Wind turbine generator cabin vibration acceleration abnormity identification method and device

Abstract

The invention discloses a wind turbine generator cabin vibration acceleration abnormity identification method and device. The method comprises the following steps: acquiring operation data of a wind turbine generator in real time; the operation data comprises vibration acceleration of the cabin in the horizontal direction and the vertical direction, and environment and equipment data; inputting environment and equipment data into the prediction model to obtain predicted values of vibration acceleration of the cabin in the horizontal direction and the vertical direction; calculating a residual error between the predicted value and a real value, and marking the real value corresponding to the residual error exceeding a preset threshold interval as abnormal; the prediction model is obtained by training historical data, the historical data comprises normal operation data in a period of historical time, and the trained prediction model is configured to predict normal values of vibration acceleration of the cabin in the horizontal direction and the vertical direction according to environment and equipment data. Compared with an existing anomaly recognition method, the method considers the influence of the environment and the equipment on the recognition result, so that the anomaly recognition result is more accurate.

Description

Technical Field

[0001] The present application relates to the technical field of wind turbine generator sets, and in particular to a method and device for identifying abnormal vibration acceleration of a wind turbine generator set cabin. Background Art

[0002] As a clean and renewable form of energy, wind power has significant advantages in reducing greenhouse gas emissions compared to traditional fossil fuel power generation, and its impact on the ecological environment is relatively small.

[0003] The vibration condition of the nacelle of a wind turbine is a key indicator for evaluating the operating status of a wind turbine and is crucial to ensuring the safety of the unit. Abnormal vibration of the nacelle will not only affect the stability of the unit and even cause shutdown and power generation loss, but long-term abnormal vibration may also damage key components such as the wind rotor and bearings, which may lead to serious safety accidents. Therefore, continuous monitoring and early warning of nacelle vibration acceleration are extremely important to ensure the safe and stable operation of wind turbines.

[0004] At present, the vibration signal of the cabin is subjected to Fourier transform to observe the vibration spectrum, thereby identifying the frequency with larger amplitude and comparing it with the natural frequency of the unit components, and then identifying the abnormal vibration acceleration of the cabin. However, the vibration acceleration signal of the cabin is easily affected by various factors such as the environment and equipment. The existing abnormality identification method has large errors in practical applications. Summary of the invention

[0005] The embodiments of the present application provide a method and device for identifying abnormal vibration acceleration of a wind turbine cabin, thereby solving the technical problem that the vibration acceleration signal of the cabin is easily affected by various factors such as the environment and equipment, and that existing abnormality identification methods have large errors in practical applications.

[0006] In a first aspect, an embodiment of the present application provides a method for identifying abnormal vibration acceleration of a wind turbine cabin, the method comprising: acquiring operating data of a wind turbine in real time; wherein the operating data includes the vibration acceleration of the cabin in the horizontal and vertical directions, as well as environmental and equipment data; inputting the environmental and equipment data into the prediction model to obtain predicted values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions; calculating the residual between the predicted value and the true value, and marking the true value corresponding to the residual that exceeds a preset threshold interval as abnormal; wherein the prediction model is obtained by training with historical data, the historical data includes normal operating data within a historical period of time, and the trained prediction model is configured to predict normal values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions based on the environmental and equipment data.

[0007] In combination with the first aspect, in a possible implementation, the prediction model is a temporal convolutional network improved by effective channel attention.

[0008] In combination with the first aspect, in a possible implementation, during the training of the prediction model, a geometric mean optimization algorithm is used to select optimal hyperparameters.

[0009] In combination with the first aspect, in a possible implementation, the method further includes: performing statistics on the distribution of the residuals to calculate the mean and standard deviation of the residuals; and determining the preset threshold interval using the mean and the standard deviation according to the 3σ criterion.

[0010] In combination with the first aspect, in a possible implementation manner, the environment and equipment data include wind speed, wind rotor speed, output power, pitch angle and yaw direction.

[0011] In combination with the first aspect, in a possible implementation manner, in the prediction model process, the historical data is divided into a training set and a test set according to a preset ratio.

[0012] In combination with the first aspect, in a possible implementation, before training the prediction model, the missing values ​​in the historical data are filled using the average values ​​of the previous and next moments.

[0013] In a second aspect, an embodiment of the present application provides a device for identifying abnormal vibration acceleration of a wind turbine cabin, the device comprising: an acquisition module for acquiring operating data of a wind turbine in real time; wherein the operating data includes the vibration acceleration of the cabin in the horizontal and vertical directions, as well as environmental and equipment data; a prediction module for inputting the environmental and equipment data into the prediction model to obtain predicted values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions; an identification module for calculating the residual between the predicted value and the true value, and marking the true value corresponding to the residual that exceeds a preset threshold interval as abnormal; wherein the prediction model is obtained by training with historical data, the historical data includes normal operating data within a historical period, and the trained prediction model is configured to predict normal values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions based on the environmental and equipment data.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; and a memory storing computer executable instructions, wherein the computer executable instructions, when executed by the one or more processors, implement the method for identifying abnormal vibration acceleration of a wind turbine cabin as described in the first aspect or any possible implementation method of the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, the method for identifying abnormal vibration acceleration of a wind turbine cabin as described in the first aspect or any possible implementation method of the first aspect is implemented.

[0016] The technical solution provided in the embodiments of the present application has at least the following technical effects: The embodiment of the present application provides a method for identifying abnormal vibration acceleration of a wind turbine cabin. After obtaining the operating data of the wind turbine, the method uses a prediction model for prediction. The prediction model can obtain a prediction value based on the environment and equipment data. The prediction value is the normal value of the vibration acceleration of the cabin in the horizontal and vertical directions. The residual between the predicted value and the true value can determine whether the vibration acceleration of the cabin is abnormal. Compared with the existing abnormality identification method, this method takes into account the impact of the environment and equipment on the identification result, making the abnormality identification result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A schematic diagram of a flow chart of a method for identifying abnormal vibration acceleration of a wind turbine nacelle provided in an embodiment of the present application; Figure 2 A schematic diagram of a process for training a prediction model provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a wind turbine nacelle vibration acceleration anomaly identification device provided in an embodiment of the present application; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] At present, the vibration signal of the cabin is subjected to Fourier transform to observe the vibration spectrum, thereby identifying the frequency with larger amplitude and comparing it with the natural frequency of the unit components, and then identifying the abnormal vibration acceleration of the cabin. However, the vibration acceleration signal of the cabin is easily affected by various factors such as the environment and equipment. The existing abnormality identification method has large errors in practical applications.

[0020] In view of the above problems, an embodiment of the present application provides a method for identifying abnormal vibration acceleration of a wind turbine nacelle, which can avoid the influence of the environment and equipment on the abnormal identification results and make the abnormal identification results more accurate.

[0021] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] Below, some terms in the embodiments of the present application are explained to facilitate understanding by those skilled in the art.

[0023] Temporal Convolutional Network (TCN) is a convolutional neural network architecture designed for sequence modeling. The temporal convolutional network is composed of multiple residual blocks stacked in sequence. The residual block is the core unit of TCN, including dilated causal convolution and residual connection.

[0024] Dilated causal convolution combines the advantages of dilated convolution and causal convolution. For time series, the output of causal convolution at time t only depends on the value at time t and before time t. Unlike traditional convolution, causal convolution is unidirectional and is a strict time-constrained model. When the input time step increases and the temporal convolution network needs to process long-distance dependencies, larger convolution kernels or deeper model layers may be required to capture long-distance information, resulting in a large increase in model parameters. To solve this problem, TCN introduced dilated convolution units. Dilated convolution adds gaps between adjacent elements of the convolution kernel to increase the receptive field of the convolution kernel (the size of the input area covered by the convolution kernel), as shown below: ; Where d is the expansion coefficient, which indicates the expansion multiple of the convolution kernel size. When = 1, the dilated causal convolution is equivalent to the ordinary convolution; k represents the size of the convolution filter. Under the same number of parameters, the dilated causal convolution can effectively capture the dependency information at a longer distance. The dilated causal convolution can effectively limit the size of the model and maintain the efficiency of calculation while capturing the long-term dependency of the sequence.

[0025] Residual connections are used in TCN to optimize the gradient propagation path. The idea of ​​residual connections is as follows: ;in, It represents the output of the input variable after passing through the convolutional network. The output is added to the input variable x, and the output value is obtained through the activation function. In practical applications, in order to ensure the consistency of dimensions, a 1×1 convolution is used to adjust the dimension of the input variable x, and then The residual structure can solve the gradient disappearance and gradient explosion problems in deep models, making it possible to train deeper networks.

[0026] The Geometric Mean Optimizer (GMO) is a meta-heuristic algorithm that simulates the unique properties of the geometric mean operator in mathematics and can be used to evaluate the fitness and diversity of individuals in the search space. The GMO specifically includes the following (1) to (8).

[0027] (1) Initialize and generate a random search individual. The speed and position of the search individual are defined as: .

[0028] ; In the formula, MF is the fuzzy membership function between individuals, which can be calculated by the following formula: .

[0029] (3) Arrange DFI in descending order and select the best top N individuals.

[0030] (4) Calculate the weighted average of the opposite best individuals of the focused individuals according to the following formula, where the weight is the DFI: ; Further, in order to make With stronger randomness, Gaussian mutations are imposed on individuals: .

[0031] (5) Update search individual: ;in, .

[0032] (6) Determine whether the termination conditions are met.

[0033] (7) If not satisfied, return to step (2).

[0034] (8) The best solution found so far is taken as the optimization result.

[0035] The present application embodiment provides a method for identifying abnormal vibration acceleration of a wind turbine nacelle. Figure 1 As shown, the method includes steps S101 to S103.

[0036] S101, obtaining the operation data of the wind turbine in real time, wherein the operation data includes the vibration acceleration of the cabin in the horizontal and vertical directions, as well as the environment and equipment data.

[0037] The environment and equipment data are the operating parameters of the wind turbine and the external environment parameters. Exemplarily, the environment and equipment data include wind speed, rotor speed, output power, blade pitch angle and yaw direction.

[0038] S102: Input the environment and equipment data into the prediction model to obtain the predicted values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions.

[0039] The prediction model is obtained through historical data training, which includes normal operating data within a historical period of time. The trained prediction model is configured to predict the normal values ​​of the vibration acceleration of the cabin in the horizontal and vertical directions based on the environment and equipment data.

[0040] Specifically, the process of training the prediction module is as follows: Figure 2 As shown, it includes steps S201 to S205.

[0041] S201. Collect historical data.

[0042] The historical data obtained in step S201 is used as the basis for training the prediction model. For example, the operation data is collected every one minute, the vibration acceleration of the cabin in the horizontal and vertical directions is used as the target variable, and the environment and equipment data are used as the dependent variable to form historical data.

[0043] S202: Preprocess historical data.

[0044] Specifically, preprocessing includes data cleaning (removing noise and outliers), filling missing values, normalization, and data segmentation (dividing training sets and test sets), etc. Among them, in order to ensure the continuity of historical data in the time dimension, the missing values ​​can be filled with the average value of the previous and next moments.

[0045] It should be noted that when performing data segmentation on historical data, the historical data is divided into a training set and a test set according to a preset ratio, for example, the preset ratio is 8:2.

[0046] S203: Build a prediction model.

[0047] Exemplarily, the prediction model is a temporal convolutional network improved by effective channel attention. The effective channel attention can automatically assign weights to information of different channels by learning input features, so that the prediction model has better expression ability for key features.

[0048] S204. Use the geometric mean optimization algorithm to select the optimal hyperparameters.

[0049] After the optimal hyperparameters are found in step S204, the performance of the prediction model can be significantly improved.

[0050] S205: training prediction module.

[0051] S103: Calculate the residual between the predicted value and the true value, and mark the true value corresponding to the residual exceeding a preset threshold range as abnormal.

[0052] The wind turbine cabin vibration acceleration abnormality identification method provided in the embodiment of the present application is to predict the predicted value corresponding to the environment and equipment data at each moment through a prediction model, and the predicted value is the normal value of the cabin vibration acceleration in the horizontal and vertical directions, that is, the method takes into account that the normal values ​​of the cabin vibration acceleration in the horizontal and vertical directions corresponding to different environment and equipment data are different. Therefore, compared with the existing abnormality identification method, this method takes into account the impact of the environment and equipment on the identification result, making the abnormality identification result more accurate, reducing the operation and maintenance costs and the losses caused by equipment damage, and ensuring the safe operation of the wind turbine.

[0053] The wind turbine nacelle vibration acceleration anomaly identification method also includes: performing statistics on the distribution of residuals, calculating the mean and standard deviation of the residuals; and determining a preset threshold interval using the mean and standard deviation according to the 3σ criterion.

[0054] Through the above steps, the preset threshold interval is determined as ; When the residual value is in When the residual value is within , it is determined that the true value corresponding to the residual value is normal; when the residual value is within When it is abnormal, the true value corresponding to the residual value is determined to be abnormal.

[0055] like Figure 3 As shown, the embodiment of the present application provides a wind turbine nacelle vibration acceleration abnormality identification device 300, which includes an acquisition module 301, a prediction module 302 and an identification module 303. The acquisition module 301 is used to acquire the operation data of the wind turbine in real time; wherein the operation data includes the vibration acceleration of the nacelle in the horizontal and vertical directions, as well as the environment and equipment data; the prediction module 302 is used to input the environment and equipment data into the prediction model to obtain the predicted value of the vibration acceleration of the nacelle in the horizontal and vertical directions; the identification module 303 is used to calculate the residual between the predicted value and the true value, and mark the true value corresponding to the residual exceeding the preset threshold interval as abnormal; wherein the prediction model is obtained by training with historical data, and the historical data includes normal operation data within a historical period, and the trained prediction model is configured to predict the normal value of the vibration acceleration of the nacelle in the horizontal and vertical directions according to the environment and equipment data.

[0056] Among them, the prediction model is a temporal convolutional network improved by effective channel attention.

[0057] Environmental and equipment data include wind speed, rotor speed, output power, pitch angle and yaw direction.

[0058] The wind turbine nacelle vibration acceleration anomaly identification device 300 also includes a selection module, which is used to select optimal hyperparameters using a geometric mean optimization algorithm during the training of the prediction model.

[0059] The wind turbine nacelle vibration acceleration anomaly identification device 300 also includes an interval determination module, which is used to perform statistics on the distribution of residuals, calculate the mean and standard deviation of the residuals, and determine the preset threshold interval using the mean and standard deviation according to the 3σ criterion.

[0060] The wind turbine nacelle vibration acceleration anomaly identification device 300 further includes a partitioning module, which is used to partition historical data into a training set and a test set according to a preset ratio during the model prediction process.

[0061] The wind turbine nacelle vibration acceleration anomaly identification device 300 also includes a preprocessing module, which is used to fill in the missing values ​​in the historical data by using the average values ​​of the previous and next moments before training the prediction model.

[0062] The devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in various modules according to their functions. When implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0063] like Figure 4 As shown, an embodiment of the present application further provides an electronic device 400, which includes: one or more processors; and a memory storing computer executable instructions, which, when executed by one or more processors, implement the method for identifying abnormal vibration acceleration of a wind turbine cabin described above in the embodiment of the present application.

[0064] The embodiment of the present application also provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, the method for identifying abnormal vibration acceleration of a wind turbine cabin described in the embodiment of the present application is implemented.

[0065] The above-mentioned storage media include but are not limited to random access memory (RAM), read-only memory (ROM), cache, hard disk drive (HDD) or memory card (Memory Card).

[0066] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other. Each embodiment focuses on the differences from other embodiments.

[0067] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present application.

Claims

1. A method for identifying abnormal vibration acceleration of a wind turbine nacelle, characterized in that: include: Acquire the operation data of the wind turbine in real time; wherein the operation data includes the vibration acceleration of the nacelle in the horizontal and vertical directions, as well as the environment and equipment data; Inputting the environment and equipment data into a prediction model to obtain predicted values ​​of vibration acceleration of the cabin in horizontal and vertical directions; Calculating the residual between the predicted value and the true value, and marking the true value corresponding to the residual exceeding a preset threshold interval as abnormal; Wherein, the prediction model is obtained through historical data training, and the historical data includes normal operating data within a historical period of time. The trained prediction model is configured to predict normal values ​​of vibration acceleration of the cabin in the horizontal and vertical directions based on the environment and equipment data.

2. The method according to claim 1, characterized in that: The prediction model is a temporal convolutional network improved by effective channel attention.

3. The method according to claim 2, characterized in that In the process of training the prediction model, the geometric mean optimization algorithm is used to select the optimal hyperparameters.

4. The method according to claim 1, characterized in that: The method further comprises: Performing statistics on the distribution of the residuals, and calculating the mean and standard deviation of the residuals; According to the 3σ criterion, the preset threshold interval is determined using the mean and the standard deviation.

5. The method according to claim 1, characterized in that The environment and equipment data include wind speed, wind rotor speed, output power, pitch angle and yaw direction.

6. The method according to claim 1, characterized in that In the prediction model process, the historical data is divided into a training set and a test set according to a preset ratio.

7. The method according to claim 1, characterized in that Before training the prediction model, the missing values ​​in the historical data are filled with the average values ​​of the previous and next moments.

8. A wind turbine nacelle vibration acceleration abnormality identification device, characterized in that: include: An acquisition module, used to acquire the operation data of the wind turbine in real time; wherein the operation data includes the vibration acceleration of the cabin in the horizontal and vertical directions, as well as the environment and equipment data; A prediction module, used for inputting the environment and equipment data into the prediction model to obtain the predicted values ​​of the vibration acceleration of the cabin in the horizontal direction and the vertical direction; An identification module, used for calculating the residual between the predicted value and the true value, and marking the true value corresponding to the residual exceeding a preset threshold interval as abnormal; Wherein, the prediction model is obtained through historical data training, and the historical data includes normal operating data within a historical period of time. The trained prediction model is configured to predict normal values ​​of vibration acceleration of the cabin in the horizontal and vertical directions based on the environment and equipment data.

9. An electronic device, characterized in that: include: one or more processors; as well as A memory storing computer executable instructions, wherein the computer executable instructions, when executed by the one or more processors, implement the method for identifying abnormal vibration acceleration of a wind turbine nacelle according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the method for identifying abnormal vibration acceleration of a wind turbine nacelle according to any one of claims 1 to 7 is implemented.

Images