Wind turbine fault early warning modeling method and device, electronic equipment and storage medium

Abstract

The application provides a wind turbine fault early warning modeling method and device, electronic equipment and storage medium. Wherein, the wind turbine fault early warning modeling method comprises obtaining the parameters of the to-be-alarmed unit component corresponding to the early warning model and the data of the parameters of the to-be-alarmed unit component; inputting the data of the parameters of the to-be-alarmed unit component into the early warning model, training the early warning model, obtaining the trained early warning model, and obtaining the fault item of the to-be-alarmed unit component. In this way, the professional threshold of the early warning model is reduced, and the user operation is simple.

Description

Technical Field

[0001] This invention relates to the field of wind turbine technology, and in particular to a method, apparatus, electronic device and storage medium for wind turbine fault early warning modeling. Background Technology

[0002] With the rapid advancement of artificial intelligence technology, the intelligent operation and maintenance model of wind farms, characterized by "unmanned operation and minimal staffing," is gradually being put on the agenda.

[0003] In related technologies, users need to learn the professional knowledge of the model in advance, master the operation of the model, and understand the data required by the model before using it. This results in a high professional threshold for users to use and learn the model, and a high level of complexity in user operation. Summary of the Invention

[0004] This application provides a method, device, electronic equipment, and storage medium for modeling fault early warning of wind turbine units. The method lowers the professional threshold for early warning models and is easy for users to operate.

[0005] This application provides a method for modeling fault early warning of wind turbine generators, including:

[0006] Acquire the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, and the data of the parameters of the components of the unit to be warned;

[0007] The parameter data of the unit component to be warned is input into the warning model, the warning model is trained, and the trained warning model and the fault entries of the unit component to be warned are obtained.

[0008] Furthermore, the acquisition of parameters of the pre-set target unit components corresponding to the early warning model and the data of the target unit components includes:

[0009] Obtain the training conditions selected by the user, including the selected wind field, the selected early warning model, and the selected training time period;

[0010] Based on the selected early warning model, determine the parameters of the pre-set components of the unit to be warned that correspond to the selected early warning model;

[0011] Based on the selected wind field and the selected training time period, data on the parameters of the components of the unit to be warned are obtained.

[0012] Furthermore, obtaining the training conditions selected by the user includes:

[0013] Receive training instructions input through the user interface;

[0014] In response to the training instruction, candidate training conditions are displayed on the user interface;

[0015] Based on the selection instruction for the candidate training conditions received through the user interface, the candidate training conditions selected by the selection instruction are obtained as the training conditions selected by the user.

[0016] Furthermore, the early warning model includes an early warning model for preset unit components, the parameters of the unit components to be warned include the parameters of the preset unit components within a preset time period in a preset wind field, and the data of the parameters of the unit components to be warned are the data of the parameters of the preset unit components.

[0017] Furthermore, the post-training early warning model includes one or more of the following: a post-training early warning model for the pitch system, a post-training early warning model for the converter, a post-training early warning model for the main control system, a post-training early warning model for the hydraulic system, a post-training early warning model for the tower, a post-training early warning model for the gearbox system, a post-training early warning model for the power grid equipment, and a post-training early warning model for the generator system.

[0018] Furthermore, after inputting the parameter data of the unit component to be warned into the warning model, training the warning model to obtain the trained warning model, and obtaining the fault entries for the unit component to be warned, the wind turbine fault warning modeling method further includes:

[0019] Generate model information for the trained early warning model, the model information including one or more of the following: number, name, version number, early warning code, and the unit component that issued the early warning;

[0020] The model information is stored according to the trained early warning model.

[0021] Furthermore, after inputting the parameter data of the unit component to be warned into the warning model, training the warning model, and obtaining the trained warning model and the fault entries of the unit component to be warned, the wind turbine fault warning modeling method further includes:

[0022] The trained early warning model is packaged using Docker to obtain a Docker package;

[0023] Send the Docker package to the fault early warning system.

[0024] Furthermore, the early warning model includes an early warning model in an active state;

[0025] Before acquiring the parameters of the wind turbine components to be warned and the data of the parameters of the wind turbine components to be warned, the wind turbine fault early warning modeling method further includes:

[0026] Receive an activation command input through a user interface, the activation command carrying indication information of a warning model in a pending activation state;

[0027] The activation command is parsed to obtain the instruction information;

[0028] In response to the activation command, obtain the early warning model of the pending activation state;

[0029] Activate the warning model in the pending state to obtain the warning model in the activated state.

[0030] Furthermore, the fault entries for the warning include the unit number, the warning time, and the unapproved status;

[0031] After inputting the parameter data of the unit component to be warned into the warning model, training the warning model to obtain the trained warning model, and obtaining the fault entries for the unit component to be warned, the method further includes:

[0032] Display the warning fault entries on the user interface;

[0033] Based on the review instruction received through the user interface for the unreviewed state, modify the warning fault entry to obtain the correct warning fault entry;

[0034] Issue the correct fault entries to the wind farm as early warnings.

[0035] Furthermore, after inputting the parameter data of the unit component to be warned into the warning model, training the warning model to obtain the trained warning model, and obtaining the fault entries for the unit component to be warned, the method further includes:

[0036] Obtain the running data of the trained model during the assessment period;

[0037] Based on the operational data, obtain the correct fault entries for the early warning within the assessment period;

[0038] The operational accuracy of the trained model is determined based on the correct fault entries and the fault entries warned during the assessment period.

[0039] Furthermore, the step of inputting the parameter data of the components of the unit to be warned into the warning model, and training the warning model to obtain the trained warning model includes:

[0040] Data cleaning is performed on the parameter data of the components of the unit to be warned, resulting in cleaned parameter data;

[0041] Select normal data from the parameters after cleaning;

[0042] The normal data is normalized to obtain normalized data;

[0043] Based on the normalized data, a training dataset is obtained;

[0044] The method further includes training the early warning model based on the training dataset to obtain a trained early warning model; after training the early warning model based on the training dataset to obtain the trained early warning model, the method further includes:

[0045] If the warning accuracy of the post-trained warning model meets the preset threshold, the post-trained warning model is used as the trained warning model.

[0046] This application provides a wind turbine fault early warning modeling device, comprising:

[0047] The acquisition module is used to acquire the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, and the data of the parameters of the components of the unit to be warned.

[0048] The modeling module is used to input the parameter data of the unit component to be warned into the warning model, train the warning model, obtain the trained warning model, and obtain the fault entries of the unit component to be warned.

[0049] This application provides an electronic device, including a processor and a memory;

[0050] Memory, used to store computer programs;

[0051] A processor, when executing a program stored in memory, implements any of the methods described above.

[0052] This application provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the method described above.

[0053] In some embodiments, the wind turbine fault early warning modeling method of this application can obtain the parameter data of the preset components of the turbine to be warned corresponding to the early warning model, without the need for manual input of the parameter data of the components of the turbine to be warned, reducing the professional threshold of the early warning model and simplifying user operation. Attached Figure Description

[0054] Figure 1 The diagram shown is a flowchart illustrating the wind turbine fault early warning modeling method provided in an embodiment of this application.

[0055] Figure 2 As shown Figure 1 The flowchart of step 110 in the wind turbine fault early warning modeling method is shown below;

[0056] Figure 3 As shown Figure 1 The flowchart shown is a schematic diagram of the issuance of correct fault entries for early warning after step 120 in the wind turbine fault early warning modeling method.

[0057] Figure 4 As shown Figure 1 The flowchart shown is a schematic diagram of the operation accuracy evaluation after step 120 in the wind turbine fault early warning modeling method.

[0058] Figure 5 The diagram shown is a schematic diagram of the module of the wind turbine fault early warning modeling device provided in the embodiment of this application;

[0059] Figure 6 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application. Detailed Implementation

[0060] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0061] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0062] To address the technical challenge of high technical barriers for users to use and learn the model, this application provides a wind turbine fault early warning modeling method. This method can acquire data on the parameters of the pre-set components of the turbine to be warned, which correspond to the early warning model. It eliminates the need for manual input of the parameters of the components to be warned, reducing the technical barriers to the early warning model, making the model less complex to use, simpler for users, and improving the user experience.

[0063] Figure 1 The diagram shown is a flowchart of the wind turbine fault early warning modeling method provided in an embodiment of this application.

[0064] The wind turbine fault early warning modeling method may include the following steps 110 to 120:

[0065] Step 110: Obtain the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, and the data of the parameters of the components of the unit to be warned.

[0066] The early warning model is developed based on years of fault early warning experience and models from wind farm operations. Models with high maturity and accuracy are selected as references. Thus, the early warning model can refer to a pre-configured model of the wind turbine components of the wind farm. These early warning models have pre-configured parameters for the components of the turbines to be warned.

[0067] When using these early warning models, data on the parameters of the pre-configured generator components to be warned are acquired. This parameter data reflects the data required for modeling. For example, the parameters of the generator components to be warned may include the temperature of the generator drive end bearing, but this is not a limitation.

[0068] There are several ways to acquire the parameter data of the aforementioned unit components subject to early warning. In one method, the parameter data of the unit components subject to early warning is acquired based on the early warning model selected by the user. In another method, the parameters of the unit components subject to early warning are acquired automatically. Detailed explanations are as follows.

[0069] Figure 2 As shown Figure 1 The flowchart of step 110 in the wind turbine fault early warning modeling method is shown.

[0070] In such Figure 2In some embodiments of step 110 shown above, step 110 may further include step 111, obtaining the training conditions selected by the user. The selected training conditions include a selected wind farm, a selected early warning model, and a selected training time period. The selected training conditions reflect the user's training needs. The selected training conditions refer to the training conditions related to the training model chosen by the user. The user selects the wind farm to be trained as the selected wind farm, the early warning model to be trained as the selected early warning model, and the training time of the early warning model in the wind farm to be trained as the selected training time period. Step 112, based on the selected early warning model, obtain the parameters of the pre-set turbine components to be warned corresponding to the selected early warning model. Step 113, based on the selected wind farm and the selected training time period, obtain the parameter data of the turbine components to be warned. Thus, the user does not need to learn the early warning model or which parameters of the turbine components to be warned should be used in the early warning model; they only need to select the wind farm, the early warning model, and the training time period, and the parameter data of the turbine components to be warned within the selected wind farm and the selected training time period will be automatically obtained subsequently. This early warning model has lower complexity, is simpler to operate, and improves the user experience.

[0071] The process of obtaining the user-selected training conditions includes the following three steps: First, receiving the training instruction input through the user interface. Second, responding to the training instruction, displaying candidate training conditions on the user interface. Third, based on the selection instruction received through the user interface, obtaining the candidate training conditions selected by the selection instruction, and using this as the user-selected training condition. The training instruction is an indication to start training. Candidate training conditions include at least one wind farm, at least one early warning model for at least one unit component, and at least one training time period. This allows the user to easily determine the selected training condition from these candidate conditions.

[0072] The early warning model may include one or more of the following: early warning model for blades, early warning model for pitch system, early warning model for converter, early warning model for main control system, early warning model for hydraulic system, early warning model for tower, early warning model for gearbox system, early warning model for power grid equipment, and early warning model for generator system.

[0073] The training conditions selected by the user can include a wind farm, an early warning model for a turbine component, and a training period. Based on the user-selected conditions, the system subsequently reads the parameter data of the turbine component to be warned, corresponding to the pre-set parameters of the early warning model, as the data required for modeling. For example, in the XXX wind farm, the early warning model for the generator drive-end bearing uses the generator drive-end bearing temperature from September 6, 2021 to September 6, 2020 as the parameter for the turbine component to be warned. This can determine whether the generator drive-end bearing temperature is abnormal. Optionally, the training period for the early warning model can be, for example, one year, thus covering the impact of seasonal temperature changes on the data of turbine components in the wind turbine. Of course, the user-selected training conditions can also include multiple wind farms, multiple early warning models for multiple turbine components, and multiple training periods. "Multiple" refers to two or more, depending on the user's needs.

[0074] In some other embodiments of the parameter data for the aforementioned unit components to be warned, the warning model includes a warning model for the unit components to be warned, and the parameters of the unit components to be warned include the parameters of preset unit components within a preset wind field and a preset time period. The data of these parameters for the unit components to be warned is the data of the parameters of the preset unit components. This allows for the pre-setting of wind fields and time periods for the warning model of the preset unit components, enabling automatic and real-time training of the warning model.

[0075] The pre-defined early warning model for the unit component can be a pre-defined model of an automatically operating unit component. For this pre-defined early warning model, a pre-defined wind field and a pre-defined training time period can be configured. The pre-defined wind field reflects the wind field range within which data on the parameters of the pre-defined unit component can be automatically trained and collected. The pre-defined time period reflects the time period within which data on the parameters of the pre-defined unit component can be automatically trained and collected. Thus, the system can operate according to the pre-defined early warning model, pre-defined wind field, and pre-defined time period. The pre-defined time period can be one day, one month, or six months, thus enabling training for different pre-defined time periods to meet various needs.

[0076] Of course, the two embodiments of step 110 above can be implemented in combination or separately, and there is no limitation here.

[0077] Step 120: Input the parameter data of the unit components to be warned into the warning model, train the warning model, obtain the trained warning model, and obtain the fault entries of the unit components to be warned.

[0078] Among them, the post-training early warning model includes one or more of the following: post-training early warning model for blades, post-training early warning model for pitch system, post-training early warning model for converter, post-training early warning model for main control system, post-training early warning model for hydraulic system, post-training early warning model for tower, post-training early warning model for gearbox system, post-training early warning model for power grid equipment, and post-training early warning model for generator system.

[0079] Step 120 above can further include the following four steps: Step 1: Clean the data of the parameters of the unit components to be warned, obtaining cleaned parameter data. Step 2: Select normal data from the cleaned parameter data. Step 2 can include deleting one or more of the following: data from non-power generation states, data from the parameters of the unit components to be warned that contain missing data, and data from the parameters of the unit components to be warned that exceed the normal threshold range. For example, deleting data from the parameters of the unit components to be warned that exceed the normal threshold range can further include calculating the mean μ and standard deviation δ of the parameter data at each time point based on the unit data from the wind farm, and deleting data outside the normal threshold range of μ±3δ. In this way, the data of the parameters of the unit components to be warned can be cleaned, retaining data usable for modeling and deleting abnormal and unusable data. Step 3: Normalize the normal data to obtain normalized data. The third step above may further include obtaining minimum and maximum parameter files based on the data of the components of the unit to be warned, and performing maximum and minimum normalization of the data of the components to be warned, so as to achieve standardization of the data of the components to be warned. The fourth step is to obtain a training dataset based on the normalized data. Based on the training dataset, the warning model is trained to obtain the trained warning model.

[0080] The data of the parameters of the aforementioned units to be warned can be data stored in the database of the SCADA (Supervisory Control And Data Acquisition) system.

[0081] The fourth step mentioned above can further include the following steps: Step 1: Dividing the normalized data into a training dataset and a test dataset according to a preset ratio. The training dataset is used to train the early warning model, and the test dataset is used to test the accuracy of the trained early warning model. Step 2: Training the early warning model based on the training dataset to obtain the trained early warning model. The preset ratio reflects that the training dataset is larger than the test dataset. The preset ratio is the ratio of the training dataset to the test dataset. Preset ratios include 6:4, 7:3, or 8:2.

[0082] Following step 4 above, the method further includes step 1: testing the trained early warning model based on the test dataset, obtaining test results, and determining the early warning accuracy of the trained model based on the test results. The test results are then compared with the labels of the test dataset to verify the accuracy of the trained early warning model and the extracted features. Step 2: if the early warning accuracy of the trained early warning model does not meet a preset threshold, the model parameters of the trained early warning model are adjusted, and the process returns to step 4 above. Each trained early warning model has different model parameters. Specific model parameters in the trained early warning model include, for example, the number of neurons and hidden layers in the neural network. Step 3: if the early warning accuracy of the trained early warning model meets the preset threshold, the trained early warning model is used as the trained early warning model. This ensures that the trained early warning model has good early warning accuracy for subsequent use. Based on the early warning model, parameter training is conducted on the components of the unit to be warned. The training process and results can be displayed in real time on the user interface for user verification. In addition to the default range of optimized model parameters, the training part of the early warning model also provides some training parameters for users to input and optimize the range of model parameters based on their experience.

[0083] The preset threshold can be determined based on the required precision. The higher the required precision, the larger the preset threshold. Optionally, the preset threshold can be greater than 80%. For example, the preset threshold could be 85%.

[0084] In this embodiment, the data of the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, can be obtained. There is no need to manually input the data of the parameters of the components to be warned, which reduces the professional threshold of the early warning model, makes the early warning model less complicated to use, simplifies user operation, and improves user experience.

[0085] Combination Figure 1 As shown, after step 120 above, the wind turbine fault early warning modeling method further includes: 1. Generating model information for the trained early warning model. This model information includes one or more of the following: model number, name, version number, early warning code, and the turbine component being warned. 2. Storing the model information according to the trained early warning model. This allows for the management of the trained early warning model's model number, name, version number, early warning code, and the turbine component being warned. The model number is a unique identifier reflecting the basic information of the early warning model. Basic information may include, for example, the turbine component being warned and the wind farm being warned by the model. Storing model information according to the trained early warning model can further include independently storing the model information in a model database according to the trained early warning model. This model database can store all trained early warning models, thus enabling unified management of the trained early warning models.

[0086] Combination Figure 1As shown, after step 120 above, the wind turbine fault early warning modeling method further includes packaging the trained early warning model into a Docker container to obtain a Docker package, and sending the Docker package to the fault early warning operation system. The Docker container contains all the environments for running the trained early warning model. Activating the trained early warning model will then enable early warning functionality. After the Docker package is packaged, it is transmitted to the fault early warning operation system. This ensures that each trained early warning model has an independent operating environment, without interference, avoiding operating environment conflicts. Furthermore, the completed trained early warning model is distributed to the fault early warning operation system in the form of a Docker package, enabling the system to subsequently issue fault warnings and dispatch maintenance tasks based on early warning work orders.

[0087] Combination Figure 1 As shown, the early warning model includes an active early warning model; after step 120 above, the wind turbine fault early warning modeling method also includes the following four steps:

[0088] The first step involves receiving an activation command input through the user interface. This command carries indication information about the warning model in a pending activation state. This indication information reflects the warning model's identifier, such as its name or number. The second step involves parsing the activation command to obtain the indication information. The third step, in response to the activation command, retrieves the warning model in a pending activation state. The fourth step activates the warning model in a pending activation state, resulting in an activated warning model. This process minimizes manual input by the user, simplifying operation and increasing activation speed. It enables wind field management and operational status management of the warning model, ensuring that the activated model can effectively perform its warning functions.

[0089] Figure 3 As shown Figure 1 The diagram shows the process of issuing correct fault entries for early warning after step 120 in the wind turbine fault early warning modeling method.

[0090] like Figure 1 and Figure 3 As shown, the fault entries for the warning include the unit number, the warning time, and the unapproved status.

[0091] Following step 120 above, the wind turbine fault early warning modeling method also includes:

[0092] Step 120: Display the warning fault entries on the user interface;

[0093] Step 140: Based on the review instruction received via the user interface for the unreviewed state, modify the warning fault entry to obtain the correct warning fault entry. Step 140 may further include changing the unreviewed state in the warning fault entry to the reviewed state to obtain the correct warning fault entry. Step 140 may further include modifying or deleting false alarm fault entries in the warning fault entry to obtain the correct warning fault entry. Thus, in addition to modifying the review status alone, it is possible to modify fault entries individually, such as false alarm fault entries, or to modify both the review status and false alarm fault entries.

[0094] Step 150: Issue the correct fault entries to the wind farm as early warnings. Following step 140 above, the method further includes generating an early warning work order for the correct entries. Step 150 may further include issuing the early warning work order to the wind farm to facilitate maintenance personnel in carrying out maintenance work.

[0095] In this embodiment, the management and display of fault entries warned by the early warning model can be realized, allowing for expert review. This manual review of the fault entries improves their accuracy. Furthermore, correctly identified fault entries are sent to the wind farm to facilitate maintenance personnel in their work. This creates an integrated workflow from model development to operation, early warning work order issuance, and maintenance.

[0096] Figure 4 As shown Figure 1 The flowchart shown is a schematic diagram of the operation accuracy evaluation process after step 120 in the wind turbine fault early warning modeling method.

[0097] Combination Figure 1 and Figure 4 As shown, after step 120, the wind turbine fault early warning modeling method further includes:

[0098] Step 131: Obtain the operational data of the trained early warning model within the assessment period. The assessment period can be set according to user needs or determined based on the number of early warning items. For example, the assessment period could be six months, or it could be when the number of early warning items exceeds a predetermined number. The predetermined number of items can be determined based on the user's requirements for the operational effectiveness of the early warning model. The higher the operational effectiveness of the early warning model, the fewer the predetermined number of items, thus allowing for timely determination of operational accuracy and subsequent adjustments. Step 141: Based on the operational data, obtain the correctly predicted fault items within the assessment period. Step 151: Based on the correctly predicted fault items and the correctly predicted fault items within the assessment period, determine the operational accuracy of the trained early warning model. This effectively combines the establishment and operation of the early warning model, achieving integrated operation. Furthermore, after the operation, the operational accuracy of the trained early warning model is evaluated, and adjustments are made to improve operational accuracy, making the trained early warning model more accurate. Simultaneously, the operational accuracy of the early warning model is evaluated to verify the model's performance.

[0099] The operational accuracy is illustrated using an assessment period of six months or when there are more than 20 warning items as an example.

[0100]

[0101] This integrated modeling and operation evaluation system encompasses the entire modeling and operation process. It requires only a wind farm, an early warning model, and a training period to train a high-precision early warning model, which can then be directly deployed and put into operation. After the evaluation period, the model's operational accuracy can be assessed. This approach reduces the difficulty of using early warning models, attracting more wind farm operation and maintenance personnel to participate in their use and promoting the rapid development of intelligent operation and maintenance.

[0102] Figure 5 The diagram shown is a schematic of the modules of the wind turbine fault early warning modeling device provided in the embodiment of this application.

[0103] like Figure 5 As shown, the wind turbine fault early warning modeling device includes:

[0104] The acquisition module 21 is used to acquire the parameters of the pre-set unit components to be warned corresponding to the warning model and the data of the parameters of the unit components to be warned;

[0105] The modeling module 22 is used to input the parameter data of the unit component to be warned into the warning model, train the warning model, obtain the trained warning model, and obtain the fault entries of the unit component to be warned.

[0106] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0107] One embodiment of this application provides a wind turbine fault early warning modeling system including one or more processors for implementing the wind turbine fault early warning modeling method described above. Alternatively, one embodiment of this application provides a wind turbine fault early warning modeling system including a wind turbine fault early warning modeling device as described above. Thus, based on years of experience in intelligent early warning and diagnosis in the wind power industry, an integrated wind turbine fault early warning modeling system has been developed, realizing integrated modeling operations such as data processing, model training, and model evaluation.

[0108] Figure 6 The diagram shown is a schematic representation of the electronic device 30 provided in an embodiment of this application. The electronic device 30 may include a processor 31, a memory 33 storing machine-executable instructions, and a communication interface 32. The processor 31 and the memory 33 can communicate via a system bus 34. Furthermore, by reading and executing the machine-executable instructions in the memory 33 corresponding to data fetch or data return logic, the processor 31 can execute the methods described above.

[0109] The method provided in this invention can be applied to electronic devices. Specifically, the electronic device can be a desktop computer, a portable computer, a server, etc. No limitation is made herein; any electronic device that can implement the embodiments of this invention falls within the protection scope of this invention.

[0110] The memory 33 mentioned in this document can be any electronic, magnetic, optical, or other physical storage device that can contain or store information such as executable instructions, data, etc. For example, machine-readable storage media can be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, storage drives (such as hard disk drives), solid-state drives, any type of storage disk (such as optical discs, DVDs, etc.), or similar storage media, or combinations thereof.

[0111] In some embodiments, a machine-readable storage medium, such as Figure 6 The memory 33 in the machine-readable storage medium stores machine-executable instructions that, when executed by the processor, implement the method described above. For example, the machine-readable storage medium can be ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0112] This application also provides a computer program stored in a machine-readable storage medium, such as... Figure 6 The memory 33 in the memory, and when the processor executes the computer program, it causes the processor 31 to perform the method described above.

[0113] The above are merely preferred embodiments of this specification and are not intended to limit this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification shall be included within the scope of protection of this specification.

[0114] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for modeling fault early warning in wind turbine units, characterized in that, include: Acquire the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, and the data of the parameters of the components of the unit to be warned; Obtain the training conditions selected by the user, wherein the selected training conditions include a wind farm, an early warning model of a unit component, and a training time period, and / or include multiple wind farms, multiple early warning models of multiple unit components, and multiple training time periods; The parameter data of the unit component to be warned is input into the warning model, the warning model is trained, and the trained warning model and the fault entries of the unit component to be warned are obtained. The method further includes: The user interface displays the fault entries for the warning; the fault entries for the warning include the unit number, the warning time, and the unapproved status; Based on the review instruction received through the user interface for the unreviewed state, modify the warning fault entry to obtain the correct warning fault entry; Issue the correct fault entries to the wind farm as early warning; Obtain the running data of the trained model within the assessment period; Based on the operational data, obtain the correct fault entries for the early warning within the assessment period; The operational accuracy of the trained model is determined based on the correct fault entries and the fault entries warned during the assessment period.

2. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, The acquisition of parameters of the pre-set components of the unit to be warned, corresponding to the early warning model, and the data of the parameters of the components of the unit to be warned, includes: Obtain the training conditions selected by the user, including the selected wind field, the selected early warning model, and the selected training time period; Based on the selected early warning model, determine the parameters of the pre-set components of the unit to be warned that correspond to the selected early warning model; Based on the selected wind field and the selected training time period, data on the parameters of the components of the unit to be warned are obtained.

3. The wind turbine fault early warning modeling method as described in claim 2, characterized in that, The process of obtaining the training conditions selected by the user includes: Receive training instructions input through the user interface; In response to the training instruction, candidate training conditions are displayed on the user interface; Based on the selection instruction for the candidate training conditions received through the user interface, the candidate training conditions selected by the selection instruction are obtained as the training conditions selected by the user.

4. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, The early warning model includes an early warning model for preset unit components. The parameters of the unit components to be warned include the parameters of the preset unit components within a preset time period in a preset wind field. The data of the parameters of the unit components to be warned are the data of the parameters of the preset unit components.

5. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, The post-training early warning model includes one or more of the following: a post-training early warning model for the blades, a post-training early warning model for the pitch system, a post-training early warning model for the converter, a post-training early warning model for the main control system, a post-training early warning model for the hydraulic system, a post-training early warning model for the tower, a post-training early warning model for the gearbox system, a post-training early warning model for the power grid equipment, and a post-training early warning model for the generator system.

6. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, After inputting the parameter data of the unit component to be warned into the warning model, training the warning model to obtain the trained warning model, and obtaining the fault entries of the unit component to be warned, the wind turbine fault warning modeling method further includes: Generate model information for the trained early warning model, the model information including one or more of the following: number, name, version number, early warning code, and the unit component that issued the early warning; The model information is stored according to the trained early warning model.

7. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, After inputting the parameter data of the unit component to be warned into the warning model, training the warning model, and obtaining the trained warning model and the fault entries of the unit component to be warned, the wind turbine fault warning modeling method further includes: The trained early warning model is packaged using Docker to obtain a Docker package; Send the Docker package to the fault early warning system.

8. The wind turbine fault early warning modeling method according to any one of claims 1 to 7, characterized in that, The early warning model includes an early warning model in an active state; Before acquiring the parameters of the wind turbine components to be warned and the data of the parameters of the wind turbine components to be warned, the wind turbine fault early warning modeling method further includes: Receive an activation command input through a user interface, the activation command carrying indication information of a warning model in a pending activation state; The activation command is parsed to obtain the instruction information; In response to the activation command, obtain the early warning model of the pending activation state; Activate the warning model in the pending state to obtain the warning model in the activated state.

9. The wind turbine fault early warning modeling method as described in claim 1, characterized in that, The step of inputting the parameter data of the components of the unit to be warned into the warning model, training the warning model, and obtaining the trained warning model includes: Data cleaning is performed on the parameter data of the components of the unit to be warned, resulting in cleaned parameter data; Select normal data from the parameters after cleaning; The normal data is normalized to obtain normalized data; Based on the normalized data, a training dataset is obtained; The early warning model is trained based on the training dataset to obtain the trained early warning model; After training the early warning model based on the training dataset to obtain the trained early warning model, the method further includes: If the warning accuracy of the post-trained warning model meets the preset threshold, the post-trained warning model is used as the trained warning model.

10. A wind turbine fault early warning modeling device, characterized in that, include: The acquisition module is used to acquire the parameters of the pre-set components of the unit to be warned, which correspond to the early warning model, and the data of the parameters of the components of the unit to be warned. Obtain the training conditions selected by the user, wherein the selected training conditions include a wind farm, an early warning model of a unit component, and a training time period, and / or include multiple wind farms, multiple early warning models of multiple unit components, and multiple training time periods; The modeling module is used to input the parameter data of the unit component to be warned into the warning model, train the warning model, obtain the trained warning model, and obtain the fault entries of the unit component to be warned. The device further includes: displaying the warning fault entries on a user interface; the warning fault entries include the unit number, warning time, and unapproved status; modifying the warning fault entries based on the approval instruction received through the user interface for the unapproved status to obtain the correct warning fault entries; issuing the correct warning fault entries to the wind farm; acquiring the operational data of the trained model within the assessment period; acquiring the correct warning fault entries within the assessment period based on the operational data; and determining the operational accuracy of the trained model based on the correct warning fault entries within the assessment period and the warning fault entries within the assessment period.

11. An electronic device, characterized in that, Including processor and memory; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method according to any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements the method as described in any one of claims 1-9.

Images