Numerical-physical hybrid test method for floating wind turbine based on real-time interpolation

Abstract

This invention relates to the field of offshore wind power generation testing technology, and discloses a numerical-physical hybrid test method for floating wind turbines based on real-time interpolation. The method involves loading an aerodynamic thrust load database; cyclically executing the following steps at a set time step: collecting motion attitude data and absolute wind field data of the floating wind turbine model; calculating the real-time relative inflow wind speed at the hub; interpolating the prototype aerodynamic thrust from the database and converting it into the target aerodynamic thrust; calculating the feedforward duty cycle reference value of the thrust actuator based on electromechanical mapping relationships; collecting the actual physical thrust to calculate the closed-loop tracking error; generating feedback compensation commands using a proportional-integral-differential algorithm; fusing the feedforward and feedback commands to generate a comprehensive drive duty cycle; and outputting digital pulse signals to the thrust actuator to adjust the actual physical thrust. This invention reduces the real-time calculation load of the system, improves the thrust simulation accuracy and dynamic tracking stability, and enhances the physical realism of the hybrid test.

Description

Technical Field

[0001] This invention relates to the field of offshore wind power generation testing technology, specifically to a numerical-physical hybrid testing method for floating wind turbines based on real-time interpolation. Background Technology

[0002] Floating wind turbines are widely used in deep-sea wind power development. Before actual deployment, they need to undergo pool model tests to verify their hydrodynamic performance and system stability. Since traditional pool tests cannot simultaneously satisfy the similarity criteria of Froude number and Reynolds number, a numerical-physical hybrid testing method is often used. This method arranges a scaled-down floating platform model in a physical pool to simulate hydrodynamics, and installs a thrust actuator on top of the model. The aerodynamic load is calculated in real time by a computer and converted into a control signal to drive the actuator to apply an equivalent aerodynamic thrust to the model.

[0003] In existing numerical-physical hybrid experiments, the control system needs to run an aerodynamic coupling analysis program in real time to solve for the aerodynamic loads at each step length. This solution process consumes computational resources, resulting in a long solution time step, which can easily cause delays in signal transmission and control output, affecting the real-time performance and continuity of the experiment.

[0004] Floating wind turbines experience spatial motion under the combined effects of wind and waves. Their swaying and rolling attitude changes alter the relative inflow velocity across the impeller sweep surface, generating aerodynamic damping effects. Some hybrid testing methods directly use absolute wind speed for load calculations, failing to incorporate the floating platform's own velocity into the relative wind speed calculation. This results in the physical thrust applied to the model not reflecting the true motion coupling state, reducing the accuracy of load simulation.

[0005] Furthermore, thrust actuators exhibit mechanical inertia and response lag when outputting thrust. Existing thrust control strategies largely rely on a single closed-loop error feedback adjustment. When sudden changes in wind speed or model attitude cause a step increase in target thrust, the control system exhibits tracking errors and slow response. When the system maintains steady-state tracking errors for an extended period, the single closed-loop control loop is prone to integral saturation, causing the calculated control commands to exceed the physical operating limits of the thrust actuator. Moreover, when changes in target thrust cause the error direction to reverse, the system's accumulated error integral value is high, preventing it from exiting saturation in a timely manner, leading to desaturation delay, overshoot and oscillation in thrust control, and affecting thrust loading accuracy and experimental reliability. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a numerical-physical hybrid test method for floating wind turbines based on real-time interpolation. This method solves the technical problems in existing numerical-physical hybrid tests for floating wind turbines, such as high real-time calculation load of complex aerodynamic loads, low dynamic tracking accuracy of thrust loading, and failure to accurately reflect the coupling effect of platform motion on relative wind speed.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation. The method cyclically executes the following experimental control process at discrete time points according to a set time step: loading an aerodynamic thrust load database; acquiring motion attitude data of the floating wind turbine model and, combined with pre-acquired absolute wind field data, calculating the real-time relative inflow velocity at the hub; interpolating the prototype aerodynamic thrust from the aerodynamic thrust load database based on the real-time relative inflow velocity, and converting it into a target aerodynamic thrust based on a scaling criterion; calculating the forward thrust of a pre-configured thrust actuator based on the target aerodynamic thrust and a pre-calibrated electromechanical mapping relationship. The system feeds a reference value for the pulse width modulation duty cycle and generates a feedforward control command; it acquires the original physical thrust applied to the floating wind turbine model by the thrust actuator through a pre-configured force sensor and filters it to obtain the actual physical thrust; it calculates the thrust closed-loop tracking error between the target aerodynamic thrust and the actual physical thrust, calculates the pulse width modulation duty cycle compensation amount using a proportional-integral-derivative control algorithm, and generates a feedback compensation command; it fuses the feedforward control command and the feedback compensation command to generate a comprehensive drive duty cycle, and converts it into a digital pulse width modulation signal to be output to the thrust actuator to adjust the actual physical thrust.

[0008] The construction process of the aerodynamic thrust load database includes: establishing a full-scale aerodynamic-hydraulic servo-elastic coupling simulation model of the floating wind turbine, and configuring a variable pitch control strategy based on generator speed feedback and a generator electromagnetic torque control strategy in the full-scale aerodynamic-hydraulic servo-elastic coupling simulation model. Environmental input conditions covering the target test conditions are set for time-domain simulation calculations, and the relative inflow wind speed and corresponding prototype aerodynamic thrust at the floating wind turbine hub are extracted at each integral time step. The extracted relative inflow wind speed and corresponding prototype aerodynamic thrust are cleaned and discretized. After removing transient distortion data during the simulation start-up phase, the remaining time-history data are divided into multiple wind speed intervals of fixed step length according to the magnitude of the relative inflow wind speed and averaged statistically, generating a static lookup table with relative inflow wind speed as the independent variable and prototype aerodynamic thrust as the dependent variable, which is then encapsulated and stored as the aerodynamic thrust load database. By constructing the aerodynamic thrust load database offline as described above, the real-time calculation load on the control hardware during the test can be reduced.

[0009] The calibration process for the pre-calibrated electromechanical mapping relationship includes: sending a pulse width modulation (PWM) drive signal with progressively increasing duty cycle to the thrust actuator in a calibration environment; and synchronously recording the steady-state physical thrust and steady-state rotational speed values ​​after maintaining a stable time at each duty cycle step. A least-squares method is used for polynomial curve fitting to extract the functional relationship between the steady-state physical thrust and steady-state rotational speed values, obtaining a thrust-rotational speed mapping model containing quadratic terms. Similarly, a rotational speed duty cycle mapping model containing quadratic terms is obtained by extracting the functional relationship between the steady-state rotational speed values ​​and the duty cycle of the PWM drive signal. The thrust-rotational speed mapping model and the rotational speed duty cycle mapping model are pre-stored for use in experiments. During experiments, the target rotational speed is calculated based on the target aerodynamic thrust combined with the thrust-rotational speed mapping model, and then the feedforward PWM duty cycle reference value is calculated based on the target rotational speed combined with the rotational speed duty cycle mapping model.

[0010] The calculation of the real-time relative inflow velocity at the hub, based on pre-acquired absolute wind field data, specifically includes: extracting the sway displacement and pitch angle from the motion attitude data; performing time-differencing operations using a first-order backward difference algorithm; and smoothing the data using a digital low-pass filter to obtain the sway velocity and pitch angular velocity of the floating wind turbine model. According to the principles of rigid body kinematics, the sway velocity of the floating wind turbine model is superimposed with the projection of the linear velocity caused by the pitch angular velocity in the horizontal direction to calculate the model-induced velocity at the hub of the floating wind turbine model. Based on the Froude similarity criterion, the model-induced velocity is converted into a prototype-induced velocity according to the square root of the geometric scaling ratio. This prototype-induced velocity is then vector-superimposed with the absolute inflow velocity value of the absolute wind field data at the current discrete time point to obtain the real-time relative inflow velocity used for querying the aerodynamic thrust load database.

[0011] The step of obtaining the prototype aerodynamic thrust by interpolating the real-time relative inflow wind speed in the aerodynamic thrust load database specifically includes: determining whether the real-time relative inflow wind speed is within the effective working wind speed range formed by the cut-in wind speed and cut-out wind speed of the preset floating wind turbine prototype; if the real-time relative inflow wind speed exceeds the effective working wind speed range, then assigning the prototype aerodynamic thrust corresponding to the current discrete time point to zero; if the real-time relative inflow wind speed is within the effective working wind speed range, then using a binary search algorithm to determine the wind speed interval where the real-time relative inflow wind speed is located in the aerodynamic thrust load database, obtaining the lower boundary discrete wind speed point value, the upper boundary discrete wind speed point value, and the prototype aerodynamic thrust corresponding to the lower boundary discrete wind speed point value and the upper boundary discrete wind speed point value respectively; and using a linear interpolation algorithm to calculate the prototype aerodynamic thrust corresponding to the real-time relative inflow wind speed at the current discrete time point. Subsequently, according to the similarity criteria of fluid mechanics and dynamics, the prototype aerodynamic thrust obtained by interpolation using the cube of the geometric scaling ratio is subjected to Froude scaling conversion to calculate the preliminary scaled aerodynamic thrust. The preliminary scaled aerodynamic thrust is compared with a preset maximum safe thrust threshold and amplitude limiting processing is performed. When the preliminary scaled aerodynamic thrust exceeds the maximum safe thrust threshold, the value of the target aerodynamic thrust is forcibly limited to the maximum safe thrust threshold; otherwise, the preliminary scaled aerodynamic thrust is directly output as the target aerodynamic thrust.

[0012] The step of acquiring the original physical thrust applied to the floating wind turbine model by the thrust actuator through a force sensor and filtering it to obtain the actual physical thrust specifically includes: smoothing the original physical thrust acquired by the force sensor and filtering out high-frequency noise using a first-order low-pass filtering algorithm to obtain the actual physical thrust; wherein, the filtering coefficient of the first-order low-pass filtering algorithm is calculated based on the set time step and the preset cutoff frequency, and the actual physical thrust is calculated by weighting the original physical thrust at the current discrete time point and the actual physical thrust at the previous discrete time point. The calculation of pulse width modulation duty cycle compensation using the proportional-integral-derivative (PID) control algorithm and the generation of feedback compensation commands specifically includes: converting the continuous-time control law into a difference equation using a backward differential discretization method; calculating the pulse width modulation duty cycle compensation based on the thrust closed-loop tracking error at the current discrete time point, the thrust closed-loop tracking error at the previous discrete time point, the set time step, and the accumulated error integral obtained by integrating the thrust closed-loop tracking error; and performing anti-saturation processing on the accumulated error integral by forcibly constraining the accumulated error integral through a preset integral limiting threshold to prevent control command over-limit and desaturation delay caused by long-term steady-state errors.

[0013] The process of fusing the feedforward control command and the feedback compensation command to generate the comprehensive drive duty cycle specifically includes: adding and superimposing the feedforward pulse width modulation duty cycle reference value included in the feedforward control command and the pulse width modulation duty cycle compensation amount included in the feedback compensation command to generate a preliminary comprehensive drive duty cycle; and performing operating range limiting processing on the preliminary comprehensive drive duty cycle according to the hardware communication protocol of the thrust actuator, setting values ​​below 10% to 10% and values ​​above 90% to 90% to obtain the comprehensive drive duty cycle. Based on a linear mapping relationship, according to the minimum and maximum high-level pulse width times identified by the electronic speed controller configured in the thrust actuator, the comprehensive drive duty cycle is mapped and calculated as the high-level pulse width time corresponding to the current discrete time point. Through the digital output channel of the pre-configured data acquisition card, a square wave signal with alternating high and low levels is generated according to a preset signal frequency. The high level is maintained for the same amount of time as the calculated high-level pulse width time within each square wave signal cycle, and is sent to the electronic speed controller of the thrust actuator to drive the thrust actuator to generate actual physical thrust.

[0014] This invention provides a numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation. It has the following beneficial effects: 1. This invention pre-constructs an aerodynamic thrust load database and, during the experiment, uses a lookup table and interpolation algorithm to obtain the prototype aerodynamic thrust based on the calculated real-time relative inflow wind speed. By using offline-generated static lookup tables to replace the real-time solution process of the aerodynamic difference equations, the real-time computational load of the control system is effectively reduced, ensuring the real-time computational efficiency and smooth operation of the numerical physics hybrid experiment.

[0015] 2. This invention employs a drive strategy that integrates feedforward control and feedback compensation. A pre-calibrated electromechanical mapping relationship of the thrust actuator is used to calculate the feedforward duty cycle reference value to accelerate system response. The actual physical thrust is acquired through a force sensor, and closed-loop error compensation is performed using a proportional-integral-derivative (PID) algorithm. Simultaneously, anti-saturation limiting constraints are applied to the error integral term to prevent control over-limits and desaturation delays caused by steady-state error accumulation, thereby improving the control accuracy and dynamic tracking stability of the physical thrust output.

[0016] 3. In calculating the real-time inflow wind speed, this invention extracts the sway and pitch motion attitude data of the floating wind turbine model to calculate the model-induced velocity, and then performs vector superposition with the absolute wind field data after scaling. By incorporating the spatial motion state of the floating platform itself into the calculation input conditions of the aerodynamic load through the wind speed calculation method, it can objectively reproduce the change of the platform motion on the relative wind speed of the impeller swept surface, thereby accurately simulating the aerodynamic damping effect existing in actual operation and improving the physical reality of the test load. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the algorithm flow of the present invention; Figure 2 This is a schematic diagram of the system architecture of the present invention; Figure 3 This is a frequency domain comparison curve of aerodynamic thrust power spectral density under turbulent wind conditions according to the present invention. Detailed Implementation

[0018] 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.

[0019] Please see the appendix Figure 2 The present invention provides a numerical-physical hybrid test method for floating wind turbines based on real-time interpolation. This method is implemented based on a corresponding test system, including: a floating wind turbine model, a motion monitoring system, a thrust actuator, a force sensor, and a control host.

[0020] The floating wind turbine model is deployed in a physical water tank environment.

[0021] The motion monitoring system is deployed in a physical water tank environment to collect motion attitude data of the floating wind turbine model in real time.

[0022] The thrust actuator is installed on the top of the floating wind turbine model and is used to apply aerodynamic thrust to the floating wind turbine model.

[0023] The force sensor is installed at the mechanical connection between the thrust actuator and the floating wind turbine model tower to measure the actual physical thrust.

[0024] The control host establishes communication connections with the motion monitoring system, the thrust actuator, and the force sensor, respectively.

[0025] The control unit includes a database module, a wind speed calculation module, an interpolation module, and a control module.

[0026] Please see the appendix Figure 1 This invention provides a numerical-physical hybrid test method for floating wind turbines based on real-time interpolation, comprising the following steps: S1. The database module of the control host loads the pre-generated aerodynamic thrust load database; S2. The motion monitoring system collects the motion attitude data of the floating wind turbine model in real time and transmits the motion attitude data to the control host. S3. The wind speed calculation module of the control host receives motion attitude data and, combined with the input absolute wind field data, calculates the real-time relative inflow wind speed at the hub height of the floating wind turbine model. S4. The interpolation module of the control host performs interpolation calculations in the aerodynamic thrust load database based on the real-time relative inflow wind speed to obtain the prototype aerodynamic thrust, and converts the prototype aerodynamic thrust into the target aerodynamic thrust based on the scaling criterion. S5. The control module of the control host calculates the basic speed control quantity of the thrust actuator based on the target aerodynamic thrust and the calibration mapping relationship, and generates feedforward control commands. S6. The force sensor collects the actual physical thrust applied to the floating wind turbine model by the thrust actuator and feeds it back to the control host. The control module calculates the deviation between the target aerodynamic thrust and the actual physical thrust. When the deviation exceeds the set threshold, the speed compensation is calculated according to the proportional-integral-derivative control algorithm to generate a feedback compensation command. S7. The control module merges the feedforward control command and the feedback compensation command to generate a pulse width modulation drive signal and outputs it to the thrust actuator. The thrust actuator adjusts the output power according to the drive signal. The test system executes steps S2 to S7 in a cycle according to a fixed time step.

[0027] The aerodynamic thrust load database loaded into the database module of the control host is pre-built during the offline preparation phase. The construction process of the aerodynamic thrust load database includes establishing a full-scale aerodynamic-hydraulic servo-elastic coupling simulation model of the floating wind turbine. Wind turbine control strategies are configured in the coupled simulation model. Specifically, the wind turbine control strategies include a variable pitch control strategy based on generator speed feedback and a generator electromagnetic torque control strategy. Below the rated wind speed, the system maintains a fixed blade pitch angle and adjusts the generator torque to track the optimal tip speed ratio; above the rated wind speed, the system adjusts the blade pitch angle to maintain the generator's rated power output. By configuring the wind turbine control strategies, the aerodynamic thrust of the floating wind turbine exhibits nonlinear changes at different wind speeds. For the specific modeling process of the aerodynamic-hydraulic servo-elastic coupling simulation model, those skilled in the art can use existing wind turbine dynamics analysis programs to build the system. The aerodynamic calculation and multibody dynamics solving algorithms are well-known technologies in this field and will not be elaborated upon here.

[0028] A series of environmental input conditions covering the target test conditions were set, and a coupled simulation model was run for time-domain simulation calculations. The environmental input conditions included discrete average wind speed gradients, turbulence intensity parameters, and matching regular or irregular wave sea state parameters. The time-history dynamic response data of the floating wind turbine was recorded during the simulation calculations. The time-history dynamic response data included information on wind speed, platform motion state, and aerodynamic loads on the wind turbine within the preset operating time.

[0029] The relative inflow wind speed and corresponding prototype aerodynamic thrust at the turbine hub are extracted from the time-series dynamic response data for each integral time step. Because floating wind turbines are subjected to the combined effects of waves and wind at sea, their platforms undergo multi-degree-of-freedom motion, causing a difference between the perceived wind speed at the turbine hub and the absolute ambient wind speed. The relative inflow wind speed is the superposition of the absolute inflow wind speed and the induced velocity generated by the floating platform's motion. The mapping relationship between the prototype aerodynamic thrust and the relative inflow wind speed is expressed as: ; In the formula, This represents the prototype's aerodynamic thrust. This indicates the relative inflow velocity at the wind turbine hub. This represents a nonlinear aerodynamic thrust function that incorporates the logic for wind turbine pitch and torque control.

[0030] The extracted relative inflow wind speed and prototype aerodynamic thrust data were cleaned and discretized. After removing transient distortion data from the simulation startup phase, the remaining time-series data were divided into intervals and averaged according to the magnitude of the relative inflow wind speed. Transient distortion data refers to data within a preset time period after the simulation starts; the preset time period ranges from tens to hundreds of seconds before the simulation, determined based on the system stabilization time. Interval division involves dividing the relative inflow wind speed from the cut-in wind speed to the cut-out wind speed into multiple wind speed intervals with fixed step sizes, ranging from 0.1 m / s to 0.5 m / s. Within each wind speed interval, the average value of the corresponding prototype aerodynamic thrust was calculated, obtaining a unique thrust value corresponding to each discrete wind speed point. A static lookup table was generated with relative inflow wind speed as the independent variable and prototype aerodynamic thrust as the dependent variable. This static lookup table was encapsulated and stored as an aerodynamic thrust load database. The aerodynamic thrust load database adopts a one-dimensional data array or key-value pair mapping structure to enable the database module of the control host to perform memory addressing and data reading during real-time testing.

[0031] Before conducting the pool test, in addition to building an aerodynamic thrust load database, it is necessary to pre-establish the electromechanical mapping relationship between the thrust actuator's control commands and actual physical outputs. This mapping relationship provides reference data for feedforward control in real-time testing. The calibration process for the electromechanical mapping relationship is as follows.

[0032] The thrust actuator is fixed on a calibration bench. The calibration bench is set up in a calm environment or a closed wind tunnel test section. A force sensor is installed in series at the base of the thrust actuator, and a speed sensor is installed at the end of the motor rotor shaft of the thrust actuator. The control module of the control host establishes an electrical signal connection with the electronic speed controller of the thrust actuator.

[0033] The control module of the main control unit sends a pulse width modulation (PWM) drive signal with an increasing duty cycle to the electronic speed controller of the thrust actuator. The duty cycle of the PWM drive signal increases progressively in fixed steps within the effective operating range. The effective operating range is set to 10% to 90% of the duty cycle, and the step size is set to 1% to 5%. After maintaining stability for a certain time at each duty cycle step, the currently set duty cycle, the steady-state speed value fed back by the speed sensor, and the steady-state physical thrust value fed back by the force sensor are recorded synchronously. The stabilization time is set to 3 to 5 seconds to ensure that the rotor speed and flow field reach a stable state. Multiple sets of discrete calibration data samples are generated through traversal testing across the entire range.

[0034] According to aerodynamic principles, the aerodynamic thrust generated by a thrust actuator is approximately proportional to the square of its rotor speed. Simultaneously, the steady-state speed of the motor exhibits a non-linear correlation with the duty cycle of the input pulse-width modulation (PWM) drive signal. Therefore, by establishing a polynomial mathematical model containing quadratic terms, the transmission relationship between the target aerodynamic thrust, the target speed, and the duty cycle reference value can be accurately characterized, allowing the required control duty cycle to be directly calculated based on the target aerodynamic thrust during experiments.

[0035] Based on the collected calibration data samples, the control unit uses the least squares method to perform polynomial curve fitting. The functional relationship between the steady-state physical thrust value and the steady-state rotational speed value is extracted to obtain the thrust-speed mapping model. The specific fitting formula is expressed as follows: ; In the formula, This indicates the target rotational speed at which the required thrust is generated. Indicates the target aerodynamic thrust. , , This represents the first empirical coefficient obtained by fitting the calibration data sample using the least squares method. Its specific value is determined by the aerodynamic shape of the thrust actuator blade and the physical characteristics of the motor.

[0036] The functional relationship between the steady-state speed value and the duty cycle of the pulse width modulation drive signal is extracted to obtain the speed duty cycle mapping model. The specific fitting formula is expressed as follows: ; In the formula, This indicates the corresponding feedforward pulse width modulation duty cycle reference value. This indicates the target rotational speed at which the required thrust is generated. , , This represents the second empirical coefficient obtained by fitting the calibration data sample using the least squares method. Its specific value is determined by the response characteristics of the electronic speed governor of the thrust actuator. For the calculation principle and coefficient solution process of polynomial curve fitting using the least squares method, those skilled in the art can consult conventional numerical analysis methods; the data fitting process is well-known in the field and will not be elaborated upon here.

[0037] The thrust-speed mapping model and the speed-duty cycle mapping model, along with their corresponding coefficient values, are pre-stored in the control module of the main control unit. During real-time water tank testing, the control module directly calls upon the aforementioned mapping models based on the target aerodynamic thrust obtained through interpolation to calculate the basic speed control quantity required for feedforward control and the corresponding pulse width modulation duty cycle reference value. Establishing this electromechanical mapping relationship allows for direct feedforward commands based on the target aerodynamic thrust, avoiding the control delay problem caused by relying solely on feedback adjustment.

[0038] During the online real-time water tank test, the control host continuously acquired the state parameters and environmental wind field parameters of the floating wind turbine model as the basis input for subsequent aerodynamic thrust calculation. The real-time high-frequency acquisition process of motion attitude and wind field environment is as follows.

[0039] The motion monitoring system collects real-time motion attitude data of the floating wind turbine model in the physical water tank. The system employs non-contact optical motion capture equipment or an inertial measurement unit (IMU). Before using the equipment, a unified global coordinate system for the water tank is established, with the X-axis parallel to the absolute wind direction set for the experiment, and this X-axis direction is used as the sway direction. When using non-contact optical motion capture equipment, multiple infrared cameras are deployed around the physical water tank, and reflective markers are installed on the platform deck or tower height of the floating wind turbine model. The infrared cameras calculate the six-degree-of-freedom motion data of the floating wind turbine model in the global coordinate system of the water tank by capturing the positional changes of the reflective markers. The six-degree-of-freedom motion data includes sway displacement, yaw displacement, heave displacement, roll angle, pitch angle, and yaw angle. When using an IMU, the sensor is fixed at the center of gravity of the floating wind turbine model, and the three-axis acceleration and angular velocity are measured and integrated to obtain the motion displacement and attitude angles. For the measurement principles and coordinate calculation algorithms of non-contact optical motion capture devices or inertial measurement units, those skilled in the art can consult the relevant instrument manuals. Their spatial positioning and attitude calculation are well-known technologies in the field and will not be elaborated here.

[0040] The wind speed calculation module of the control host synchronously reads pre-set absolute wind field data. This absolute wind field data is a time-series file of natural wind speeds generated based on the environmental conditions of the target sea area. The natural wind speed time-series file contains instantaneous wind speed values ​​that vary over time, generated according to a preset wind spectrum. The preset wind spectrum uses either the Kemal or von Kármán wind spectrum. Within each test cycle, the wind speed calculation module extracts the absolute inflow wind speed value at the corresponding moment according to the current system time. This numerical data input method eliminates the instability and spatial non-uniformity in wind speed output from physical wind turbine equipment.

[0041] The control host is equipped with a hardware timer, using a preset time step as the interrupt period, to synchronize the data sampling frequency of the motion monitoring system with the data reading frequency of the wind speed calculation module. The time step is set to a range of 0.01 seconds to 0.05 seconds, determined by the highest cutoff frequency of the floating wind turbine's aerodynamic load and the computing power of the control host. At each discrete time point corresponding to the time step, the control host timestamps the acquired motion attitude data of the floating wind turbine model with the current absolute inflow wind speed value, generating the basic input data packet for the current calculation cycle.

[0042] For floating wind turbines, the changes in aerodynamic loads are mainly caused by relative motion in the downwind direction. This downwind relative motion is primarily composed of the superposition of two parts: the overall translation of the platform along the wind direction and the translation of the tower hub along the wind direction caused by the platform's rotation. Specifically, the platform's translation along the wind direction corresponds to sway motion, and the translation of the tower top caused by the platform's rotation corresponds to pitching motion. Therefore, to accurately calculate the downwind aerodynamic response of the wind turbine, the longitudinal motion component, which has the most significant impact on the flow field, is extracted from the basic input data package. Its mathematical expression is as follows: ; In the formula, Represents the current discrete time point The basic input data packet column vector, This indicates the floating wind turbine model at the current discrete time point. The oscillation displacement, This indicates the floating wind turbine model at the current discrete time point. pitch angle, This represents the current discrete time point in the natural wind speed time series file. The absolute inflow wind speed value, This represents the current discrete time point, expressed in increments of time step. This represents the matrix transpose operation. The time synchronization alignment mechanism ensures that the dynamic response of the model in the physical water tank matches the numerical wind field input in the time dimension, avoiding control phase delay caused by asynchronous sampling. The control host temporarily stores the basic input data packet in a memory buffer for the wind speed calculation module to call and execute the relative wind speed calculation.

[0043] After reading the basic input data packet from the memory buffer, the wind speed calculation module of the control host calculates the relative wind speed by combining the scaling criterion and kinematic principles. The kinematic calculation process of the relative inflow wind speed at the hub is as follows.

[0044] The wind speed calculation module of the control host performs time-dimensional difference calculations on the sway displacement and pitch angle in the basic input data packet to obtain the sway velocity and pitch angle velocity of the floating wind turbine model at the current discrete time point. To meet the real-time calculation requirements of the experiment, the difference calculation adopts a first-order backward difference algorithm, that is, the derivative at the current time point is approximated by dividing the difference between the sampled value at the current discrete time point and the sampled value at the previous discrete time point by the time step. Since the raw data collected by the sensor contains measurement noise, direct difference calculation will amplify the noise interference and thus cause oscillations in subsequent control commands. The wind speed calculation module uses a digital low-pass filter to smooth the difference results. The cutoff frequency of the digital low-pass filter is set according to the upper limit of the wave excitation frequency, and the value range is set to 1 Hz to 5 Hz. For the design and implementation of the digital low-pass filter, those skilled in the art can use conventional Butterworth filters or Chebyshev filters, and their signal filtering algorithms are well-known technologies in the field, and will not be described in detail here.

[0045] The wind speed calculation module calculates the model-induced velocity at the top hub of the floating wind turbine model based on the principles of rigid body kinematics. The longitudinal motion of the floating platform in wind and wave environments manifests as the overall horizontal translation and rotation about its center of gravity. The top hub is fixed to the platform, and its horizontal velocity component is the superposition of the platform's sway velocity and the projection of the linear velocity caused by the platform's roll in the horizontal direction. The formula for calculating the model-induced velocity is as follows: ; In the formula, Indicates the current discrete time point Model-induced velocity at the hub of the floating wind turbine model Indicates the current discrete time point The oscillation velocity of the model after filtering. This represents the vertical height from the hub center of the floating wind turbine model to the center of gravity of the floating platform. Its value is determined in advance based on the structural dimensions of the floating wind turbine model. Indicates the current discrete time point pitch angle, Indicates the current discrete time point The model's pitch angular velocity after filtering. It represents the current discrete time point with increasing time steps.

[0046] To maintain data matching with the prototype-scale aerodynamic thrust load database pre-built during the offline preparation phase, the wind speed calculation module, based on the Froude similarity criterion, converts the calculated model-induced velocities into prototype-scale-induced velocities. The pool physical experiment is scaled down according to the principles of geometric similarity and equal Froude numbers. Since the Froude criterion ensures that the ratios of inertial forces and gravity are equal between the model and the prototype, velocity physical quantities need to be scaled up proportionally according to the square root of the geometric scaling ratio. The conversion formula for prototype-induced velocities is as follows: ; In the formula, This indicates the value at the current discrete time point after conversion to the prototype scale. prototype induced velocity, Indicates the current discrete time point Model-induced velocity at the hub of the floating wind turbine model This indicates the geometric scale ratio set for the experiment. The geometric scale ratio is determined based on the wave-generating capacity of the physical water tank and the actual physical dimensions of the offshore floating wind turbine, and its value ranges from 40 to 80. It represents the current discrete time point with increasing time steps.

[0047] The wind speed calculation module vector-superimposes the absolute inflow wind speed value with the calculated prototype induced velocity to obtain the relative inflow wind speed for querying the database at the current moment. When the floating wind turbine model moves in the downwind direction due to wave action, the relative wind speed felt at the hub decreases; when the floating wind turbine model moves in the upwind direction, the relative wind speed increases. The displacement direction and wind speed direction in the basic input data package have been uniformly calibrated onto the X-axis of the pool's global coordinate system. The formula for calculating the relative inflow wind speed is as follows: ; In the formula, Indicates the current discrete time point The relative inflow velocity, This represents the current discrete time point in the natural wind speed time series file. The absolute inflow wind speed value, whose dimensions correspond to the prototype scale. This indicates the value at the current discrete time point after conversion to the prototype scale. prototype induced velocity, This represents the current discrete time point, incremented by the time step. The calculated relative inflow wind speed reflects the impact of the multi-degree-of-freedom motion of the platform caused by waves on the aerodynamic load of the wind turbine rotor. The control unit transmits the relative inflow wind speed to the subsequent interpolation module for target aerodynamic thrust matching.

[0048] The interpolation module of the control host receives the relative inflow wind speed from the wind speed calculation module and performs table lookup and interpolation calculations based on the pre-loaded aerodynamic thrust load database in memory.

[0049] Based on the actual physical operating logic of a wind turbine, the turbine operates and generates actual aerodynamic thrust only within a preset effective operating wind speed range. Therefore, the interpolation module determines whether the relative inflow wind speed at the current discrete time point is within the effective operating wind speed range of the floating wind turbine model. The lower limit of the effective operating wind speed range is set as the cut-in wind speed, which ranges from 3 m / s to 5 m / s, and its specific value is determined by the startup characteristics of the floating wind turbine prototype. The upper limit of the effective operating wind speed range is set as the cut-out wind speed, which is generally set to 25 m / s, and its specific value is determined by the safe shutdown protection strategy of the floating wind turbine prototype. When the relative inflow wind speed is less than the cut-in wind speed or greater than the cut-out wind speed, the interpolation module directly assigns the prototype aerodynamic thrust corresponding to the current discrete time point to zero, simulating the turbine's shutdown and unloading state.

[0050] When the relative inflow wind speed is within the effective operating wind speed range, the interpolation module performs memory addressing in the aerodynamic thrust load database. Since the actual calculated relative inflow wind speed is a continuously changing floating-point number, it typically cannot directly match the discretely distributed wind speed nodes in the database. The interpolation module uses the relative inflow wind speed as the search target and employs a binary search algorithm to determine the wind speed interval within the aerodynamic thrust load database. It then obtains the lower boundary discrete wind speed point value, the upper boundary discrete wind speed point value, and the prototype aerodynamic thrust corresponding to the lower boundary discrete wind speed point and the upper boundary discrete wind speed point, respectively. For the memory addressing process of the binary search algorithm, those skilled in the art can consult conventional computer data structure documentation; its array retrieval algorithm is a well-known technique in the field and will not be elaborated upon here.

[0051] After determining the wind speed range and corresponding boundary values, the interpolation module uses a linear interpolation algorithm to calculate the precise thrust value corresponding to the relative inflow wind speed at the current discrete time point. While the overall relationship between aerodynamic thrust and wind speed is non-linear, the thrust variation curve can be approximated as linear within a pre-defined, small, fixed-step wind speed range. The linear interpolation algorithm has low computational complexity, enabling floating-point operations to be performed within the control host's time step, meeting the hardware processing requirements for real-time control in pool testing. The specific calculation formula for linear interpolation is as follows: ; In the formula, Indicates the current discrete time point The prototype aerodynamic thrust obtained through interpolation calculation. This represents the prototype aerodynamic thrust corresponding to the discrete wind speed points at the lower boundary. This represents the prototype aerodynamic thrust corresponding to the discrete wind speed points at the upper boundary. This represents the discrete wind speed values ​​at the lower boundary. This represents the numerical values ​​of discrete wind speed points at the upper boundary. Indicates the current discrete time point The relative inflow velocity, It represents the current discrete time point with increasing time steps.

[0052] The interpolation module temporarily stores the calculated prototype aerodynamic thrust as the basic data input for subsequent scaling conversion. The calculation method, based on a pre-stored database lookup table combined with in-memory interpolation, avoids the real-time calculation process of aerodynamic analysis software during pool testing, reducing the computational burden on the control host and minimizing system latency.

[0053] After completing the lookup and interpolation calculations for the relative inflow velocity, the interpolation module of the control host obtains the prototype aerodynamic thrust at the current discrete time point. Since the physical pool test is designed based on the Froude similarity criterion, there is a fixed physical scaling relationship between the floating wind turbine model and the prototype. The interpolation module needs to convert the thrust at the prototype scale into the target aerodynamic thrust suitable for the physical pool environment.

[0054] In marine engineering pool tests, the motion of the floating platform is primarily governed by gravity and inertial forces; therefore, the test system follows the Froude similarity criterion. According to the similarity criteria of fluid mechanics and dynamics, the physical scaling ratio of a force is equal to the fluid density ratio multiplied by the cube of the geometric scaling ratio. Since both the actual offshore prototype wind turbine and the floating wind turbine model in the pool operate in a standard atmospheric environment, their air densities are approximately equal, i.e., the fluid density ratio is 1. Therefore, the conversion ratio between the prototype aerodynamic thrust and the scaled-down aerodynamic thrust is equal to the cube of the geometric scaling ratio set for the test. By directly performing Froude scaling conversion on the calculated prototype aerodynamic thrust, the test system avoids the Reynolds number distortion problem caused by direct geometric scaling of the physical blades, ensuring that the floating wind turbine model can withstand aerodynamic forces that meet the requirements of dynamic equivalence in the pool. The specific formula for the scaling conversion performed by the interpolation module is as follows: ; In the formula, Indicates the current discrete time point The preliminary scaled-down aerodynamic thrust obtained from the conversion. Indicates the current discrete time point The prototype aerodynamic thrust obtained through interpolation calculation. This indicates the geometric scale ratio set for the experiment. The geometric scale ratio is determined based on the wave-generating capacity of the physical water tank and the actual physical dimensions of the offshore floating wind turbine, and its value ranges from 40 to 80. It represents the current discrete time point with increasing time steps.

[0055] After calculating the preliminary scaled aerodynamic thrust, the interpolation module performs amplitude limiting safety protection on the preliminary scaled aerodynamic thrust. Because the thrust actuator installed at the top of the tower has a physical output limit, when extreme wind and wave excitation conditions cause a sudden change in the instantaneous relative inflow wind speed, resulting in the calculated preliminary scaled aerodynamic thrust exceeding the hardware's execution capacity, directly issuing unconstrained control commands would cause the drive motor to overload, overheat, or even be damaged. The interpolation module compares the calculated preliminary scaled aerodynamic thrust with a preset maximum safe thrust threshold and outputs the final target aerodynamic thrust. The calculation formula for the amplitude limiting process is as follows: ; In the formula, Indicates the current discrete time point The final output target aerodynamic thrust after amplitude limiting processing This represents the function that takes the minimum value. Indicates the current discrete time point The preliminary scaled-down aerodynamic thrust obtained from the conversion. This represents the preset maximum safe thrust threshold. The specific value of the maximum safe thrust threshold is determined comprehensively based on the rated power of the thrust actuator, the blade limiting speed, and the test data of the maximum available static thrust calibrated on the force measuring bench. To maintain a safety margin in system control, the maximum safe thrust threshold is set to 80% to 90% of the maximum available static thrust of the thrust actuator. When the initial scaled-down aerodynamic thrust exceeds the maximum safe thrust threshold, the interpolation module limits the target aerodynamic thrust value to the maximum safe thrust threshold, thereby achieving physical constraints and hardware protection for the thrust actuator.

[0056] After scaling and limiting, the target aerodynamic thrust conforms to the hardware conditions of the pool test execution system in terms of physical dimensions and numerical range, reflecting the equivalent aerodynamic load that the floating wind turbine model should experience at the current discrete time point. The interpolation module of the control host outputs the target aerodynamic thrust to the control module as the core input reference for generating the basic speed control quantity and feedforward control command.

[0057] While issuing feedforward control commands, the control module of the control host obtains the actual aerodynamic thrust generated by the thrust actuator in order to construct a closed-loop feedback control circuit.

[0058] A force sensor mounted on the base of the thrust actuator collects the raw physical thrust data output by the actuator in real time. The force sensor is a high-frequency dynamic tension / compression sensor. During the water tank test, the thrust actuator operates amidst mechanical vibrations of the drive motor, aerodynamic disturbances generated by the rotating blades cutting through the airflow, and elastic resonance of the floating wind turbine model tower structure itself. These physical phenomena cause high-frequency noise to be mixed into the raw physical thrust data collected by the force sensor. If the raw physical thrust data containing high-frequency noise is directly introduced into the closed-loop feedback system, it will cause high-frequency oscillations in the control commands output by the control module, which in turn will cause frequent acceleration and deceleration of the thrust actuator motor, resulting in hardware fatigue and system instability.

[0059] To extract the thrust signal reflecting the low-frequency variation trend of aerodynamic loads, the control module of the control host performs filtering processing on the acquired raw physical thrust data. Considering the timeliness requirements of the algorithm calculation in the water tank test control system, the control module adopts a first-order low-pass filtering algorithm to smooth the raw physical thrust data. The first-order low-pass filtering algorithm only depends on the current sampled value and the previous filtered output value, consuming little memory and computing resources of the control host, and can complete the calculation within a set millisecond-level discrete time step. The specific calculation formula for the first-order low-pass filtering is as follows: ; In the formula, Indicates the current discrete time point Actual physical thrust after filtering This represents the data collected by the force sensor at the current discrete time point. The original physical thrust data, Indicates at the previous discrete time point Actual physical thrust after filtering This represents the time step between two adjacent discrete time points, and its value range is set to 0.01 seconds to 0.05 seconds. These represent the filter coefficients of a first-order low-pass filter. It represents the current discrete time point with increasing time steps.

[0060] Filter coefficients This determines the filter's ability to attenuate high-frequency noise and its response delay to valid signals. The control module uses the system's time step. The filter coefficients are calculated using the set cutoff frequency. The specific calculation formula is as follows: ; In the formula, These represent the filter coefficients of a first-order low-pass filter. This represents the time step between two adjacent discrete time points. Pi is a constant. This represents the cutoff frequency of a first-order low-pass filter. The specific value is determined based on the physical characteristics of the tower's resonant frequency and the motor's high-frequency noise, typically ranging from 1 Hz to 5 Hz. The filter coefficient is determined through the aforementioned physical relationships. The value typically ranges from 0.05 to 0.2. When the ambient wind and wave excitation frequency is low, the cutoff frequency can be reduced. Then select the smaller one To enhance the smoothing effect; when it is necessary to capture higher-frequency aerodynamic load abrupt changes, the cutoff frequency can be increased. Then select the larger one This value is used to reduce the phase delay of the control system.

[0061] The actual physical thrust, after first-order low-pass filtering, eliminates interference from mechanical vibration and high-frequency aerodynamic disturbances, while retaining the aerodynamic load response caused by changes in relative inflow wind speed. The control module of the control host temporarily stores the filtered actual physical thrust in memory as the basic input for calculating thrust error and performing feedback adjustment. For the data acquisition communication protocol and hardware interface configuration of the force sensor, those skilled in the art can consult the relevant sensor datasheets; the signal acquisition and analog-to-digital conversion processes are well-known technologies in the field and will not be elaborated upon here.

[0062] After acquiring the target aerodynamic thrust and the filtered actual physical thrust, the control module of the control host calculates the final drive signal of the thrust actuator through a digital closed-loop control algorithm. The closed-loop error calculation and the discretization implementation process of the PID (Proportional-Integral-Derivative) control law are as follows.

[0063] The control module of the control host synchronously reads the data transmitted by the preceding module in each calculation cycle to calculate the thrust closed-loop tracking error. The thrust closed-loop tracking error represents the difference between the ideal aerodynamic force that the floating wind turbine model should currently experience in the physical water tank and the actual aerodynamic force applied by the thrust actuator. The formula for calculating the thrust closed-loop tracking error is as follows: ; In the formula, Indicates the current discrete time point The thrust closed-loop tracking error, Indicates the current discrete time point The final output target aerodynamic thrust after scaling and limiting is Indicates the current discrete time point Actual physical thrust after filtering It represents the current discrete time point with increasing time steps.

[0064] The control module performs discretized PID control calculations based on the thrust closed-loop tracking error. The control system for the water tank test is a sampled-sampling control system based on a digital computer, which cannot directly handle continuous-time domain calculus and integral operations. The control module employs a backward differential discretization method to convert the continuous-time PID control law into a difference equation that can be executed within each time step, in order to calculate the pulse width modulation duty cycle compensation. The discrete PID calculation formula is as follows: ; In the formula, Indicates the current discrete time point Pulse width modulation duty cycle compensation amount, Represents the proportional gain coefficient. Represents the integral gain coefficient. Represents the differential gain coefficient. Indicates the current discrete time point The thrust closed-loop tracking error, Indicates at the previous discrete time point The thrust closed-loop tracking error, Indicates the current discrete time point The cumulative value of the integral of the error. This represents the time step between two adjacent discrete time points. The value of the time step is set to range from 0.01 seconds to 0.05 seconds. It represents the current discrete time point with increasing time steps.

[0065] proportional gain coefficient Integral gain coefficient Differential gain coefficient The specific value depends on the electromechanical response characteristics of the thrust actuator. The control module applies a step target thrust command to the thrust actuator on a static test bench, records the rise time, overshoot, and steady-state error of the actual physical thrust, and performs parameter calibration using the Ziegler-Nichols tuning method. After bench calibration, the proportional gain coefficient... Integral gain coefficient Differential gain coefficient The specific value range is usually set between 0.1 and 5.0.

[0066] In the above formula, the update logic for the cumulative error integral at the current discrete time point is as follows: ; In the formula, Indicates the current discrete time point The cumulative value of the integral of the error. Indicates at the previous discrete time point The cumulative value of the integral of the error. Indicates the current discrete time point The thrust closed-loop tracking error, This represents the time step between two adjacent discrete time points. It represents the current discrete time point with increasing time steps.

[0067] The control module performs anti-saturation processing on the accumulated integral value of the error. When the thrust closed-loop tracking error persists for a long time, the accumulated integral value of the error will continue to increase, causing the control command to exceed the physical operating limits of the thrust actuator. This leads to desaturation delay and system overshoot when the error reverses. The control module sets an integral limiting threshold and uses a limiting algorithm to forcibly constrain the accumulated integral value of the error. The formula for calculating the integral limit is as follows: ; In the formula, Indicates the current discrete time point The final accumulated error integral value used for calculation after anti-saturation treatment This represents the function that takes the maximum value. This represents the function that takes the minimum value. This indicates the preset integration threshold. Integration threshold The specific value is calculated by dividing the maximum allowable duty cycle adjustment margin of the thrust actuator under rated operating conditions by the integral gain coefficient. Calculation determined.

[0068] In the thrust control system, feedforward control, based on a pre-established electromechanical mapping relationship, can directly provide the basic drive signal according to the target aerodynamic thrust, providing the main feedforward driving force of the system and achieving rapid response. Meanwhile, PID feedback control relies on the real-time calculated thrust closed-loop tracking error to dynamically compensate for unmodeled disturbances and steady-state errors caused by motor response dead zones, mechanical friction, and changes in the aerodynamic environment of the water tank. The combination of these two approaches achieves a balance between response speed and control accuracy. The control module sums and superimposes the feedforward pulse width modulation duty cycle reference value obtained through interpolation with the pulse width modulation duty cycle compensation amount calculated by the closed-loop feedback to generate the comprehensive drive duty cycle. The calculation formula is as follows: ; In the formula, Indicates the current discrete time point The overall drive duty cycle output to the thrust actuator Indicates the current discrete time point The reference value of the feedforward pulse width modulation duty cycle is calculated based on the electromechanical mapping relationship. Indicates the current discrete time point Pulse width modulation duty cycle compensation amount, It represents the current discrete time point with increasing time steps.

[0069] The control module limits the operating range of the integrated drive duty cycle according to the hardware communication protocol of the thrust actuator. When the integrated drive duty cycle is below 10%, it is set to 10%; when it is above 90%, it is set to 90%. The control module converts the processed integrated drive duty cycle into a low-level hardware electrical signal and sends it to the electronic speed controller of the thrust actuator to complete the closed-loop aerodynamic thrust control cycle at the current discrete time point. Feedforward control provides the basic control input for the thrust actuator, and PID feedback control, based on this, corrects for system errors caused by model aerodynamic characteristics and environmental disturbances in real time.

[0070] After completing the calculation and limiting of the comprehensive drive duty cycle, the control module of the control host converts the digital quantity into a low-level hardware electrical signal that the thrust actuator can recognize, so as to drive the hardware device to generate actual aerodynamic thrust.

[0071] The thrust actuator employs a miniature fan assembly consisting of a brushless DC motor and propeller blades, and is equipped with an electronic speed controller for drive control. The electronic speed controller adjusts the motor speed by receiving a digital pulse width modulation (PWM) signal with a preset frequency and target pulse width. In the physical principle of brushless motor control, PWM technology adjusts the equivalent average voltage output to the stator windings of the motor by switching the high-frequency switching transistors on and off, thereby achieving stepless speed regulation. The control module uses the system's data acquisition card to map the comprehensive drive duty cycle obtained through digital calculations into a PWM signal conforming to the hardware communication protocol. The frequency of the PWM signal is determined according to the hardware specifications of the electronic speed controller, typically ranging from 50 Hz to 400 Hz.

[0072] The control module calculates the high-level pulse width time corresponding to the current discrete time point based on a linear mapping relationship. The formula for calculating the high-level pulse width time is as follows: ; In the formula, Indicates the current discrete time point The high-level pulse width time of the generated digital pulse width modulation signal, This indicates the minimum high-level pulse width time corresponding to the motor idling speed identified by the electronic speed controller. This value is calibrated according to the underlying firmware program of the electronic speed controller and is usually set to 1000 microseconds. This indicates the maximum high-level pulse width and duration corresponding to the motor operating at full load as identified by the electronic speed controller. This value is calibrated according to the underlying firmware program of the electronic speed controller and is usually set to 2000 microseconds. Indicates the current discrete time point The overall drive duty cycle output to the thrust actuator is calculated as a decimal between 0.1 and 0.9 after the preceding operating range limiting process. It represents the current discrete time point with increasing time steps.

[0073] After calculating the high-level pulse width, the control module drives the digital output channel of the data acquisition card within the control host to generate a square wave signal with alternating high and low levels according to the set signal frequency. To ensure high-precision pulse width output, the data acquisition card uses an internal high-frequency hardware timer to generate a fundamental clock cycle, and a counter is used to compare and achieve level switching. Within each square wave signal cycle, the physical time to maintain the high level is equal to the calculated... The electronic speed controller receives the square wave signal via a signal line, analyzes its high-level duration, and adjusts the on / off state of the internal power switching transistors accordingly to output a corresponding drive AC current to the stator windings of the brushless DC motor. For the inverter circuit topology inside the electronic speed controller and the electromagnetic commutation principle of the brushless DC motor, those skilled in the art can consult conventional motor control manuals; the underlying motor drive technology is well-known in the field and will not be elaborated upon here.

[0074] The rotor of the brushless DC motor rotates under the action of the driving current, driving the propeller blades fixed on the motor shaft to rotate synchronously. The rotating propeller blades accelerate the air medium in front, generating actual physical thrust in the direction of the thrust actuator's shaft according to aerodynamic principles. This physical thrust acts directly on the hub position at the top of the floating wind turbine model, reproducing the equivalent aerodynamic load in a physical water tank environment.

[0075] The actual physical thrust generated by the thrust actuator during operation is collected in real time by a force sensor installed on its base and fed back to the control module of the control host for closed-loop error calculation at the next discrete time point. The digital calculation of the control signal, the physical conversion of the electrical signal, the thrust work of the actuator, and the data acquisition of the force sensor work together to complete the closed-loop control process of the aerodynamic thrust in the floating wind turbine model pool test.

[0076] Please see the appendix Figure 3To further verify the effectiveness of the feedforward and closed-loop feedback coordinated control algorithm in the control host of this invention from a frequency domain perspective, the Fast Fourier Transform (FFT) algorithm was used to perform frequency domain transformation on the long-duration aerodynamic thrust time-domain data recorded in the pool experiment. The frequency response characteristics and energy distribution of aerodynamic thrust under different control strategies were compared and analyzed. The wind and wave combined working condition set in the experiment not only included low-frequency natural wind turbulence energy, but also included the aerodynamic forces induced by the pitching frequency band of the floating wind turbine model caused by regular wave excitation. The horizontal axis represents frequency (Hertz, Hz), and the vertical axis represents the aerodynamic thrust power spectral density (N). 2 To clearly illustrate the significant range between low-frequency high energy and high-frequency small peak values ​​(Hz), a logarithmic (Log) axis is used on the vertical axis. The thick black solid line in the figure represents the calculated power spectral density of the "target aerodynamic thrust"; the black dashed line represents the actual physical thrust power spectral density measured using "conventional open-loop control"; and the black dotted line represents the actual physical thrust power spectral density measured using "the closed-loop control of this invention".

[0077] from Figure 3 The thick black solid line (target aerodynamic thrust) shows that the ideal aerodynamic load exhibits two distinct energy peaks in the frequency domain: the first peak is located in the extremely low frequency band around 0.05 Hz, mainly corresponding to the broadband load disturbance caused by the turbulent components of the natural wind field; the second peak is concentrated in the high frequency band around 0.6 Hz, which is the aerodynamic peak induced by the pitching motion of the floating wind turbine model caused by wave excitation in the physical water tank, resulting in periodic high-frequency changes in the relative inflow wind speed of the tower top hub. This high-frequency coupled dynamic response is the key focus of the floating wind turbine water tank test.

[0078] Combination Figure 3 The curves show that when using the traditional open-loop control strategy (black dashed line), the system exhibits a significant "low-pass filtering" effect due to the physical inertia of the thrust actuator drive motor and mechanical rotor, the response dead zone, and the lack of a feedback adjustment mechanism. In the low-frequency range below 0.1Hz, the traditional open-loop control can still maintain a certain load tracking, but there is a large overall attenuation of energy amplitude (represented by the bottom curve in the figure). In the high-frequency range above 0.4Hz (especially the wave-induced aerodynamic frequency band of 0.6Hz), the actual thrust energy of the traditional open-loop control shows severe attenuation and distortion, making it completely unable to reproduce the high-frequency aerodynamic coupling load at the wave frequency, thus rendering the pool test meaningless in terms of dynamic response equivalence.

[0079] In contrast, by adopting the feedforward and closed-loop feedback coordinated control scheme based on real-time interpolation (black dashed line) of this invention, the control module achieves rapid response of the system's main drive through feedforward commands, and utilizes PID closed-loop feedback to perform high-frequency dynamic compensation for control errors, thus broadening the effective operating bandwidth of the thrust actuator. Figure 3 As can be seen, to avoid complete curve overlap and affect recognizability in the logarithmic coordinate system, the black dotted line visually presents a "companion curve" shape that is perfectly parallel to the thick black solid line (the two maintain a fixed proportional relationship in numerical terms). This parallel fit eliminates steady-state error and exhibits extremely high frequency domain following fidelity. More importantly, in the high-frequency wave excitation band of around 0.6Hz, the closed-loop control of this invention can accurately capture and reproduce the energy peak of induced aerodynamic force, avoiding high-frequency dynamic distortion caused by mechanical hysteresis. In addition, the slight burrs in the black dotted line curve are due to the system truly reflecting the frequency domain noise floor of the small multiplicative white noise introduced by the force sensor in the pool experiment, possessing the texture of real physical experimental data. This frequency domain verification result fully demonstrates that the numerical-physical hybrid experimental method proposed in this invention can meet the stringent high-frequency response requirements of complex wind and wave coupled dynamics experiments of deep-sea floating wind turbines.

Claims

1. A numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation, characterized in that, Includes the following steps: Load the aerodynamic thrust load database; The following control process is executed cyclically at each discrete time point according to a set time step: The motion attitude data of the floating wind turbine model is collected, and combined with the pre-acquired absolute wind field data, the real-time relative inflow wind speed at the hub of the floating wind turbine model is calculated. The prototype aerodynamic thrust is obtained by interpolation in the aerodynamic thrust load database based on the real-time relative inflow wind speed, and then converted into the target aerodynamic thrust based on the scaling criterion. Based on the target aerodynamic thrust and the pre-calibrated electromechanical mapping relationship, calculate the feedforward pulse width modulation duty cycle reference value of the pre-configured thrust actuator and generate feedforward control commands; The original physical thrust applied to the floating wind turbine model by the thrust actuator is collected by a pre-configured force sensor and filtered to obtain the actual physical thrust; Calculate the thrust closed-loop tracking error between the target aerodynamic thrust and the actual physical thrust, and use the proportional-integral-derivative control algorithm to calculate the pulse width modulation duty cycle compensation amount and generate feedback compensation commands. The feedforward control command and the feedback compensation command are combined to generate a comprehensive drive duty cycle, which is then converted into a digital pulse width modulation signal and output to the thrust actuator to adjust the actual physical thrust.

2. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The process of constructing the aerodynamic thrust load database includes: A full-scale aero-hydraulic servo-elastic coupling simulation model of a floating wind turbine is established, and a floating wind turbine control strategy is configured in the full-scale aero-hydraulic servo-elastic coupling simulation model. The floating wind turbine control strategy includes a variable pitch control strategy based on generator speed feedback and a generator electromagnetic torque control strategy. Set environmental input conditions to cover the target test conditions and perform time-domain simulation calculations to extract the relative inflow wind speed and the corresponding prototype aerodynamic thrust at the hub of the floating wind turbine at each integral time step. The extracted relative inflow wind speed and corresponding prototype aerodynamic thrust are cleaned and discretized. After removing transient distortion data from the simulation start-up phase, the remaining time history data are divided into multiple wind speed intervals with fixed step sizes according to the magnitude of the relative inflow wind speed and averaged statistically. A static lookup table with relative inflow wind speed as the independent variable and prototype aerodynamic thrust as the dependent variable is generated and encapsulated and stored as the aerodynamic thrust load database.

3. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The calibration process for the pre-calibrated electromechanical mapping relationship includes: In the calibration environment, a pulse width modulation drive signal with progressively increasing duty cycle is sent to the thrust actuator. After each duty cycle step is held for a stable time, the steady-state physical thrust value and steady-state speed value are recorded synchronously. The least squares method is used to perform polynomial curve fitting, and the functional relationship between the steady-state physical thrust value and the steady-state speed value is extracted to obtain a thrust-speed mapping model containing quadratic terms. The functional relationship between the steady-state speed value and the duty cycle of the pulse width modulation drive signal is extracted to obtain a speed duty cycle mapping model containing quadratic terms. The thrust-speed mapping model and the speed duty cycle mapping model are pre-stored for use in the experiment to calculate the target speed based on the target aerodynamic thrust and the thrust-speed mapping model, and then to calculate the feedforward pulse width modulation duty cycle reference value based on the target speed and the speed duty cycle mapping model.

4. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The calculation of the real-time relative inflow wind speed at the wheel hub, based on pre-acquired absolute wind field data, specifically includes: The sway displacement and pitch angle are extracted from the motion posture data. A first-order backward difference algorithm is used to perform time-difference operations, and a digital low-pass filter is used for smoothing to obtain the sway velocity and pitch angle velocity of the floating wind turbine model. Based on the principles of rigid body kinematics, the sway velocity of the floating wind turbine model is superimposed with the projection of the linear velocity caused by the pitch angular velocity of the floating wind turbine model in the horizontal direction to calculate the model-induced velocity at the hub of the floating wind turbine model. Based on the Froude similarity criterion, the model-induced velocity is converted into the prototype-induced velocity according to the square root of the geometric scaling ratio; The absolute inflow wind speed at the current discrete time point is vector-superimposed with the prototype induced velocity to obtain the real-time relative inflow wind speed for querying the aerodynamic thrust load database.

5. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The step of obtaining the prototype aerodynamic thrust by interpolating the real-time relative inflow wind speed in the aerodynamic thrust load database specifically includes: Determine whether the real-time relative inflow velocity is within the effective working velocity range formed by the cut-in velocity and cut-out velocity of the preset floating fan prototype. If the real-time relative inflow velocity exceeds the effective working velocity range, the prototype aerodynamic thrust corresponding to the current discrete time point will be set to zero. If the real-time relative inflow wind speed is within the effective working wind speed range, then the binary search algorithm is used to determine the wind speed range where the real-time relative inflow wind speed is located in the aerodynamic thrust load database, and obtain the lower boundary discrete wind speed point value, the upper boundary discrete wind speed point value, and the prototype aerodynamic thrust corresponding to the lower boundary discrete wind speed point value and the upper boundary discrete wind speed point value respectively. The prototype aerodynamic thrust corresponding to the real-time relative inflow wind speed at the current discrete time point is calculated using a linear interpolation algorithm.

6. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The conversion of target aerodynamic thrust based on the scaling criterion specifically includes: Based on the similarity criteria of fluid mechanics and dynamics, the prototype aerodynamic thrust obtained by interpolation of the cube of the geometric scaling ratio is subjected to Froude scaling conversion to calculate the preliminary scaled aerodynamic thrust. The initial scaled-down aerodynamic thrust is compared with a preset maximum safe thrust threshold and amplitude limiting processing is performed. When the initial scaled-down aerodynamic thrust exceeds the maximum safe thrust threshold, the value of the target aerodynamic thrust is forcibly limited to the maximum safe thrust threshold; otherwise, the initial scaled-down aerodynamic thrust is directly output as the target aerodynamic thrust.

7. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The process of acquiring the original physical thrust applied to the floating wind turbine model by the thrust actuator through a force sensor and filtering it to obtain the actual physical thrust specifically includes: The original physical thrust collected by the force sensor is smoothed and high-frequency noise is filtered out using a first-order low-pass filtering algorithm to obtain the actual physical thrust. The filtering coefficients of the first-order low-pass filtering algorithm are calculated based on the set time step and the preset cutoff frequency, and the actual physical thrust is calculated by weighting the original physical thrust at the current discrete time point and the actual physical thrust at the previous discrete time point.

8. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The process of calculating the pulse width modulation duty cycle compensation amount and generating feedback compensation commands using the proportional-integral-derivative control algorithm specifically includes: The continuous-time control law is converted into a difference equation using a backward differential discretization method. Based on the thrust closed-loop tracking error at the current discrete time point, the thrust closed-loop tracking error at the previous discrete time point, the set time step, and the cumulative error integral obtained by integrating the thrust closed-loop tracking error, the pulse width modulation duty cycle compensation is calculated by combining the preset proportional gain coefficient, the preset integral gain coefficient, and the preset differential gain coefficient. Anti-saturation processing is performed on the accumulated error integral value. The accumulated error integral value is forcibly constrained by a preset integral limit threshold to prevent control command over-limit and desaturation delay caused by long-term steady-state error.

9. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The process of fusing the feedforward control command and the feedback compensation command to generate the comprehensive drive duty cycle specifically includes: The feedforward pulse width modulation duty cycle reference value included in the feedforward control command is added together with the pulse width modulation duty cycle compensation amount included in the feedback compensation command to generate a preliminary comprehensive drive duty cycle. According to the hardware communication protocol of the thrust actuator, the working range of the preliminary integrated drive duty cycle is limited, setting values ​​below 10% to 10% and values ​​above 90% to 90%, thus obtaining the integrated drive duty cycle.

10. The numerical-physical hybrid experimental method for floating wind turbines based on real-time interpolation according to claim 1, characterized in that, The step of converting the integrated drive duty cycle into a digital pulse width modulation signal and outputting it to the thrust actuator to adjust the actual physical thrust specifically includes: Based on the linear mapping relationship, according to the minimum high-level pulse width time and the maximum high-level pulse width time identified by the electronic speed controller configured in the thrust actuator, the comprehensive drive duty cycle is mapped and calculated as the high-level pulse width time corresponding to the current discrete time point; The digital output channel of the pre-configured data acquisition card generates a square wave signal with alternating high and low levels according to a preset signal frequency. The high level is maintained for the same duration as the calculated high-level pulse width within each square wave signal cycle, and is then sent to the electronic speed controller of the thrust actuator to drive the thrust actuator to generate actual physical thrust.

Images