A method for monitoring the posture of a mixed-tower wind turbine tower based on operating state feedback

Abstract

This invention relates to the field of wind turbine structural health monitoring, and discloses an adaptive monitoring method for the attitude of hybrid-tower wind turbines based on operational status feedback. The method includes: synchronously acquiring the wind turbine operational data stream and the original tower attitude monitoring signal; determining the transient aerodynamic thrust based on the operational data stream, and using this thrust to determine the dynamic propagation velocity of the mechanical excitation wave generated by the stress stiffening effect; determining the propagation delay based on the dynamic propagation velocity, and performing a reverse time shift on the monitoring signal to achieve time-domain synchronization; reducing the feedback weight of the measurement update link of the solution model under the unit's operational adjustment state, and outputting the tower attitude data. This invention establishes a mapping relationship between aerodynamic loads and mechanical wave propagation characteristics, achieving precise decoupling between the unit's operational excitation and the tower's actual response, eliminating monitoring data distortion caused by spatiotemporal misalignment, and improving the reliability of safety monitoring for large-megawatt units.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine generator structural health monitoring technology, and particularly relates to an adaptive monitoring method for the attitude of hybrid tower wind turbines based on operational status feedback. Background Technology

[0002] Currently, as global wind turbines evolve towards large megawatts and ultra-flexible support structures, hybrid tower structures, formed by splicing a lower concrete tower and an upper steel tower, have become the mainstream solution in the industry. By arranging sensor arrays at key cross-sections of the tower, monitoring signals are collected to reconstruct the spatial dynamic attitude of the tower.

[0003] During the operation of a wind turbine generator set, the main control system maintains power stability through actions such as pitch, yaw, or speed regulation. These actions induce strong amplitude vibrations within the tower structure. Due to the low fundamental frequency of the hybrid tower structure, the self-induced disturbance signal induced by the unit's regulation overlaps with the tower's own attitude response frequency, resulting in the raw monitoring data collected by sensors containing spurious responses generated by operational excitations. For example, Chinese invention patent application CN121520142A discloses a method and system for dynamic monitoring of wind turbine generator set tower settlement and tilt. Through frequency domain analysis, the signal is decomposed into high-frequency dynamic signals using a preset cutoff frequency. The response and low-frequency steady-state deformation calculation, settlement and tilt rate calculation, and signal domain processing logic have limitations in large flexible heterogeneous support structures: the frequency domain decoupling mechanism ignores the physical time delay of mechanical excitation wave in the heterogeneous material transmission process, causing spatiotemporal misalignment between the algorithm adjustment window and the forced vibration range of the structure, resulting in phase hysteresis and monitoring distortion. The scheme does not consider the stress stiffening effect under large load conditions, that is, the aerodynamic thrust sudden change instantaneously changes the equivalent modulus of the tower material, causing the excitation wave propagation velocity to deviate from the static preset value. The data confidence of the compensation model decreases under extreme variable load conditions, making it difficult to meet the convergence requirements of the attitude solution matrix for ultra-high hybrid tower units of 120 meters and above.

[0004] Therefore, how to utilize the unit's operating status feedback to construct an alignment mechanism that can characterize the mechanical wave propagation hysteresis and aerodynamic stress coupling of the mixed tower, so as to achieve precise decoupling between operating excitation and the actual attitude response of the tower, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] The present invention aims to solve the problem that the transient excitation interference induced by the active adjustment action of the unit and the real-time calculation of the tower attitude have coupling distortion in the physical spatiotemporal domain, which makes it difficult to balance the monitoring accuracy and the real-time response.

[0006] In this technical solution, a method for adaptive monitoring of the attitude of a hybrid wind turbine tower based on operational status feedback includes the following steps: Step S101: Synchronously acquire the operating data stream of the wind turbine generator corresponding to the timestamp and the original monitoring signal of the tower attitude collected by the sensor array deployed at different elevations of the hybrid tower wind turbine tower; the operating data stream includes instantaneous pitch angle and generator speed. Step S102: Determine the instantaneous aerodynamic thrust of the wind turbine rotor acting on the top of the nacelle based on the operating data stream; Step S103: Using the instantaneous aerodynamic thrust and the preset equivalent stiffness mapping relationship, determine the dynamic propagation wave velocity of the mechanical excitation wave generated by the stress stiffening effect in the tower. Step S104: Determine the propagation delay of the mechanical excitation wave from the cabin to the location of each sensor based on the physical topological distance of each sensor in the sensor array relative to the cabin and the dynamic propagation wave speed. Step S105: The original tower attitude monitoring signal is reversed by using the propagation delay to synchronize the original monitoring signal with the operation data stream on the time axis. Step S106: When the running data stream indicates that the wind turbine generator is in the action adjustment state, the feedback weight of the measurement update link in the attitude calculation model is reduced, the attitude calculation model after feedback weight correction is used to process the synchronized original monitoring signal, and the real-time attitude monitoring data of the tower is output.

[0007] Preferably, step S103 includes: step S1031, mapping the instantaneous aerodynamic thrust to the transient bending stress distribution of each level section of the tower; step S1032, linearly correcting the original elastic modulus of the tower material according to the transient bending stress distribution to obtain the transient equivalent modulus of the tower under load; step S1033, determining the dynamic propagation wave velocity of the mechanical excitation wave inside the tower based on the ratio of the transient equivalent modulus to the square root of the tower material density.

[0008] Preferably, step S105 includes: step S1051, using the timestamp of the running data stream as the starting reference, performing a time-reverse translation on the original tower attitude monitoring signal, with the translation amount being the propagation delay; step S1052, resampling the translated original tower attitude monitoring signal using a linear interpolation algorithm to generate an aligned monitoring sequence that is synchronized with the running data stream on a physical time scale.

[0009] Preferably, before outputting the real-time attitude monitoring data of the tower, the method further includes: step S401, taking the aligned monitoring sequence as the observation input and the running data stream as the system control excitation input, and substituting them into the adaptive Kalman filter model; step S402, determining the process noise covariance matrix in the adaptive Kalman filter model according to the fluctuation rate of the running data stream, so as to realize the dynamic gain adjustment of the tower attitude monitoring data.

[0010] Preferably, the prediction model of the adaptive Kalman filter model incorporates a correction coefficient, which is used to quantify the impact of the operating excitation disturbances generated by the unit's actions on the convergence of the attitude solution matrix; the value of the correction coefficient increases nonlinearly with the increase of the pitch rate in the operating data stream.

[0011] Preferably, the correction coefficient is configured with an asymmetric recovery time constant; when the operating excitation disturbance decreases, the numerical recovery slope of the correction coefficient is less than its rising slope when the disturbance increases, so that the recovery speed of the feedback weight matches the structural vibration residual sway characteristics of the tower after the unit operation ends.

[0012] Preferably, the sensor array includes a first sensor group deployed on top of the steel section of the concrete tower, a second sensor group deployed at the interface between the concrete section and the steel section, and a third sensor group deployed above the foundation ring.

[0013] Preferably, for the second sensor group, the propagation delay includes an interface propagation correction term when the mechanical excitation wave passes through the connection interface; the interface propagation correction term is accumulated into the propagation delay based on the acoustic impedance difference between the materials on both sides of the connection interface and the dynamic change of the interface preload caused by the instantaneous aerodynamic thrust.

[0014] Preferably, the real-time attitude monitoring data of the output tower includes: feeding back the real-time attitude monitoring data to the main control system of the wind turbine generator set, and triggering the unit to reduce load or shut down when the monitored attitude deviation exceeds the preset safety boundary threshold.

[0015] Compared with existing technologies, the adaptive monitoring method for the attitude of hybrid wind turbine towers based on operational status feedback has the following advantages: 1. In the adaptive monitoring of tower attitude of hybrid wind turbines, an evaluation operator for operation-induced disturbances is constructed by utilizing the main control operating parameters of the wind turbine generator. This achieves logical decoupling between the mechanical disturbances generated by the unit's adjustment actions and the actual attitude response of the tower. This mechanism transforms the active adjustment processes such as pitch or yaw of the unit into prior known terms in the sensing domain. When the unit performs drastic actions, the algorithm dynamically lowers the feedback coefficients in the measurement update link based on the value of the evaluation operator, thereby blocking the pollution of the attitude calculation matrix by transient excitation interference. This approach avoids the distortion of monitoring data caused by self-induced disturbances of the unit, resolves the inherent contradiction between the noise filtering capability and the response phase hysteresis of traditional static filtering, and reduces the probability of system erroneous shutdown.

[0016] 2. Combining the physical characteristics of the hybrid tower structure, the axial length of the steel tower and the sensor installation elevation are used as the physical benchmark for time axis alignment. This eliminates the timing misalignment between the issuance of the main control command and the acquisition of physical fluctuations by the sensor. By establishing a propagation model of mechanical waves between heterogeneous materials and discontinuous interfaces, this scheme can accurately calculate the physical propagation hysteresis time of the excitation energy from the nacelle to the sensor position. This is used as the translation parameter of the evaluation operator. This spatiotemporal alignment mechanism ensures that the algorithm's anti-disturbance actions accurately cover the physical instant of stress distortion and modal superposition inside the tower, eliminates phase distortion caused by delay, and makes the attitude calculation results more consistent with the real physical motion trajectory of large flexible structures.

[0017] 3. Based on the stress stiffening effect induced by aerodynamic thrust, the propagation velocity of mechanical waves is dynamically corrected to enhance the robustness of the system under extreme aerodynamic load changes. When the wind turbine encounters strong gusts or performs emergency pitch control, the upper steel tower experiences local stiffness changes due to the huge instantaneous thrust, causing the actual propagation velocity of the mechanical excitation wave to deviate from the static constant. This scheme uses the generator speed and pitch angle to deduce the instantaneous thrust of the wind turbine in real time, and then calculates the dynamic wave velocity correction factor to achieve nonlinear compensation for the propagation hysteresis time. This dynamic hedging mechanism, which deeply integrates the aerodynamic-structural coupling characteristics of the wind turbine, ensures that the monitoring system can still maintain the convergence and high confidence of the attitude solution matrix under extreme load conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the overall steps of the hybrid tower wind turbine attitude adaptive monitoring method based on operational status feedback according to the present invention. Figure 2 This is a state evolution and control logic diagram of the monitoring method of the present invention under different operating conditions. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, bottom, transverse, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between components in a specific state (as shown in the accompanying drawings). They are only for the convenience of describing this invention and do not require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated.

[0021] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0022] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0023] An adaptive monitoring method for the tower attitude of a hybrid wind turbine based on operational status feedback includes the following steps: Step S101: Synchronously acquire the operating data stream of the wind turbine generator corresponding to the timestamp and the original monitoring signal of the tower attitude collected by the sensor array deployed at different elevations of the hybrid tower wind turbine tower; the operating data stream includes instantaneous pitch angle and generator speed. Step S102: Determine the instantaneous aerodynamic thrust of the wind turbine rotor acting on the top of the nacelle based on the operating data stream; Step S103: Using the instantaneous aerodynamic thrust and the preset equivalent stiffness mapping relationship, determine the dynamic propagation wave velocity of the mechanical excitation wave generated by the stress stiffening effect in the tower. Step S104: Determine the propagation delay of the mechanical excitation wave from the cabin to the location of each sensor based on the physical topological distance of each sensor in the sensor array relative to the cabin and the dynamic propagation wave speed. Step S105: The original tower attitude monitoring signal is reversed by using the propagation delay to synchronize the original monitoring signal with the operation data stream on the time axis. Step S106: When the running data stream indicates that the wind turbine generator is in the action adjustment state, the feedback weight of the measurement update link in the attitude calculation model is reduced, the attitude calculation model after feedback weight correction is used to process the synchronized original monitoring signal, and the real-time attitude monitoring data of the tower is output.

[0024] Preferably, step S103 includes: step S1031, mapping the instantaneous aerodynamic thrust to the transient bending stress distribution of each level section of the tower; step S1032, linearly correcting the original elastic modulus of the tower material according to the transient bending stress distribution to obtain the transient equivalent modulus of the tower under load; step S1033, determining the dynamic propagation wave velocity of the mechanical excitation wave inside the tower based on the ratio of the transient equivalent modulus to the square root of the tower material density.

[0025] Preferably, step S105 includes: step S1051, using the timestamp of the running data stream as the starting reference, performing a time-reverse translation on the original tower attitude monitoring signal, with the translation amount being the propagation delay; step S1052, resampling the translated original tower attitude monitoring signal using a linear interpolation algorithm to generate an aligned monitoring sequence that is synchronized with the running data stream on a physical time scale.

[0026] Preferably, before outputting the real-time attitude monitoring data of the tower, the method further includes: step S401, taking the aligned monitoring sequence as the observation input and the running data stream as the system control excitation input, and substituting them into the adaptive Kalman filter model; step S402, determining the process noise covariance matrix in the adaptive Kalman filter model according to the fluctuation rate of the running data stream, so as to realize the dynamic gain adjustment of the tower attitude monitoring data.

[0027] Preferably, the prediction model of the adaptive Kalman filter model incorporates a correction coefficient, which is used to quantify the impact of the operating excitation disturbances generated by the unit's actions on the convergence of the attitude solution matrix; the value of the correction coefficient increases nonlinearly with the increase of the pitch rate in the operating data stream.

[0028] Preferably, the correction coefficient is configured with an asymmetric recovery time constant; when the operating excitation disturbance decreases, the numerical recovery slope of the correction coefficient is less than its rising slope when the disturbance increases, so that the recovery speed of the feedback weight matches the structural vibration residual sway characteristics of the tower after the unit operation ends.

[0029] Preferably, the sensor array includes a first sensor group deployed on top of the steel section of the concrete tower, a second sensor group deployed at the interface between the concrete section and the steel section, and a third sensor group deployed above the foundation ring.

[0030] Preferably, for the second sensor group, the propagation delay includes an interface propagation correction term when the mechanical excitation wave passes through the connection interface; the interface propagation correction term is accumulated into the propagation delay based on the acoustic impedance difference between the materials on both sides of the connection interface and the dynamic change of the interface preload caused by the instantaneous aerodynamic thrust.

[0031] Preferably, the real-time attitude monitoring data of the output tower includes: feeding back the real-time attitude monitoring data to the main control system of the wind turbine generator set, and triggering the unit to reduce load or shut down when the monitored attitude deviation exceeds the preset safety boundary threshold.

[0032] Example 1: In the structural monitoring of high-power concrete-steel hybrid tower wind turbine generators with a height exceeding 120 meters during operation, when the unit performs pitch control, the blade angle adjustment causes aerodynamic thrust changes at the top of the tower. The mechanical vibration energy induced by this unit action is transmitted along the flexible steel tower. Due to the stiffness change and impedance mismatch in the steel-concrete transition section of the hybrid tower, the excitation wave is reflected and superimposed on modes inside the structure, resulting in strong amplitude transient disturbances mixed in the original signal collected by the sensor. If a static low-pass filtering strategy is adopted, the filtering depth is usually increased in order to filter out the interference signal that coincides with the first-order natural frequency of the tower, thereby introducing phase hysteresis in the attitude calculation link, causing a time misalignment between the tower clearance data sensed by the main control system and the structural displacement.

[0033] In the above-mentioned operating conditions, the method of this invention simultaneously acquires the operating data of the wind turbine generator corresponding to the timestamp, as well as the raw monitoring signals collected by the sensor array deployed at different elevations of the hybrid-tower wind turbine tower. The operating data includes the instantaneous pitch angle and the generator speed. Based on the operating data, the instantaneous aerodynamic thrust F of the wind turbine rotor acting on the top of the nacelle is determined. thrust By utilizing instantaneous aerodynamic thrust and a preset equivalent stiffness mapping relationship, the dynamic propagation wave velocity of the mechanical excitation wave generated by the stress stiffening effect within the tower is determined, and the axial length L of the upper steel tower of the wind turbine generator is extracted. sThe static carrier velocity V0 of the steel tower and the installation elevation of the sensors are used to address the baseline drift of the material caused by the diurnal temperature difference in the atmospheric boundary layer. Surface temperature data is acquired from the outer wall of the tower. Based on the physical law of the negative correlation between temperature and the elastic modulus of materials in solid heat transfer, linear temperature compensation is applied to the reference value of the static carrier velocity of the steel tower using the surface temperature data to determine the actual static carrier velocity of the steel tower under the current environment. To address the misalignment between microsecond-level propagation hysteresis and the physical scale of the conventional sampling period, a high-frequency phase-locked loop oversampling link based on a field-programmable gate array (FPGA) is initiated. Multiple high-frequency sub-sampling points are inserted within each cycle of the reference trigger clock. The effective time resolution of the front-end digital conversion device has been improved to the microsecond level. Although the macroscopic modal response of the tower system belongs to the low-frequency range, the stress stiffening effect caused by extreme pitching actions makes the minute changes in the transmission hysteresis over a span of hundreds of meters fall only in the sub-millisecond range. Conventional 100Hz low-frequency sampling beats will inevitably produce millisecond-level time domain blind zones, thus swallowing the minute distortion characteristics of wave velocity nonlinearity. Therefore, it is necessary to forcibly attach a high-frequency phase-locked loop circuit to construct a microsecond-level time scale, establish the physical prerequisites for capturing transient wave velocity fluctuations, and open up the cross-scale data conversion chain from the underlying microsecond-level electronic beats to the macroscopic low-frequency mechanical strain inversion calculation, based on the formula. Determine the dynamic propagation hysteresis time of the excitation wave from the cabin to each sensor location. , of which F thrust For instantaneous aerodynamic thrust, L s V is the conduction path length, V0 is the static carrier velocity of the steel tower, and K is the structural sensitivity coefficient. To determine the dynamic propagation delay duration, and then utilize the dynamic propagation delay duration. The original monitoring signal is reversed in time to align it with the operational data on the time axis. When the operational data indicates that the wind turbine is in an action adjustment state, the feedback weight of the measurement update link in the attitude calculation model is reduced, and the attitude calculation model after the feedback weight is corrected is used to process the aligned original monitoring signal.

[0034] A hedging mechanism based on aerodynamic-structural coupling characteristics is constructed. It utilizes the stress stiffening effect induced by aerodynamic thrust to correct the propagation velocity of mechanical waves. By delaying and shifting the timing of the operation-induced disturbance evaluation operator, the algorithm's weighted disturbance suppression action is anchored to the instant the mechanical excitation wave reaches the sensor location. After the unit completes pitch control and enters the vibration recovery phase, a nonlinear scheduling function incorporating trigger thresholds and asymmetric time constants ensures that the recovery speed of the feedback weights during disturbance reduction matches the structural vibration characteristics of the tower after the unit's operation ends. This suppresses matrix singularity oscillations during the calculation process, ultimately ensuring that the output tower attitude data falls within the transient region of the actuator's action. Maintaining high fidelity and real-time performance without delay, this system addresses the monitoring distortion caused by impedance mismatch and nonlinear stiffness changes in mixed-tower wind turbines under unsteady gusts and emergency shutdown conditions. The exponential scheduling function configures the lower bound of the dead zone as the unit's normal idle thrust threshold and the upper bound as half of the peak aerodynamic thrust during violent movements. When the thrust variation rate falls within the upper and lower bound dead zone intervals, it indicates that the structure has entered the vibration residual swing stage. At this point, the weight is paused and linearly recovers. The specific value of the asymmetric recovery time constant is calculated and extracted by multiplying the first-order natural mode period of the tower by the logarithmic decay rate of the residual swing damping. This constrains the recovery slope of the feedback weight to synchronously match the physical envelope curve of the residual vibration energy dissipated over time in the tower structure.

[0035] Example 2: This example establishes a multibody dynamics and finite element analysis joint simulation verification platform for a high-power concrete-steel hybrid tower wind turbine generator. By inputting the physical topology parameters and material elastic properties of the 120-meter-high hybrid tower structure, the forced vibration process of the wind turbine encountering a gust of 25.0 m / s at a rated wind speed of 15.5 m / s and implementing emergency pitch control is simulated. The data source comes from the physical displacement time sequence at the sensor installation location output by the platform, and Gaussian white noise with a signal-to-noise ratio of 20 dB and mechanical vibration interference with a frequency in the range of 0.5 Hz to 2.5 Hz are superimposed on it. The sampling frequency f of the key parameter is set. s The frequency is set to 100 Hz. The basis for this parameter is as follows: In order to capture the first three modal characteristics of the tower, the upper limit of its highest structural response frequency is 5.0 Hz. According to the sampling theorem and with a 10-fold oversampling margin reserved to reduce phase distortion caused by digital filtering, the lower limit of the sampling frequency is set to 50 Hz. Combined with the computing load boundary of the main control processor, 100 Hz is selected.

[0036] During the disturbance period of 1.25 to 3.50 seconds after the pitch control is initiated, the original monitoring signal S raw Including a forced oscillation component with an amplitude of 0.45 meters, the system acquires the corresponding operational data stream in real time and calculates the instantaneous aerodynamic thrust F of the wind turbine. thrustThe dynamic propagation wave velocity of the steel tower section was increased from 750.5 kN to 980.2 kN. Based on the positively correlated monotonic function relationship between stress stiffening effect and material equivalent stiffness, the dynamic propagation wave velocity of the steel tower section increased from 5100.2 m / s to 5145.8 m / s. Using this wave velocity and the sensor installation elevation at 90 meters, the dynamic propagation hysteresis time of the mechanical excitation wave from the nacelle to the sensor was determined. The transition occurs within the range of 15.1 milliseconds to 16.5 milliseconds, utilizing the resulting dynamic propagation hysteresis duration. Performing a reverse time shift on the original monitoring signal, due to the sub-millisecond decimal places in the dynamic propagation hysteresis duration, requires interpolation algorithms to reconstruct waveform point data for non-integer sampling periods when performing a reverse time shift on the discrete-time monitoring sequence. The spatial discrete point numerical smoothing reconstruction process inevitably involves truncation errors and effectively possesses low-pass filtering physical characteristics, leading to a consequent amplitude attenuation of the high-frequency physical real energy corresponding to the abrupt edge mapping of the excitation wave within the original monitoring signal. To address the attenuation of high-frequency physical characteristics caused by sub-millisecond shift operations, a collaborative procedure for time-series resampling and frequency domain compensation is constructed, based on the Xiang... Based on the sampling theorem and the inherent low-pass filtering physical characteristics of discrete signal interpolation, the amplitude attenuation ratio of the original monitoring signal before and after translation is calculated. A high-pass compensation operator based on the reciprocal of the amplitude attenuation ratio is constructed and applied to the translated digital signal sequence to compensate for the physical high-frequency characteristic components at the abrupt edge of the excitation wave. A hardware-level precise time protocol interface circuit is introduced to receive the synchronous pulse hardwired signal periodically sent by the main control system. The crystal counter inside the local processor performs microsecond-level zeroing and reset according to the synchronous pulse hardwired signal to eliminate the time axis reference deviation caused by the jitter of heterogeneous communication bus transmission.

[0037] Performance verification was performed using control groups. Control group one employed a static low-pass filter with a cutoff frequency of 0.5 Hz, while control group two used a fixed propagation delay shift of 16.0 milliseconds without adjusting the feedback weight. In this invention's sample group, when the wind turbine was in motion adjustment mode, the feedback weight of the measurement update link in the attitude calculation model was reduced from 0.85 to 0.15. At the most severe disturbance time of 2.10 seconds, the root mean square error (RMS) of the attitude data output by control group one relative to the actual displacement was 12.4%, and the phase lag reached 155.0 milliseconds. Control group two, due to the lack of correction for wave velocity variations caused by stress stiffening, still had a 4.2 millisecond synchronization residual after shifting, with an RMS error of 6.8%. In contrast, this invention's sample group, through adaptive weakening of the feedback weight, reduced the RMS error of its output attitude data to 1.1%, and the phase lag was less than 10.0 milliseconds. When the instantaneous aerodynamic thrust F... thrustAfter the rate of change drops below 15.0 kN / s and the unit enters the oscillation recovery phase, the feedback weight reverts to the normal value based on a recovery time constant of 2.5 seconds. By changing the external wind speed gradient, in a test sequence where the gust intensity decreases from 25.0 m / s to 20.0 m / s, the corresponding dynamic propagation hysteresis time is... The variation range was synchronously reduced from 1.4 milliseconds to 0.8 milliseconds. The root mean square error of the sample group of this invention was maintained below 1.5% throughout the entire wind speed gradient range. By mapping the physical time scale of the operational data feedback to the transmission scale of the structural stress wave, the anti-disturbance window of the monitoring system and the actual arrival time of the mechanical wave were physically aligned in situ.

[0038] Example 3: During the structural monitoring of a large-megawatt concrete-steel hybrid tower wind turbine, the aerodynamic thrust caused by the turbine's pitch control changes by 300 kN within 0.5 seconds. The induced mechanical excitation wave experiences a velocity drift due to the stress response characteristics of the steel tower material. To determine the structural sensitivity coefficient K, a calibration procedure for the equivalent stiffness mapping relationship was established during the static load testing phase before turbine operation. By controlling the yaw system to position the rotor at a preset windward angle, the equivalent static load sequence of aerodynamic thrust was obtained. Synchronous triggers deployed at the nacelle and sensor locations were used to measure the propagation velocity of the excitation wave in the tower. A first-order regression operator was applied to the load and velocity sequence, determining the structural sensitivity coefficient K to be 0.00015 N. -1 .

[0039] The system uses the SCADA clock of the main control system as the global time reference. A circular buffer with a length of 1024 sampling points is allocated in the local processor memory to temporarily store the raw monitoring signals acquired by the sensor array, utilizing the dynamic propagation hysteresis duration. Determine the time index offset for retrieving historical monitoring frames from the circular buffer, and calculate the instantaneous aerodynamic thrust F. thrust The first derivative determines the attenuation ratio of the feedback weight. When the aerodynamic thrust change rate is in the range of 20 kN / s to 100 kN / s, the feedback weight W decreases from the initial value of 0.85 to 0.15 according to a linear attenuation rule. The calculation of the feedback weight W is based on the following formula: Where W is the feedback weight, W base The initial values ​​are the normal weights. This is the gain adjustment operator, with a value of 0.007. The instantaneous aerodynamic thrust change rate is used as the basis for the precise value of the aforementioned gain adjustment operator. The value is determined by inversion calibration based on the pre-set megawatt-class rotor full-size ground test dataset of controlled impact limit test. The stress stiffening coefficient value is generated by performing uniaxial stepped loading ultrasonic penetration testing in batches according to the selected steel and concrete material grades. The extreme values ​​of the sound velocity drift gradient of each stress segment are extracted and fitted. Based on the above two quantitative control constants, the physical generalization support basis is covered for various heterogeneous installed capacity platforms.

[0040] The attitude calculation model lowers the feedback weight of the measurement update link, guiding the state prediction value to converge towards the constraints of the structural dynamics model, suppressing forced vibration interference in the original signal. As the thrust fluctuation represented by the operational data stream decreases, the feedback weight W regresses to its normal value through an exponential scheduling function containing a nonlinear dead zone. The monitoring system recovers to the observation state within 120 milliseconds, and the output tower attitude data maintains real-time performance within the load transient range. This solves the monitoring distortion problem caused by physical impedance mismatch and nonlinear stiffness changes in large flexible hybrid tower structures under complex aerodynamic loads. To ensure the physical consistency of the equivalent stiffness mapping relationship, a transient response model of the tower material under complex stress fields is established. The system will acquire the instantaneous aerodynamic thrust F. thrust Mapped to transient bending stress distribution at each level of the tower cross section The cross-sectional geometric parameters of the tower are used to calculate the moment of inertia of the material, determine the distribution of the bending moment generated by the top thrust along the tower axis, and, based on the principles of elasticity, determine the original elastic modulus E of the tower material. base Stress correction was applied to obtain the transient equivalent modulus E of the tower under load. trans Using the formula Determine the nonlinear increment of material stiffness, where E trans E is the transient equivalent modulus. base The original elastic modulus, The stress stiffening factor is... For the cross-sectional stress, and further based on the transient equivalent modulus E trans With tower material density The square root ratio is used to calculate the dynamic propagation wave velocity, anchoring the calculation process of mechanical wave propagation delay to the underlying mechanical constitutive relationship of the tower structure.

[0041] Example 4: During the commissioning of a concrete-steel hybrid tower wind turbine, to address the differences in mechanical wave propagation characteristics caused by the tower section wall thickness and concrete strength grade, the system calibrates the baseline value of the structural sensitivity coefficient K through controlled excitation. The main control system drives the blades to generate step pitch control movements with a frequency of 1 Hz to 3 Hz to produce a top impact load. Simultaneously, pressure sensors are used to acquire the bending moment response at the blade root and convert it into an instantaneous aerodynamic thrust F acting on the top of the nacelle. thrustThe sequence records the initial time deviation of the excitation wave arriving at the sensor at different elevation positions. Five gradient loading and unloading tests are applied within the thrust range of 200 kN to 600 kN. The least squares method is used to linearly fit the observed wave velocity change with the thrust amplitude.

[0042] After calibrating the structural sensitivity coefficient K, the monitoring system maintains temporal consistency of multi-source heterogeneous data before attitude calculation through data synchronization and buffer initialization procedures. The system uses a synchronization protocol to align the local processor's clock with the wind turbine's main control system clock. Based on a propagation delay limit of 16.5 milliseconds, a 2048-frame deep first-in-first-out (FIFO) data stack is created in the static random access memory (SRAM) to cache the raw monitoring signals acquired by the sensor array. Using the pitch trigger flag in the running data stream as a logical starting point, the system backtracks from the data stack to retrieve data based on the propagation delay... The corrected monitoring data frame.

[0043] Example 5: During the commissioning of the wind turbine tower monitoring system, the system initiated an initial calibration process for the internal parameters of the attitude calculation model, taking into account the damping characteristics of the tower structure and the sensor installation angle. By acquiring the displacement sequence of the tower under background environmental excitation, the natural frequencies of the structure were extracted from the displacement sequence using discrete Fourier transform. and modal damping ratio Combined with the preset structural amplitude standard deviation With sampling period Calculate the components Q of the state transition covariance matrix Q. ii Q ii The calculation formula is as follows: , where Q ii For process noise covariance components, The standard deviation of structural amplitude. The modal damping ratio, The natural frequency of the structure, The sampling period is determined by using the acquired Q. ii An initial value for the covariance of the attitude calculation model is set to constrain the fluctuation range of the predicted state values. A dynamic gain adjustment procedure for the process noise covariance matrix is ​​established to eliminate the broadband operating disturbances of the tower caused by the active adjustment actions of the unit. Based on the principle of variance characterizing signal dispersion in random signal analysis, the deviation sequence of wind turbine speed control commands is collected, and the discrete variance of the deviation sequence within a preset sliding window is calculated. This variance is multiplied by the initial value of the normal process noise covariance to determine the dynamic increment of the process noise covariance at the current moment, and accumulated into the process noise covariance matrix of the attitude calculation model. This enables the attitude calculation model to actively reduce the prediction confidence weight when the wind turbine control system issues drastic action commands, thus constraining attitude divergence in the transient range of strong loads.

[0044] To address the azimuth deviation caused by the sensor array being installed on the inner wall of the tower, the system reads the yaw absolute encoder value in the static monitoring window after the yaw motor stops, and simultaneously acquires the three-dimensional distribution of gravity acceleration collected by the sensor on the sensitive axis. The coordinate transformation matrix is ​​reconstructed using the projection relationship between the gravity vector direction and the physical principal axis of the tower, aligning the sensor's local coordinate system to the geometric center axis of the tower. The output attitude monitoring data eliminates the coordinate projection residual caused by structural deviation, keeping the zero-point drift rate below the 0.8% threshold.

[0045] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback, characterized in that, Includes the following steps: Step S101: Synchronously acquire the operating data stream of the wind turbine generator corresponding to the timestamp and the original monitoring signal of the tower attitude collected by the sensor array deployed at different elevations of the hybrid tower wind turbine tower; the operating data stream includes instantaneous pitch angle and generator speed. Step S102: Determine the instantaneous aerodynamic thrust of the wind turbine rotor acting on the top of the nacelle based on the operating data stream; Step S103: Using the instantaneous aerodynamic thrust and the preset equivalent stiffness mapping relationship, determine the dynamic propagation wave velocity of the mechanical excitation wave generated by the stress stiffening effect in the tower. Step S104: Determine the propagation delay of the mechanical excitation wave from the cabin to the location of each sensor based on the physical topological distance of each sensor in the sensor array relative to the cabin and the dynamic propagation wave speed. Step S105: The original tower attitude monitoring signal is reversed by using the propagation delay to synchronize the original monitoring signal with the operation data stream on the time axis. Step S106: When the running data stream indicates that the wind turbine generator is in the action adjustment state, the feedback weight of the measurement update link in the attitude calculation model is reduced, the attitude calculation model after feedback weight correction is used to process the synchronized original monitoring signal, and the real-time attitude monitoring data of the tower is output.

2. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 1, characterized in that, Step S103 includes: Step S1031, mapping the instantaneous aerodynamic thrust to the transient bending stress distribution of each level section of the tower; Step S1032, linearly correcting the original elastic modulus of the tower material according to the transient bending stress distribution to obtain the transient equivalent modulus of the tower under load; Step S1033, determining the dynamic propagation wave velocity of the mechanical excitation wave inside the tower based on the ratio of the transient equivalent modulus to the square root of the tower material density.

3. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 1, characterized in that, Step S105 includes: Step S1051, using the timestamp of the running data stream as the starting reference, performing a time-reverse translation on the original tower attitude monitoring signal, with the translation amount being the propagation delay; Step S1052, resampling the translated original tower attitude monitoring signal using a linear interpolation algorithm to generate an aligned monitoring sequence that is synchronized with the running data stream on a physical time scale.

4. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 3, characterized in that, Before outputting the real-time attitude monitoring data of the tower, the process further includes: step S401, taking the alignment monitoring sequence as the observation input and the running data stream as the system control excitation input, and substituting them into the adaptive Kalman filter model; step S402, determining the process noise covariance matrix in the adaptive Kalman filter model based on the fluctuation rate of the running data stream, thereby realizing dynamic gain adjustment of the tower attitude monitoring data.

5. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 4, characterized in that, The adaptive Kalman filter model introduces a correction coefficient into its prediction model. This correction coefficient is used to quantify the impact of operational excitation disturbances caused by unit actions on the convergence of the attitude solution matrix. The value of the correction factor increases non-linearly with the increase of the pitch rate in the running data stream.

6. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 5, characterized in that, The correction coefficient is configured with an asymmetric recovery time constant; when the operating excitation disturbance decreases, the numerical recovery slope of the correction coefficient is less than its rising slope when the disturbance increases, so that the recovery speed of the feedback weight matches the structural vibration residual sway characteristics of the tower after the unit operation ends.

7. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 1, characterized in that, The sensor array includes a first sensor group deployed on top of the steel section of the concrete tower, a second sensor group deployed at the interface between the concrete section and the steel section, and a third sensor group deployed above the foundation ring.

8. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 7, characterized in that, For the second sensor group, the propagation delay includes an interface propagation correction term when the mechanical excitation wave passes through the connection interface; the interface propagation correction term is accumulated into the propagation delay based on the difference in acoustic impedance between the materials on both sides of the connection interface and the dynamic change of the interface preload caused by the instantaneous aerodynamic thrust.

9. The method for adaptive monitoring of tower attitude of hybrid wind turbines based on operational status feedback according to claim 1, characterized in that, The real-time attitude monitoring data of the output tower includes: feeding back the real-time attitude monitoring data to the main control system of the wind turbine generator set. When the monitored attitude deviation exceeds the preset safety boundary threshold, the main control system triggers the unit to reduce load or shut down.

Images