Yaw bearing gear protection device, method, and yaw load optimization method

Abstract

This invention discloses a yaw bearing gear protection device, method, and yaw load optimization method, relating to the field of wind turbine technology. By setting first and second markers on both sides of the repaired yaw bearing teeth, and setting first and second marker detection elements on the yaw pinion, a trigger signal is issued after detecting the tooth corresponding to the first and second markers entering or exiting the meshing area with the yaw pinion. The main control system, based on the yaw direction of the yaw bearing, outputs a load reduction command if the yaw bearing rotates towards the meshing area, and outputs a load reduction cancellation command otherwise. The load output of the yaw motor is controlled by the yaw inverter to achieve load reduction or load reduction cancellation. By monitoring the repaired yaw bearing online, the yaw drive load control is optimized, effectively preventing tooth breakage again due to excessive load on the repaired yaw bearing gear, improving the reliability of the yaw bearing, and extending the service life of the wind turbine.

Description

Technical Field

[0001] This invention relates to the field of wind turbine technology, and in particular to a yaw bearing gear protection device, method, and yaw load optimization method. Background Technology

[0002] To maximize wind energy capture, the rotor must always face the wind direction, which requires a yaw system. The yaw system consists of three parts: yaw detection and control, yaw drive, and yaw braking. The yaw drive consists of a yaw motor, a yaw gearbox, and a yaw bearing.

[0003] If a wind turbine experiences a broken tooth in its yaw bearing, the tooth is typically repaired by welding. While the repaired tooth can ensure the basic normal operation of the yaw system, its load-bearing capacity is weaker than that of a normal gear. Under certain high-load conditions, even after the repaired tooth has withstood a significant load, there is still a risk of it breaking again.

[0004] Therefore, it is particularly important to reduce the load on the repaired gear teeth and avoid the recurrence of tooth breakage. Summary of the Invention

[0005] The purpose of this invention is to provide a yaw bearing gear protection device, method, and yaw load optimization method, which enables online monitoring of the repaired yaw bearing, optimizes yaw drive load control, and effectively prevents the yaw bearing gear teeth from breaking again due to bearing large loads after repair.

[0006] To address the aforementioned technical problems, this invention provides a yaw bearing gear protection device, comprising a main control system, a yaw frequency converter, a yaw motor, a yaw bearing, a yaw pinion, a first marker and a second marker disposed on both sides of the repair gear teeth of the yaw bearing, and a first marker detection element and a second marker detection element disposed on the yaw pinion. The first marker detection element and the second marker detection element are used to issue a trigger signal after detecting that the gear teeth corresponding to the first marker and the second marker have entered or exited the meshing area with the yaw pinion, and to output a load reduction command after determining that the yaw bearing is rotating towards the meshing area, and to output a load reduction cancellation command after the yaw bearing is moving away from the meshing area, based on the yaw direction of the yaw bearing. The yaw frequency converter is used to control the load output of the yaw motor according to the load reduction command and the load reduction cancellation command.

[0007] Among them, the number of teeth between the first marker, the second marker and the repaired tooth is equal.

[0008] The number of teeth between the first marker, the second marker and the repair gear teeth is 1 to 3.

[0009] The first and second markers are located on the side of the yaw bearing or on the rotating shaft of the yaw bearing.

[0010] The first marker detection component and the second marker detection component are welded, welded or bolted to the yaw gear.

[0011] The marker detection device is a proximity switch or a photoelectric switch, and the first marker and the second marker are proximity switch baffles or photoelectric reflectors.

[0012] It also includes a yaw motor encoder installed on the yaw motor and an alarm device for detecting the operating status of the yaw motor encoder. The yaw motor encoder is used to measure the yaw angle position information and send it to the main control system. The alarm device is used to output a fault prompt to the main control system after detecting an abnormal operating status of the yaw motor encoder.

[0013] It also includes a display module connected to the main control system, used to display the yaw direction, speed of the yaw bearing, and the output load of the yaw motor.

[0014] In addition, embodiments of this application also provide a yaw bearing gear protection method, applied to the above-mentioned yaw bearing gear protection device, including:

[0015] S1: Determine whether a trigger signal is received from the first marker detection unit or the second marker detection unit. The first marker detection unit and the second marker detection unit are used to issue a trigger signal after detecting that the gear tooth corresponding to the first marker and the second marker enters the meshing area with the yaw pinion. The first marker and the second marker are set on both sides of the repair gear tooth of the yaw bearing, and the first marker detection unit and the second marker detection unit are set on the yaw pinion.

[0016] If so, S2: Determine whether the rotation direction of the yaw bearing is towards the meshing area;

[0017] If yes, S3: Generate and output a load reduction command, controlling the yaw inverter to reduce the output load of the driven yaw motor according to the load reduction command; otherwise, S4: Generate and output a load reduction cancellation command, controlling the yaw inverter to restore the output load of the yaw motor according to the load reduction cancellation command.

[0018] In addition, this application embodiment also provides a yaw load optimization method, applied to the above-mentioned yaw bearing gear protection device and the above-mentioned yaw bearing gear protection method, including:

[0019] S11: Obtain historical wind direction data of wind turbine units, perform preprocessing operations on historical operation data and tag data, and output preprocessed data. Preprocessing operations include outlier removal, missing value imputation and normalization.

[0020] S12: Perform feature construction and feature selection on the preprocessed data to extract key temporal features for characterizing wind direction;

[0021] S13: Use an LSTM network to extract time-dependent features, use a CNN module to extract local structural change features, construct a CNN-LSTM deep learning model, and train the CNN-LSTM deep learning model.

[0022] S14: Input the wind direction information of the current time period into the CNN-LSTM deep learning model to predict the wind direction, and output the wind direction prediction information for the future time period.

[0023] S15: Adjust the yaw direction of the wind turbine's yaw bearing based on wind direction forecast information.

[0024] The yaw bearing gear protection device, method, and yaw load optimization method provided in this invention have the following advantages compared with the prior art:

[0025] The yaw bearing gear protection device, method, and yaw load optimization method provided in this invention embodiment, by setting a first marker and a second marker on both sides of the repaired gear teeth of the yaw bearing, and setting a first marker detection element and a second marker detection element on the yaw pinion, a trigger signal is issued after detecting that the gear teeth corresponding to the first and second markers enter or exit the meshing area with the yaw pinion. The main control system outputs a load reduction command after determining that the yaw bearing is rotating towards the meshing area, and outputs a load reduction cancellation command otherwise, based on the yaw direction of the yaw bearing. The load output of the yaw motor is controlled by the yaw frequency converter to realize the load reduction or load reduction cancellation operation. By monitoring the repaired yaw bearing online, the yaw drive load control is optimized, effectively preventing the yaw bearing gear from breaking again due to bearing a large load after repair, improving the reliability of the yaw bearing and extending the service life of the wind turbine. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A schematic diagram of the connection structure of an embodiment of the yaw bearing gear protection device provided by the present invention;

[0028] Figure 2 A schematic diagram of the structure of a yaw bearing is provided as an embodiment of the yaw bearing gear protection device provided by the present invention.

[0029] Figure 3 A schematic diagram of the yaw drive assembly structure of one embodiment of the yaw bearing gear protection device provided by the present invention;

[0030] Figure 4 A schematic diagram of the step flow structure of an embodiment of the yaw bearing gear protection method provided by the present invention;

[0031] Figure 5 A schematic diagram of the step flow structure of an embodiment of the yaw load optimization method provided by the present invention;

[0032] Among them, 1-Yaw motor, 2-Yaw gearbox, 3-Yaw bearing, 4-Yaw pinion, 5-First marker detection piece, 6-Second marker detection piece, 7-First marker, 8-Second marker. Detailed Implementation

[0033] 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.

[0034] Please refer to Figures 1-5 , Figure 1 A schematic diagram of the connection structure of an embodiment of the yaw bearing gear protection device provided by the present invention; Figure 2 A schematic diagram of the structure of a yaw bearing is provided as an embodiment of the yaw bearing gear protection device provided by the present invention. Figure 3 A schematic diagram of the yaw drive assembly structure of one embodiment of the yaw bearing gear protection device provided by the present invention; Figure 4 A schematic diagram of the step flow structure of an embodiment of the yaw bearing gear protection method provided by the present invention; Figure 5 This is a schematic diagram of the step flow structure of an embodiment of the yaw load optimization method provided by the present invention.

[0035] In one specific embodiment, the yaw bearing gear protection device includes a main control system, a yaw frequency converter, a yaw motor 1, a yaw bearing 3, a yaw pinion 4, a first marker 7 and a second marker 8 disposed on both sides of the repaired gear teeth of the yaw bearing 3, and a first marker detection element 5 and a second marker detection element 6 disposed on the yaw pinion 4. The first marker detection element 5 and the second marker detection element 6 are used to issue a trigger signal after detecting that the gear teeth corresponding to the first marker 7 and the second marker 8 have entered or exited the meshing area with the yaw pinion 4. The main control system receives the trigger signal and, based on the yaw direction of the yaw bearing 3, outputs a load reduction command after determining that the yaw bearing 3 is rotating towards the meshing area, and outputs a load reduction cancellation command after the yaw bearing 3 is moving away from the meshing area. The yaw frequency converter is used to control the load output of the yaw motor 1 according to the load reduction command and the load reduction cancellation command.

[0036] By setting a first marker 7 and a second marker 8 on both sides of the repaired gear teeth of the yaw bearing 3, and setting a first marker detection element 5 and a second marker detection element 6 on the yaw pinion 4, a trigger signal is issued after the gear teeth corresponding to the first marker 7 and the second marker 8 enter or exit the meshing area with the yaw pinion 4. The main control system outputs a load reduction command after determining that the yaw bearing 3 is rotating towards the meshing area, and outputs a load reduction cancellation command otherwise, based on the yaw direction of the yaw bearing 3. The load output of the yaw motor 1 is controlled by the yaw frequency converter to realize the load reduction or load reduction cancellation operation. By monitoring the repaired yaw bearing 3 online, the yaw drive load control is optimized, effectively preventing the yaw bearing 3 gear from breaking again due to bearing a large load, improving the reliability of the yaw bearing 3 and extending the service life of the wind turbine.

[0037] In this application, by setting the first marker 7 and the second marker 8 on the yaw bearing 3 so that it rotates with the yaw bearing 3, the rotation position of the yaw bearing 3 is determined by detecting the position of the first marker 7 and the second marker 8, and it is determined whether it has rotated into the position of the repair gear tooth. There are no restrictions on the type, quantity, setting position and setting method of the first marker 7 and the second marker 8.

[0038] Preferably, the number of teeth between the first marker 7, the second marker 8, and the repaired gear teeth is equal.

[0039] By ensuring that the number of teeth between the first marker 7, the second marker 8, and the repair gear teeth is equal, the time from the issuance of the load reduction command for the yaw bearing 3 to the engagement of the repair gear teeth is equal regardless of whether the rotation is forward or reverse. This prevents the actual load on the repair gear teeth from changing and avoids the situation where the actual load on the repair gear teeth is greater than the design parameters due to the short time from load reduction to engagement, thus reducing the reliability of use.

[0040] In this application, the number of teeth between the first marker 7, the second marker 8 and the repair gear is not limited, and the number of teeth between the first marker 7, the second marker 8 and the repair gear is 1 to 3.

[0041] For example, if the number of teeth between the first marker 7, the second marker 8 and the repair tooth is 2, the first marker 7 and the second marker 8 are set at the positions of the two teeth before and after the repair tooth for position detection, etc.

[0042] This application does not limit the way the first marker 7 and the second marker 8 are set. The first marker 7 and the second marker 8 are set on the side of the yaw bearing 3 or on the rotating shaft of the yaw bearing 3.

[0043] If the rotation axis of the yaw bearing 3 is vertical, the first marker 7 and the second marker 8 can be set on the side of the repair gear teeth, that is, set directly above the yaw bearing 3.

[0044] Alternatively, the first marker 7 and the second marker 8 can be set on the rotation axis of the yaw bearing 3, so that they rotate with the rotation axis of the yaw bearing 3. In this way, the height of the first marker 7 and the second marker 8 can be flexibly set.

[0045] In addition, markers can be set at other locations on the yaw bearing 3, or other methods can be used to set the markers.

[0046] In this application, a marker detection device is used to detect the position of the first marker 7 and the second marker 8. Point-to-point detection can be used, that is, marker detection devices are set at both ends of the meshing area. Point-to-point detection can also be used, that is, marker detection devices are set in the meshing area, and the first marker 7 and the second marker 8 can be detected as long as they enter the detection range. Alternatively, other structures can be used.

[0047] Alternatively, in this application, the first marker 7 and the second marker 8 can be set in the yaw bearing 3 and integrated into a workpiece to be tested. The first marker 7 and the second marker 8 are installed at both ends of the workpiece to be tested. When passing through the meshing area, they generate acoustic, optical, and other signals, or reflect the acoustic, optical, and other signals and are detected by the marker detection component, thereby realizing the generation of the corresponding trigger signal.

[0048] This application does not limit the type, structure, and setting method of the marker detection component. The first marker detection component 7 and the second marker detection component 6 are welded, welded, or bolted to the yaw gear 4, or connected in other ways.

[0049] The first marker detection element 5 and the second marker detection element 6 are proximity switches or photoelectric switches, or other detectors. The first marker 7 and the second marker 8 are proximity switch baffles or photoelectric reflectors, or other sensors or signal generating devices, such as reflective strips, as long as they can be accurately detected.

[0050] To further improve the operational reliability of the entire device, in one embodiment, the yaw bearing 3 gear protection device also includes a yaw motor 1 encoder installed on the yaw motor 1 and an alarm device for detecting the operating status of the yaw motor 1 encoder. The yaw motor 1 encoder is used to measure yaw angle position information and send it to the main control system. The alarm device is used to output a fault prompt to the main control system after detecting an abnormal operating status of the yaw motor 1 encoder.

[0051] By installing an encoder for yaw motor 1 and an alarm device for detecting the operating status of the encoder, the yaw angle position information can be measured by the encoder, allowing the main control system to obtain the operating position information of yaw bearing 3 in real time. This improves the accuracy of the yaw bearing 3's position determination. Furthermore, the reliability of operation can be further improved by cross-verifying the position information of yaw bearing 3 through the aforementioned marker detection device, the first marker 7, and the second marker 8. If the position information of yaw bearing 3 detected by the marker detection device is inconsistent with the position information of yaw bearing 3 obtained by the encoder of yaw motor 1, one of them will inevitably have failed, allowing maintenance personnel to perform rapid maintenance.

[0052] This application does not limit the type of encoder and alarm device for the yaw motor 1.

[0053] To further improve the management efficiency of the device operation, in one embodiment, the yaw bearing 3 gear protection device also includes a display module connected to the main control system, which is used to display the yaw direction, speed and output load of the yaw bearing 3 and the yaw motor 1.

[0054] The display module displays the yaw direction and speed of the yaw bearing 3 and the output load of the yaw motor 1 in real time. The operating status of the yaw bearing 3 can be obtained in real time and compared with the commands of the relevant main control system in real time to determine whether the control command is consistent with the actual status, thereby determining whether the equipment is operating normally and improving management efficiency.

[0055] This application does not limit the type or structure of the display module.

[0056] In addition, this application can also use communication modules, such as WiFi modules, 4G communication modules, 5G communication modules, etc., to remotely acquire data and issue control commands, thereby improving the reliability and flexibility of operation.

[0057] It should be noted that if there are multiple repair teeth in the yaw bearing 3 in this application, the first marker 7 and the second marker 8 are set in the manner described above. If there is an intersection, such as when the marker corresponding to the previous repaired tooth enters the range of the marker corresponding to the next repaired tooth, the two can be considered as one area and the area is defined as a new repair area. The load reduction operation can be performed when the repair area enters the meshing area.

[0058] In this application, the load reduction operation and tooth repair inspection of the yaw bearing 3 are generally targeted at a maximum of about 10 teeth of the yaw bearing 3 that need to be repaired. Too many repaired teeth will cause the yaw bearing 3 to frequently enter the load reduction operation, which will greatly reduce the overall load capacity. Therefore, it is recommended to replace the bearing directly after more than 10 teeth are repaired.

[0059] Furthermore, during the load reduction operation of the yaw bearing 3 in this application, the actual load is not limited to a small amount. The full yaw load is mainly determined based on the rated torque of the yaw motor 1, such as a rated torque of 20 N·m for a single yaw motor 1. The maximum load of the repaired gear teeth is calculated as 80% of the rated load of the motor (16 N·m). Of course, if the reliability of the repair and the structural strength can be guaranteed, and after testing, a larger load can be used for operation. The actual operating load can be customized or executed according to a preset percentage. After a certain period of execution, the maximum load can be appropriately reduced to ensure its reliability.

[0060] In one embodiment, the yaw bearing 3 gear protection device includes a main control system, a yaw frequency converter, a yaw motor 1, a yaw motor 1 encoder, a yaw gearbox 2, a yaw bearing 3, a yaw pinion 4, a proximity switch, and a proximity switch baffle.

[0061] The encoder is connected to the yaw motor 1 to detect the position of the gear in the repaired yaw bearing 3.

[0062] The main control system is connected to the proximity switch and the yaw inverter, receiving the trigger signal from the proximity switch and controlling the loading and unloading of the yaw inverter. The yaw inverter is connected to the main control system and yaw motor 1, receiving the loading and unloading commands from the main control system and controlling the load output of yaw motor 1. The encoder of yaw motor 1 is installed inside yaw motor 1 to measure the yaw angle position. Yaw motor 1 and yaw gearbox 2 are integrated and connected to yaw pinion 4 to drive yaw bearing 3. The proximity switch and proximity switch baffle are connected to the main control system to determine the position of the gear teeth of the repaired yaw bearing 3.

[0063] A proximity switch, namely proximity switch A (left) and proximity switch B (right), is installed on each side of the yaw gear 4 to detect the position of the gear teeth of the repaired yaw bearing 3. In addition, a proximity switch baffle is welded to the normal gear on both sides of the gear teeth of the repaired yaw bearing 3 to trigger proximity switches A and B, namely proximity switch baffle A and proximity switch baffle B.

[0064] Assuming the repaired yaw bearing 3 has the Nth tooth, the proximity switch baffles A and B are positioned directly above the (N-2)th and (N+2)th teeth, respectively. Assuming the current yaw direction is clockwise, when the repaired yaw bearing 3's tooth is about to engage with the yaw pinion 4, proximity switch baffle A, located at the (N-2)th tooth, triggers proximity switch A. The main control system receives the trigger signal from proximity switch A, thus determining that the repaired yaw bearing 3's tooth is about to engage with the yaw pinion 4. The main control system then issues a yaw load reduction control command to the yaw inverter. Upon receiving the load reduction command, the yaw inverter reduces the load output of the yaw motor 1, preventing the repaired yaw bearing 3's tooth from bearing a large load. When the proximity switch baffle B installed at the N+2th gear tooth triggers the proximity switch B, the main control system receives the trigger signal from the proximity switch B. Therefore, the main control system determines that after the gear tooth of the repaired yaw bearing 3 has completed meshing with the yaw pinion 4, the main control system cancels the yaw load reduction control command, and the yaw inverter restores the load output of the yaw motor 1.

[0065] The current yaw direction is counterclockwise. When the teeth of the repaired yaw bearing 3 are about to mesh with the yaw pinion 4, the proximity switch baffle A installed at the (N-2)th tooth triggers the proximity switch B. At this time, the main control system receives the trigger signal from proximity switch B. Therefore, the main control system determines that the teeth of the repaired yaw bearing 3 are about to mesh with the yaw pinion 4, and sends a yaw load reduction control command to the yaw inverter. After receiving the load reduction command from the main control system, the yaw inverter reduces the load output of the yaw motor 1 to avoid the teeth of the repaired yaw bearing 3 bearing a large load. When the proximity switch baffle B installed at the (N-2)th tooth triggers the proximity switch A, the main control system receives the trigger signal from proximity switch A. Therefore, the main control system determines that after the teeth of the repaired yaw bearing 3 have completed meshing with the yaw pinion 4, the main control system cancels the yaw load reduction control command, and the yaw inverter restores the load output of the yaw motor 1.

[0066] The proximity switch and yaw encoder can mutually verify the position of the repaired yaw bearing 3 gear. The working principle of the yaw encoder is as follows: after the engine room is aligned to north, the yaw encoder is zeroed through the main control system, at which point the corresponding position of yaw bearing 3 is zero degrees. Assuming that yaw bearing 3 has N teeth, and the repaired yaw bearing 3 has the Mth tooth, the corresponding angle is M×360 / N.

[0067] When the yaw encoder fails, and the main control system cannot confirm the position of the teeth of the repaired yaw bearing 3 through the yaw encoder, the position of the teeth can be confirmed by a proximity switch installed on the edge of the teeth. The main control system will issue an alarm signal after the yaw encoder fails, but the unit can operate normally. The alarm signal will be cleared after the yaw encoder failure is resolved. Conversely, when the proximity switch fails, and the main control system cannot confirm the position of the teeth of the repaired yaw bearing 3 through the proximity switch, the position of the teeth can be confirmed by the yaw encoder. The main control system will issue an alarm signal after the proximity switch fails, but the unit can operate normally. The alarm signal will be cleared after the proximity switch failure is resolved.

[0068] In addition, embodiments of this application also provide a yaw bearing gear protection method, applied to the above-mentioned yaw bearing gear protection device, including:

[0069] S1: Determine whether a trigger signal is received from the first marker detection unit or the second marker detection unit. The first marker detection unit and the second marker detection unit are used to issue a trigger signal after detecting that the gear tooth corresponding to the first marker and the second marker enters the meshing area with the yaw pinion. The first marker and the second marker are set on both sides of the repair gear tooth of the yaw bearing, and the first marker detection unit and the second marker detection unit are set on the yaw pinion.

[0070] If so, S2: Determine whether the rotation direction of the yaw bearing is towards the meshing area;

[0071] If yes, S3: Generate and output a load reduction command, controlling the yaw inverter to reduce the output load of the driven yaw motor according to the load reduction command; otherwise, S4: Generate and output a load reduction cancellation command, controlling the yaw inverter to restore the output load of the yaw motor according to the load reduction cancellation command.

[0072] In addition, this application embodiment also provides a yaw load optimization method, applied to the above-mentioned yaw bearing gear protection device and the above-mentioned yaw bearing gear protection method, including:

[0073] S11: Acquire historical wind direction data of wind turbines, perform preprocessing operations on historical operating data and tag data, and output preprocessed data. Preprocessing operations include outlier removal, missing value imputation and normalization. Wind direction data acquisition can be achieved using integrated wind speed and direction sensors, ultrasonic anemometers, laser anemometers and Doppler radar, etc. Generally, ultrasonic anemometers are mainly used to measure wind speed and direction data.

[0074] S12: Perform feature construction and feature selection on the preprocessed data to extract key temporal features for characterizing wind direction;

[0075] S13: Use an LSTM network to extract time-dependent features, use a CNN module to extract local structural change features, construct a CNN-LSTM deep learning model, and train the CNN-LSTM deep learning model.

[0076] S14: Input the wind direction information of the current time period into the CNN-LSTM deep learning model to predict the wind direction, and output the wind direction prediction information for the future time period.

[0077] S15: Adjust the yaw direction of the wind turbine's yaw bearing based on wind direction forecast information.

[0078] In the wind direction prediction process of this application, historical wind direction data is first preprocessed, then wind direction feature processing is performed to extract key temporal features that characterize wind direction, then a CNN layer is used for data classification, recognition and prediction, then a prediction model is built with an LSTM layer as the main body, data training is performed, and finally the wind direction is predicted and output.

[0079] In one embodiment, the wind direction prediction process is as follows:

[0080] 1. Wind direction data preprocessing.

[0081] def __init__(self, seq_length=2 forecast_horizon=1 / 12, n_directions=16):

[0082] parameter:

[0083] seq_length: Length of the input sequence (in hours)

[0084] forecast_horizon: Forecast step size

[0085] n_directions: Number of wind direction categories (16 directions)

[0086] """

[0087] self.seq_length = seq_length

[0088] self.forecast_horizon = forecast_horizon

[0089] self.n_directions = n_directions

[0090] self.scaler = StandardScaler()

[0091] 2. Wind direction characteristics.

[0092] def load_and_preprocess(self, filepath):

[0093] Loading and preprocessing data

[0094] # Assume the CSV file contains the following: wind_direction, wind_speed, temperature, pressure, humidity

[0095] data = pd.read_csv(filepath)

[0096] # Processing wind direction data

[0097] self.n_directions == 16:

[0098] # Convert 360 degrees to 16 directions (0-15)

[0099] data['wind_dir_class'] = ((data['wind_direction'] + 11.25) % 360 / 22.5).astype(int)

[0100] # Select Feature Columns

[0101] feature_columns = ['wind_speed', 'temperature', 'pressure', 'humidity']

[0102] # Standardization Features

[0103] features_scaled = self.scaler.fit_transform(data[feature_columns])

[0104] return features_scaled, data['wind_dir_class'].values

[0105] def create_sequences(self, features, labels):

[0106] """Creating sequence data"""

[0107] X, y = [], []

[0108] for i in range(len(features) - self.seq_length -self.forecast_horizon +1):

[0109] X.append(features[i:i + self.seq_length])

[0110] y.append(labels[i + self.seq_length + self.forecast_horizon - 1])

[0111] return np.array(X), np.array(y)

[0112] 3. Establish a model.

[0113] class CNNLSTMWindDirection(nn.Module):

[0114] "CNN-LSTM Wind Direction Prediction Model"

[0115] def __init__(self, input_dim, hidden_dim, num_layers, n_directions,

[0116] cnn_channels=32, kernel_size=3, dropout=0.3):

[0117] """

[0118] parameter:

[0119] input_dim: Input feature dimension

[0120] hidden_dim: LSTM hidden layer dimension

[0121] num_layers: Number of LSTM layers

[0122] n_directions: Number of output categories

[0123] cnn_channels: Number of CNN channels

[0124] kernel_size: CNN convolution kernel size

[0125] dropout: Dropout probability

[0126] """

[0127] super(CNNLSTMWindDirection, self).__init__()

[0128] # CNN part

[0129] self.conv1 = nn.Conv1d(

[0130] in_channels=input_dim,

[0131] out_channels=cnn_channels,

[0132] kernel_size=kernel_size,

[0133] padding=kernel_size / / 2 )

[0135] self.bn1 = nn.BatchNorm1d(cnn_channels)

[0136] self.conv2 = nn.Conv1d(

[0137] in_channels=cnn_channels,

[0138] out_channels=cnn_channels*2,

[0139] kernel_size=kernel_size,

[0140] padding=kernel_size / / 2 )

[0142] self.bn2 = nn.BatchNorm1d(cnn_channels*2)

[0143] self.pool = nn.MaxPool1d(kernel_size=2)

[0144] self.relu = nn.ReLU()

[0145] self.dropout = nn.Dropout(dropout)

[0146] # LSTM part

[0147] self.lstm = nn.LSTM(

[0148] input_size=cnn_channels*2,

[0149] hidden_size=hidden_dim,

[0150] num_layers=num_layers,

[0151] batch_first=True,

[0152] dropout=dropout if num_layers > 1 else 0,

[0153] bidirectional=True )

[0155] # Fully Connected Layer

[0156] self.fc1 = nn.Linear(hidden_dim * 2, 64) # Bidirectional LSTM, so *2

[0157] self.fc2 = nn.Linear(64, n_directions)

[0158] def forward(self, x):

[0159] # x shape: (batch_size, seq_len, input_dim)

[0160] # Adjusting dimensions for 1D convolution: (batch, channels, seq_len)

[0161] x = x.permute(0, 2, 1)

[0162] # CNN layer

[0163] x = self.relu(self.bn1(self.conv1(x)))

[0164] x = self.dropout(x)

[0165] x = self.relu(self.bn2(self.conv2(x)))

[0166] x = self.pool(x) # Pooling reduces sequence length

[0167] # Adjusting dimensions for LSTM: (batch, seq_len, channels)

[0168] x = x.permute(0, 2, 1)

[0169] # LSTM layer

[0170] lstm_out, (hidden, cell) = self.lstm(x)

[0171] # Take the output of the last time step

[0172] lstm_out = lstm_out[:, -1, :]

[0173] # Fully Connected Layer

[0174] x = self.relu(self.fc1(lstm_out))

[0175] x = self.dropout(x)

[0176] x = self.fc2(x)

[0177] `return x` outputs the wind direction prediction result.

[0178] It should be noted that there are no restrictions on the type of network used for feature extraction in this application, nor on the type of network used for local structural change features, the model constructed, or the number of training iterations.

[0179] After preprocessing historical wind direction data, a model is built and trained, and finally the current direction is predicted. This reduces the frequent turning of the yaw bearing, the continuous starting of the yaw system, and the inertial impact load, thereby reducing the maximum load on the gears and improving the reliability of the yaw bearing.

[0180] In one embodiment, wind direction prediction is used to optimize the yaw load control of the wind turbine, reduce the load on the repaired yaw bearing gear, and prevent the recurrence of tooth breakage.

[0181] After training, the wind direction prediction model first processes wind direction data over a period of time. Then, the processed data and selected historical wind direction data are input into a CNN-LSTM. The CNN-LSTM fits the trained data and outputs predicted wind direction values ​​through a fully connected neural network, ultimately achieving prediction of the wind direction time series. Finally, the main control system optimizes the yaw control strategy based on the predicted wind direction data, reducing unnecessary frequent forward and reverse yaw movements, minimizing yaw impact loads, and preventing the repaired yaw bearing teeth from breaking again.

[0182] Among them, Long Short-Term Memory (LSTM) networks are one of the better deep learning architectures for time series learning tasks, excelling at classifying, processing, and predicting time series data. Convolutional Neural Networks (CNNs), on the other hand, aim to extract features from objects using a specific model and then classify, identify, predict, or make decisions based on those features.

[0183] Combining CNN and LSTM for wind direction prediction presents several advantages. While LSTM-based models offer high accuracy, they suffer from significant errors during abrupt wind shifts. Furthermore, large historical step sizes lead to excessive training time and memory consumption, reducing algorithm efficiency. CNN, on the other hand, excels at feature extraction and deep feature mining. Therefore, combining CNN and LSTM retains the strengths of both, increases model depth, and further improves wind direction prediction accuracy.

[0184] In one embodiment, the yaw load optimization method includes:

[0185] 1. Obtain historical wind direction data for wind turbines, and perform preprocessing operations on historical operating data and tag data, including outlier removal, missing value imputation, and normalization;

[0186] 2. Perform feature construction and feature selection on the preprocessed historical operation data, and extract key time-series features to characterize wind direction. The main features include wind speed, temperature, humidity, and air pressure.

[0187] 3. Construct a CNN-LSTM deep learning model, use the LSTM network to extract time-dependent features, combine the CNN module to extract local structural change features, and establish a wind direction prediction model;

[0188] 4. After the model training is completed, it is used as the input of the wind turbine yaw load optimization control method. Under the condition of frequent wind direction changes, it reduces unnecessary frequent forward and reverse yaw movements, reduces the generation of yaw impact load, and avoids the yaw bearing gear from breaking teeth again after repair.

[0189] The above method can enable online monitoring of the repaired yaw bearing, optimize yaw drive load control, and effectively prevent the yaw bearing gear from breaking again due to bearing large loads after repair.

[0190] In one embodiment, the yaw optimization process is as follows:

[0191] Calculate the current yaw error angle and determine if the yaw error angle is greater than or equal to 6°. If not, the crew does not yaw. If so, the crew calculates the average yaw error angle for the next 5 minutes based on the CNN-LSTM model prediction. The crew then determines if the yaw error angle is greater than or equal to 6° again. If so, the crew yaws; otherwise, the crew does not yaw. After the crew yaws, the crew determines if the yaw error angle is less than or equal to 3°. If so, the crew returns to the initial yaw angle error detection. Otherwise, the crew calculates the average yaw error angle for the next 5 minutes based on the CNN-LSTM model prediction.

[0192] This application does not limit the yaw threshold for the yaw error angle of the aircraft, and it can be customized. It also does not limit the yaw angle error threshold after yaw adjustment.

[0193] In summary, the yaw bearing gear protection device, method, and yaw load optimization method provided in this embodiment of the invention, by setting a first marker and a second marker on both sides of the repaired gear teeth of the yaw bearing, and setting a first marker detection element and a second marker detection element on the yaw pinion, a trigger signal is issued after the gear teeth corresponding to the first and second markers enter or exit the meshing area with the yaw pinion. The main control system outputs a load reduction command after determining that the yaw bearing is rotating towards the meshing area, and outputs a load reduction cancellation command otherwise, based on the yaw direction of the yaw bearing. The load output of the yaw motor is controlled by the yaw frequency converter to realize the load reduction or load reduction cancellation operation. By monitoring the repaired yaw bearing online, the yaw drive load control is optimized, effectively preventing the yaw bearing gear from breaking again due to bearing a large load after repair, improving the reliability of the yaw bearing and extending the service life of the wind turbine.

[0194] The yaw bearing gear protection device, method, and yaw load optimization method provided by this invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make several improvements and modifications to this invention without departing from the principles of this invention, and these improvements and modifications also fall within the protection scope of the claims of this invention.

Claims

1. A yaw bearing gear protection device, characterized in that, The system includes a main control system, a yaw inverter, a yaw motor, a yaw bearing, a yaw pinion, a first marker and a second marker disposed on both sides of the repair tooth of the yaw bearing, and a first marker detection element and a second marker detection element disposed on the yaw pinion. The first marker detection element and the second marker detection element are used to issue a trigger signal after detecting that the tooth corresponding to the first marker and the second marker enters or exits the meshing area with the yaw pinion. The main control system receives the trigger signal and, based on the yaw direction of the yaw bearing, outputs a load reduction command after determining that the yaw bearing is rotating towards the meshing area, and outputs a load reduction cancellation command after the yaw bearing is moving away from the meshing area. The yaw inverter is used to control the load output of the yaw motor according to the load reduction command and the load reduction cancellation command.

2. The yaw bearing gear protection device as described in claim 1, characterized in that, The number of teeth between the first marker, the second marker, and the repaired gear tooth is equal.

3. The yaw bearing gear protection device as described in claim 2, characterized in that, The number of teeth between the first marker, the second marker and the repair gear is 1 to 3.

4. The yaw bearing gear protection device as described in claim 3, characterized in that, The first marker and the second marker are disposed on the side of the yaw bearing or on the rotating shaft of the yaw bearing.

5. The yaw bearing gear protection device as described in claim 4, characterized in that, The first marker detection component and the second marker detection component are welded, welded, or bolted to the yaw gear.

6. The yaw bearing gear protection device as described in claim 5, characterized in that, The marker detection device is a proximity switch or a photoelectric switch, and the first marker and the second marker are proximity switch baffles or photoelectric reflectors.

7. The yaw bearing gear protection device as described in any one of claims 1-6, characterized in that, It also includes a yaw motor encoder installed on the yaw motor and an alarm device for detecting the operating status of the yaw motor encoder. The yaw motor encoder is used to measure yaw angle position information and send it to the main control system. The alarm device is used to output a fault prompt to the main control system after detecting an abnormal operating status of the yaw motor encoder.

8. The yaw bearing gear protection device as described in claim 7, characterized in that, It also includes a display module connected to the main control system, used to display the yaw direction and speed of the yaw bearing and the output load of the yaw motor.

9. A method for protecting yaw bearing gears, characterized in that, The device applied to the yaw bearing gear protection device as described in any one of claims 1-8 includes: S1: Determine whether a trigger signal is received from the first marker detection unit or the second marker detection unit. The first marker detection unit and the second marker detection unit are used to issue a trigger signal after detecting that the gear tooth corresponding to the first marker and the second marker enters the meshing area with the yaw pinion. The first marker and the second marker are set on both sides of the repair gear tooth of the yaw bearing, and the first marker detection unit and the second marker detection unit are set on the yaw pinion. If so, S2: Determine whether the rotation direction of the yaw bearing is towards the meshing area; If yes, S3: Generate and output a load reduction command, controlling the yaw inverter to reduce the output load of the driven yaw motor according to the load reduction command; otherwise, S4: Generate and output a load reduction cancellation command, controlling the yaw inverter to restore the output load of the yaw motor according to the load reduction cancellation command.

10. A method for optimizing yaw load, characterized in that, Applied to, as in any one of claims 1-8, the yaw bearing gear protection device and the yaw bearing gear protection method as in claim 9, comprising: S11: Obtain historical wind direction data of wind turbine units, perform preprocessing operations on historical operation data and tag data, and output preprocessed data. Preprocessing operations include outlier removal, missing value imputation and normalization. S12: Perform feature construction and feature selection on the preprocessed data to extract key temporal features for characterizing wind direction; S13: Use an LSTM network to extract time-dependent features, use a CNN module to extract local structural change features, construct a CNN-LSTM deep learning model, and train the CNN-LSTM deep learning model. S14: Input the wind direction information of the current time period into the CNN-LSTM deep learning model to predict the wind direction, and output the wind direction prediction information for the future time period. S15: Adjust the yaw direction of the wind turbine's yaw bearing based on wind direction forecast information.

Images