PLC control-based tower drum vibration active suppression method and system

Abstract

The embodiment of the application relates to the technical field of tower vibration suppression, and provides a tower vibration active suppression method and system based on PLC control, real-time acquisition of tower acceleration signals, extraction of window vibration energy and dominant frequency characteristics, construction of a composite evaluation system containing performance degradation and model mismatch dimensions, utilization of index normalization and nonlinear weighted fusion technology, intelligent quantification of urgency of re-identification, triggering of a lightweight online identification and control parameter recalculation process only when system control performance significantly decreases and it is confirmed that the decrease is caused by model parameter drift, otherwise maintaining a low-power consumption monitoring mode, which effectively solves the performance attenuation problem caused by the fact that a traditional fixed parameter controller cannot adapt to the time-varying of tower structure characteristics, avoids excessive occupation of PLC computing resources by a persistent complex algorithm, realizes efficient and robust control of a time-varying structure system on a limited computing power platform, and ensures vibration reduction effect in the whole life cycle of a fan.

Description

Technical Field

[0001] This invention relates to the field of tower vibration suppression technology, and in particular to a method and system for active tower vibration suppression based on PLC control. Background Technology

[0002] As wind turbines develop towards larger capacity and taller towers, the tower, as a key supporting component, has seen a significant increase in structural flexibility. Under complex and variable wind loads, the tower is highly susceptible to large-scale swaying, which not only affects the power generation efficiency of the wind turbine but also accelerates fatigue damage to the tower base flanges and bolts, and in severe cases, even threatens structural safety. Therefore, the industry has widely adopted active tower vibration suppression systems based on Programmable Logic Controllers (PLCs) to attenuate structural vibration by applying reverse control forces.

[0003] However, existing PLC-based tower vibration suppression schemes face numerous technical problems during long-term operation. For example, traditional control strategies typically rely on a static system model determined during the design or commissioning phase, where control parameters such as feedback gain remain fixed once set. This approach ignores the time-varying nature of the tower's structural dynamics. In actual service environments, the tower's natural frequency and damping characteristics slowly drift due to the combined effects of ambient temperature fluctuations, changes in foundation soil stiffness, material aging, and blade icing. When there is a significant deviation between the actual object and the static model within the PLC, the controller with fixed parameters will fail to maintain the expected vibration reduction effect, leading to performance degradation and potentially even system instability due to severe model mismatch. Furthermore, some improved schemes attempt to introduce parameter update mechanisms, but limited by the PLC's limited computing resources, the decision logic for triggering updates is often overly simplistic. Existing triggering mechanisms typically rely solely on a single vibration amplitude threshold for binary judgment, failing to accurately distinguish whether the vibration exceeding limits is caused by external instantaneous strong wind disturbances or by internal model mismatch. This lack of collaborative judgment mechanism is prone to false triggering. For example, it may mistakenly start the high-computing-power model identification process when only routine control is needed to suppress gust interference, resulting in the waste of valuable computing resources. Or it may fail to respond in time when the model does fail, making it impossible to achieve a balance between resource constraints and adaptive requirements. Summary of the Invention

[0004] The present invention aims to solve at least one of the problems existing in the prior art, and to provide a method and system for active suppression of tower vibration based on PLC control.

[0005] One aspect of the present invention provides a PLC-controlled active suppression method for tower vibration, comprising: Obtain the raw acceleration signal of the tower at the current moment; The raw acceleration signal at the current moment is filtered, buffered, and its features are extracted to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. A composite event decision is made based on window vibration energy and window dominant frequency to obtain identification trigger flags; In response to the identification trigger flag being true, lightweight online identification is performed on the time series of the filtered acceleration signal and the time series of the control force to obtain a new discrete state space matrix. The control parameters are adaptively recalculated based on the new discrete state space matrix to obtain the new controller gain. Based on the new controller gain, vibration suppression control is applied to the filtered acceleration signal to obtain the control output command for the current cycle, and the tower is controlled according to the control output command for the current cycle.

[0006] Another aspect of the present invention provides a PLC-controlled active tower vibration suppression system, comprising: The raw signal acquisition module is used to acquire the raw acceleration signal of the tower at the current moment; The filtering, data buffering, and feature extraction module is used to filter, buffer, and extract features from the raw acceleration signal at the current moment to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. The composite event decision module is used to make composite event decisions based on window vibration energy and window dominant frequency to obtain identification trigger flags. The lightweight online identification module is used to perform lightweight online identification on the time series of the filtered acceleration signal and the time series of the control force in response to the identification trigger flag being true, so as to obtain a new discrete state space matrix. The adaptive recalculation module for control parameters is used to adaptively recalculate control parameters based on the new discrete state space matrix to obtain a new controller gain. The vibration suppression control module is used to perform vibration suppression control on the filtered acceleration signal based on the new controller gain to obtain the control output command for the current cycle, and control the tower according to the control output command for the current cycle.

[0007] Compared with existing technologies, this invention acquires tower acceleration signals in real time and extracts window vibration energy and dominant frequency characteristics to construct a composite evaluation system that includes performance degradation and model mismatch dimensions. By using index normalization and nonlinear weighted fusion technology, the urgency of re-identification is intelligently quantified. Only when the system control performance significantly declines and it is confirmed that it is due to model parameter drift is the lightweight online identification and control parameter recalculation process triggered. Otherwise, a low-power monitoring mode is maintained. This mechanism effectively solves the problem that traditional fixed parameter controllers cannot adapt to the performance degradation caused by the time-varying characteristics of tower structures. At the same time, it avoids the excessive occupation of PLC computing resources by continuous complex algorithms, realizes efficient and robust control of time-varying structural systems on a limited computing power platform, and ensures the vibration reduction effect throughout the entire life cycle of the wind turbine. Attached Figure Description

[0008] One or more embodiments are illustrated by way of example with the corresponding pictures in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0009] Figure 1 This is a flowchart of a PLC-controlled active tower vibration suppression method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the data flow of a PLC-controlled active tower vibration suppression method according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating the process of filtering, data buffering, and feature extraction of the original acceleration signal at the current moment to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency in the active suppression method for tower vibration based on PLC control according to an embodiment of the present invention. Figure 4 A flowchart illustrating the active suppression method for tower vibration based on PLC control according to an embodiment of the present invention, which uses composite event decision-making based on window vibration energy and window dominant frequency to obtain the identification trigger flag; Figure 5 This is a flowchart illustrating the active vibration suppression method for towers based on PLC control according to an embodiment of the present invention, which uses a new controller gain to perform vibration suppression control on a filtered acceleration signal to obtain the control output command for the current cycle. Figure 6 This is a block diagram of a PLC-controlled active vibration suppression system for towers according to an embodiment of the present invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details are presented in the various embodiments of the present invention to facilitate a better understanding of the invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and with various variations and modifications based on the following embodiments. The division of the various embodiments below is for ease of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined with and referenced by each other without contradiction.

[0011] As indicated in the specification and claims of this invention, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0012] While this invention makes various references to certain modules in systems according to embodiments of the invention, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0013] This invention uses flowcharts to illustrate the operations performed by the system according to embodiments of the invention. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps can be processed in reverse order or simultaneously, as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0014] Existing PLC-based active tower vibration suppression systems typically employ static models and fixed parameters determined during the design phase. These systems cannot effectively adapt to the drift in dynamic characteristics of the tower caused by environmental changes or structural aging. Furthermore, limited by the computational power of PLCs, they struggle to support continuous and complex model identification, leading to gradual deterioration or even failure of vibration reduction performance over long-term operation. Therefore, this invention proposes a PLC-controlled active tower vibration suppression method. This method aims to achieve on-demand allocation of computational resources through intelligent evaluation of system status, thereby maintaining high fidelity of the control model with low resource consumption. Specifically, the technical solution of this invention first acquires and filters the tower acceleration signal in real time, and calculates the window vibration energy and window dominant frequency through data buffering and feature extraction. Then, based on the window vibration energy and window dominant frequency, a composite event decision is made, using normalization and weighted fusion techniques to quantify the urgency of performance degradation and model mismatch to generate an identification trigger flag. Only when the identification trigger flag is true, a lightweight online identification process is responsively initiated, updating the discrete state space matrix using the latest signal sequence, and adaptively recalculating the controller gain accordingly. Finally, based on the new controller gain obtained from the adaptive recalculation, a control output command for the current cycle is generated, and vibration suppression is implemented according to the corresponding updated control parameters. This invention, through this closed-loop monitoring-decision-update mechanism, effectively solves the problem of poor adaptability of fixed-parameter controllers to time-varying systems during tower suppression, while avoiding redundant calculations that consume PLC resources.

[0015] Figure 1 This is a flowchart of a PLC-controlled active suppression method for tower vibration according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the data flow in a PLC-controlled active suppression method for tower vibration according to an embodiment of the present invention. (In conjunction with...) Figure 1 and Figure 2According to an embodiment of the present invention, a PLC-based active vibration suppression method for towers includes the following steps: S100, acquiring the original acceleration signal of the tower at the current moment; S200, filtering, buffering, and extracting features from the original acceleration signal at the current moment to obtain a filtered acceleration signal, a window vibration energy, and a window dominant frequency; S300, performing composite event decision-making based on the window vibration energy and the window dominant frequency to obtain an identification trigger flag; S400, responding to the identification trigger flag being true, performing lightweight online identification on the time series of the filtered acceleration signal and the time series of the control force to obtain a new discrete state space matrix; S500, performing adaptive recalculation of control parameters based on the new discrete state space matrix to obtain a new controller gain; S600, performing vibration suppression control on the filtered acceleration signal based on the new controller gain to obtain a control output command for the current period, and controlling the tower according to the control output command for the current period.

[0016] Specifically, in step S100, the raw acceleration signal of the tower at the current moment is acquired. It is understood that during operation, the wind turbine tower is subjected to continuous multi-source disturbances such as random wind loads, blade rotation excitation, and grid load fluctuations. Its vibration response exhibits significant time-varying and nonlinear characteristics, making it difficult to achieve precise vibration attenuation through control strategies lacking real-time observation data. Therefore, in the technical solution of this invention, the raw acceleration signal of the tower at the current moment is acquired to capture the instantaneous motion state and dynamic response of the tower structure in the current time slice. This provides real and unprocessed physical data support for subsequent signal filtering, energy feature extraction, and real-time calculation of the control law, ensuring that the active suppression action can accurately match the current structural vibration phase and amplitude.

[0017] More specifically, in a concrete example of the present invention, step S100 first relies on high-precision accelerometers installed at specific locations on the top of the tower or inside the nacelle to sense changes in the inertial force of the structure. These sensors are rigidly connected along the main vibration direction of the tower, such as in the downwind or crosswind direction. The sensors convert the sensed mechanical vibration into continuously varying analog electrical signals, such as standard voltage or current signals, and transmit them to the analog input port of the programmable logic controller (PLC) via interference-resistant transmission lines. Subsequently, an analog-to-digital converter (ADC) discretizes the analog signal transmitted from the sensor within a preset sampling period, quantizing the continuous voltage or current amplitudes into digital codes. The sampling process strictly follows the sampling theorem of signal processing, ensuring that the sampling rate covers the effective bandwidth of the tower's modal frequencies of interest, thereby generating a series of raw acceleration values ​​arranged in chronological order and temporarily storing them in the controller's memory address as the input source for subsequent algorithm processing.

[0018] Specifically, in step S200, the original acceleration signal at the current moment is filtered, buffered, and feature-extracted to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. It is understood that the original acceleration signal acquired on-site inevitably contains high-frequency electromagnetic noise, mechanical clutter, and DC bias, and the sampled value at a single moment cannot characterize the dynamic evolution trend or periodic characteristics of the tower structure over a period of time, making it difficult to directly use for evaluating the overall control performance or model matching degree of the system. Therefore, in the technical solution of this invention, the original acceleration signal at the current moment is further filtered, buffered, and feature-extracted to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. This removes interference components unrelated to the mode of interest, constructs a time series window that reflects recent historical states, and quantifies the energy index characterizing vibration strength and the frequency index characterizing structural characteristics from it. This provides a clean signal with a high signal-to-noise ratio for subsequent control law calculations, and provides statistically significant state criteria for event-driven intelligent decision-making mechanisms. It effectively avoids false triggering caused by instantaneous noise or non-dominant mode interference, and ensures that the control system accurately perceives the true dynamic behavior of the structure.

[0019] Figure 3 This is a flowchart illustrating the process of filtering, buffering, and feature extraction of the original acceleration signal at the current moment to obtain the filtered acceleration signal, windowed vibration energy, and windowed dominant frequency in a PLC-controlled active vibration suppression method for towers according to an embodiment of the present invention. Figure 3 As shown, step S200 includes: S210, performing bandpass filtering on the original acceleration signal at the current moment to obtain a filtered acceleration signal; S220, storing the filtered acceleration signal in a first-in-first-out circular buffer; S230, calculating the root mean square of the time series of the filtered acceleration signal in the circular buffer as the window vibration energy; S240, performing a fast Fourier transform on the time series of the filtered acceleration signal in the circular buffer to obtain the window dominant frequency.

[0020] In step S210, the raw acceleration signal at the current moment is bandpass filtered to obtain a filtered acceleration signal. It is understood that wind farms typically have complex electromagnetic environments and mechanical operating conditions, which inevitably leads to the inclusion of high-frequency measurement noise, high-frequency vibration components of the nacelle, and low-frequency DC components caused by sensor installation tilt angles or zero-point drift in the acquired raw acceleration signal. If these interference components participate directly in control calculations without processing, it will cause deviations or even saturation in the control commands. Therefore, in the technical solution of this invention, the raw acceleration signal at the current moment is further bandpass filtered to obtain a filtered acceleration signal, thereby accurately capturing the target frequency band containing key tower mode information, while effectively attenuating various interference signals and non-target mode components outside the passband. This significantly improves the signal-to-noise ratio of the feedback signal, eliminates the influence of DC bias on the steady-state accuracy of the control system, and prevents the actuator from generating ineffective actions or excessive wear due to tracking high-frequency noise.

[0021] More specifically, in a specific example of the present invention, the filtering process in step S210 first determines the target mode frequency range to be suppressed based on the dynamic design parameters of the tower structure. This determines the low-frequency cutoff frequency and high-frequency cutoff frequency of the bandpass filter. The low-frequency cutoff frequency is used to isolate DC bias and extremely low-frequency drift, while the high-frequency cutoff frequency is used to filter out sensor noise and higher-order structural modes. Subsequently, a digital filter model matching the tower's vibration frequency characteristics is constructed in the computation unit, such as a second-order Butterworth or Chebyshev digital filter designed using the bilinear transform method, and discretized into specific difference equations. Within each control cycle, the discrete raw acceleration values ​​acquired at the current moment are substituted into these difference equations, and recursive calculations are performed using historical input and output data to output a smooth acceleration sequence that retains only the dominant vibration characteristics of the tower in real time.

[0022] In step S220, the filtered acceleration signal is stored in a first-in-first-out (FIFO) circular buffer. It is understood that a macroscopic assessment of the tower's vibration state, such as the strength of vibration energy or the drift of the dominant frequency, cannot be derived solely from instantaneous sampling values ​​at a single moment; rather, it requires statistical analysis of the signal's evolution over a past period. Furthermore, frequent large-scale data relocation or reallocation in industrial controllers consumes significant computational cycles, leading to a decrease in real-time performance. Therefore, in the technical solution of this invention, the filtered acceleration signal is further stored in a FIFO circular buffer to construct a sliding data window of fixed length that dynamically shifts over time. This allows for real-time input of new data and automatic discarding of outdated data without moving the physical storage location of existing data. In this way, a continuous and real-time historical vibration sequence can be maintained with extremely low time and space complexity, providing complete time-slice data support for subsequent root mean square (RMS) calculations and spectral analysis.

[0023] More specifically, in a concrete example of the present invention, step S220 first pre-allocates a continuous fixed-length address space as a data carrier in the storage space of the programmable logic controller. The length of this address space is precisely set according to the product of the window time width required for subsequent feature extraction and the signal sampling frequency. Simultaneously, a write pointer or index variable pointing to the starting address of this storage space is initialized. Whenever a new filtered acceleration value is generated, it is directly written to the storage unit pointed to by the current write pointer, overwriting the oldest data at that location. Subsequently, the write pointer is incremented, and a modulo operation is performed on the total length of the buffer, so that when the pointer reaches the end of the storage space, it automatically wraps back to the starting position. Through this circular addressing mechanism, the buffer always stores acceleration data from the most recent N sampling cycles, achieving continuous updating and seamless connection of the data stream.

[0024] In step S230, the root mean square (RMS) of the time series of the filtered acceleration signal in the annular buffer is calculated as the window vibration energy. It is understood that the tower's vibration manifests as a reciprocating motion around its equilibrium position, and its instantaneous acceleration signal exhibits alternating positive and negative characteristics. A simple arithmetic mean approaches zero, thus losing its physical meaning, and a single peak value is highly susceptible to transient impacts and cannot accurately reflect the magnitude of the vibration energy and fatigue accumulation effect continuously borne by the structure over a period of time. Therefore, in the technical solution of this invention, the RMS of the time series of the filtered acceleration signal in the annular buffer is further calculated as the window vibration energy. This allows for macroscopic quantification of the vibration intensity within the current time window from a statistical perspective, transforming the fluctuating time-domain signal into a single, non-negative effective value index. This accurately characterizes the average kinetic energy level of the tower within the monitoring period, providing a robust quantitative basis for subsequent judgments on whether the control system meets the preset vibration reduction performance standards, and avoiding misjudgments due to local signal fluctuations.

[0025] More specifically, in a specific example of the present invention, step S230 first iterates through all N filtered acceleration sampling points currently stored in the annular buffer to obtain complete time series data. Then, a pipelined numerical operation is performed according to the mathematical definition of root mean square (RMS): each discrete acceleration sampling value is squared one by one to eliminate sign differences and convert it into an instantaneous energy representation; all values ​​obtained from the squared operation are summed to obtain the total energy integral within the time window; this total energy integral is divided by the data length N of the annular buffer to calculate the average power; finally, the square root operation is performed on this average power, and the result is defined as the window vibration energy. Its value directly reflects the actual sway intensity of the tower under the current wind conditions and is used as a key performance evaluation parameter in subsequent event decision logic. In a specific scenario, the sampling frequency can be set to 100Hz, the annular buffer length N can be set to 1024, corresponding to a time window of approximately 10 seconds, and the performance threshold can be selected as 1.5 times the RMS value of the tower's vibration under rated operating conditions.

[0026] In step S240, a Fast Fourier Transform (FFT) is performed on the time series of the filtered acceleration signal in the annular buffer to obtain the dominant frequency of the window. It is understood that while the acceleration waveform of the tower structure in the time domain directly reflects the vibration amplitude, it is difficult to directly reveal the inherent dynamic characteristics of the structure. Especially when the structural stiffness changes due to foundation loosening or icing, the resulting natural frequency drift is hidden in the complex time-domain waveform and cannot be captured by simple time-domain statistics, thus making it impossible to determine whether the current control model has failed due to parameter mismatch. Therefore, in the technical solution of this invention, a FFT is further performed on the time series of the filtered acceleration signal in the annular buffer to obtain the dominant frequency of the window. This maps the discrete time series to the frequency domain, analyzes the distribution of the current vibration energy at different frequency components, and accurately locates the peak frequency where the energy is most concentrated. This allows for real-time monitoring of the actual dominant frequency of the tower and comparison with the pre-stored theoretical model frequency in the controller, providing a physically meaningful frequency domain criterion for identifying model mismatch risks.

[0027] More specifically, in a specific example of the present invention, step S240 first non-destructively reads the latest N-point filtered acceleration data sequence from the circular buffer and copies it to a temporary operand array. To suppress spectral leakage caused by data truncation, a window function is applied to the time-domain data in the temporary operand array, for example, multiplying by a Hanning or Hamming window coefficient, so that the amplitude at both ends of the signal is smoothly attenuated to zero. Subsequently, a fast Fourier transform algorithm with a base-2 or mixed base is applied to perform a butterfly operation on the windowed real number sequence, decomposing it into a complex spectral sequence with real and imaginary parts. Next, the magnitude or power spectral density corresponding to each frequency point is calculated, and the amplitude data is traversed throughout the entire spectral range to search for the position of the spectral peak with the largest amplitude. Finally, based on the index position of the spectral peak in the operand array and the ratio of the sampling frequency to the number of data points N, the corresponding physical frequency value is calculated, and this physical frequency value is determined as the current window dominance frequency for subsequent model mismatch evaluation.

[0028] Specifically, in step S300, a composite event decision is made based on the window vibration energy and the window dominant frequency to obtain the identification trigger flag. It is understandable that the technical flaw of the event triggering mechanism in existing schemes stems from its use of simple "OR" logic for decision-making. This logic is inherently rigid and binary; it cannot quantify the severity of the problem, treating minor and severe exceedances equally. This is too crude in complex industrial wind power scenarios, easily leading to unnecessary waste of computational resources. A more critical weakness of this logic is that it completely ignores the deep causal relationship between performance degradation and model mismatch. For example, performance degradation may be triggered by only a brief, extreme external disturbance (such as a sudden strong gust of wind). In this case, the model itself is not mismatched, and immediately triggering model re-identification is not only unhelpful but may even introduce errors.

[0029] Therefore, in the technical solution of this invention, a composite event decision is further performed based on window vibration energy and window dominant frequency to obtain identification trigger flags. This constructs a dynamic weighted trigger decision process based on synergistic effects. Through continuous quantization and nonlinear fusion, intelligent assessment of the urgency of re-identification is achieved. This mechanism aims to accurately capture the most urgent update demand when model mismatch is the root cause of performance degradation in the physical scenario of tower vibration suppression. At this critical moment, two alarm conditions will occur simultaneously. This effectively avoids the problem that the original "OR" logic cannot distinguish these drastically different industrial scenarios, resulting in a lack of intelligence and robustness in its decision-making mechanism. It accurately allocates valuable PLC computing resources to the most needed situations, ensuring that high-consumption update tasks are only initiated when model parameter drift substantially affects the control effect.

[0030] Figure 4 This is a flowchart illustrating a PLC-controlled active vibration suppression method for towers according to an embodiment of the present invention, which uses a composite event decision based on window vibration energy and window dominant frequency to obtain an identification trigger flag. For example... Figure 4 As shown, step S300 includes: S310, normalizing and quantifying the severity of window vibration energy and window dominant frequency to obtain performance degradation score and model mismatch score; S320, performing urgency-weighted fusion on performance degradation score and model mismatch score to obtain re-identification urgency score; S330, generating the identification trigger flag based on re-identification urgency score and urgency threshold.

[0031] In step S310, the window vibration energy and window dominant frequency are normalized and quantified to obtain performance degradation scores and model mismatch scores. It is understood that the reason for performing index normalization and severity quantization first is that the original binary judgment cannot reflect the degree to which the system state deviates from the normal range, and simple threshold logic is difficult to numerically distinguish between slight parameter fluctuations and catastrophic system failures, causing the controller to be unable to perceive the urgency level of the problem. Therefore, in the technical solution of this invention, the window vibration energy and window dominant frequency are further normalized and quantified to obtain performance degradation scores and model mismatch scores, thereby transforming discrete alarm signals into continuous, measurable scores. This transforms the abstract performance difference into a concrete numerical value, providing a precise hierarchical quantification basis for subsequent adaptive control decisions and avoiding the misallocation or abuse of computing resources due to a lack of refined assessment of fault severity.

[0032] More specifically, in a specific example of the invention, the process of step S310 involves monitoring the window vibration energy in real time. With window dominance frequency This is achieved through processing. Specifically, the performance degradation score is calculated using the following formula. It will cause vibration energy to exceed the preset performance threshold. Quantify the relative proportions: .

[0033] in, Score for performance degradation. For window vibration energy, The preset performance threshold, This is a function of maximum value. It should be understood that a higher performance degradation score directly indicates that the current control effect of the active suppression system is further from the desired target, thus providing a quantitative input for subsequent urgency assessment regarding the severity of the current problem. For example, under conditions of sudden increase in wind shear, if the actual monitored vibration energy only exceeds the performance threshold by 10%, the calculated performance degradation score is 0.1, which might indicate that no intervention is necessary. However, if under extreme gust impact, the vibration energy surges to three times the performance threshold, the calculated score jumps directly to 2.0. This huge numerical difference directly warns that the system is currently on the verge of extreme danger and loss of control.

[0034] Meanwhile, the model mismatch score is calculated using another formula. Model mismatch score The calculation will use the measured dominant vibration frequency, i.e., the window dominant frequency. Compared to the current model reference frequency stored in the PLC, i.e., the current model frequency. The deviation, relative to the preset frequency tolerance, i.e., the frequency mismatch tolerance. Normalize the model. Model mismatch score. The calculation process can be expressed as: .

[0035] in, The model mismatch score, For window-dominant frequency, For the current model frequency, This represents the tolerance for frequency mismatch. Model mismatch score. This study quantifies the drift of the natural frequency, a core parameter describing the tower's dynamic characteristics. This provides a forward-looking quantitative basis for assessing the risk of future performance degradation, rather than simply responding to adverse consequences that have already occurred. For example, when blade icing significantly alters the overall mass distribution of the tower during winter, its actual natural frequency may shift by 0.05 Hz relative to the model's reference frequency. If the preset frequency mismatch tolerance is only 0.01 Hz, then the model mismatch score... The calculated score is as high as 5.0. This significantly high score accurately conveys to the controller the physical fact that the model is severely distorted, rather than simply conveying noise interference. Through this two-dimensional quantification mechanism, the current control state and potential model risks can be comprehensively and quantitatively perceived.

[0036] That is, the window vibration energy and window dominant frequency are normalized and their severity quantified to obtain the performance degradation score and model mismatch score, including: normalizing and quantifying the window vibration energy and window dominant frequency using the following formula: ; ; in, Score for performance degradation. For window vibration energy, The preset performance threshold, It is a function with maximum value. The model mismatch score, For window-dominant frequency, For the current model frequency, This represents the tolerance for frequency mismatch.

[0037] In step S320, the performance degradation score and model mismatch score are weighted and fused based on urgency to obtain a re-identification urgency score. It is understood that simple linear combinations or logical judgments cannot capture the special importance of concurrent performance degradation and model mismatch, and simply summing a single indicator is insufficient to accurately pinpoint the root cause of the fault that truly requires parameter updates. Therefore, in the technical solution of this invention, the performance degradation score and model mismatch score are further weighted and fused based on urgency to obtain a re-identification urgency score, thereby intelligently identifying the scenario where performance deterioration is caused by model inaccuracy—the scenario most in need of intervention. This allows for precise focusing on the fault scenario with the strongest causal correlation, assigning it the highest update priority, thus ensuring that computing resources are used to solve the fundamental problem, improving the intelligence level of decision-making and resource utilization efficiency.

[0038] More specifically, in one particular example of the invention, a urgency-weighted fusion and synergy modeling are performed, which abandons the traditional linear weighted summation method and instead uses a nonlinear fusion formula to calculate the overall re-identification urgency score. This mechanism not only includes a linear weighted term for the two scores but also introduces a nonlinear synergistic effect term. This synergistic effect term acts like a smart amplifier; when performance degradation and model mismatch occur simultaneously, its gain is far greater than the linear superposition of the two occurring independently. For example, when only external gusts cause excessive vibration, although the performance degradation score is high, the model mismatch score is zero or extremely low, and the synergistic term contributes almost nothing, keeping the total score at a low level and avoiding misjudgment. However, when foundation loosening leads to natural frequency drift and subsequently resonance, both scores increase simultaneously, and the synergistic term rapidly amplifies the total score through multiplication, quickly triggering the update mechanism.

[0039] The performance degradation score and model mismatch score are weighted and fused according to urgency to obtain the re-identification urgency score, including: weighting and fusing the performance degradation score and model mismatch score according to urgency using the following formula: ; in, To further identify the urgency score, and These are the performance degradation score and the model mismatch score, respectively. , , These are the performance degradation weight, model mismatch weight, and synergy term weight, used to adjust the importance of each factor. By adjusting these weights, the system's sensitivity preference to model bias or performance fluctuations can be flexibly customized according to the environmental characteristics of different wind fields. For example, in a specific scenario, the performance degradation weight, model mismatch weight, and synergy term weight can be set to 0.4, 0.4, and 0.2, respectively, to balance the impact of performance degradation, model mismatch, and their synergistic effect on the urgency of re-identification.

[0040] In step S330, the identification trigger flag is generated based on the re-identification urgency score and urgency threshold. It is understood that a simple numerical score is insufficient to directly translate into specific actions of the controller; a clear risk tolerance boundary must be set as the final criterion for initiating high-energy-consuming computing tasks. Otherwise, the system will frequently malfunction due to endless minor fluctuations. Therefore, in the technical solution of this invention, the identification trigger flag is further generated based on the re-identification urgency score and urgency threshold. This upgrades the system from a simple, rigid binary logic judgment to an intelligent decision-making system capable of quantifying the severity of problems and deeply understanding the causal relationships between various factors. This makes the adaptive triggering mechanism for vibration suppression more robust and resource-efficient, intelligently distinguishing between performance fluctuations caused by instantaneous external disturbances and systemic performance degradation caused by internal model mismatch, and prioritizing and accurately allocating limited PLC computing resources to the latter. This not only avoids unnecessary computational overhead and potential erroneous updates but also ensures that the control system can perform the most effective self-correction at critical moments, thereby maintaining the continuous, efficient, and stable operation of the active vibration suppression system throughout the entire wind turbine lifecycle.

[0041] More specifically, in one particular example of the invention, the final decision is made through dynamic threshold determination. That is, the calculated re-identification urgency score is used... With a preset urgency threshold The urgency threshold represents the boundary of the overall system risk that the operator can tolerate, serving as a final and reliable decision gate. For example, for offshore wind turbines located in typhoon-prone areas, operators may set a lower urgency threshold to remain highly sensitive to any minor model deviations; while for turbines in inland low-wind-speed areas, the urgency threshold can be appropriately increased to reduce maintenance frequency. Within each control cycle, the PLC performs a re-identification of the urgency score. Compared with the preset urgency threshold The comparison logic only outputs an identification trigger flag when the comprehensive risk assessment result confirms that the risk exceeds the acceptable range. If the result is true, a high-cost identification and update process is initiated. This ensures the effectiveness of the system's adaptive capabilities while avoiding overreaction to fluctuations in normal operating conditions. In a specific scenario, the urgency threshold can be set to 1.0. That is, when the re-identification urgency score is greater than or equal to 1.0, a model update is triggered, while when the re-identification urgency score is less than 1.0, the monitoring mode is maintained.

[0042] In step S400, in response to the identification trigger flag being true, a lightweight online identification is performed on the time series of the filtered acceleration signal and the time series of the control force to obtain a new discrete state space matrix. It is understood that the activation of the identification trigger flag reveals that the current physical dynamic characteristics of the tower have deviated significantly from the pre-stored mathematical model within the controller. Continuing to use the old model for control will fail to generate the correct reverse damping force and may even exacerbate vibration due to phase errors. Therefore, in the technical solution of this invention, in response to the identification trigger flag being true, a lightweight online identification is further performed on the time series of the filtered acceleration signal and the time series of the control force to obtain a new discrete state space matrix. This allows for the reconstruction of the mathematical model describing the current dynamic behavior of the tower within the limited computation cycle of the PLC using the latest input / output observation data. This enables the accurate capture of subtle changes in the natural frequency and damping ratio caused by ambient temperature, structural aging, or foundation loosening, providing a mathematical benchmark consistent with physical realities for subsequent calculations of the optimal control law.

[0043] More specifically, in a specific example of the present invention, in step S400, the PLC's main control program first detects that the identification trigger flag is true, then calls the system identification algorithm stored in a specific function block, and locks a long-time data buffer to obtain the acceleration response sequence synchronously recorded over a past period as the system output, and the control force sequence applied to the active mass damper as the system input. To accommodate the PLC's computational resource constraints, this process constructs a second-order difference equation, i.e., an ARX model structure, that can approximate the single-mode or multi-mode characteristics of the tower. The parameter vector to be identified is set to include regression coefficients reflecting the system's inertia, damping, and stiffness characteristics. Subsequently, the recursive least squares (RLS) algorithm is used to perform iterative calculations. This algorithm avoids the cumbersome calculation of high-dimensional matrix inversion, instead correcting the parameter estimates by using sampled data one by one. In each iteration, the current updated gain vector is first calculated using the covariance matrix of the previous time step, then the estimated parameter vector is corrected using the prediction error and the gain vector, and finally the covariance matrix is ​​updated for the calculation of the next time step, until the parameters converge. Once the stable difference equation coefficients are obtained, these coefficients are mapped and filled into the system matrix A and input matrix B of the discrete state-space expression according to the standard transformation method of modern control theory, thereby completing the transformation from input-output data to state-space model, and the flag bits are reset to end the identification process.

[0044] Specifically, in step S500, the control parameters are adaptively recalculated based on the new discrete state space matrix to obtain a new controller gain. It is understood that the effectiveness of the control algorithm is highly dependent on the accuracy of the controlled object's dynamic model. A fixed set of state feedback gains can only achieve optimal pole configuration under specific stiffness and damping combinations. Once the identification step confirms that the tower's state space matrix has shifted due to the evolution of structural physical characteristics, the original control parameters will no longer match the current system poles, causing the calculated control force to deviate from the optimal solution in phase and amplitude, and potentially even triggering vibration divergence due to phase lag. Therefore, in the technical solution of this invention, the control parameters are further adaptively recalculated based on the new discrete state space matrix to obtain a new controller gain, thereby mapping the updated physical model parameters to the optimal control law parameters in real time, reconstructing the optimal coupling relationship between the control signal and the structural response. This ensures that the suppression torque output by the active mass damper always acts along the gradient direction that causes the system's vibration energy to decay fastest under the current wind and structural conditions, restoring and maintaining the theoretically optimal vibration reduction performance of the system design.

[0045] More specifically, in a specific example of the present invention, step S500 first establishes a control objective based on a linear quadratic regulator (LQR), setting a weighted cost function aimed at minimizing the mean square value of the tower top displacement and the energy consumption of the control input. Given the computational bottleneck of directly solving high-dimensional algebraic Riccati equations within an industrial controller with a millisecond-level scan cycle, this embodiment employs an online interpolation calculation strategy based on feature indexing. Key parameters characterizing the core characteristics of the system, such as eigenvalues ​​corresponding to stiffness changes or squared natural frequencies, are extracted from the newly identified discrete state-space matrix and used as retrieval keys. These retrieval keys are then used to search a pre-set multidimensional gain scheduling table in the PLC's non-volatile memory area. This multidimensional gain scheduling table contains an offline pre-calculated set of optimal gains covering the entire lifecycle stiffness variation range. After locating the numerical range into which the new parameter falls, polynomial or linear interpolation operations are performed, and a new controller gain matrix matching the current model is accurately calculated through mathematical synthesis. Finally, the calculation results are written to the controller's execution memory area, and the model baseline parameters used for subsequent mismatch judgment are updated synchronously to complete the closed-loop adaptive adjustment.

[0046] Specifically, in step S600, based on the new controller gain, vibration suppression control is applied to the filtered acceleration signal to obtain the control output command for the current cycle. The tower is then controlled according to the control output command for the current cycle. It is understood that the optimal controller gain calculated by the adaptive algorithm only represents a logical-level strategy update. If it is not applied in real-time to the current physical feedback signal, a reverse damping force that conforms to the current structural characteristics cannot be generated, resulting in the mathematical model optimization failing to translate into physical vibration reduction effectiveness. Therefore, in the technical solution of this invention, vibration suppression control is further applied to the filtered acceleration signal based on the new controller gain to obtain the control output command for the current cycle. This combines the latest control strategy with the real-time structural motion state to calculate the precise drive command required by the actuator within the current control cycle. In this way, the tower can be controlled using the control output command for the current cycle, driving the active mass damper to generate an inertial force with matched amplitude and strictly opposite phase, thereby maximizing the dissipation of the tower's vibration energy at the physical level and ensuring the immediate effectiveness and stable operation of the closed-loop control system after the model update.

[0047] Figure 5 This is a flowchart illustrating the active vibration suppression method for towers based on PLC control according to an embodiment of the present invention, which uses a new controller gain to perform vibration suppression control on a filtered acceleration signal to obtain the control output command for the current cycle. Figure 5 As shown, step S600 includes: S610, performing state estimation on the filtered acceleration signal and the control signal of the previous cycle to obtain the current state vector of the system; S620, generating the control output command for the current cycle based on the current state vector of the system and the new controller gain.

[0048] In step S610, state estimation is performed on the filtered acceleration signal and the control signal from the previous cycle to obtain the current state vector of the system. It is understood that in the engineering practice of wind turbine tower vibration control, due to cost and installation / maintenance limitations, acceleration sensors are typically only installed at the tower top or nacelle to sense structural response. Key state variables such as displacement and velocity, which directly reflect structural safety, are difficult to obtain directly using low-cost hardware. Furthermore, simply deriving displacement by performing a second numerical integration on the acceleration signal will result in severe data distortion due to the cumulative effect of low-frequency noise and zero-point drift from the sensor, causing the state feedback-based control strategy to fail. Therefore, in the technical solution of this invention, state estimation is further performed on the filtered acceleration signal and the control signal from the previous cycle to obtain the current state vector of the system. This utilizes the observer technique in modern control theory to fuse known control input information with measured acceleration output information, mathematically reconstructing in real time the complete dynamic state of the system that is either not directly measurable or has high measurement noise. This effectively overcomes the limitations of sensor configuration and eliminates drift errors in numerical integration, providing the controller with high-precision structural displacement and velocity feedback, ensuring that the active suppression system can apply precise reverse damping control based on the tower's actual motion posture.

[0049] More specifically, in a concrete example of the present invention, step S610 first instantiates a full-dimensional state observer or steady-state Kalman filter model in the arithmetic unit of the programmable logic controller. This model is loaded with the latest discrete system state matrix and output observation matrix. During the execution of each control cycle, based on the system state estimate from the previous moment and the actual control force applied to the active mass damper in the previous cycle, a one-step prediction is performed using the system dynamics equations to calculate the prior estimate of the tower state at the current moment. Subsequently, the theoretical acceleration output corresponding to the prior estimate is calculated, and it is differentially analyzed with the acceleration signal actually acquired at the current moment and processed by bandpass filtering to obtain the observation residual. Next, the observation residual is multiplied by a pre-designed observer gain matrix to perform closed-loop correction on the prior estimate, thereby obtaining a posterior state vector that converges to the true physical state. This posterior state vector contains the decoupled real-time displacement, velocity, and acceleration components of the tower top, which are directly used as input variables for the full-state feedback control law and for the synthesis of subsequent control commands.

[0050] In step S620, a control output command for the current cycle is generated based on the system's current state vector and the new controller gain. It is understood that the core of state feedback control lies in mapping the system's internal dynamic variables to executable commands from external actuators in real time. Simple state vector estimates and controller gain matrices are merely intermediate mathematical variables; without algebraic synthesis conforming to the control law, they cannot directly drive the physical device to generate the required reverse force. Therefore, in the technical solution of this invention, a control output command for the current cycle is further generated based on the system's current state vector and the new controller gain. This allows for a mathematical synthesis of the updated control strategy and the current structural motion state, based on the principle of negative feedback control. This enables the calculation of the precise torque or thrust value required by the active mass damper to achieve optimal vibration reduction at the current instant, ensuring that the control action strictly adheres to the preset dynamic suppression target.

[0051] More specifically, in a specific example of the present invention, step S620 performs matrix algebra operations in the arithmetic logic unit of the controller, strictly following the standard feedback form of the state-space method. This operation projects a multi-dimensional state vector containing information such as tower displacement and velocity onto the control input space, and generates the final control command through linear weighted combination. Specifically, step S620 includes: generating the control output command for the current cycle using the following formula: ; in, For the control output command of the current cycle, The new controller gain is the new controller gain matrix obtained through adaptive recalculation. This is the current state vector of the system output by the state observer. This represents the current period. For example, in a scenario where a sudden gust of wind causes a significant downwind displacement of the tower top, the state vector... The significant displacement and velocity components were captured in real time, combined with gains optimized for the current high wind speed conditions. Calculate a reverse control command with a large amplitude. ,use The driving mass block accelerates rapidly in the upwind direction to generate a huge reaction inertial force. In another scenario, if the tower's natural frequency decreases due to icing in winter, the previous steps have already... The gain is updated to match the low-frequency characteristics. At this time, the generated control output command will be strictly locked to apply damping at the new resonant frequency, avoiding the risk of structural instability caused by control force phase error due to frequency mismatch, and ensuring that every action of the actuator can accurately dissipate the vibration energy of the structure.

[0052] In summary, the active tower vibration suppression method based on PLC control according to the embodiments of the present invention is explained. It acquires tower acceleration signals in real time and extracts window vibration energy and dominant frequency characteristics to construct a composite evaluation system that includes performance degradation and model mismatch dimensions. By using index normalization and nonlinear weighted fusion technology, the urgency of re-identification is intelligently quantified. Only when the system control performance significantly declines and it is confirmed that it is due to model parameter drift is the lightweight online identification and control parameter recalculation process triggered. Otherwise, a low-power monitoring mode is maintained. This mechanism effectively solves the problem that traditional fixed parameter controllers cannot adapt to the performance degradation caused by the time-varying characteristics of tower structures. At the same time, it avoids the excessive occupation of PLC computing resources by continuous complex algorithms, realizes efficient and robust control of time-varying structural systems on a limited computing power platform, and ensures the vibration reduction effect throughout the entire life cycle of the wind turbine.

[0053] The present invention also provides a PLC-controlled active tower vibration suppression system.

[0054] Figure 6 This is a block diagram of a PLC-controlled active vibration suppression system for towers according to an embodiment of the present invention. Figure 6 As shown, the active tower vibration suppression system 100 based on PLC control according to an embodiment of the present invention includes: a raw signal acquisition module 110, used to acquire the raw acceleration signal of the tower at the current moment; a filtering, data buffering and feature extraction module 120, used to filter, buffer and extract features from the raw acceleration signal at the current moment to obtain a filtered acceleration signal, window vibration energy and window dominant frequency; a composite event decision module 130, used to make composite event decisions based on window vibration energy and window dominant frequency to obtain an identification trigger flag; a lightweight online identification module 140, used to perform lightweight online identification on the time series of the filtered acceleration signal and the time series of the control force in response to the identification trigger flag being true to obtain a new discrete state space matrix; a control parameter adaptive recalculation module 150, used to perform adaptive recalculation of control parameters based on the new discrete state space matrix to obtain a new controller gain; and a vibration suppression control module 160, used to perform vibration suppression control on the filtered acceleration signal based on the new controller gain to obtain a control output command for the current period, and control the tower according to the control output command for the current period.

[0055] For example, the composite event decision module 130 includes: The index normalization and severity quantization unit is used to normalize and quantify the window vibration energy and window dominant frequency to obtain the performance degradation score and model mismatch score. The urgency-weighted fusion unit is used to perform urgency-weighted fusion of performance degradation score and model mismatch score to obtain re-identification urgency score; The identification trigger flag generation unit is used to generate the identification trigger flag based on the re-identification urgency score and urgency threshold.

[0056] For example, the vibration suppression control module 160 includes: The state estimation unit is used to perform state estimation on the filtered acceleration signal and the control signal of the previous cycle to obtain the current state vector of the system. The current cycle control output generation unit is used to generate control output commands for the current cycle based on the current state vector of the system and the new controller gain.

[0057] The specific implementation method of the PLC-controlled active tower vibration suppression system 100 provided in this embodiment of the invention can be found in the description of the PLC-controlled active tower vibration suppression method in this embodiment of the invention, and will not be repeated here.

[0058] The PLC-controlled active tower vibration suppression system 100 according to embodiments of the present invention can be implemented in various types of computing devices or control units. For example, it can be a programmable logic controller deployed in the main control cabinet of a wind turbine generator or the control cabinet at the bottom of the tower, a digital signal processor assisting the PLC in high-precision vibration signal acquisition and preprocessing, or a high-performance industrial computer integrated into the nacelle control unit and the active mass damper drive system. In one possible implementation, the PLC-controlled active tower vibration suppression system 100 according to embodiments of the present invention can be integrated into the computing device as a software module and / or a hardware module. For example, the PLC-controlled active tower vibration suppression system 100 can be an intelligent control function block in the computing device or PLC operating environment. This software module is configured to perform filtering, buffering, and feature extraction of the original acceleration signal; composite event triggering decision based on window energy and dominant frequency; lightweight online identification of the system model in response to trigger flags; and adaptive recalculation of controller gain and generation of state feedback control law based on the updated model. Alternatively, it can be a dedicated wind turbine tower vibration reduction adaptive control algorithm program developed for the computing device. Of course, the PLC-based tower vibration active suppression system 100 can also be one of the many hardware modules of the computing device or control unit, or it can be embedded in a field-programmable gate array circuit to accelerate the spectrum analysis and matrix operation process of the fast Fourier transform in parallel, or it can be a vibration monitoring and active control signal processing integrated circuit for a specific application.

[0059] Those skilled in the art will understand that the above embodiments are specific implementations of the present invention, and in practical applications, various changes can be made in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for actively suppressing tower vibration based on PLC control, characterized in that, include: Obtain the raw acceleration signal of the tower at the current moment; The raw acceleration signal at the current moment is filtered, buffered, and its features are extracted to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. A composite event decision is made based on window vibration energy and window dominant frequency to obtain identification trigger flags; In response to the identification trigger flag being true, lightweight online identification is performed on the time series of the filtered acceleration signal and the time series of the control force to obtain a new discrete state space matrix. The control parameters are adaptively recalculated based on the new discrete state space matrix to obtain the new controller gain. Based on the new controller gain, vibration suppression control is applied to the filtered acceleration signal to obtain the control output command for the current cycle, and the tower is controlled according to the control output command for the current cycle.

2. The active tower vibration suppression method based on PLC control according to claim 1, characterized in that, The raw acceleration signal at the current moment is filtered, buffered, and its features are extracted to obtain the filtered acceleration signal, windowed vibration energy, and windowed dominant frequency, including: The original acceleration signal at the current moment is bandpass filtered to obtain the filtered acceleration signal; The filtered acceleration signal is stored in a first-in-first-out circular buffer. The root mean square of the time series of the filtered acceleration signal in the circular buffer is calculated as the window vibration energy; Fast Fourier Transform is performed on the time series of the filtered acceleration signal in the circular buffer to obtain the window dominant frequency.

3. The active tower vibration suppression method based on PLC control according to claim 1, characterized in that, Composite event decision-making based on window vibration energy and window dominant frequency to obtain identification trigger flags, including: The window vibration energy and window dominant frequency are normalized and their severity is quantified to obtain the performance degradation score and model mismatch score. The performance degradation score and the model mismatch score are weighted and fused by urgency to obtain the re-identification urgency score; The identification trigger flag is generated based on the re-identification urgency score and urgency threshold.

4. The active tower vibration suppression method based on PLC control according to claim 3, characterized in that, The window vibration energy and window dominant frequency are normalized and their severity quantified to obtain performance degradation scores and model mismatch scores, including: The window vibration energy and window dominant frequency are normalized and their severity quantified using the following formula: ； ； in, Score for performance degradation. For window vibration energy, The preset performance threshold, It is a function with maximum value. The model mismatch score, For window-dominant frequency, For the current model frequency, This represents the tolerance for frequency mismatch.

5. The active tower vibration suppression method based on PLC control according to claim 3, characterized in that, The performance degradation score and model mismatch score are weighted and fused based on urgency to obtain the re-identification urgency score, including: The performance degradation score and model mismatch score are weighted and fused based on urgency using the following formula: ； in, To further identify the urgency score, and These are the performance degradation score and the model mismatch score, respectively. , , These are the performance degradation weight, model mismatch weight, and synergy term weight, respectively.

6. The active tower vibration suppression method based on PLC control according to claim 1, characterized in that, Based on the new controller gain, vibration suppression control is applied to the filtered acceleration signal to obtain the control output command for the current cycle, including: The current state vector of the system is obtained by performing state estimation on the filtered acceleration signal and the control signal of the previous cycle. Based on the system's current state vector and the new controller gain, generate the control output command for the current cycle.

7. The active tower vibration suppression method based on PLC control according to claim 6, characterized in that, Based on the system's current state vector and the new controller gain, generate the control output command for the current cycle, including generating the control output command for the current cycle using the following formula: ； in, For the control output command of the current cycle, For the new controller gain, Let this be the current state vector of the system. Indicates the current period.

8. A PLC-controlled active vibration suppression system for towers, characterized in that, include: The raw signal acquisition module is used to acquire the raw acceleration signal of the tower at the current moment; The filtering, data buffering, and feature extraction module is used to filter, buffer, and extract features from the raw acceleration signal at the current moment to obtain the filtered acceleration signal, window vibration energy, and window dominant frequency. The composite event decision module is used to make composite event decisions based on window vibration energy and window dominant frequency to obtain identification trigger flags. The lightweight online identification module is used to perform lightweight online identification on the time series of the filtered acceleration signal and the time series of the control force in response to the identification trigger flag being true, so as to obtain a new discrete state space matrix. The adaptive recalculation module for control parameters is used to adaptively recalculate control parameters based on the new discrete state space matrix to obtain a new controller gain. The vibration suppression control module is used to perform vibration suppression control on the filtered acceleration signal based on the new controller gain to obtain the control output command for the current cycle, and control the tower according to the control output command for the current cycle.

9. The active tower vibration suppression system based on PLC control according to claim 8, characterized in that, The complex event decision-making module includes: The index normalization and severity quantization unit is used to normalize and quantify the window vibration energy and window dominant frequency to obtain the performance degradation score and model mismatch score. The urgency-weighted fusion unit is used to perform urgency-weighted fusion of performance degradation score and model mismatch score to obtain re-identification urgency score; The identification trigger flag generation unit is used to generate the identification trigger flag based on the re-identification urgency score and urgency threshold.

10. The active tower vibration suppression system based on PLC control according to claim 8, characterized in that, The vibration suppression control module includes: The state estimation unit is used to perform state estimation on the filtered acceleration signal and the control signal of the previous cycle to obtain the current state vector of the system. The current cycle control output generation unit is used to generate control output commands for the current cycle based on the current state vector of the system and the new controller gain.

Images