Wind turbine life testing methods and devices, electronic equipment, and storage media

Abstract

This application relates to the field of nuclear power plant equipment safety technology, and in particular to a method, apparatus, electronic device, and storage medium for detecting the lifespan of wind turbines. According to an embodiment of this application, the wind turbine lifespan detection method involves: acquiring a pre-trained wind turbine lifespan detection model; acquiring target operating state data of the wind turbine; extracting features from the target operating state data to obtain target operating state features; and performing wind turbine lifespan detection on the target operating state features using the pre-trained wind turbine lifespan detection model to obtain target wind turbine lifespan data. This improves the accuracy of wind turbine lifespan detection.

Description

Technical Field

[0001] This application relates to the field of wind turbine testing, and in particular to a method and apparatus for wind turbine life testing, electronic equipment, and storage medium. Background Technology

[0002] Wind turbine life testing is a key technology for nuclear power plants to improve wind energy utilization efficiency and ensure the safe operation of wind turbine equipment. However, with the increasing complexity of the wind turbine operating environment, the diversity of wind turbine failure modes, and the increase in wind turbine data, the difficulty of life testing for wind turbine equipment in nuclear power plants is constantly increasing, and there may be issues with the safe operation of wind turbine equipment in nuclear power plants.

[0003] Currently, wind turbine lifespan assessment typically uses machine learning models (such as support vector machines, random forests, or linear regression) to predict the remaining service life of wind turbine equipment. However, this method requires manual selection of feature indicators from wind turbine operating data and ignores the temporal characteristics of wind turbine operation. This limits the accuracy of wind turbine lifespan assessment to the selection of feature indicators, resulting in low accuracy. Therefore, improving the accuracy of wind turbine lifespan assessment remains a pressing challenge for the industry. Summary of the Invention

[0004] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method, apparatus, electronic equipment, and storage medium for wind turbine life testing, which can improve the accuracy of wind turbine life testing.

[0005] The wind turbine life testing method according to the first aspect of this application includes:

[0006] Obtain a pre-trained wind turbine life detection model; wherein, the pre-training process of the wind turbine life detection model includes:

[0007] Obtain prior operating status data of the wind turbine;

[0008] The wind turbine life distribution is predicted by using a preset Weibull cumulative distribution function to predict the prior operating state data, and the prior life distribution data is obtained.

[0009] The initial wind turbine life detection model is trained based on the prior life distribution data and the prior operating status data to obtain the pre-trained wind turbine life detection model.

[0010] Acquire the target operating status data of the wind turbine;

[0011] Feature extraction is performed on the target operating state data to obtain the target operating state features;

[0012] The wind turbine lifespan data is obtained by performing wind turbine lifespan detection on the target operating state characteristics using the pre-trained wind turbine lifespan detection model.

[0013] According to some embodiments of this application, training the initial wind turbine life detection model based on the prior life distribution data and the prior operating state data to obtain a pre-trained wind turbine life detection model includes:

[0014] Obtain the initial wind turbine life detection model;

[0015] The wind turbine lifespan is predicted by using the wind turbine lifespan detection model to analyze the prior operating state data and obtain the predicted wind turbine lifespan data.

[0016] The predicted lifespan distribution data is obtained by using the Weibull cumulative distribution function to predict the lifespan distribution of the predicted wind turbine lifespan data.

[0017] Obtain the actual wind turbine life data from the prior operating state data, and calculate the target loss value of the predicted wind turbine life data, the predicted life distribution data, the actual wind turbine life data, and the prior life distribution data according to a preset loss function.

[0018] The model parameters of the wind turbine life detection model are updated based on the target loss value. Then, the wind turbine life prediction is performed on the prior operating state data using the wind turbine life detection model until the wind turbine life detection model meets the preset training conditions, thus obtaining the pre-trained wind turbine life detection model.

[0019] According to some embodiments of this application, the loss function includes a neural network loss function and a Weibull loss function;

[0020] The step of calculating the target loss value of the predicted wind turbine lifespan data, the predicted lifespan distribution data, the actual wind turbine lifespan data, and the prior lifespan distribution data according to a preset loss function includes:

[0021] The life loss value between the predicted wind turbine life data and the actual wind turbine life data is calculated based on the neural network loss function.

[0022] The lifetime distribution loss value between the predicted lifetime distribution data and the prior lifetime distribution data is calculated based on the Weibull loss function.

[0023] The target loss value is obtained by integrating the lifetime loss value and the lifetime distribution loss value through preset weight hyperparameters.

[0024] According to some embodiments of this application, the step of predicting the wind turbine life distribution data by using a preset Weibull cumulative distribution function on the prior operating state data to obtain prior life distribution data includes:

[0025] The wind turbine shape characterization parameters are determined based on the prior operating state data.

[0026] Based on the wind turbine shape characterization parameters and the prior operating state data, determine the wind turbine dimensional characterization parameters;

[0027] Based on the wind turbine shape characterization parameters and the wind turbine size characterization parameters, the wind turbine lifetime distribution is predicted from the prior operating state data to obtain the prior lifetime distribution data.

[0028] According to some embodiments of this application, determining the wind turbine dimensional characterization parameters based on the wind turbine shape characterization parameters and the prior operating state data includes:

[0029] Obtain the wind turbine failure time, number of wind turbine failures, and wind turbine failure variables from the prior operating status data;

[0030] The scale characterization parameters of the wind turbine are obtained by performing scale characterization calculations on the wind turbine failure time, the number of wind turbine failures, the wind turbine failure variables and the wind turbine shape characterization parameters.

[0031] According to some embodiments of this application, the step of performing wind turbine life detection on the target operating state characteristics using the pre-trained wind turbine life detection model to obtain target wind turbine life data includes:

[0032] The wind turbine health status is detected by the target operating state characteristics using the wind turbine life detection model, and wind turbine health status data is obtained.

[0033] Based on the wind turbine health status data, the remaining lifespan of the target wind turbine is detected according to the target operating status characteristics, and the lifespan data of the target wind turbine is obtained.

[0034] According to some embodiments of this application, the step of detecting the wind turbine health status by using the wind turbine life detection model to detect the target operating state characteristics and obtain wind turbine health status data includes:

[0035] The wind turbine operating health features are extracted from the target operating status features using the wind turbine life detection model.

[0036] The health status of the wind turbine is assessed based on a preset linear function to obtain a wind turbine health assessment score.

[0037] The health status data of the wind turbine is determined based on the wind turbine health assessment score.

[0038] According to some embodiments of this application, the step of extracting wind turbine operating health features from the target operating state features using the wind turbine life detection model to obtain wind turbine operating health features includes:

[0039] The target operating state characteristics are encoded using the wind turbine life detection model to obtain the encoded operating characteristics;

[0040] The encoded running features are weighted by health features to obtain weighted healthy running features;

[0041] The weighted health operation characteristics are decoded to obtain the wind turbine operation health characteristics.

[0042] According to some embodiments of this application, the step of detecting the remaining lifespan of the target wind turbine based on the wind turbine health status data to obtain the target wind turbine lifespan data includes:

[0043] Obtain the health weight and health bias term of the wind turbine health data;

[0044] Based on the health status weight, the health status bias term and the wind turbine health status data, wind turbine operation fault detection is performed on the target operating status characteristics to obtain wind turbine operation fault data.

[0045] Based on the wind turbine operation fault data, the lifespan data of the target wind turbine is determined.

[0046] According to some embodiments of this application, the step of extracting features from the target operating state data to obtain target operating state features includes:

[0047] Perform Fourier transform processing on the target operating state features to obtain at least one operating state spectrum feature;

[0048] The operating state spectral feature corresponding to the maximum amplitude is selected from at least one of the operating state spectral features as the target operating state feature.

[0049] According to some embodiments of this application, the step of selecting the operating state spectrum feature corresponding to the maximum amplitude from at least one of the operating state spectrum features as the target operating state feature includes:

[0050] Divide at least one of the operating state spectral features into frequency bands to obtain at least one operating state frequency band;

[0051] The operating state frequency band corresponding to the maximum amplitude is selected from at least one of the operating state frequency bands as the target operating state feature.

[0052] According to some embodiments of this application, before performing feature extraction on the target operating state data to obtain the target operating state features, the method further includes:

[0053] Acquire initial running status data and perform data cleaning on the initial running status data to obtain cleaned running status data;

[0054] The cleaning operation status data is normalized to obtain normalized operation status data;

[0055] The normalized operating status data is divided into data segments by using a preset wind turbine operating time period to obtain the target operating status data.

[0056] The wind turbine life detection device according to a second aspect embodiment of this application includes:

[0057] The wind turbine life detection model acquisition module is used to acquire a pre-trained wind turbine life detection model.

[0058] The wind turbine operating status data acquisition module is used to acquire the target operating status data of the wind turbine;

[0059] The feature extraction module is used to extract features from the target running state data to obtain target running state features;

[0060] The wind turbine life detection module is used to detect the wind turbine life of the target operating state characteristics through the pre-trained wind turbine life detection model, and obtain the target wind turbine life data.

[0061] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the wind turbine life detection method as described in any one of the embodiments of the first aspect of this application.

[0062] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program that is executed by a processor to implement the wind turbine life detection method as described in any one of the embodiments of the first aspect of this application.

[0063] The wind turbine life detection method, apparatus, electronic device, and storage medium according to embodiments of this application have at least the following beneficial effects: obtaining a pre-trained wind turbine life detection model; wherein the pre-training process of the wind turbine life detection model includes: acquiring prior operating state data of the wind turbine; predicting the wind turbine life distribution based on the prior operating state data using a preset Weibull cumulative distribution function to obtain prior life distribution data; training an initial wind turbine life detection model based on the prior life distribution data and the prior operating state data to obtain a pre-trained wind turbine life detection model; acquiring target operating state data of the wind turbine; extracting features from the target operating state data to obtain target operating state features; and performing wind turbine life detection on the target operating state features using the pre-trained wind turbine life detection model to obtain target wind turbine life data. This improves the accuracy of wind turbine life detection.

[0064] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0065] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0066] Figure 1 A schematic flowchart illustrating the wind turbine life testing method provided in this application embodiment;

[0067] Figure 2 for Figure 1 The flowchart of step S101 in the text;

[0068] Figure 3 for Figure 2 The flowchart of step S202 in the text;

[0069] Figure 4 for Figure 1 The flowchart of step S101 in the text;

[0070] Figure 5 for Figure 4 The flowchart of step S404 in the document;

[0071] Figure 6 This is another schematic flowchart of the fan life detection method provided in the embodiments of this application;

[0072] Figure 7 for Figure 1 The flowchart of step S103 in the process;

[0073] Figure 8 for Figure 7 The flowchart of step S702 in the process;

[0074] Figure 9 for Figure 1 The flowchart of step S104 in the process;

[0075] Figure 10 for Figure 9 The flowchart of step S901 in the process;

[0076] Figure 11 for Figure 10 The flowchart of step S1001 in the text;

[0077] Figure 12 For Figure 9 The flowchart of step S902 in the text;

[0078] Figure 13 This is a schematic diagram of the structure of the wind turbine life detection device provided in the embodiments of this application;

[0079] Figure 14 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0080] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0081] In the description of this application, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0082] In the description of this application, it should be understood that the orientation descriptions, such as up, down, left, right, front, and back, are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0083] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0084] In the description of this application, it should be noted that, unless otherwise explicitly defined, terms such as "setting," "installation," and "connection" should be interpreted broadly. Those skilled in the art can reasonably determine the specific meaning of the above terms in this application based on the specific content of the technical solution. Furthermore, the identification of specific steps in the following text does not imply a limitation on the order of steps or execution logic. The execution order and logic between each step should be understood and inferred from the content described in the embodiments.

[0085] Wind turbine life testing is a key technology for nuclear power plants to improve wind energy utilization efficiency and ensure the safe operation of wind turbine equipment. However, with the increasing complexity of the wind turbine operating environment, the diversity of wind turbine failure modes, and the increase in wind turbine data, the difficulty of life testing for wind turbine equipment in nuclear power plants is constantly increasing, and there may be issues with the safe operation of wind turbine equipment in nuclear power plants.

[0086] Currently, wind turbine lifespan assessment typically uses machine learning models (such as support vector machines, random forests, or linear regression) to predict the remaining service life of wind turbine equipment. However, this method requires manual selection of feature indicators from wind turbine operating data and ignores the temporal characteristics of wind turbine operation. This limits the accuracy of wind turbine lifespan assessment to the selection of feature indicators, resulting in low accuracy. Therefore, improving the accuracy of wind turbine lifespan assessment remains a pressing challenge for the industry.

[0087] Therefore, using a wind turbine life detection model to detect the target operating status data of wind turbines can improve the accuracy of wind turbine life detection. It can automatically determine the key characteristic indicators related to wind turbine life detection without the need for manual selection of characteristic indicators of wind turbine operating data. Furthermore, it combines the time series of wind turbine operation, which helps to improve the accuracy of wind turbine life detection.

[0088] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method, apparatus, electronic equipment, and storage medium for wind turbine life testing, which can improve the accuracy of wind turbine life testing.

[0089] The following explanation is based on the attached diagram:

[0090] Reference Figure 1 The wind turbine life testing method according to the embodiments of this application may include, but is not limited to:

[0091] Step S101: Obtain the pre-trained wind turbine life detection model;

[0092] Step S102: Obtain the target operating status data of the wind turbine;

[0093] Step S103: Extract features from the target operating state data to obtain the target operating state features;

[0094] Step S104: The wind turbine lifespan is detected by the target operating status characteristics through a pre-trained wind turbine lifespan detection model to obtain the target wind turbine lifespan data.

[0095] The wind turbine life detection method described in steps S101 to S104 of this application requires obtaining a pre-trained wind turbine life detection model. The pre-training process of the wind turbine life detection model includes: acquiring prior operating state data of the wind turbine; predicting the wind turbine life distribution using a preset Weibull cumulative distribution function to obtain prior life distribution data; training an initial wind turbine life detection model based on the prior life distribution data and the prior operating state data to obtain a pre-trained wind turbine life detection model; acquiring target operating state data of the wind turbine; extracting features from the target operating state data to obtain target operating state features; and performing wind turbine life detection on the target operating state features using the pre-trained wind turbine life detection model to obtain target wind turbine life data. This improves the accuracy of wind turbine life detection.

[0096] In some embodiments, step S101 specifically involves the pre-trained wind turbine life detection model being trained using prior life distribution data and prior operating state data. This wind turbine life detection model can be a neural network model based on an attention mechanism.

[0097] Specifically, the pre-training process of the wind turbine life detection model includes: acquiring prior operating state data of the wind turbine; predicting the wind turbine life distribution using a preset Weibull cumulative distribution function to obtain prior life distribution data; and training the initial wind turbine life detection model based on the prior life distribution data and prior operating state data to obtain the pre-trained wind turbine life detection model.

[0098] Furthermore, the prior operating status data of the wind turbine can be the prior operating knowledge of the wind turbine collected from the sensors of the wind turbine generator set in the nuclear power plant. This prior operating knowledge can come from a large amount of experimental data, industry standards, or long-term operating experience summaries of previous wind turbines of the same type. This data reflects the operating status and performance indicators of the wind turbine at different points in time, and the prior operating status data can include, but is not limited to: wind turbine bearing temperature, wind turbine motor temperature, wind turbine vibration data, wind speed, wind turbine blade speed, wind turbine internal pressure data, torque and force on the wind turbine shaft, lubricating oil viscosity, ambient humidity and ambient temperature, etc.

[0099] Reference Figure 2 According to some embodiments of this application, step S101 uses a preset Weibull cumulative distribution function to predict the wind turbine life distribution of the prior operating state data, obtaining prior life distribution data, which may include, but is not limited to:

[0100] Step S201: Determine the wind turbine shape characterization parameters based on prior operating state data;

[0101] Step S202: Determine the wind turbine dimensional characterization parameters based on the wind turbine shape characterization parameters and prior operating status data;

[0102] Step S203: Based on the wind turbine shape characterization parameters and wind turbine size characterization parameters, predict the wind turbine life distribution from the prior operating state data to obtain the prior life distribution data.

[0103] In some embodiments of this application, the Weibull cumulative distribution function is a commonly used statistical tool that can predict the lifespan distribution of a wind turbine based on prior operating data. In step 101, this calculation process includes several key operations:

[0104] In step S201, specifically, the wind turbine shape characterization parameter is one of the key parameters of the Weibull distribution, which describes the wind turbine failure trend over time.

[0105] Furthermore, when the fan shape characterization parameter k = 1, it indicates that the fan failure rate remains constant over time, and the fan may experience random failures; when the fan shape characterization parameter k > 1, it indicates that the fan failure rate increases over time, and the fan failure may be related to wear or aging; when the fan shape characterization parameter k < 1, it indicates that the fan failure rate decreases over time, which may indicate that the fan is in the early stage of failure.

[0106] For example, for Class A fans, statistical analysis of historical data shows that their shape characterization parameters are usually between 1.5 and 2.5, indicating that the fan failure rate increases or decreases over time, and fan failures may be related to wear or aging.

[0107] In this embodiment, the wind turbine shape characterization parameters are determined based on prior operating status data, which facilitates subsequent prediction of when the wind turbines of nuclear power plants need maintenance or replacement, and helps to improve the efficiency and reliability of wind power generation.

[0108] Step S202 in some embodiments, refer to Figure 3 According to some embodiments of this application, in step S202, determining the wind turbine dimensional characterization parameters based on the wind turbine shape characterization parameters and prior operating state data may include, but is not limited to:

[0109] Step S301: Obtain the wind turbine failure time, number of wind turbine failures, and wind turbine failure variables from the prior operating status data;

[0110] Step S302: Perform scale characterization calculations on the wind turbine failure time, number of wind turbine failures, wind turbine failure variables, and wind turbine shape characterization parameters to obtain wind turbine scale characterization parameters.

[0111] In some embodiments of this application, when determining the wind turbine scale characterization parameters, it is necessary to obtain the wind turbine failure time, number of wind turbine failures, and wind turbine failure variables from the prior operating state data to obtain the wind turbine's operating status, which is the basis for subsequent wind turbine lifetime distribution prediction. In step 202, this calculation process includes several key operations:

[0112] In step S301, specifically, the fan failure time refers to the specific time point or time period from the start of operation of the fan to the occurrence of a failure.

[0113] Specifically, the number of wind turbine failures records the total number of wind turbine failures that occurred within a specific time period.

[0114] Specifically, the wind turbine failure variable refers to the number of wind turbine failures plus incomplete operational failure samples.

[0115] Specifically, the wind turbine failure time, number of wind turbine failures, and wind turbine failure variables are all obtained from prior operating status data.

[0116] In step S302, specifically, the wind turbine scale characterization parameter represents the scale of the wind turbine lifetime distribution or wind turbine failure time in the Weibull distribution. The wind turbine scale characterization parameter acts to amplify or reduce the Weibull distribution curve, but it does not affect the shape of the distribution, and the smaller the wind turbine scale characterization parameter, the faster the wind turbine failure rate increases.

[0117] Specifically, by recording the wind turbine failure time, the specific time points and time cycles from the installation and operation of the wind turbine to the occurrence of failure can be observed. The distribution of wind turbine failures can be observed, the number of wind turbine failures helps to understand the failure frequency of the wind turbine, and the wind turbine failure variables help us to more comprehensively understand the failure modes of the wind turbine. Together, these data reflect the actual lifespan of the wind turbine under different operating conditions.

[0118] Specifically, the scale characterization parameters of wind turbines are calculated by using the wind turbine scale characterization parameter formula. The formula considers the weighted average of all wind turbine failure times and is suitable for handling incomplete wind turbine data. That is, at the end of the last observation period, there may not be enough data to determine the failure status of all wind turbine components, thus improving the accuracy of subsequent wind turbine life detection.

[0119] The embodiments of this application shown in steps S301 to S302 can be used together with the wind turbine shape characterization parameters in the subsequent Weibull distribution function to predict the remaining service life of the wind turbine. This realizes the transformation of wind turbine operating status data into a scientific prediction of wind turbine life, providing important information support for the operation and management of wind turbines in nuclear power plants.

[0120] In step S203, specifically, the prior life distribution data refers to the shape and scale of the Weibull distribution map of the prior life of the wind turbine, which is jointly determined by the wind turbine shape characterization parameter and the wind turbine scale characterization parameter.

[0121] Specifically, the cumulative distribution function of the Weibull distribution can be used to predict the lifespan distribution of wind turbines based on their operating status data. This function gives the probability that a wind turbine will fail before a given time point or time period, enabling the prediction of the failure probability of different wind turbines at different time points, thereby achieving the prediction of the remaining service life of the wind turbines.

[0122] Furthermore, the prediction process takes into account various operating conditions of the wind turbine, such as temperature, vibration, and rotational speed, which may affect the turbine's lifespan. By combining these conditions with the cumulative distribution function of the Weibull distribution, the lifespan distribution of the wind turbine under specific operating conditions can be predicted more accurately.

[0123] In this embodiment, the wind turbine life distribution is predicted based on the wind turbine shape characterization parameters and wind turbine size characterization parameters, reflecting the expected life of the wind turbine under different operating conditions. This enables a quantitative assessment of the wind turbine's health status, facilitating the improvement of the accuracy of wind turbine life detection in the future.

[0124] The embodiments of this application shown in steps S201 to S203 can provide a scientific basis for predicting the lifespan of wind turbines, helping maintenance teams to better plan maintenance strategies, thereby improving the operating efficiency and reliability of wind turbines. Ultimately, this prior lifespan distribution data will be used to train and validate the wind turbine lifespan prediction model, ensuring that the model can accurately predict the remaining service life of the wind turbine.

[0125] In some specific embodiments of this application, the parameters characterizing the wind turbine dimensions can be expressed by the following formulas:

[0126]

[0127] Where η represents the wind turbine dimensional characterization parameter, n represents the wind turbine fault variable, and t m Let β represent the time of the m-th fan failure, β represent the fan shape characteristic parameter, and r represent the number of fan failures.

[0128] Furthermore, the prior lifetime distribution data can be expressed by the following formula:

[0129]

[0130] Where F(t) represents the prior lifetime distribution data of the wind turbine failure time t, t represents the wind turbine failure time, η represents the wind turbine size characterization parameter, and β represents the wind turbine shape characterization parameter.

[0131] In step S101, this embodiment of the application trains the initial wind turbine life detection model based on prior life distribution data and prior operating status data. This allows for continuous improvement of the wind turbine life detection model's predictive performance by combining prior data, which helps to increase the accuracy of subsequent models in predicting the remaining life of the wind turbine. This process may include the following sub-steps.

[0132] Reference Figure 4 According to some embodiments of this application, step S101 trains the initial wind turbine life detection model based on prior life distribution data and prior operating state data to obtain a pre-trained wind turbine life detection model, which may include, but is not limited to:

[0133] Step S401: Obtain the initial wind turbine life detection model;

[0134] Step S402: The wind turbine lifespan is predicted based on the prior operating status data using the wind turbine lifespan detection model, and the predicted wind turbine lifespan data is obtained.

[0135] Step S403: The wind turbine life distribution is predicted by using the Weibull cumulative distribution function to obtain the predicted life distribution data.

[0136] Step S404: Obtain the actual wind turbine life data from the prior operating status data, and calculate the target loss value of the predicted wind turbine life data, predicted life distribution data, actual wind turbine life data, and prior life distribution data according to the preset loss function.

[0137] Step S405: Update the model parameters of the wind turbine life detection model based on the target loss value, and return to execute the wind turbine life prediction based on the prior operating state data through the wind turbine life detection model until the wind turbine life detection model meets the preset training conditions and obtains the pre-trained wind turbine life detection model.

[0138] In some embodiments, step S401 specifically refers to an unoptimized neural network model for wind turbine life detection, which may be an attention-based neural network model.

[0139] Specifically, the initial wind turbine life detection model can predict the life of the wind turbine based on its prior operating status data and output the wind turbine life prediction results.

[0140] In some embodiments, step S402 specifically involves using this model to process the prior operating state data using an attention mechanism, thereby extracting prior health features from the prior operating data, assessing the health status of the wind turbine based on the prior health features, and predicting the lifespan of the wind turbine based on the health status of the wind turbine.

[0141] Specifically, the wind turbine life detection model can learn the relationship between prior operating status data and wind turbine life, identify key features that affect wind turbine life, and thus improve the accuracy of wind turbine life distribution prediction.

[0142] For example, a wind turbine lifespan assessment model may learn that the risk of wind turbine failure increases at specific temperature and vibration levels, or that the lifespan of a wind turbine may be shortened when the wind speed is too high. Using this information, the wind turbine lifespan assessment model can predict the remaining service life of the wind turbine and generate predicted wind turbine lifespan data.

[0143] In some embodiments, step S403 specifically refers to the predicted lifetime distribution data, which is determined by the shape and scale of the predicted prior wind turbine lifetime Weibull distribution map, which is determined by the wind turbine shape characterization parameter and the wind turbine scale characterization parameter.

[0144] Specifically, using the Weibull cumulative distribution function to predict the distribution of wind turbine lifespan data helps to more comprehensively understand the statistical characteristics of wind turbine lifespan, including key indicators such as average lifespan and failure rate. This allows for scoring of the predicted wind turbine lifespan data, and the scoring can determine whether the initial wind turbine lifespan detection model needs to be trained.

[0145] In some embodiments, step S404 specifically refers to the actual wind turbine lifespan data, which is the specific operating time data of the wind turbine during the entire period from installation and commissioning to eventual failure or decommissioning.

[0146] For example, if a wind turbine fails due to bearing damage after 5 years and 3 months of operation after installation, then the 5 years and 3 months timeframe represents the turbine's true lifespan. Similarly, for wind turbines that are decommissioned without failure but due to other reasons (such as technological updates or reduced efficiency), the total operating time from installation to decommissioning also represents their true lifespan. Furthermore, when a wind turbine's power generation decreases from 10 million kilowatt-hours per year initially to 9 million kilowatt-hours in the third year, the change in power generation can also reflect the turbine's aging process.

[0147] Reference Figure 5 According to some embodiments of this application, step S404, which calculates the target loss value of the predicted wind turbine lifespan data, predicted lifespan distribution data, actual wind turbine lifespan data, and prior lifespan distribution data based on a preset loss function, may include, but is not limited to:

[0148] Step S501: Calculate the life loss value between the predicted wind turbine life data and the actual wind turbine life data based on the neural network loss function.

[0149] Step S502: Calculate the lifetime distribution loss value between the predicted lifetime distribution data and the prior lifetime distribution data based on the Weibull loss function;

[0150] Step S503: Integrate the lifetime loss value and lifetime distribution loss value using preset weight hyperparameters to obtain the target loss value.

[0151] In some embodiments of this application, the target loss value is a key indicator for measuring the difference between the predicted value and the true value, used to guide parameter optimization during model training. In the wind turbine lifespan detection model training process, it is necessary not only to measure the accuracy of a single predicted value through the loss function, but also to measure the overall wind turbine lifespan distribution. Therefore, the loss function includes the neural network loss function and the Weibull loss function. In step S404, this calculation process includes several key operations:

[0152] In some embodiments, step S501 specifically refers to the life loss value as the loss value between the predicted wind turbine life data and the actual wind turbine life data.

[0153] Specifically, the loss function for a neural network can be the MES (Mean Squared Error) function.

[0154] In this embodiment, the life loss value between the predicted wind turbine life data and the actual wind turbine life data is calculated based on the neural network loss function. This facilitates the continuous adjustment of the wind turbine life detection model parameters based on the life loss function to minimize the life loss value, thereby improving the accuracy of wind turbine life detection.

[0155] In some embodiments, step S502 specifically refers to the lifetime distribution loss value, which is the loss value between the predicted lifetime distribution data and the prior lifetime distribution data.

[0156] Specifically, the Weibull loss function, as a prior constraint, is used to measure the consistency between the predicted lifetime distribution and the prior lifetime distribution based on historical data, enabling further evaluation of the accuracy of the lifetime distribution model in predicting wind turbine lifetime.

[0157] In this embodiment, the lifetime distribution loss value between the predicted lifetime distribution data and the prior lifetime distribution data is calculated based on the Weibull loss function. This helps to ensure that the model not only performs well in individual predictions, but also accurately reflects the lifetime characteristics of the wind turbine in the overall distribution.

[0158] In some embodiments, step S503 specifically involves the target loss function including a neural network loss function and a Weibull loss function.

[0159] Specifically, the lifetime loss value and lifetime distribution loss value can be integrated using preset weight hyperparameters to balance the contributions of the two loss values ​​to the target loss value, thus obtaining the final target loss value. This integration of the loss function allows the wind turbine lifetime detection model to simultaneously consider the accuracy of individual wind turbine lifetime prediction and the consistency of the overall wind turbine lifetime distribution during optimization. The weight hyperparameters can be adjusted according to the specific requirements of the actual task.

[0160] The embodiments of this application provided through steps S501 to S503 integrate the performance of the wind turbine life detection model in predicting the lifespan of a single wind turbine and the overall wind turbine lifespan distribution. During the training process of the wind turbine life detection model, an optimization target is provided to the model, and the model will continuously update the parameters to minimize the target loss value, thereby reducing the prediction error of the wind turbine life detection model and improving the accuracy of wind turbine lifespan prediction.

[0161] In some embodiments, step S405 specifically involves adjusting the connection weights between neurons in the wind turbine life detection model using a backpropagation algorithm. This allows for adjusting model parameters by minimizing the loss function. After updating the parameters, the model continuously learns based on operational status feature data and known life labels, repeating the above process until the model's performance meets preset training conditions (such as achieving a certain accuracy or the target loss value being less than a preset loss threshold), thus obtaining a pre-trained wind turbine life detection model.

[0162] For example, the predicted lifespan of a wind turbine corresponds to the characteristic data of its operating status (such as high speed, large vibration amplitude, and high temperature). The model can adjust the connection weights so that when the model encounters similar characteristic data of operating status, it can predict a shorter lifespan of the wind turbine.

[0163] In this embodiment, a pre-trained wind turbine lifespan detection model is obtained through multiple iterations and parameter updates. This model can accurately predict the remaining service life of the wind turbine, and its prediction results match the prior wind turbine lifespan data, ensuring the accuracy of wind turbine lifespan detection and thus supporting the sustainable development of the wind power industry in nuclear power plants.

[0164] According to the embodiments of this application provided by steps S401 to S405, the wind turbine life detection model can be continuously adjusted and optimized during the training process to reduce prediction errors. Furthermore, by using a comprehensive loss function, the wind turbine life detection model can learn more comprehensively the wind turbine fault characteristics related to wind turbine life, thereby improving the accuracy of wind turbine life detection.

[0165] In some specific embodiments of this application, the loss function can be expressed by the following formula:

[0166]

[0167] in, This indicates the predicted lifespan data for wind turbines. The loss function is derived from the actual wind turbine lifespan data y, where n represents the number of samples of prior operating state data, and y i This represents the actual lifespan data of the i-th wind turbine. Let F(y) represent the i-th predicted wind turbine lifespan data, where λ represents the weight hyperparameter used to balance the contributions of the neural network loss function and the Weibull loss function in the loss function. i F(y) represents the prior life distribution data for the i-th real wind turbine life data. i ) represents the predicted life distribution data of the i-th predicted wind turbine life data.

[0168] In some embodiments, step S102 specifically refers to the target operating status data, which is the operating status data of the fan in the nuclear power plant that needs to be tested for fan life. This data can be collected from the sensors of the fan generator set in the nuclear power plant, such as the fan bearing temperature, motor temperature, vibration data, wind speed, fan blade speed, fan internal pressure data, torque and force on the fan shaft, lubricating oil viscosity, ambient humidity, and ambient temperature.

[0169] Reference Figure 6According to some embodiments of this application, before step S103 extracts features from the target operating state data to obtain the target operating state features, the wind turbine life detection method may also include, but is not limited to:

[0170] Step S601: Obtain initial running status data and perform data cleaning on the initial running status data to obtain cleaned running status data;

[0171] Step S602: Normalize the cleaning operation status data to obtain normalized operation status data;

[0172] Step S603: Divide the normalized operating status data into data segments based on the preset wind turbine operating time period to obtain the target operating status data.

[0173] In some embodiments of this application, by preprocessing the wind turbine's operating status data, the operating status data can be cleaned and normalized, thereby improving the data quality of the wind turbine's operating status data. Before step S103, the cleaning and normalization process includes the following key operations:

[0174] In some embodiments, step S601 specifically refers to the initial operating state data, which is data that has not been preprocessed, and the initial operating state data may come from various sensors of the fan (such as temperature sensors, vibration sensors, speed sensors, etc.).

[0175] Specifically, data cleaning aims to remove errors and inconsistencies from data to improve data quality. This includes identifying and handling missing values ​​in the initial operating status data, which can be done through interpolation or deleting wind turbine operation records containing missing values. Simultaneously, it involves identifying and handling outliers in the initial operating status data. These values ​​may be extreme values ​​caused by sensor malfunctions or external interference, which can be identified and processed using statistical methods. Furthermore, the initial operating status data undergoes operations such as removing duplicate records, correcting erroneous data, and formatting data to obtain cleaned operating status data.

[0176] In this embodiment, by cleaning the initial operating status data, it is ensured that the operating status data used for analysis is accurate and reliable, which facilitates the improvement of the accuracy of wind turbine life detection in the future.

[0177] In some embodiments, step S602 specifically involves normalization, which refers to scaling the cleaning operation status data to a specific range, typically 0 to 1 or -1 to 1, to obtain normalized operation status data.

[0178] In this embodiment, by normalizing the cleaning operation status data, it is possible to avoid data of different dimensions and magnitudes affecting the detection results of the subsequent wind turbine life detection model, and to ensure that each data has the same weight in the life detection process.

[0179] In some embodiments, step S603 specifically refers to the wind turbine operating time period, which is an analysis method based on wind turbine operating data of different time series. This method divides the wind turbine operating data of different time periods into windows of fixed size, with each window containing wind turbine operating status data for a certain period.

[0180] In this embodiment, the normalized operating status data is divided into data segments by a preset wind turbine operating time period. This facilitates the subsequent Fourier transform and frequency band division of the wind turbine operating status data for different time periods, improving the efficiency of feature extraction. It also helps the subsequent wind turbine life detection model to capture the features of wind turbine operating data for different time periods, thereby capturing the dynamic changes in wind turbine operating status and improving the efficiency of wind turbine life prediction.

[0181] The embodiments of this application provided through steps S601 to S603 provide a high-quality data foundation for subsequent feature extraction and wind turbine life detection models to detect the remaining life of wind turbines, which helps to improve the accuracy and reliability of wind turbine life detection.

[0182] Reference Figure 7 According to some embodiments of this application, step S103 extracts features from the target operating state data to obtain target operating state features, which may include, but are not limited to:

[0183] Step S701: Perform Fourier transform processing on the target running state features to obtain at least one running state spectrum feature;

[0184] Step S702: Select the operating state spectrum feature corresponding to the maximum amplitude from at least one operating state spectrum feature as the target operating state feature.

[0185] In some embodiments of this application, the target operating state features not only include detailed information about the wind turbine's operating state, but also achieve efficient extraction of wind turbine operating data features by focusing on the most significant part of the wind turbine amplitude signal. In step S103, this process includes the following key operations:

[0186] In some embodiments, step S701 specifically refers to the operating state spectrum features extracted from the target operating state data of the wind turbine through Fourier transform processing technology, which represent the characteristics of the wind turbine's vibration or fluctuation at different frequencies. The operating state spectrum features include, but are not limited to, the vibration frequencies of the wind turbine bearings, blades, and gears, as well as the amplitudes corresponding to the vibration frequencies.

[0187] Specifically, the time-domain signal is converted into a frequency-domain signal by performing a Fourier transform on the target's operating state data.

[0188] In this embodiment, by performing Fourier transform processing on the target operating state characteristics, at least one operating state spectrum characteristic is obtained. The operating state spectrum characteristic can be obtained from the wind turbine's operating data, which can reflect the dynamic characteristics of the wind turbine during operation and is key information for identifying the wind turbine's health status and predicting potential faults.

[0189] Reference Figure 8 According to some embodiments of this application, step S702, which selects the operating state spectrum feature corresponding to the maximum amplitude from at least one operating state spectrum feature as the target operating state feature, may include, but is not limited to:

[0190] Step S801: Divide the frequency band of at least one operating state spectral feature to obtain at least one operating state frequency band;

[0191] Step S802: Select the operating state frequency band corresponding to the maximum amplitude from at least one operating state frequency band as the target operating state feature.

[0192] In some embodiments of this application, the operating state spectrum feature corresponding to the largest amplitude is selected as the target operating state feature from at least one operating state spectrum feature. This allows the model to focus only on the most significant part of the wind turbine amplitude signal during wind turbine life detection, improving the efficiency of subsequent wind turbine life detection. In step S702, this process includes the following key operations:

[0193] In some embodiments, step S801 specifically refers to the operating frequency band, which in spectrum analysis divides the frequency range of a signal into several consecutive intervals or bands, each interval or band representing a specific frequency range.

[0194] For example, the spectral characteristics of the operating state can be divided into the low-frequency band, the mid-frequency band, and the high-frequency band of the operating state.

[0195] Furthermore, at least one operating state spectral characteristic is divided into frequency bands, that is, the frequency range in the spectrum is divided into several frequency bands, each representing a specific frequency interval of the wind turbine's operating state. The frequency band division can be based on the wind turbine's physical characteristics, common failure modes, or the needs of spectrum analysis, with the aim of simplifying complex spectrum data into a more easily analyzed and understood frequency band structure.

[0196] In some embodiments, step S802 specifically refers to the target operating state feature, which is key information extracted from the wind turbine's operating data that can represent the wind turbine's current operating state.

[0197] For example, the target operating status characteristics can be the vibration level, temperature, pressure, and speed of each fan in a nuclear power plant.

[0198] Specifically, the operating state frequency band corresponding to the maximum amplitude is selected from at least one operating state frequency band as the target operating state feature. The purpose of this step is to identify the frequency ranges with the most concentrated energy in the spectrum, which may be closely related to the wind turbine's key operating data or potential faults. The frequency band with the maximum amplitude may indicate the most significant vibration signals during wind turbine operation, which may be caused by normal operation or potential faults of the wind turbine.

[0199] The embodiments of this application shown via steps S801 to S802 provide crucial data support for subsequent fault diagnosis and life prediction of wind turbines by highlighting the most significant vibration signals in each operating frequency band.

[0200] As shown in the embodiments of this application via steps S701 to S702, the target operating state characteristics not only include time-domain information of wind turbine operation, but also frequency-domain information, which facilitates the subsequent wind turbine life detection model to learn the operating state characteristics of the wind turbine from multiple perspectives, thereby improving the accuracy and reliability of wind turbine life prediction.

[0201] In some specific embodiments of this application, the operating state spectrum can be divided into a low-frequency band (0-10Hz), a mid-frequency band (10-50Hz), and a high-frequency band (50-100Hz). If the amplitude of 30Hz in the mid-frequency band is significantly higher than that of other frequencies, it may indicate that the bearing components of the wind turbine have significant vibration signals at this frequency. This may be due to the specific operating mechanism of the wind turbine or potential bearing failure. Taking the maximum amplitude corresponding to each frequency band as the target operating state feature makes it easier to predict the maintenance needs and remaining service life of the wind turbine if the amplitude of 30Hz is highly correlated with the early bearing failure of the wind turbine in the subsequent wind turbine life detection model training.

[0202] Reference Figure 9 According to some embodiments of this application, step S104 performs wind turbine life detection on the target operating state characteristics using a pre-trained wind turbine life detection model to obtain target wind turbine life data, which may include, but is not limited to:

[0203] Step S901: The wind turbine health status is detected by the target operating status characteristics through the wind turbine life detection model to obtain wind turbine health status data;

[0204] Step S902: Based on the wind turbine health status data, the remaining lifespan of the target wind turbine is detected according to the characteristics of the target operating status, and the lifespan data of the target wind turbine is obtained.

[0205] In some embodiments of this application, a pre-trained wind turbine life detection model is used to detect the target operating state characteristics of the wind turbine. This eliminates the need for manual selection of feature indicators for wind turbine operating data and incorporates the operating state of the wind turbine at different times during the life detection process, effectively improving the accuracy of wind turbine life detection. In step S104, this process includes the following key operations:

[0206] Step S901 in some embodiments, refer to Figure 10 According to some embodiments of this application, step S901 uses a wind turbine lifespan detection model to detect the wind turbine health status of the target operating state characteristics and obtain wind turbine health status data, which may include, but is not limited to:

[0207] Step S1001: Extract wind turbine operating health features from the target operating status features using the wind turbine life detection model to obtain wind turbine operating health features;

[0208] Step S1002: Evaluate the health status of the wind turbine based on the preset linear function to obtain a wind turbine health assessment score.

[0209] Step S1003: Determine the health status data of the wind turbine based on the wind turbine health assessment score.

[0210] In some embodiments of this application, the wind turbine health status is detected by using a wind turbine lifespan detection model to assess the target operating state characteristics. This quantifies the wind turbine health status detection process and effectively measures the remaining lifespan of subsequent wind turbines. In step S901, this process includes the following key operations:

[0211] Step S1001 in some embodiments, refer to Figure 11 According to some embodiments of this application, in step S1001, the wind turbine operating health features are extracted from the target operating state features using the wind turbine life detection model to obtain wind turbine operating health features, which may include, but are not limited to:

[0212] Step S1101: The target operating state characteristics are encoded using the wind turbine life detection model to obtain the encoded operating characteristics;

[0213] Step S1102: Perform health feature weighting on the encoded running features to obtain weighted health running features;

[0214] Step S1103: Decode the weighted health operation characteristics to obtain the wind turbine operation health characteristics.

[0215] In some embodiments of this application, by extracting wind turbine operating health characteristics, features irrelevant to lifespan detection can be removed from the wind turbine, and features indicative of the wind turbine's health status can be extracted, thereby improving the accuracy of subsequent wind turbine lifespan detection. In step S1001, this process includes the following key operations:

[0216] In some embodiments, step S1101 specifically refers to encoding operational features, which means representing target operational state features through a fixed-dimensional vector space.

[0217] Specifically, the wind turbine life detection model includes an embedding layer and a bidirectional GRU (Bidirectional Gated Recurrent Unit) network layer. The embedding layer and the bidirectional GRU network layer enable the encoding of target operating state characteristics. Each GRU contains an update gate, a reset gate, and a candidate activation mechanism.

[0218] Furthermore, firstly, the target running state features at each time step are converted into fixed-dimensional vectors through an embedding layer to map the target running state features to a high-dimensional space. Secondly, in the forward GRU, the embedded vector data is processed according to the time series from past to future. The update gate determines the vector data to be retained in the past, the reset gate determines the vector data to be discarded in the past, and the candidate mechanism is used to generate new candidate states for the target running state features. The new candidate states combine the input of the current time series and the gated vector data to form the forward running features. Then, in the backward GRU, for the forward running features, the update gate, reset gate, and candidate mechanism are combined according to the time series from the future to the past to determine the forward running features to be retained or discarded, and the backward running features are output. Finally, the forward running features and the backward running features are integrated to output the encoded running features.

[0219] In this embodiment, the target operating status features are encoded by the wind turbine life detection model, providing a standardized input representation for subsequent capture of time series related information. Combined with bidirectional GRU, the model can capture the operating features of the time series from two directions and can simultaneously consider the information before and after each operating feature in the time series. This provides a rich information foundation for subsequent health status assessment and wind turbine life prediction, enabling the wind turbine life detection model to more accurately predict the health status and remaining service life of the wind turbine.

[0220] In some embodiments, step S1102 specifically involves the weighted health operation features reflecting the degree of influence of each coded operation feature on the identification of the wind turbine's health status. The weighted health operation features not only include information from the coded operation features but also incorporate the importance of each time step in different time series.

[0221] Specifically, the wind turbine life distribution model uses an attention mechanism to identify coded operating features that make significant contributions to assessing the health status of wind turbines. Different weights are assigned to different coded operating features to reflect the degree of contribution of the coded operating features to the health status of wind turbines. This achieves the combination of attention weights and coded operating features to generate weighted health operating features.

[0222] For example, if the wind turbine life detection model identifies that the peak vibration of the wind turbine is closely related to the health status of the wind turbine over a period of time, then the coded operating characteristics of that period will be given greater weight.

[0223] In this embodiment, by applying health feature weighting to the coded operating features, the model can focus on the coded operating features that have a significant impact on the health status of the wind turbine, thereby more accurately identifying potential faults in the wind turbine and facilitating the improvement of the accuracy of wind turbine life detection in the future.

[0224] In some embodiments, step S1103 specifically refers to the characteristics related to the degree of wind turbine operating health.

[0225] For example, the vibration frequency, fan speed, generator power, and sound signals of the fan during operation.

[0226] Specifically, the wind turbine life detection model includes a decoding layer. The decoding layer determines the hidden state of the current time step by using the hidden state of the previous time step output by the bidirectional GRU, the output of the previous time step, and the weighted healthy operation features. The hidden state is then activated by the ReLU function to obtain the wind turbine's operating health features.

[0227] In this embodiment, by decoding the weighted healthy operation features, the wind turbine operating health features are obtained. This can convert the weighted healthy operation features into wind turbine health status features, highlight the features that have the greatest impact on the wind turbine health status, and take into account the impact of each time step, which makes it easier to determine the specific time point when the wind turbine fails and improves the accuracy of wind turbine life detection.

[0228] Through steps S1101 to S1103 shown in the embodiments of this application, the wind turbine life detection model can extract indicative features of health status from the original target operating state features, and transform these features into a form that the model can use. It also considers the impact of each time step, which facilitates the subsequent determination of the specific time point when the wind turbine fails, thereby providing a scientific basis for the maintenance and life prediction of the wind turbine.

[0229] In some embodiments, step S1002 specifically involves using a wind turbine health assessment score to measure the operational health of the wind turbine at different points in time.

[0230] Specifically, this linear function can calculate the weights for each wind turbine's operational health feature, and finally, combined with the bias term, it is used to transform multiple wind turbine operational health features into a quantifiable health assessment score.

[0231] Furthermore, by using a linear function to weight and sum the vibration level, operating temperature, and speed of each component of the fan, a comprehensive fan health assessment score for each component is calculated. The fan health assessment score can intuitively reflect the health status of the fan; the higher the score, the lower the probability of the fan failing.

[0232] Wind turbine health assessment scores provide a quantitative reference for turbine maintenance and lifespan prediction, enabling operators to optimize turbine operation and maintenance strategies based on data-driven approaches. This method combines data analytics and domain knowledge, providing an effective tool for intelligent wind turbine management.

[0233] In this embodiment, the health status of each wind turbine can be evaluated based on a preset linear function, resulting in a wind turbine health assessment score. The wind turbine health assessment score provides a quantitative indicator for wind turbine life prediction. The higher the score, the better the health status of the wind turbine. The lower the score, the more intuitively the possibility of wind turbine failure, which helps to improve the efficiency of subsequent wind turbine life prediction.

[0234] In some embodiments, step S1003 specifically involves wind turbine health data reflecting the current health characteristics of each component of the wind turbine, used to identify the possibility of wind turbine failure.

[0235] Specifically, by comparing the wind turbine health assessment score with a preset health threshold, the degree of failure of different components of the wind turbine can be determined. Furthermore, if the wind turbine health assessment score is less than the health threshold, it indicates that there is a risk of failure in different components of the wind turbine. If the wind turbine health assessment score is greater than or equal to the health threshold, it indicates that the wind turbine can continue to operate normally.

[0236] In this embodiment, the health status data of the wind turbine is determined based on the wind turbine health assessment score, realizing the quantitative assessment of wind turbine fault detection. This enables intuitive monitoring of the wind turbine's operating status and provides an effective method for intelligent management of wind turbines.

[0237] Through steps S1001 to S1003 shown in the embodiments of this application, the wind turbine life detection model can intuitively display the health status of the wind turbine and provide a data basis for subsequent prediction of the remaining service life of the wind turbine, which helps to improve the efficiency of subsequent wind turbine life detection.

[0238] Step S902 in some embodiments, refer to Figure 12 According to some embodiments of this application, step S902, which involves detecting the remaining lifespan of the wind turbine based on the wind turbine health status data to obtain the target wind turbine lifespan data, may include, but is not limited to:

[0239] Step S1201: Obtain the health weight and health bias of the wind turbine health data;

[0240] Step S1202: Based on the health level weight, health level bias term and wind turbine health level data, perform wind turbine operation fault detection on the target operating status characteristics to obtain wind turbine operation fault data;

[0241] Step S1203: Determine the target wind turbine lifespan data based on the wind turbine operation fault data.

[0242] In some embodiments of this application, detecting the remaining lifespan of a wind turbine based on its health status data involves analyzing the wind turbine's faults within the target operating state characteristics to determine its health status and predict its remaining lifespan. In step S902, this process includes the following key operations:

[0243] In some embodiments, step S1201 specifically involves the health level weight reflecting the degree of influence of different component characteristics of the wind turbine on the overall health status of the wind turbine, while the bias term is used to adjust the influence of the wind turbine health level data on the overall health status of the wind turbine, ensuring that the wind turbine health level data can accurately reflect the actual health status of the wind turbine as a whole.

[0244] In this embodiment, by acquiring the health weights and health biases of the wind turbine health data, the wind turbine health data can be quantified. Furthermore, by combining the weights and biases, a comprehensive assessment of the wind turbine health data can be achieved, which helps to improve the accuracy of subsequent wind turbine life prediction.

[0245] In some embodiments, step S1202 specifically refers to the wind turbine operation fault data, which refers to fault characteristics that pose a risk of failure to wind turbine operation.

[0246] Specifically, by weighting and summing the health status data of each wind turbine during operation, and identifying signs of wind turbine failure risk or performance degradation, data features related to wind turbine failure can be determined.

[0247] This involves analyzing the wind turbine's operating data to identify potential failure modes or signs of performance degradation. For example, a sudden increase in the turbine's vibration level could be an early warning sign of a malfunction. This analysis yields wind turbine operational fault data that reveals potential problems that may arise during operation.

[0248] In this embodiment, the target operating status characteristics are analyzed to detect wind turbine operation faults based on health level weights, health level bias terms, and wind turbine health level data. This combination of the current operating health status of the wind turbine and fault data during operation provides comprehensive data support for wind turbine life prediction.

[0249] In some embodiments, step S1203 specifically refers to the target wind turbine life data, which is the current remaining life data of the wind turbine.

[0250] Specifically, the target lifespan data of the wind turbine can be determined by activating the wind turbine operation fault data through the Sigmoid function of the wind turbine life detection model.

[0251] Furthermore, if the wind turbine's operational failure data shows multiple signs of performance degradation, it may mean that the wind turbine has a short remaining lifespan and requires more frequent maintenance or possible replacement.

[0252] In this embodiment, the target wind turbine life data is determined based on the wind turbine operation failure data, which provides important information for wind turbine health management and life prediction, and makes it easier for nuclear power plants to make more accurate wind turbine maintenance or replacement decisions based on the target wind turbine life data.

[0253] By taking into account the current health status of the wind turbine and the fault characteristics during its operation through steps S1201 to S1203 in the embodiments of this application, comprehensive data support is provided for wind turbine life prediction, which effectively improves the accuracy of wind turbine life prediction and makes it easier for nuclear power plants to make more accurate wind turbine maintenance or replacement decisions based on the target wind turbine life data.

[0254] By using steps S901 to S902 provided in the embodiments of this application, the health status data of the wind turbine is automatically identified through the life detection model, eliminating the need for manual selection of characteristic indicators of the wind turbine operation data. Furthermore, the wind turbine's operating status at different times is incorporated into the life detection, effectively improving the accuracy of the wind turbine life detection.

[0255] In some specific embodiments of this application, if the target operating state characteristics are the temperature, vibration, rotational speed, wind speed, power, etc. of various components during wind turbine operation, the health features are weighted and decoded to output the characteristics related to the health status of the wind turbine operation, such as the wind turbine vibration level, bearing temperature, and wind turbine rotational speed. A health score is assigned to these characteristics. For example, if the vibration level is 0.05g, the bearing temperature is 85℃, and the wind turbine rotational speed is 1500RPM, these characteristics are weighted using a wind turbine lifespan detection model to assess the current health score of the wind turbine as 85 points. Based on this current health score and the operating health characteristics, the weights and biases of the health status are calculated. Assuming the weight is 0.7 and the bias is 5, the bias and weight are activated using the Sigmoid function, resulting in a remaining lifespan of 8 years for the wind turbine.

[0256] It should be noted that the embodiments of this application first obtain a pre-trained wind turbine life detection model, which can understand and identify the complex relationship between wind turbine operating status and wind turbine life. Moreover, the pre-trained model does not require collecting and analyzing data from scratch when predicting wind turbine life, thus improving the efficiency of wind turbine life detection. It also obtains the target operating status data of the wind turbine and extracts the actual data of the current wind turbine operation, providing data support for subsequent wind turbine life detection. Secondly, by extracting features from the target operating status data, the target operating status features are obtained, which can select key features related to wind turbine life detection, making it easier for subsequent support models to improve the accuracy of wind turbine life detection. Finally, the wind turbine life detection is performed on the target operating status features through the pre-trained wind turbine life detection model to obtain the target wind turbine life data. There is no need to manually select feature indicators of wind turbine operating data, and the wind turbine operating status at different times is combined in the wind turbine life detection, which effectively improves the accuracy of wind turbine life detection.

[0257] Reference Figure 13 The wind turbine life testing device according to the second aspect of this application may include, but is not limited to:

[0258] The wind turbine life detection model acquisition module 1301 is used to acquire a pre-trained wind turbine life detection model.

[0259] The wind turbine operating status data acquisition module 1302 is used to acquire the target operating status data of the wind turbine.

[0260] Feature extraction module 1303 is used to extract features from target running state data to obtain target running state features;

[0261] The wind turbine life detection module 1304 is used to detect the life of the target wind turbine by using a pre-trained wind turbine life detection model to detect the characteristics of the target operating state and obtain the life data of the target wind turbine.

[0262] It is evident that the content of the above-described wind turbine life testing method embodiments is applicable to the embodiments of this wind turbine life testing device. The specific functions implemented by this wind turbine life testing device embodiment are the same as those of the above-described wind turbine life testing method embodiments, and the beneficial effects achieved are also the same as those achieved by the above-described wind turbine life testing method embodiments.

[0263] Reference Figure 14 , Figure 14 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0264] The processor 1401 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0265] The memory 1402 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1402 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1402 and is called and executed by the processor 1401 to execute the wind turbine life detection method of the embodiments of this application.

[0266] The input / output interface 1403 is used to implement information input and output;

[0267] The communication interface 1404 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0268] Bus 1405 transmits information between various components of the device (e.g., processor 1401, memory 1402, input / output interface 1403, and communication interface 1404);

[0269] The processor 1401, memory 1402, input / output interface 1403 and communication interface 1404 are connected to each other within the device via bus 1405.

[0270] This application also provides a computer program product, which includes a computer program. The processor of a computer device reads and executes the computer program, causing the computer device to perform the aforementioned wind turbine life detection method.

[0271] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in this disclosure and the foregoing drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “including,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatuses.

[0272] It should be understood that in this disclosure, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, which may include, but is not limited to, any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0273] It should be understood that in the description of the embodiments of this application, "multiple" means two or more, "greater than", "less than", "exceeding" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself.

[0274] In the several embodiments provided in this disclosure, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0275] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0276] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0277] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and may include, but is not limited to, several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this disclosure. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0278] It should also be understood that the various implementation methods provided in this application can be combined arbitrarily to achieve different technical effects.

[0279] The above is a detailed description of the embodiments of this disclosure. However, this disclosure is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this disclosure. All such equivalent modifications or substitutions are included within the scope defined by the claims of this disclosure.

Claims

1. A method for detecting the life of a fan, characterized by, include: Obtain a pre-trained wind turbine life detection model; wherein, the pre-training process of the wind turbine life detection model includes: Obtain prior operating status data of the wind turbine; The wind turbine lifespan distribution is predicted by performing a preset Weibull cumulative distribution function on the prior operating state data, resulting in prior lifespan distribution data, including: The wind turbine shape characterization parameters are determined based on the prior operating state data. Based on the wind turbine shape characterization parameters and the prior operating state data, determine the wind turbine dimensional characterization parameters; Based on the wind turbine shape characterization parameters and the wind turbine size characterization parameters, the wind turbine lifetime distribution is predicted from the prior operating state data to obtain the prior lifetime distribution data; The initial wind turbine life detection model is trained based on the prior life distribution data and the prior operating state data to obtain a pre-trained wind turbine life detection model, including: Obtain the initial wind turbine life detection model; The wind turbine lifespan is predicted by using the wind turbine lifespan detection model to analyze the prior operating state data and obtain the predicted wind turbine lifespan data. The predicted lifespan distribution data is obtained by using the Weibull cumulative distribution function to predict the lifespan distribution of the predicted wind turbine lifespan data. Obtain the actual wind turbine lifespan data from the prior operating state data, and calculate the target loss value of the predicted wind turbine lifespan data, the predicted lifespan distribution data, the actual wind turbine lifespan data, and the prior lifespan distribution data according to a preset loss function, including: The loss function includes the neural network loss function and the Weibull loss function; The life loss value between the predicted wind turbine life data and the actual wind turbine life data is calculated based on the neural network loss function. The lifetime distribution loss value between the predicted lifetime distribution data and the prior lifetime distribution data is calculated based on the Weibull loss function. The target loss value is obtained by integrating the lifetime loss value and the lifetime distribution loss value through preset weight hyperparameters. The model parameters of the wind turbine life detection model are updated based on the target loss value. The wind turbine life detection model is then used to predict the wind turbine life based on the prior operating state data until the wind turbine life detection model meets the preset training conditions, thus obtaining the pre-trained wind turbine life detection model. Acquire the target operating status data of the wind turbine; Feature extraction is performed on the target operating state data to obtain the target operating state features; The wind turbine lifespan is detected by the pre-trained wind turbine lifespan detection model based on the target operating state characteristics, resulting in target wind turbine lifespan data, including: The wind turbine health status is detected by the target operating state characteristics using the wind turbine life detection model, and wind turbine health status data is obtained. Based on the wind turbine health status data, the remaining lifespan of the target wind turbine is detected according to the target operating status characteristics, and the lifespan data of the target wind turbine is obtained.

2. The method of claim 1, wherein, The step of determining the wind turbine dimensional characterization parameters based on the wind turbine shape characterization parameters and the prior operating state data includes: Obtain the wind turbine failure time, number of wind turbine failures, and wind turbine failure variables from the prior operating status data; The scale characterization parameters of the wind turbine are obtained by performing scale characterization calculations on the wind turbine failure time, the number of wind turbine failures, the wind turbine failure variables and the wind turbine shape characterization parameters.

3. The method of claim 1, wherein, The step of detecting the wind turbine health status by using the wind turbine lifespan detection model to identify the target operating state characteristics and obtaining wind turbine health status data includes: The wind turbine operating health features are extracted from the target operating status features using the wind turbine life detection model. The health status of the wind turbine is assessed based on a preset linear function to obtain a wind turbine health assessment score. The health status data of the wind turbine is determined based on the wind turbine health assessment score.

4. The method of claim 3, wherein, The step of extracting wind turbine operating health features from the target operating state features using the wind turbine life detection model to obtain wind turbine operating health features includes: The target operating state characteristics are encoded using the wind turbine life detection model to obtain the encoded operating characteristics; The encoded running features are weighted by health features to obtain weighted healthy running features; The weighted health operation characteristics are decoded to obtain the wind turbine operation health characteristics.

5. The method of claim 1, wherein, The step of detecting the remaining lifespan of the target wind turbine based on the wind turbine health status data to obtain the target wind turbine lifespan data includes: Obtain the health weight and health bias term of the wind turbine health data; Based on the health status weight, the health status bias term and the wind turbine health status data, wind turbine operation fault detection is performed on the target operating status characteristics to obtain wind turbine operation fault data. Based on the wind turbine operation fault data, the lifespan data of the target wind turbine is determined.

6. The method of claim 1, wherein, The step of extracting features from the target operating state data to obtain target operating state features includes: Perform Fourier transform processing on the target operating state features to obtain at least one operating state spectrum feature; The operating state spectral feature corresponding to the maximum amplitude is selected from at least one of the operating state spectral features as the target operating state feature.

7. The method of claim 6, wherein, The step of selecting the operating state spectrum feature corresponding to the maximum amplitude from at least one of the operating state spectrum features as the target operating state feature includes: Divide at least one of the operating state spectral features into frequency bands to obtain at least one operating state frequency band; The operating state frequency band corresponding to the maximum amplitude is selected from at least one of the operating state frequency bands as the target operating state feature.

8. The method of claim 7, wherein, Before extracting features from the target operating state data to obtain the target operating state features, the method further includes: Acquire initial running status data and perform data cleaning on the initial running status data to obtain cleaned running status data; The cleaning operation status data is normalized to obtain normalized operation status data; The normalized operating status data is divided into data segments by using a preset wind turbine operating time period to obtain the target operating status data.

9. A blower life detection apparatus, characterized by, A method for implementing any one of claims 1 to 8 includes: A wind turbine life detection model acquisition module is used to acquire a pre-trained wind turbine life detection model; wherein, the pre-training process of the wind turbine life detection model includes: Obtain prior operating status data of the wind turbine; The wind turbine lifespan distribution is predicted by performing a preset Weibull cumulative distribution function on the prior operating state data, resulting in prior lifespan distribution data, including: The wind turbine shape characterization parameters are determined based on the prior operating state data. Based on the wind turbine shape characterization parameters and the prior operating state data, determine the wind turbine dimensional characterization parameters; Based on the wind turbine shape characterization parameters and the wind turbine size characterization parameters, the wind turbine lifetime distribution is predicted from the prior operating state data to obtain the prior lifetime distribution data; The initial wind turbine life detection model is trained based on the prior life distribution data and the prior operating state data to obtain a pre-trained wind turbine life detection model, including: Obtain the initial wind turbine life detection model; The wind turbine lifespan is predicted by using the wind turbine lifespan detection model to analyze the prior operating state data and obtain the predicted wind turbine lifespan data. The predicted lifespan distribution data is obtained by using the Weibull cumulative distribution function to predict the lifespan distribution of the predicted wind turbine lifespan data. Obtain the actual wind turbine lifespan data from the prior operating state data, and calculate the target loss value of the predicted wind turbine lifespan data, the predicted lifespan distribution data, the actual wind turbine lifespan data, and the prior lifespan distribution data according to a preset loss function, including: The loss function includes the neural network loss function and the Weibull loss function; The life loss value between the predicted wind turbine life data and the actual wind turbine life data is calculated based on the neural network loss function. The lifetime distribution loss value between the predicted lifetime distribution data and the prior lifetime distribution data is calculated based on the Weibull loss function. The target loss value is obtained by integrating the lifetime loss value and the lifetime distribution loss value through preset weight hyperparameters. The model parameters of the wind turbine life detection model are updated based on the target loss value. The wind turbine life detection model is then used to predict the wind turbine life based on the prior operating state data until the wind turbine life detection model meets the preset training conditions, thus obtaining the pre-trained wind turbine life detection model. The wind turbine operating status data acquisition module is used to acquire the target operating status data of the wind turbine; The feature extraction module is used to extract features from the target running state data to obtain target running state features; The wind turbine life detection module is used to detect the wind turbine life of the target operating state characteristics through the pre-trained wind turbine life detection model, and obtain the target wind turbine life data. The wind turbine life detection module specifically includes: The wind turbine health status is detected by the target operating state characteristics using the wind turbine life detection model, and wind turbine health status data is obtained. Based on the wind turbine health status data, the remaining lifespan of the target wind turbine is detected according to the target operating status characteristics, and the lifespan data of the target wind turbine is obtained.

10. An electronic device, comprising: include: The device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the wind turbine life detection method as described in any one of claims 1 to 8.

11. A computer readable storage medium, characterized in that, The storage medium stores a program, which is executed by a processor to implement the wind turbine life detection method as described in any one of claims 1 to 8.

Images