Optimization method for intelligent yaw system of wind generating set

Abstract

The invention relates to the technical field of new energy, and discloses an optimization method for an intelligent yaw system of a wind generating set, and the method comprises the steps: building the intelligent yaw system, enabling the system to be internally provided with seven modules, and enabling a multi-sensor data collection module to collect wind direction, wind speed and yaw angle data in real time, the data fusion module performs Kalman filtering fusion processing to obtain a high-confidence state estimation value, the intelligent decision output module generates an optimal yaw control instruction by adopting a multi-mode control strategy and a parameter self-adaptive mechanism based on fused data, an execution mechanism is driven to realize accurate wind alignment, and the system integrated hardware-in-the-loop test module is used for realizing accurate wind alignment. Omnibearing verification can be carried out before deployment, after actual operation, operation indexes are continuously monitored through the performance evaluation module, control parameters and fusion algorithm parameters are adjusted online through the optimization module on the basis of methods such as machine learning, closed-loop optimization is formed, finally, yaw precision, response speed and system stability are remarkably improved, and maintenance cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of new energy technology, specifically to an optimization method for an intelligent yaw system of a wind turbine generator set. Background Technology

[0002] In wind power generation, the dynamic changes in wind direction exhibit significant randomness and complexity, posing a severe challenge to the efficient wind capture of wind turbines. As the core subsystem ensuring that the wind turbine always aligns with the prevailing wind direction, the existing intelligent yaw system, while having achieved basic functions, still suffers from three technical bottlenecks: First, the yaw control accuracy is limited by the performance boundaries of multi-source sensor fusion algorithms, making it difficult to maintain optimal wind orientation under complex conditions such as turbulence and gusts; second, the system response lags behind sudden changes in wind direction, leading to a decrease in wind energy utilization; and third, key components (such as sensors and controllers) are exposed to harsh environments such as vibration and salt spray for extended periods, resulting in frequent failures and high maintenance costs. These shortcomings directly affect the power generation efficiency and economic viability of wind turbines, necessitating systemic optimization through technological innovation. Summary of the Invention

[0003] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an optimization method for an intelligent yaw system for wind turbine generators. This method has the advantages of high-precision multi-source data fusion, rapid dynamic response, strong environmental adaptability, and self-optimization capabilities. It solves the problems of insufficient accuracy, slow response, poor equipment reliability, and high operation and maintenance costs of traditional yaw systems under complex wind conditions.

[0004] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: an optimization method for an intelligent yaw system of a wind turbine generator set, comprising the following steps: Step 1: Establish an intelligent yaw system. The system has built-in multi-sensor data acquisition module, data fusion module, intelligent decision output module, hardware control module, hardware-in-the-loop testing module, performance evaluation module, and optimization module. Step 2: The multi-sensor data acquisition module is responsible for controlling the wind direction sensor, wind speed sensor, and position / yaw angle sensor to perform synchronous or asynchronous data acquisition, and completing the preliminary verification and caching of the data; Step 3: The data fusion module performs time-series alignment, outlier removal, and noise filtering on the raw data. It then uses the Kalman filter algorithm to perform optimal fusion of multi-source sensor information to obtain high-confidence estimates of wind direction, wind speed, and yaw state. Step 4: The intelligent decision output module calculates the target yaw angle and the required yaw rate based on the fused data, and outputs the optimal yaw control command according to the multi-mode control strategy and parameter adaptive mechanism to drive the actuator. Step 5: The hardware control module receives the control command, and after signal conditioning and power amplification, drives the yaw motor to perform the corresponding action to achieve precise wind alignment in the cabin. Step Six: Hardware-in-the-Loop Testing Module. In a laboratory environment, the dynamic characteristics of the wind turbine and the wind field environment are simulated through a simulation model. The control cabinet is subjected to comprehensive functional verification, extreme testing and fault injection testing to identify and fix potential problems in advance. Step 7: After the system is actually deployed, the performance evaluation module continuously collects operational data on yaw error, response delay, and power generation, and evaluates the system control effect. Step 8: The optimization module, based on performance evaluation results and operational data, fine-tunes control parameters and fusion algorithm parameters online through parameter identification, machine learning, or optimization algorithms to achieve continuous improvement in system performance.

[0005] Preferably, the multi-sensor data acquisition module includes a wind direction data acquisition unit, a wind speed data acquisition unit, and a location data acquisition unit; The wind direction data acquisition unit collects real-time wind direction angles by installing high-precision ultrasonic wind direction sensors on the top and tail sections of the nacelle. Its sampling frequency is ≥10Hz; The wind speed data acquisition unit obtains wind speed by fusing measurements from a cup-type anemometer and a lidar anemometer. and turbulence intensity ; The position data acquisition unit measures the yaw angle α and yaw rate using an absolute encoder. .

[0006] Preferably, the data fusion module includes a data preprocessing unit, a Kalman filter fusion unit, and an anomaly diagnosis and processing unit; The data preprocessing unit performs preliminary processing on the raw data, including unit conversion, outlier removal, and low-pass filtering for preliminary noise reduction. The Kalman filter fusion unit executes the Kalman filter algorithm to establish the system state equation and observation equation, and performs optimal fusion of multi-source data to obtain the corrected wind direction. Wind speed and yaw angle The estimated value is specifically achieved by defining the system state vector and the observation vector, where: State vector:

[0007] In the formula, This represents the true wind direction angle at time k. This represents the actual wind speed at time k. This represents the actual yaw angle of the wind turbine nacelle at time k. This represents the actual yaw rate of the cabin at time k; Observation vector:

[0008] In the formula, This indicates the wind direction sensor at time k. This represents the wind speed sensor reading at time k. This represents the yaw angle measurement value of the position sensor at time k; The optimal estimate is calculated iteratively through prediction and update steps; The anomaly diagnosis and processing unit monitors the data stream in real time, identifies anomalies through threshold judgment and statistical testing, and triggers smoothing processing, data removal, or redundant sensor switching mechanisms.

[0009] Preferably, the intelligent decision output module includes a desired state calculation unit, a multi-mode control strategy unit, and a parameter adaptive unit; The desired state calculation unit calculates the target yaw angle and yaw rate based on the fused data, and outputs the optimal control command, specifically including: (1) Target generation: Based on the real-time high-precision wind direction angle provided by the data fusion module, calculate the target yaw angle required for the wind turbine plane to be perpendicular to the direction of the wind; (2) Decision optimization: Taking into account the current mechanical load of the yaw system, the response characteristics of the actuator and the trend of wind direction change, the optimal yaw rate setting value that balances response speed and stability is calculated. (3) Command preprocessing: Convert the above target angle and angular velocity into the desired trajectory or set point that the underlying control law can track, and provide input for the subsequent multi-mode control strategy unit; The multi-mode control strategy unit automatically selects or weights and fuses predictive control and sliding mode control algorithms based on real-time monitored wind conditions and unit status, specifically: (1) Predictive control: Based on historical wind direction data, the future wind direction trend is predicted by using an LSTM time series model to generate feedforward control variables; (2) Sliding mode control: Rapid convergence is achieved by forcing the system state to move along a preset sliding surface. Deviation definition:

[0010] In the formula, Indicates the desired yaw angle. Indicates the actual yaw angle. Indicates the deviation in yaw angle; Sliding surface design:

[0011] In the formula, and All of these represent design parameters used to adjust the dynamic characteristics of deviations; Through the control law:

[0012] In the formula, Represents a symbolic function. This represents the sliding mode gain, which, through a discontinuous control law, forces the system state to move along the sliding surface, achieving rapid convergence. The parameter adaptive unit adjusts according to real-time wind speed. With turbulence intensity The system uses wind condition information to dynamically adjust the controller parameters.

[0013] Preferably, the hardware-in-the-loop test module includes a high-fidelity simulation unit, a test execution unit, a fault injection unit, and a result analysis unit; The high-fidelity simulation unit constructs a wind turbine dynamics model and a wind field environment model based on data collected by multiple sensors, and simulates the wind turbine response under different operating conditions. The test execution unit connects the simulation model and the actual control cabinet through a hardware interface to achieve real-time data interaction and simulate the transmission of control commands and status feedback during actual operation. The fault injection unit can simulate sensor faults, actuator faults, and communication faults to test the system's fault tolerance. The result analysis unit records the yaw accuracy, response time, and control stability indicators during the test process, generates a test report, and identifies potential problems such as control logic vulnerabilities and parameter configuration defects.

[0014] Preferably, the performance evaluation module continuously collects yaw error data after the actual wind turbine is deployed. Response delay and power generation The data.

[0015] Preferably, the performance evaluation module uses quantitative indicators, combined with preset thresholds and the analytic hierarchy process (AHP), to rate the system's control effectiveness, economy, and reliability as excellent / good / medium / poor, and generates a performance evaluation report to provide a basis for the optimization module to make decisions.

[0016] Preferably, the optimization module identifies changes in the dynamic characteristics of the system based on the performance evaluation report, updates the parameters of the wind turbine dynamic model, trains a reinforcement learning model using historical operating data, finds the optimal combination of control strategies under different wind conditions through trial and error learning, and fine-tunes key parameters online based on the optimization results to achieve dynamic optimization of system performance.

[0017] Preferably, the hardware control module includes a signal conditioning unit, a power driving unit, and a status monitoring unit; The signal conditioning unit filters, amplifies, and isolates the control commands sent by the intelligent decision output module, converting them into standard signals that meet the requirements of the actuator. The power drive unit controls the start, stop, speed and direction of the yaw motor according to the conditioned signal, and drives the nacelle to rotate through the gearbox and yaw bearing to achieve precise wind alignment; The status monitoring unit monitors the motor current, temperature and yaw angle feedback signals online to determine whether the actuator is operating normally. When an abnormal state of overload or over-temperature is detected, the protection mechanism is automatically triggered and fed back to the intelligent decision module.

[0018] Preferably, the system supports redundant sensor deployment and a fault self-diagnosis mechanism. When abnormal or failed data is detected from a sensor, it automatically switches to a backup sensor or an estimated value to ensure the continuous and stable operation of the system.

[0019] Compared with the prior art, the present invention provides an optimization method for an intelligent yaw system of a wind turbine generator set, which has the following beneficial effects: 1. This invention improves the yaw accuracy of the intelligent yaw system by 20% to 30% by employing advanced sensor fusion technology and optimized control algorithms, ensuring that the wind turbine is always in the optimal wind direction. Through comprehensive optimization of the system structure and control strategy, the response time to wind direction changes is shortened by 15% to 20%, thereby effectively reducing power generation losses caused by sudden changes in wind direction. At the same time, the improved system design adopts highly reliable components and redundant architecture, which greatly enhances its stability and anti-interference ability, while reducing the failure rate and maintenance costs, ultimately achieving efficient, safe and economical operation throughout the entire life cycle. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 An optimization method for an intelligent yaw system of a wind turbine generator set includes the following steps: Step 1: Establish an intelligent yaw system. The system has built-in multi-sensor data acquisition module, data fusion module, intelligent decision output module, hardware control module, hardware-in-the-loop testing module, performance evaluation module, and optimization module. Step 2: The multi-sensor data acquisition module is responsible for controlling the wind direction sensor, wind speed sensor and position / yaw angle sensor to perform synchronous or asynchronous data acquisition, setting a unified sampling frequency, data format and communication protocol, and completing the preliminary verification and caching of data. Step 3: The data fusion module performs preprocessing on the raw data, including time alignment, outlier removal, and noise filtering. It then uses the Kalman filter algorithm to perform optimal fusion of multi-source sensor information to obtain high-confidence estimates of wind direction, wind speed, and yaw state. Step 4: The intelligent decision output module calculates the target yaw angle and the required yaw rate based on the fused high-quality data, and outputs the optimal yaw control command according to the multi-mode control strategy and parameter adaptive mechanism to drive the actuator. Step 5: The hardware control module receives the control command, and after signal conditioning and power amplification, drives the yaw motor to perform the corresponding action to achieve precise wind alignment in the cabin. Step Six: Hardware-in-the-Loop Testing Module. In a laboratory environment, a high-fidelity simulation model is used to simulate the dynamic characteristics of the wind turbine and the wind field environment. The control cabinet (controller) is subjected to comprehensive functional verification, extreme testing and fault injection testing to identify and fix potential problems in advance. Step 7: After the system is actually deployed, the performance evaluation module continuously collects operational data such as yaw error, response delay, and power generation, and performs quantitative evaluation of the system control effect. Step 8: Based on the performance evaluation results and operational data, the optimization module fine-tunes the control parameters and fusion algorithm parameters online through parameter identification, machine learning, or optimization algorithms to achieve continuous improvement in system performance. The execution sequence of each module is as follows: the multi-sensor data acquisition module transmits the acquired and preliminarily verified raw data to the data fusion module; after completing data preprocessing, Kalman filter fusion, and anomaly diagnosis, the data fusion module outputs high-confidence fused data to the intelligent decision output module; the intelligent decision output module calculates and generates the optimal control command based on the fused data and sends it to the hardware control module for execution; the hardware-in-the-loop testing module connects to the control cabinet to conduct closed-loop testing before system deployment; after system deployment and operation, the performance evaluation module collects operating data and feeds it back to the optimization module, which then fine-tunes the algorithm parameters in the data fusion and intelligent decision modules, forming a closed-loop continuous optimization process.

[0023] The multi-sensor data acquisition module includes a wind direction data acquisition unit, a wind speed data acquisition unit, and a location data acquisition unit; The wind direction data acquisition unit collects real-time wind direction angles by installing high-precision ultrasonic wind direction sensors on the top and tail sections of the nacelle. Its sampling frequency is ≥10Hz; The wind speed data acquisition unit obtains wind speed by fusing measurements from a cup-type anemometer and a lidar anemometer. and turbulence intensity ; The position data acquisition unit measures the yaw angle α and yaw rate using an absolute encoder. ; Set the sampling frequency (e.g., 10Hz), data format, and communication protocol (e.g., CAN bus, Modbus) for all the above acquisition units to ensure the real-time performance and consistency of the data.

[0024] The advantages are: by using the multi-source sensor acquisition of the above-mentioned dual ultrasonic wind direction at the bow and stern of the cabin, dual wind speed by cup / liDAR, and absolute encoder position, as well as unified frequency sampling, CAN / Modbus protocol and timestamp alignment mechanism, the real-time performance, consistency and completeness of wind direction, wind speed and yaw angle data are greatly improved. This provides high signal-to-noise ratio and time-difference-free basic data for subsequent fusion, and achieves the beneficial effect of reducing the probability of yaw error caused by sensor drift and data packet loss.

[0025] The data fusion module includes a data preprocessing unit, a Kalman filter fusion unit, and an anomaly diagnosis and processing unit; The data preprocessing unit performs preliminary processing on the raw data, including unit conversion (such as converting voltage signals to wind speed values), outlier removal, and low-pass filtering for preliminary noise reduction. The Kalman filter fusion unit executes the Kalman filter algorithm to establish the system state equation and observation equation, and performs optimal fusion of multi-source data to obtain the corrected wind direction. Wind speed and yaw angle The estimated value is specifically achieved by defining the system state vector and the observation vector, where: State vector:

[0026] In the formula, The actual wind direction angle at time k (unit: °); Represents the actual wind speed at time k (unit: m / s); The value represents the actual yaw angle of the wind turbine nacelle at time k (unit: °). The value represents the actual yaw rate of the cabin at time k (unit: ° / s), used to describe the dynamic characteristics of the yaw motion; Observation vector:

[0027] In the formula, The value measured by a wind direction sensor (such as an ultrasonic wind direction sensor deployed on the top or tail of the cabin) at time k (unit: °). The value measured by the wind speed sensor (such as a cup-type anemometer or lidar anemometer) at time k (unit: m / s). The yaw angle measurement (in °) of the position sensor (such as an absolute encoder) at time k is represented. The optimal estimate is calculated iteratively through prediction and update steps; The anomaly diagnosis and processing unit monitors the data stream in real time and makes judgments based on thresholds (such as wind speed mutation rate). / Anomalies can be identified using statistical tests (such as the chi-square test), triggering smoothing, data removal, or redundant sensor switching mechanisms.

[0028] The advantages are: through the unit conversion, outlier removal and low-pass filtering of the above data preprocessing unit to purify the original data, the Kalman filter fusion unit establishes the state equation and observation equation to achieve optimal fusion of multi-source data, and the anomaly diagnosis and processing unit identifies and processes abnormal data through threshold judgment and statistical test, thereby achieving denoising, completion and accurate fusion of multi-sensor data, and outputting high-confidence wind direction, wind speed and yaw state estimates, providing accurate state basis for the intelligent decision output module, and avoiding the effect of single sensor error or abnormal data affecting control accuracy.

[0029] The intelligent decision output module calculates the target yaw angle based on the fused high-quality data. With yaw rate Output optimal control commands, specifically including: The multi-mode control strategy unit automatically selects or weights and fuses predictive control and sliding mode control algorithms based on real-time monitored wind conditions (stable / turbulent) and unit status, specifically: (1) Predictive control: Based on historical wind direction data, the future wind direction trend is predicted by LSTM time series model, and feedforward control is generated. By adjusting the yaw in advance when the wind direction change is slow, the lag of wind direction change → detection of deviation → re-control can be reduced, and the response speed can be improved. (2) Sliding mode control: Sliding mode control is a control method that is highly robust to disturbances (such as wind turbulence and mechanical errors). It achieves rapid convergence by forcing the system state to move along a preset sliding surface. Deviation definition:

[0030] In the formula, This represents the desired yaw angle (which can be determined by the feedforward target of predictive control or the optimal wind angle). This represents the actual yaw angle (obtained by fusing from position sensors or Kalman filters). Indicates the deviation in yaw angle; Sliding surface design:

[0031] In the formula, (positive number) and (usually a positive integer, such as) =1) both represent design parameters used to adjust the dynamic characteristics of the deviation (such as introducing the derivative or nonlinear term of the deviation) so that the system state (yaw deviation) converges to the sliding surface more quickly; Through the control law:

[0032] In the formula, Represents the symbolic function ( Output 1 when >0. <0 outputs -1), This represents the sliding mode gain (the strength of the control variable). Through this discontinuous control law, the system state is forced to move along the sliding surface, eventually converging rapidly to the desired yaw angle (making the system...). →0), achieving rapid convergence; The parameter adaptive unit adjusts according to real-time wind speed. With turbulence intensity The system uses wind condition information to dynamically adjust controller parameters (such as sliding mode gain K) to ensure optimal control performance under different operating conditions.

[0033] The advantages are: through the above-mentioned target instruction generation - multi-mode MPC / SMC switching - parameter adaptive biological three-layer architecture, and by utilizing the complementary advantages of LSTM look-ahead prediction and sliding mode robust control, and with the assistance of online gain self-tuning, the yaw step response overshoot is <1° and the settling time is <4s. Even under the extreme condition of turbulence intensity I=0.18, the steady-state error can still be controlled within ±0.5°, taking into account speed, stability and disturbance rejection, and ultimately reducing the power generation loss caused by tracking lag.

[0034] The hardware-in-the-loop test module includes a high-fidelity simulation unit, a test execution unit, a fault injection unit, and a result analysis unit; The high-fidelity simulation unit constructs a wind turbine dynamics model (including physical characteristics of the nacelle, yaw bearing, transmission mechanism, etc.) and a wind field environment model (including wind conditions such as steady-state wind, gusts, and turbulence) based on data collected by multiple sensors, simulating the wind turbine response under different operating conditions; The test execution unit connects the simulation model to the actual control cabinet (controller) through hardware interfaces (such as I / O modules and communication buses) to achieve real-time data interaction and simulate the transmission of control commands and status feedback during actual operation. The fault injection unit can simulate sensor faults (such as data drift and signal loss), actuator faults (such as motor jamming and abnormal noise from the gearbox), and communication faults (such as data delay and packet loss) to test the fault tolerance of the system. The results analysis unit records yaw accuracy, response time, and control stability indicators during the test process, generates a test report, identifies potential problems such as control logic vulnerabilities and parameter configuration defects, and provides a basis for system optimization.

[0035] The advantages are: through the complete test chain of high-fidelity wind turbine-wind field-actuator modeling + real-time interface closed loop + fault injection - automatic result analysis, extreme / fault conditions such as sensor disconnection and motor phase loss can be predicted in advance in the laboratory, thus moving the risk of on-site commissioning forward, shortening the installation and commissioning cycle, and ensuring that the controller has been optimized to the best parameters before commissioning, thereby reducing the on-site failure rate and the cost of repeated tower climbing maintenance.

[0036] The performance evaluation module continuously collects yaw error data after the actual wind turbine is deployed. Response delay and power generation The data.

[0037] The performance evaluation module uses quantitative indicators, preset thresholds, and the analytic hierarchy process (AHP) to rate the system's control effectiveness, economy, and reliability as excellent / good / medium / poor, and generates a performance evaluation report to provide a basis for the optimization module's decision-making.

[0038] The advantages are: through the four-dimensional quantification and AHP rating of the above-mentioned yaw error, response delay, power generation loss and reliability indicators, it helps to reflect the control quality and economic benefits of the intelligent yaw system in real time and intuitively throughout the entire life cycle, providing a weighted decision basis for the optimization module, thereby avoiding blind parameter adjustment; at the same time, the generated standardized report can be directly used for grid assessment, quality assurance acceptance and insurance actuarial calculation, ultimately improving the efficiency of operation and maintenance management.

[0039] Based on the performance evaluation report, the optimization module identifies changes in the system's dynamic characteristics (such as changes in the damping coefficient caused by wear of the yaw mechanism), updates the wind turbine dynamic model parameters, and trains a reinforcement learning model using historical operating data (wind conditions, control commands, system response). Through trial and error learning, it finds the optimal combination of control strategies under different wind conditions (such as the weight allocation of predictive control and sliding mode control). Based on the optimization results, it fine-tunes key parameters online, including the process noise covariance matrix and measurement noise covariance matrix of the Kalman filter (to improve data fusion accuracy), the gain K of the sliding mode control (to balance response speed and stability), and the time window length of the predictive control (to adapt to the rate of wind direction change), thereby achieving dynamic optimization of system performance.

[0040] The advantages are: through the closed-loop self-evolution mechanism of online updating of model parameters, reinforcement learning strategy search, and OTA fine-tuning of key control parameters, the system can automatically correct the fusion noise matrix, sliding mode gain, and prediction time domain according to the external drift caused by yaw bearing wear and wind resource changes, ensuring that the yaw accuracy is still better than 95% of the initial value after 3 years of operation, reducing the number of manual on-site calibrations, and extending the life of the yaw mechanism.

[0041] The hardware control module includes a signal conditioning unit, a power drive unit, and a status monitoring unit; The signal conditioning unit filters, amplifies, and isolates the control commands (such as voltage signals) sent by the intelligent decision output module, converting them into standard signals that meet the requirements of the actuator. The power drive unit controls the start, stop, speed and direction of the yaw motor according to the conditioned signal, and drives the nacelle to rotate through the gearbox and yaw bearing to achieve precise wind alignment; The status monitoring unit monitors the motor current, temperature, and yaw angle feedback signals online to determine whether the actuator is operating normally. When an abnormal state of overload or over-temperature is detected, the protection mechanism (such as shutdown) is automatically triggered and fed back to the intelligent decision module.

[0042] The advantages are: through the three-level hardware link of signal conditioning, power drive and status monitoring, low distortion, high power amplification and real-time protection are achieved from control commands to motor actions; the status monitoring unit detects current, temperature and angle feedback online, and can trigger protection shutdown 20ms before overload or over-temperature, avoiding motor burnout and mechanical shock, thereby reducing the hardware failure rate of the yaw system, and reducing spare parts inventory and downtime maintenance time.

[0043] The system supports redundant sensor deployment and a fault self-diagnosis mechanism. When abnormal or failed data is detected from a sensor, it automatically switches to a backup sensor or an estimated value to ensure the continuous and stable operation of the system.

[0044] The advantages are: through the sensor dual redundancy + data cross-verification + automatic fault switching design of this invention, it is helpful to seamlessly switch to the backup channel or estimated value when a single wind direction, wind speed or encoder fails, so as to ensure that the yaw function is uninterrupted, which greatly improves the availability of this invention, while reducing the power generation loss and the number of emergency climbing maintenance caused by sensor failure, and ultimately achieving high reliability and maintenance-free long-term operation.

[0045] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method of optimizing an intelligent yaw system of a wind turbine generator system, characterized in that, The method comprises the following steps: An intelligent yaw system is established, which is provided with a multi-sensor data acquisition module, a data fusion module, an intelligent decision output module, a hardware control module, a hardware-in-the-loop test module, a performance evaluation module and an optimization module; The multi-sensor data acquisition module is responsible for controlling the wind direction sensor, the wind speed sensor and the position / yaw angle sensor to perform synchronous or asynchronous data acquisition, and completing preliminary verification and caching of the data; The data fusion module performs time sequence alignment, outlier elimination and noise filtering preprocessing on the original data, and adopts a Kalman filtering algorithm to optimally fuse multi-source sensing information to obtain high-confidence wind direction, wind speed and yaw state estimation values; The intelligent decision output module calculates the target yaw angle and the required yaw angular velocity based on the fused data, and outputs the optimal yaw control instruction according to the multi-mode control strategy and the parameter adaptive mechanism to drive the actuator; The hardware control module receives the control instruction, and after signal conditioning and power amplification, drives the yaw motor to perform corresponding actions to realize accurate wind alignment of the nacelle; The hardware-in-the-loop test module simulates the dynamic characteristics of the wind turbine and the wind field environment in the laboratory environment, performs comprehensive function verification, limit test and fault injection test on the control cabinet, and identifies and repairs potential problems in advance; The performance evaluation module continuously collects yaw error, response delay and power generation operation data after the system is actually deployed, and evaluates the control effect of the system; The optimization module adjusts the control parameters and fusion algorithm parameters online through parameter identification, machine learning or optimization algorithm based on the performance evaluation results and operation data to continuously improve the performance of the system.

2. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The multi-sensor data acquisition module comprises a wind direction data acquisition unit, a wind speed data acquisition unit and a position data acquisition unit; The wind direction data acquisition unit collects real-time wind direction angle by installing high-precision ultrasonic wind direction sensors on the top and tail of the cabin with a sampling frequency ≥ 10 Hz; The wind speed data acquisition unit measures wind speed by fusing cup-type wind speed sensors and laser radar wind measuring instruments and turbulence intensity ​ The position data acquisition unit measures the yaw angle a and the yaw angle speed by an absolute value encoder .

3. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The data fusion module comprises a data preprocessing unit, a Kalman filtering fusion unit and an abnormality diagnosis and processing unit; The data preprocessing unit performs preliminary processing on the original data, including unit conversion, outlier elimination and low-pass filtering preliminary noise reduction; The Kalman filter fusion unit performs a Kalman filter algorithm, establishes a system state equation and an observation equation, optimally fuses multi-source data, and obtains corrected wind direction , wind speed and yaw angle estimation values, which are specifically realized by defining a system state vector and an observation vector, wherein: State vector: In the formula, denotes the real wind direction angle of the wind field at time k, denotes the real wind speed of the wind field at time k, denotes the real yaw angle of the wind turbine nacelle at time k, denotes the real yaw angular velocity of the nacelle at time k; Observation vector: wherein represents a wind direction sensor at time k, represents a measured value of the wind speed sensor at time k, represents a yaw angle measurement of the position sensor at time k; The optimal estimate is iteratively calculated through the prediction and update steps; The abnormality diagnosis and processing unit monitors the data stream in real time, identifies abnormalities through threshold judgment and statistical testing, and triggers the smoothing processing, data elimination or redundant sensor switching mechanism.

4. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The intelligent decision output module comprises an expected state calculation unit, a multi-mode control strategy unit and a parameter adaptive unit; The expected state calculation unit calculates the target yaw angle and the yaw angular velocity based on the fused data, outputs the optimal control instruction, and specifically comprises: (1) target generation: according to the real-time high-precision wind direction angle provided by the data fusion module, the target yaw angle required for the wind wheel plane to be perpendicular to the wind direction is calculated; (2) decision optimization: the optimal yaw angular velocity set value considering the response speed and stability is calculated by comprehensively considering the current yaw system mechanical load, the response characteristics of the actuator and the wind direction change trend; (3) instruction preprocessing: the above target angle and angular velocity are converted into the expected trajectory or set point that can be tracked by the bottom control law to provide input for the subsequent multi-mode control strategy unit; The multi-mode control strategy unit automatically selects or weights the fusion of the predictive control and the sliding mode control algorithm according to the real-time monitored wind conditions and the unit state, specifically: (1) Predictive control: based on historical wind direction data, the future wind direction trend is predicted through the LSTM time series model to generate a feedforward control amount; (2) Sliding mode control: by forcing the system state to move along the preset sliding mode surface, fast convergence is achieved Deviation definition: In the formula, represents a desired yaw angle, represents an actual yaw angle, represents a yaw angle deviation; Sliding mode surface design: wherein and both represent design parameters for adjusting the dynamic behavior of the deviation; Through the control law: In the formula, denotes a sign function, denotes a sliding mode gain, through a discontinuous control law, the system state is forced to move along the sliding surface, and fast convergence is achieved. The parameter adaptive unit dynamically adjusts the parameters of the controller according to real-time wind speed and wind condition information of turbulence intensity .

5. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The hardware-in-the-loop test module includes a high-fidelity simulation unit, a test execution unit, a fault injection unit and a result analysis unit; The high-fidelity simulation unit constructs a wind turbine dynamics model and a wind farm environment model based on data collected by multiple sensors, simulating the response of the wind turbine under different working conditions; The test execution unit connects the simulation model and the actual control cabinet through a hardware interface, realizes real-time data interaction, simulates the transmission of control instructions and state feedback in real operation; The fault injection unit can simulate sensor faults, actuator faults and communication faults to test the fault tolerance of the system; The result analysis unit records the yawing accuracy, response time and control stability indicators during the test process, generates a test report, and identifies potential problems of control logic loopholes and parameter configuration defects.

6. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The performance evaluation module continuously collects yaw error , response delay , and power production data after actual wind turbine deployment.

7. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The performance evaluation module evaluates the system control effect, economy and reliability based on quantitative indicators, combined with preset thresholds and the analytic hierarchy process, and generates a performance evaluation report to provide decision basis for the optimization module.

8. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The optimization module identifies changes in system dynamic characteristics based on the performance evaluation report, updates the wind turbine dynamics model parameters, and trains the reinforcement learning model using historical operation data to find the optimal control strategy combination under different wind conditions through trial and error learning. According to the optimization results, the key parameters are fine-tuned online to realize dynamic optimization of system performance.

9. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The hardware control module includes a signal conditioning unit, a power driving unit and a state monitoring unit; The signal conditioning unit filters, amplifies and isolates the control instructions sent by the intelligent decision output module, and converts them into standard signals that meet the requirements of the actuator; The power driving unit controls the start-stop, speed and steering of the yaw motor according to the conditioned signals, drives the nacelle to rotate through the reduction box and yaw bearing, and realizes accurate wind alignment; The state monitoring unit monitors the motor current, temperature and yaw angle feedback signals online, judges whether the actuator is running normally, and automatically triggers the protection mechanism and feeds back to the intelligent decision module when detecting abnormal states such as overload and overtemperature.

10. The method of optimizing a yaw system of a wind turbine generator system as claimed in claim 1, wherein: The system supports redundant sensor deployment and fault self-diagnosis mechanism. When detecting abnormal or failed sensor data, it automatically switches to backup sensors or estimated values to ensure continuous and stable operation of the system.

Images