By employing an independent modal space control strategy, the fused generalized modal commands are mapped to independent physical execution commands for each damper, and the actuation phase of each damper is controlled collaboratively to eliminate array internal losses.

[0014] Furthermore, the modal decoupling algorithm is an intrinsic orthogonal decomposition algorithm, which performs covariance matrix decomposition on the array vibration signal of the state sensing unit to extract the generalized coordinates and mode shape features of each dominant mode in real time; the digital twin computing engine is built based on a finite element reduced-order model or a long short-term memory network, and the control module is configured to perform dual time-scale collaborative control:

[0015] The feedforward branch uses minutes as the time scale and, based on the digital twin computing engine and combined with forward-looking environmental data, anticipates the dynamic response of the tower and generates the initial values ​​of the reference inertial capacity coefficient and damping for each damper. The feedback branch uses seconds as the time scale and takes real-time modal residuals and wind speed fluctuations as inputs to generate generalized modal force increments and inertial capacity fine-tuning commands for high-frequency correction.

[0016] Furthermore, the independent modal space control strategy maps and distributes the total generalized control force required for each mode to the damper actuators at each height through a pseudo-inverse algorithm; and generates controlled time delay by coordinating the inertial capacitance coefficients of each damper to achieve coordinated push-pull phase actuation between arrays, thereby eliminating the internal loss of control force under multimodal vibration.

[0017] Furthermore, the inertial-capacitance unit is a spiral pipe coiled around the outside of the cylinder. The two ends of the spiral pipe are respectively connected to the upper chamber and the lower chamber, forming a bypass passage for the flow of magnetorheological fluid. The flow regulating valve is connected in series with the spiral pipe. By adjusting the flow cross-sectional area, the effective liquid column mass of the magnetorheological fluid flowing through the bypass passage is changed, thereby realizing the continuous adjustment of the inertial-capacitance coefficient. In the damper group of each height level, the magnetorheological fluid inertial-capacitance dampers are evenly arranged circumferentially along the inner wall of the tower, and the upper and lower ends of each damper are respectively hinged to the inner wall of the tower through ball joints or universal joints.

[0018] Furthermore, the intelligent feedback algorithm is an enhanced fuzzy neural network; the input layer variables of the enhanced fuzzy neural network include: the generalized coordinate deviation of the dominant mode, the modal velocity, the current dominant mode frequency, the wind speed fluctuation value, and the ambient temperature extracted by the modal decoupling algorithm; the control objective of the enhanced fuzzy neural network is to minimize the generalized modal coordinate deviation and generalized modal acceleration of each dominant vibration mode of the tower.

[0019] Furthermore, the control module is also configured with an online model evolution mechanism: extracting the real structural response fed back by the sensor module after vibration suppression; when there is a deviation between the real response and the digital twin pre-simulation response, using the recursive least squares method to identify the drift of the actual natural frequency of the tower online, synchronously updating the stiffness matrix and mass matrix parameters in the digital twin calculation engine, and dynamically adjusting the rules and weight parameters of the intelligent feedback algorithm.

[0020] Furthermore, the system also has a hierarchical redundancy fault-tolerant mechanism: when the forward signal is lost or a communication failure occurs, the system automatically degrades to a pure feedback physical regulation mode; when the external excitation is lower than a set threshold, the system enters a low-power sleep mode; when the system fails due to power failure, the magnetorheological fluid-hydraulic inertial-capacitive damper operates in a passive damping mode.

[0021] An adaptive magnetorheological hydraulic inertial capacitive damper array control method based on digital twin drive and modal space decoupling, characterized in that the method includes the following steps:

[0022] Step 1: Multi-source signal synchronous sensing. The state sensing unit continuously acquires real-time physical vibration response signals of multiple points on the tower, and the environmental detection unit proactively acquires wind field vectors and ambient temperature excitation signals.

[0023] Step 2: Online modal decoupling and state reconstruction. The control module performs online modal order reduction on the physical vibration signal and extracts the generalized coordinates and transient features of each dominant mode of the tower.

[0024] Step 3: Dual timescale collaborative decision-making. The feedforward branch, based on a digital twin engine and combined with forward-looking wind field data, anticipates the tower's dynamic response and generates generalized modal reference commands; the feedback branch, based on an intelligent feedback algorithm, generates generalized modal correction commands according to real-time modal residuals.

[0025] Step 4: Spatial mapping and command allocation. The feedforward and feedback generalized modal commands are fused together. The total generalized modal control force is mapped to the physical execution commands of each height damper using the independent modal spatial control strategy, and the actuation phase difference between the arrays is calculated.

[0026] Step 5: Array coordination and physical energy consumption. The execution module independently adjusts the excitation coil current and flow regulating valve opening of each magnetorheological hydraulic inertial capacitive damper to dynamically change the damping coefficient and inertial capacitive coefficient of each damper in a push-pull coordination mode, thereby eliminating array internal losses.

[0027] Step Six: Closed-loop evaluation and model evolution, continuously monitor the vibration suppression effect. When there is a deviation between the actual response and the pre-simulated response, identify the actual natural frequency drift of the tower online and adaptively update the parameters of the digital twin model and the weight parameters of the intelligent feedback algorithm.

[0028] Furthermore, in step two, the online modal order reduction adopts the intrinsic orthogonal decomposition algorithm to perform covariance matrix decomposition on the multi-point physical vibration signal and extract the generalized coordinates of the first-order bending, second-order bending and torsional modes in real time; in step three, the time scale of the feedforward branch is on the order of minutes and the time scale of the feedback branch is on the order of seconds; in step four, the independent modal space control strategy distributes the generalized control force required for each mode to each height damper through a pseudo-inverse matrix algorithm, and generates controlled time delay by coordinating the inertial capacitance coefficient of each damper to achieve push-pull phase coordination between different height damper groups.

[0029] Furthermore, in step six, the online identification adopts the recursive least squares method; the method also includes step seven: high-toughness degradation and fault-tolerant guarantee, when the external stimulus is lower than the set threshold, the system enters a low-power sleep mode; when a look-ahead signal is lost or a communication failure occurs, the system automatically degrades to a pure feedback physical regulation mode or a passive damping mode according to the hierarchical redundancy mechanism.

[0030] Beneficial effects:

[0031] This invention combines a digital twin computing engine with feedforward control. The control module can use wind field data acquired in advance by the environmental detection unit to pre-simulate the tower dynamic response and generate reference control commands before the structural response occurs. This changes the working mode of traditional semi-active control systems that rely on feedback after response, and effectively reduces the time lag of control commands relative to external excitation.

[0032] This invention uses a spiral pipe-type hydraulic inertial-capacitance unit to replace the solid mass block of the traditional tuned mass damper. With a smaller liquid mass, a larger equivalent inertial-capacitance coefficient is generated through the hydraulic amplification effect of the spiral pipe. This significantly reduces the additional static load applied to the tower structure by the damping device, which is beneficial to reducing the tower construction cost.

[0033] This invention achieves independent and continuous adjustment of two parameters in a single magnetorheological hydraulic inertial-capacitive damper by adjusting the damping coefficient and inertial-capacitive coefficient through two independent control channels: the excitation coil and the flow regulating valve. This enables the device to adapt to changes in the dynamic characteristics of the tower over time, thereby solving the frequency detuning problem present in fixed-parameter devices.

[0034] This invention employs an intrinsic orthogonal decomposition algorithm to perform online modal decoupling of multi-point physical vibration signals, extracts the generalized coordinates of each dominant mode, and maps and distributes the generalized modal control force to each height damper through an independent modal spatial control strategy. At the same time, it coordinates the actuation phase of each damper, enabling dampers at different elevations in the array to act in concert according to the spatial distribution of each vibration mode, avoiding the cancellation of control forces of each damper in space, and improving the comprehensive suppression efficiency of the multi-point damper array for multimodal vibration.

[0035] This invention establishes an online model evolution mechanism. The control module uses the recursive least squares method to continuously identify changes in the actual natural frequency of the tower and updates the stiffness matrix and mass matrix parameters of the digital twin computing engine and the weight parameters of the intelligent feedback algorithm in real time. This enables the system to track changes in the dynamic characteristics of the tower caused by factors such as material aging and temperature changes, and maintain the consistency between the control parameters and the actual state of the structure throughout the entire life cycle of the tower.

[0036] This invention establishes a hierarchical redundancy fault-tolerant mechanism. When the forward signal is lost or communication fails, the system automatically degrades to a pure feedback regulation mode. When the system is powered off, the magnetorheological fluid-inertial capacitive damper operates in a passive damping mode, ensuring the basic operational safety of the system under various fault conditions. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is an overall block diagram of the adaptive magnetorheological fluid-hydraulic inertial-capacitive damper array system described in this invention.

[0039] Figure 2 This is a flowchart illustrating the operation of the control method described in this invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Example 1:

[0042] like Figure 1 As shown, an adaptive magnetorheological hydraulic inertial capacitive damper array system based on digital twin drive and modal space decoupling is presented. The system consists of three parts: a sensor module, a control module, and an execution module. The three parts are connected via industrial Ethernet for millisecond-level low-latency communication. The system is illustrated using a 120-meter-high steel wind turbine tower as an application example.

[0043] The sensor module comprises two types: a state sensing unit and an environmental detection unit. The state sensing unit consists of several triaxial accelerometers and displacement sensors, respectively located on the platform below the nacelle at the top of the tower, at two-thirds of the tower height, and at the height of each damper array. A total of nine triaxial accelerometers are installed to simultaneously acquire the X, Y, and Z-axis vibration acceleration responses at multiple points on the tower. Displacement sensors are installed at the top of the tower to monitor horizontal displacement at the top. The environmental detection unit includes a lidar wind field scanner, an ultrasonic anemometer, and a temperature sensor, installed at the top of the nacelle. The lidar wind field scanner proactively acquires wind speed vector field data within a range of 0 to 200 meters in front of the tower, providing advance warning time for feedforward control. The temperature sensor collects ambient temperature data. All sensor signals are connected to an industrial-grade data acquisition box located at the bottom of the tower. After signal conditioning, low-pass filtering, and analog-to-digital conversion, the signals are transmitted to the control module in real time at a 100Hz sampling rate.

[0044] The execution module includes two damper groups, totaling seven magnetorheological-hydraulic inertial-capacitive dampers, arranged at different elevations inside the tower. The first damper group is located at a height of 80 meters inside the tower, a position where the first-order modal strain energy is relatively high. It consists of four magnetorheological-hydraulic inertial-capacitive dampers evenly arranged in a ring at 90° intervals along the inner wall of the tower. The upper suspension point of each damper is located at a height of 81 meters, and the lower suspension point is located at a height of 79 meters. Both ends are hinged to the inner wall of the tower via self-lubricating ball bearings, ensuring that each damper transmits only axial force. The second damper group is located at a height of 60 meters inside the tower, a position where the second-order modal strain energy is relatively high. It consists of three magnetorheological-hydraulic inertial-capacitive dampers evenly arranged in a ring at 120° intervals along the inner wall of the tower. The upper suspension point is located at a height of 61 meters, and the lower suspension point is located at a height of 59 meters. Both ends are also hinged to the inner wall of the tower via ball bearings.

[0045] The specific structure of each magnetorheological fluid inertial capacitive damper is as follows: the cylinder is filled with magnetorheological fluid, one end of the piston rod extends to the outside of the cylinder, and the other end is equipped with a piston, which divides the inner cavity of the cylinder into an upper chamber and a lower chamber; an excitation coil is provided on the piston to generate an adjustable magnetic field to change the yield strength of the magnetorheological fluid, thereby adjusting the magnitude of the damping force; the inertial capacitive unit is a spiral pipe wound around the outside of the cylinder, with both ends of the spiral pipe connected to the upper chamber and the lower chamber respectively, forming a bypass passage for the flow of magnetorheological fluid; a flow regulating valve is connected in series at the inlet of the spiral pipe, and the flow regulating valve is a high-speed needle valve or a proportional butterfly valve whose flow cross-sectional area can be precisely controlled by a high-frequency servo motor or a stepper motor. The control module outputs two independent control signals to the excitation coil and the flow regulating valve respectively: the first control signal adjusts the current flowing through the excitation coil to change the magnetic field strength of the magnetorheological fluid at the piston orifice, thereby continuously adjusting the damping coefficient; the second control signal adjusts the opening of the flow regulating valve to change the effective mass of the magnetorheological fluid flowing through the bypass passage of the spiral pipe, achieving independent and continuous adjustment of the inertial volume coefficient. When the piston moves relative to the cylinder, the magnetorheological fluid flows back and forth between the upper and lower chambers through the bypass passage of the spiral pipe. Its liquid column inertial effect generates an equivalent inertial volume force. By changing the opening of the flow regulating valve to adjust the effective mass of the liquid column participating in the flow, the inertial volume coefficient can be continuously adjusted, generating a large equivalent inertial volume coefficient with a small actual liquid mass through the hydraulic amplification effect of the spiral pipe.

[0046] The control module consists of an industrial control computer and related hardware installed in the control cabinet at the bottom of the tower. It is connected to the data acquisition box via an industrial Ethernet switch. The output end integrates a multi-channel analog output board and a high-frequency response high-power current amplifier, which are connected to the excitation coils of the seven dampers and the flow regulating valve drivers, respectively. The control module software is developed based on a Python environment and embeds a digital twin computing engine and a spatial decoupling algorithm. It is configured to sequentially execute four core operations: modal decoupling, dual-timescale collaborative decision-making, spatial mapping allocation, and online model evolution, as detailed below:

[0047] (1) Online Modal Decoupling: The control module uses the intrinsic orthogonal decomposition algorithm to perform online mode reduction processing on the physical vibration signals from the 9-channel accelerometers: a covariance matrix is ​​constructed for the multi-point vibration signals, eigenvalue decomposition is performed on the covariance matrix, the eigenvectors corresponding to the largest eigenvalues ​​are extracted as the dominant mode shapes of the tower, and the physical space vibration signals are projected onto the directions of each dominant mode shape to obtain the generalized coordinate time history of each mode. Through this process, the generalized coordinates of the first-order bending, second-order bending and torsional modes can be separated in real time, and the chaotic physical vibration signals at multiple points can be transformed into generalized state variables in the modal space, eliminating the response identification bias caused by multi-modal coupling, and providing the decoupled modal state for subsequent decision-making.

[0048] (2) Dual timescale collaborative decision-making: such as Figure 2As shown, based on the decoupled modal generalized coordinates, the control module initiates parallel simulations of the feedforward and feedback branches. The feedforward branch runs a real-time digital twin constructed based on a finite element reduced-order model on a minute-by-minute timescale. The digital twin computing engine retrieves the future wind load prediction sequence obtained from lidar scanning, uses the modal states decoupled by the current intrinsic orthogonal decomposition algorithm as initial conditions, and pre-simulates the future dynamic response of the tower in the twin space. It calculates in advance the initial values ​​of the reference inertia coefficients and damping values ​​of each damper required to cope with macroscopic wind load changes, and pre-sets the parameters of each damper on a minute-by-minute timescale to eliminate the control lag of the large inertia structure. The feedback branch operates on a second-scale timescale and employs an enhanced fuzzy neural network to process high-frequency transient responses. The network uses the dominant mode generalized coordinate deviation, modal velocity, current dominant mode frequency, wind speed fluctuation, and ambient temperature extracted by the intrinsic orthogonal decomposition algorithm as input layer variables. The control objective is to minimize the generalized modal coordinate deviation and generalized modal acceleration of each dominant mode of the tower. After fuzzifying the input variables, an adaptive neural fuzzy inference system is used for online inference, outputting real-time generalized modal force increments and inertial capacitance fine-tuning commands for turbulent disturbances on a second-scale timescale, and performing high-frequency corrections to the feedforward reference commands.

[0049] (3) Spatial mapping and array coordination: The feedforward and feedback generalized modal commands are fused and superimposed in the independent modal space control unit to obtain the total generalized control force required for each mode. The control module maps the total generalized modal control force from the modal space to the physical space through a pseudo-inverse matrix algorithm, and distributes it into independent physical execution commands for 7 execution ends, namely the precise current value of each excitation coil and the opening value of each flow regulating valve. In view of the spatial characteristics of the second-order bending mode with reverse amplitude at different heights, the system automatically calculates the required actuation phase difference between the damper group at 60 meters and the damper group at 80 meters through the independent modal space control strategy. By coordinating the adjustment of the inertial capacitance coefficient of each damper to generate controlled time delay, the two sets of damper arrays are driven to actuate in a push-pull cooperative mode, so that the control forces of each damper are superimposed in space rather than canceled, thus eliminating the internal loss of control force under multimodal vibration. The current amplifier of the execution module applies current commands to each excitation coil, and the stepper motor drives each flow regulating valve to be precisely positioned, so as to realize the independent real-time adjustment of the damping coefficient and inertial capacitance coefficient of each damper.

[0050] (4) Online model evolution and hierarchical fault tolerance: such as Figure 2As shown, while the array works in coordination, the sensor module continuously feeds back the latest real structural response to the control module. The control module compares the measured response with the digital twin pre-simulation response. If there is a steady-state deviation between the two, the system calls the recursive least squares method to identify the deviation data online, extracts the drift of the tower's actual natural frequency, and automatically updates the stiffness matrix and mass matrix parameters in the digital twin calculation engine to ensure the continuous accuracy of the feedforward pre-simulation. At the same time, the system periodically evaluates the residual energy of each mode and dynamically adjusts the fuzzy rules and weight parameters of the enhanced fuzzy neural network to keep the feedback correction command matched with the current dynamic characteristics of the tower, thereby continuously tracking the natural frequency drift caused by factors such as material aging and temperature changes throughout the entire life cycle of the tower.

[0051] In addition, the system has a hierarchical redundancy fault-tolerant mechanism: when the external excitation is lower than the set threshold, the system control loop enters a low-power sleep mode, and each damper maintains its current parameter state; when the lidar look-ahead signal is lost or a partial communication failure occurs, the system automatically shields the feedforward branch and degrades to a pure feedback physical adjustment mode driven only by the enhanced fuzzy neural network, continuing to correct the real-time vibration response; when the system is completely powered off and fails, the excitation coils of each damper are de-energized, the magnetorheological fluid returns to a viscous state, and all seven magnetorheological fluid inertial capacitive dampers continue to operate in passive damping mode, ensuring the basic operational safety of the tower under various fault conditions.

[0052] Example 2:

[0053] This embodiment provides a control method based on the system described in Embodiment 1 above, such as... Figure 2 As shown, the following steps are executed cyclically.

[0054] Step 1, Multi-source signal synchronous sensing: After the system starts, the state sensing unit synchronously collects 9 acceleration signals and tower top displacement signals at a sampling rate of 100Hz. After baseline drift compensation and low-pass filtering preprocessing, the signals are sent to the control module. The environmental detection unit synchronously acquires lidar wind speed vector field, ultrasonic wind speed and direction, and ambient temperature data, and transmits them to the feedforward branch of the control module in real time.

[0055] Step 2, Online Modal Decoupling and State Reconstruction: The control module calls the intrinsic orthogonal decomposition algorithm on the preprocessed multi-point physical vibration signal to construct the vibration signal covariance matrix and perform eigenvalue decomposition. It extracts the generalized coordinates and mode shape features of the first-order bending, second-order bending and torsional modes in real time, and completes the state reconstruction from physical space to modal space.

[0056] Step 3, Dual Time Scale Collaborative Decision Making: The digital twin computing engine of the feedforward branch receives future wind load data acquired by lidar, and uses the current modal generalized coordinates as initial conditions to predict the tower dynamic response within a minute-scale time scale, generating the reference inertial capacity coefficient and initial damping value for each damper; the enhanced fuzzy neural network of the feedback branch uses the current modal generalized coordinate deviation, modal velocity, dominant modal frequency, wind speed fluctuation value and ambient temperature as inputs, and infers and outputs the generalized modal force increment and inertial capacity fine-tuning command within a second-scale time scale.

[0057] Step 4, Spatial Mapping and Command Allocation: The independent modal spatial control unit integrates the feedforward and feedback generalized modal commands, and maps the total generalized modal control force into the excitation current command and flow regulating valve opening command of each of the 7 dampers through a pseudo-inverse matrix algorithm, and calculates the actuation phase difference between the damper group at 60 meters and the damper group at 80 meters.

[0058] Step 5, Array Coordination and Physical Energy Dissipation: The execution module receives independent commands from each damper, the current amplifier applies excitation current to the excitation coil of each damper, and the stepper motor drives each flow regulating valve to position at a specified opening degree. The damping coefficient and inertia coefficient of each damper are independently adjusted in push-pull coordination mode to eliminate array internal losses and dissipate multimodal vibration energy.

[0059] Step Six, Closed-Loop Evaluation and Model Evolution: The sensor module feeds back the new round of real structural response to the control module, which compares it with the digital twin pre-simulation results. If there is a frequency deviation, the recursive least squares method is used to identify the actual natural frequency drift of the tower online, and the stiffness matrix and mass matrix parameters of the digital twin calculation engine are updated adaptively. The residual energy of each mode is evaluated simultaneously, and the rules and weight parameters of the enhanced fuzzy neural network are dynamically adjusted. After completing one evolution iteration, the process returns to Step One and repeats.

[0060] Step 7, High-Resilience Degradation and Fault Tolerance: The system continuously monitors the look-ahead signal status and communication link status in each control cycle; when the look-ahead signal is lost or communication fails, it automatically degrades to a pure feedback physical regulation mode; when the external excitation is lower than the set threshold, it enters a low-power sleep mode; when the system is powered off, each magnetorheological fluid inertial capacitive damper automatically switches to passive damping mode.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive magnetorheological hydraulic inertial capacitive damper array system based on digital twin drive and modal space decoupling, used for multimodal vibration control of wind turbine towers, characterized in that, include: The sensor module includes: a state sensing unit for acquiring real-time vibration response signals at multiple points on the tower, and an environmental detection unit for proactively acquiring external environmental excitation data; The execution module includes at least two damper groups, each damper group comprising at least two magnetorheological hydraulic inertial-capacitive dampers; each damper group is arranged at different heights in the tower; each magnetorheological hydraulic inertial-capacitive damper includes: a cylinder filled with magnetorheological fluid, a piston dividing the cylinder cavity into an upper chamber and a lower chamber, an excitation coil disposed on the piston, and an inertial-capacitive unit connecting the upper chamber and the lower chamber; the inertial-capacitive unit is provided with a flow regulating valve; the excitation coil and the flow regulating valve respectively receive independent control signals to achieve independent and continuous adjustment of the damping coefficient and the inertial-capacitive coefficient; The control module has its input terminal connected to the sensor module and its output terminal connected to the execution module. The control module is configured as follows: Based on the modal decoupling algorithm, the physical vibration signal collected by the state sensing unit is reduced in order online, and the generalized coordinates of each dominant mode of the tower are extracted. Based on the digital twin computing engine, the feedforward generalized modal reference command is generated by combining the generalized coordinates with the look-ahead data of the environmental detection unit. Based on the intelligent feedback algorithm, feedback generalized mode correction instructions are generated according to the real-time modal residuals; By employing an independent modal space control strategy, the fused generalized modal commands are mapped to independent physical execution commands for each damper, and the actuation phase of each damper is controlled collaboratively to eliminate array internal losses.

2. The system according to claim 1, characterized in that, The modal decoupling algorithm is an intrinsic orthogonal decomposition algorithm. The intrinsic orthogonal decomposition algorithm performs covariance matrix decomposition on the array vibration signal of the state sensing unit and extracts the generalized coordinates and mode shape features of each dominant mode in real time. The digital twin computing engine is built based on a finite element reduced-order model or a long short-term memory network, and the control module is configured to perform dual-timescale collaborative control: The feedforward branch uses minutes as the time scale. Based on the digital twin computing engine and combined with forward-looking environmental data, it anticipates the dynamic response of the tower and generates the initial values ​​of the reference inertial coefficient and damping of each damper. The feedback branch uses seconds as the time scale and takes real-time modal residuals and wind speed fluctuations as inputs to generate generalized modal force increments and inertial capacitance fine-tuning commands for high-frequency correction.

3. The system according to claim 1, characterized in that, The independent modal space control strategy maps and distributes the total generalized control force required for each mode to the damper actuators at each height through a pseudo-inverse algorithm; and generates controlled time delay by coordinating the inertial capacitance coefficients of each damper to achieve coordinated push-pull phase actuation between arrays, thereby eliminating control force internal friction under multimodal vibration.

4. The system according to claim 1, characterized in that, The inertial-capacitance unit is a spiral pipe coiled around the outside of the cylinder. The two ends of the spiral pipe are respectively connected to the upper chamber and the lower chamber, forming a bypass passage for the flow of magnetorheological fluid. The flow regulating valve is connected in series with the spiral pipe. By adjusting the flow cross-sectional area, the effective liquid column mass of the magnetorheological fluid flowing through the bypass passage is changed, thereby realizing the continuous adjustment of the inertial-capacitance coefficient. In the damper group of each height level, the magnetorheological fluid inertial-capacitance dampers are evenly arranged circumferentially along the inner wall of the tower, and the upper and lower ends of each damper are respectively hinged to the inner wall of the tower by ball joints or universal joints.

5. The system according to claim 1, characterized in that, The intelligent feedback algorithm is an enhanced fuzzy neural network; The input layer variables of the enhanced fuzzy neural network include: the generalized coordinate deviation of the dominant mode, the modal velocity, the frequency of the current dominant mode, the wind speed fluctuation value, and the ambient temperature extracted by the modal decoupling algorithm; The control objective of the enhanced fuzzy neural network is to minimize the generalized modal coordinate deviation and generalized modal acceleration of each dominant vibration mode of the tower.

6. The system according to claim 1, characterized in that, The control module is also equipped with an online model evolution mechanism: extracting the real structural response fed back by the sensor module after vibration suppression; when there is a deviation between the real response and the digital twin pre-simulation response, using the recursive least squares method to identify the drift of the actual natural frequency of the tower online, synchronously updating the stiffness matrix and mass matrix parameters in the digital twin calculation engine, and dynamically adjusting the rules and weight parameters of the intelligent feedback algorithm.

7. The system according to claim 1, characterized in that, The system also has a hierarchical redundancy fault-tolerance mechanism: when the forward signal is lost or a communication failure occurs, the system automatically degrades to a pure feedback physical regulation mode; when the external excitation is lower than a set threshold, the system enters a low-power sleep mode; when the system fails due to power failure, the magnetorheological fluid inertial capacitive damper operates in a passive damping mode.

8. A method for controlling an adaptive magnetorheological fluid-hydraulic inertial-capacitive damper array based on digital twin drive and modal space decoupling, according to any one of claims 1 to 7, characterized in that, The method includes the following steps: Step 1: Multi-source signal synchronous sensing. The state sensing unit continuously acquires real-time physical vibration response signals of multiple points on the tower, and the environmental detection unit proactively acquires wind field vectors and ambient temperature excitation signals. Step 2: Online modal decoupling and state reconstruction. The control module performs online modal order reduction on the physical vibration signal and extracts the generalized coordinates and transient features of each dominant mode of the tower. Step 3: Dual timescale collaborative decision-making. The feedforward branch, based on a digital twin engine and combined with forward-looking wind field data, anticipates the tower's dynamic response and generates generalized modal reference commands; the feedback branch, based on an intelligent feedback algorithm, generates generalized modal correction commands according to real-time modal residuals. Step 4: Spatial mapping and command allocation. The feedforward and feedback generalized modal commands are fused together. The total generalized modal control force is mapped to the physical execution commands of each height damper using the independent modal spatial control strategy, and the actuation phase difference between the arrays is calculated. Step 5: Array coordination and physical energy consumption. The execution module independently adjusts the excitation coil current and flow regulating valve opening of each magnetorheological hydraulic inertial capacitive damper to dynamically change the damping coefficient and inertial capacitive coefficient of each damper in a push-pull coordination mode, thereby eliminating array internal losses. Step Six: Closed-loop evaluation and model evolution, continuously monitor the vibration suppression effect. When there is a deviation between the actual response and the pre-simulated response, identify the actual natural frequency drift of the tower online and adaptively update the parameters of the digital twin model and the weight parameters of the intelligent feedback algorithm.

9. The method according to claim 8, characterized in that, In step two, the online modal reduction uses an intrinsic orthogonal decomposition algorithm to decompose the covariance matrix of the multi-point physical vibration signal and extract the generalized coordinates of the first-order bending, second-order bending, and torsional modes in real time. In step three, the time scale of the feedforward branch is on the order of minutes, and the time scale of the feedback branch is on the order of seconds. In step four, the independent modal space control strategy uses a pseudo-inverse matrix algorithm to distribute the generalized control force required for each mode to each height damper, and generates controlled time delay by coordinating the inertial capacitance coefficient of each damper to achieve push-pull phase coordination between different height damper groups.

10. The method according to claim 8, characterized in that, In step six, the online identification adopts the recursive least squares method; the method also includes step seven: high-toughness degradation and fault-tolerant protection. When the external stimulus is lower than the set threshold, the system enters a low-power sleep mode; when a forward signal is lost or a communication failure occurs, the system automatically degrades to a pure feedback physical regulation mode or a passive damping mode according to the hierarchical redundancy mechanism.

Images