Data preprocessing method and device based on AI large model

Abstract

The invention relates to a data preprocessing method and device based on an AI large model, and is applied to the technical field of wind power data processing, and the method comprises the steps: obtaining a fault diagnosis instruction for any part in a wind turbine generator; identifying the fault diagnosis instruction through a pre-constructed wind power major language model to obtain fault diagnosis content; accessing a wind power operation database according to the fault diagnosis content, and calling a target knowledge graph corresponding to the fault diagnosis content; based on the target knowledge graph, extracting target operation data which is associated with the fault diagnosis content and belongs to one or more data sources in a preset historical time period from the wind power operation database; and processing the target operation data based on the target knowledge graph to obtain target preprocessing data for fault detection. According to the method, multiple data sources are obtained by calling the knowledge graph through the large model, the utilization rate of operation data information is improved, and then the accuracy of fault diagnosis is improved.

Description

Technical Field

[0001] The present application relates to the technical field of wind power data processing, and in particular to a data preprocessing method and device based on an AI large model. Background Art

[0002] Wind turbines generate a large amount of operating data during operation, including SCADA (Supervisory Control And Data Acquisition) data such as temperature, pressure, power, and speed, and CMS (Condition Monitoring System) data reflecting the vibration of each component. Some units are also equipped with sound and video monitoring systems for important parts to obtain sound and image data. These data contain rich information and are an important basis for evaluating the performance of wind turbines, monitoring their health status, and conducting fault diagnosis and prediction. By processing and analyzing the operating data of wind turbines, real-time monitoring, fault warning, performance optimization, and other functions of wind turbines can be realized, thereby improving the operating efficiency and reliability of wind farms and reducing operation and maintenance costs.

[0003] For the large amount of operating data generated by wind turbines during operation, in order to train artificial intelligence models with higher accuracy and reliability, preprocessing the original operating data to extract more effective features is often more effective than conventional algorithm optimization.

[0004] In the related art, the operation data of the wind turbines related to fault diagnosis is obtained, and after data cleaning, it is input into the neural network model to diagnose the fault type. However, the data source obtained by this method is relatively single, and the information of the operation data is not fully utilized, resulting in low accuracy of fault diagnosis. Summary of the invention

[0005] In order to solve the above technical problems, the present application provides a data preprocessing method, device, electronic device and storage medium based on an AI (artificial intelligence) big model.

[0006] According to the first aspect of the present application, a data preprocessing method based on an AI large model is provided, comprising: Obtain fault diagnosis instructions for any component in the wind turbine; The fault diagnosis instruction is identified by using a pre-built large language model for wind power industry to obtain the fault diagnosis content; According to the fault diagnosis content, access the wind power operation database and call the target knowledge graph corresponding to the fault diagnosis content; Based on the target knowledge graph, extracting target operation data associated with the fault diagnosis content within a preset historical time period and belonging to one or more data sources from the wind power operation database; Based on the target knowledge graph, the target operation data is preprocessed to obtain target preprocessed data for fault detection.

[0007] Optionally, preprocessing the target operation data based on the target knowledge graph to obtain target preprocessed data for fault detection includes: Based on the target knowledge graph, determining the type of the target operation data; According to the type of the target operating data, the target operating data is preprocessed to obtain target preprocessed data for fault detection.

[0008] Optionally, preprocessing the target operation data according to the type of the target operation data includes: If the target operating data belongs to slice data, outlier removal is performed on the target operating data; If the target operation data is time series data, a smoothing filter process is performed on the target operation data.

[0009] Optionally, preprocessing the target operation data based on the target knowledge graph includes: Acquiring operating status data for judging the operating status of the fan; Determining the operating condition of the wind turbine generator system according to the operating condition data; If the operating condition of the wind turbine is different from the fault diagnosis requirement, the target operation data under the operating condition different from the fault diagnosis requirement is eliminated.

[0010] Optionally, the method for constructing the wind power professional large language model includes: Acquire wind power professional corpus, and convert the wind power professional corpus into question-answer pair data; A general large language model is obtained, and the general large language model is fine-tuned using the question-answer pair data to obtain the wind power professional large language model.

[0011] According to a second aspect of the present application, a data preprocessing device based on an AI large model is provided, comprising: An instruction acquisition module, used to acquire a fault diagnosis instruction for any component in the wind turbine generator set; An instruction recognition module is used to recognize the fault diagnosis instruction through a pre-built large language model of wind power industry to obtain fault diagnosis content; The knowledge graph calling module is used to access the wind power operation database according to the fault diagnosis content and call the target knowledge graph corresponding to the fault diagnosis content; A data extraction module, used to extract target operation data associated with the fault diagnosis content within a preset historical time period and belonging to one or more data sources from the wind power operation database; A data preprocessing module is used to preprocess the target operation data based on the target knowledge graph to obtain target preprocessed data for fault detection.

[0012] Optionally, the data preprocessing module is specifically used to determine the type of the target operation data based on the target knowledge graph; and preprocess the target operation data according to the type of the target operation data to obtain target preprocessed data for fault detection.

[0013] Optionally, the data preprocessing module is specifically used to determine the type of the target operation data based on the target knowledge graph; if the target operation data belongs to slice data, remove outliers from the target operation data; if the target operation data belongs to time series data, perform smoothing filtering on the target operation data.

[0014] Optionally, the data preprocessing module is specifically used to obtain operating status data for judging the operating status of the wind turbine; based on the operating status data, judge the operating condition of the wind turbine; if the operating condition of the wind turbine is different from the fault diagnosis requirement, then eliminate the target operating data under the operating condition different from the fault diagnosis requirement.

[0015] Optionally, the data preprocessing device based on the AI ​​big model further includes: The wind power professional large language model construction module is used to obtain wind power professional corpus and convert the wind power professional corpus into question-answer pair data; obtain a general large language model and use the question-answer pair data to fine-tune the general large language model to obtain the wind power professional large language model.

[0016] According to a third aspect of the present application, an electronic device is provided, comprising: a processor, wherein the processor is configured to execute a computer program stored in a memory, wherein the computer program implements the method described in the first aspect when executed by the processor.

[0017] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first aspect is implemented.

[0018] According to a fifth aspect of the present application, a computer program product is provided. When the computer program product is run on a computer, the computer is enabled to execute the method described in the first aspect.

[0019] Compared with the prior art, the technical solution provided by the embodiments of the present application has the following advantages: Through the pre-built large language model of wind power industry, the fault diagnosis instruction is identified, the wind power operation database is accessed, and the target knowledge graph corresponding to the fault diagnosis content is called. The target operation data related to the fault diagnosis content is mined using the target knowledge graph. The target operation data can belong to one data source or multiple data sources, so as to obtain the cross information between different data sources and realize the all-round perception of a certain component. After extracting the target operation data from the wind power operation database, the target operation data is analyzed using the target knowledge graph, and different data preprocessing processes can be performed for different target operation data. It can be seen that different data preprocessing schemes can be provided for different fault diagnosis instructions through the large language model and knowledge graph of wind power industry, so that the processed operation data is more suitable for the fault diagnosis model, thereby effectively improving the accuracy and reliability of subsequent fault diagnosis. In addition, the embodiment of the present application has a relatively low dependence on historical operation data and has a certain generalization ability. In the absence of historical operation data (such as a newly commissioned wind turbine), the accuracy of the fault diagnosis model can also be effectively increased. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0022] Figure 1 This is a schematic diagram of a data preprocessing method based on an AI large model applicable to an embodiment of the present application; Figure 2 A flow chart of a data preprocessing method based on an AI big model in an embodiment of the present application; Figure 3A A schematic diagram of data preprocessing using a partial knowledge graph of secondary planetary gear train damage in an embodiment of the present application; Figure 3B A schematic diagram of performing data preprocessing on the temperature of the bottom cabinet of a fan tower by using the knowledge graph of the temperature of the bottom cabinet of a fan tower in an embodiment of the present application; Figure 4 This is a structural schematic diagram of a data preprocessing device based on an AI large model in an embodiment of the present application; Figure 5 A schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the scheme of the present application will be further described below. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0024] In the following description, many specific details are set forth to facilitate a full understanding of the present application, but the present application may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only part of the embodiments of the present application, rather than all of the embodiments.

[0025] In recent years, AI big model technology has continued to develop. The widespread application of transformer architecture can almost replace traditional neural network architectures such as RNN (Recurrent Neural Network), CNN (Convolutional Neural Networks), LSTM (Long Short-Term Memory), and achieve better results in practice in various fields. At the same time, the promotion of Agent technology has well linked big models with professional knowledge and tools in professional fields, effectively preventing big models from producing hallucinations in specific fields. RAG (Retrieval Enhanced Generation) technology further optimizes the search logic of the big model database, making the model output results more reliable and further lowering the threshold for use. For this reason, AI big models have been frequently implemented in many power-related companies, providing long-term and effective assistance to companies, and also promoting the digital and intelligent transformation of the national energy and power industry.

[0026] However, the general large model has the advantage of understanding human language semantics, but it is weak in understanding professional scenarios and answering questions, and is prone to hallucinations and giving irrelevant answers. In addition, the general large model is weak in data understanding and professional knowledge processing. Even if the answer is correct, it cannot give a reasonable explanation, has poor readability, and is difficult to be directly applied to the industrial field.

[0027] Based on this, the embodiment of the present application constructs a large language model for wind power that is applicable to the professional field, mounts the expert knowledge base and mechanism model through Agent technology, and implements real-time calling. Then, the content related to text information in the knowledge base is analyzed through RAG technology to ensure that the correct results can be output during the professional knowledge question and answer and data processing of the large model. And it can give certain explanations according to the established knowledge base to assist relevant staff in further operations.

[0028] See also Figure 1 , Figure 1 This is a schematic diagram of a data preprocessing method based on an AI large model in an embodiment of the present application. The system inputs instructions for a fault diagnosis or status monitoring project, which can be manually input in real time or automatically input regularly according to on-site needs. After the wind power professional large language model obtains the instruction, it calls the corresponding target knowledge graph, and uses the target knowledge graph to retrieve the target operation data related to the instruction in the recent period from the wind power operation database. The target operation data can belong to one or more data sources. In this way, by comprehensively considering multiple data sources, the utilization rate of the operation data information is improved to ensure the correctness and calculation accuracy of the early warning. The target operation data is preprocessed using the target knowledge graph to obtain target preprocessed data for fault detection. For different fault diagnosis instructions, different knowledge graphs can be called to perform different data preprocessing processes. In this way, features are purposefully extracted for different fault types, thereby improving the accuracy and reliability of the fault diagnosis model judgment.

[0029] See also Figure 2 , Figure 2 This is a flow chart of a data preprocessing method based on an AI big model in an embodiment of the present application, which may include the following steps: Step S202: Obtain a fault diagnosis instruction for any component in the wind turbine generator system.

[0030] A wind turbine includes multiple components, such as a wind rotor, a generator, a tower, and a control system. For each component, a corresponding fault diagnosis instruction can be sent to perform fault diagnosis on the component. For example, a fault diagnosis instruction can be "gearbox vibration abnormality detection".

[0031] Step S204: using the pre-built large language model for wind power industry, the fault diagnosis instruction is identified to obtain the fault diagnosis content.

[0032] The current capabilities of the general large language model are mainly focused on solving general language problems. It lacks industry-related knowledge accumulation and has weak professional scene understanding and question-answering capabilities. In order to improve the scene understanding and question-answering capabilities of wind power professionals, a wind power professional large language model can be constructed, that is, a large language model that can be applied to the wind power field. Optionally, a wind power professional large language model can be generated by collecting wind power professional corpus based on the general large language model.

[0033] Specifically, wind power professional corpus can be obtained, such as relevant books, technical documents or corpus from other sources, and the wind power professional corpus can be processed to convert the wind power professional corpus into question-answer pair data, which can be used for fine-tuning the general large language model. A general large language model is obtained, and the general large language model is fine-tuned using the question-answer pair data to obtain a wind power professional large language model, which has the ability to understand and reason about the wind power professional knowledge involved in the corpus.

[0034] The wind power professional language model can identify fault diagnosis instructions and obtain fault diagnosis content. For example, for the fault diagnosis instruction of "gearbox vibration abnormality detection", it can be identified that the fault diagnosis content is "abnormality detection related to gearbox vibration". That is, the user wants to detect abnormalities related to gearbox vibration, for example, the gearbox rotation imbalance, rotation misalignment, secondary planetary gear system damage, etc.

[0035] Step S206: access the wind power operation database according to the fault diagnosis content, and call the target knowledge graph corresponding to the fault diagnosis content.

[0036] The wind power operation database contains the operation data generated by the wind turbine during operation. The operation data includes multiple data sources, such as SCADA data such as temperature, pressure, power, wind speed, CMS data of the vibration of each component, and sound and image data of some components. By accessing the wind power operation database, you can obtain the operation data related to the fault diagnosis content.

[0037] According to the fault diagnosis content, the wind power professional large language model can call the knowledge graph related to the fault diagnosis content from the pre-built knowledge graph library, that is, the target knowledge graph. The knowledge graph library contains multiple knowledge graphs, and each component of the wind turbine can correspond to a knowledge graph, or each fault diagnosis content corresponds to a knowledge graph. Assuming that each fault diagnosis content corresponds to a knowledge graph, the target knowledge graph contains knowledge related to the fault diagnosis content. Therefore, the target knowledge graph corresponding to the fault diagnosis content can be called, and the target knowledge graph can be used to extract the target operation data related to the fault diagnosis content from the wind power operation database, and data preprocessing suggestions can be made for the target operation data later.

[0038] Step S208, based on the target knowledge graph, extracting target operation data associated with the fault diagnosis content within a preset historical time period and belonging to one or more data sources from the wind power operation database.

[0039] As mentioned above, the operation data of wind turbines in the wind power operation database involves multiple types and multiple data sources. Based on the target knowledge graph, more operation data associated with the fault diagnosis content can be mined. For example, for the fault diagnosis instruction of "gearbox vibration abnormality detection", the gearbox vibration data can be obtained, and the damage of the secondary planetary gear train can be diagnosed through the gearbox vibration data. However, in addition to this, data such as gearbox temperature and gearbox sound can assist in determining whether the fault exists. Therefore, when pre-constructing the target knowledge graph, the gearbox vibration data, gearbox temperature data, gearbox sound data, which belong to different data sources, are set as operation data related to the damage of the secondary planetary gear train. When using the target knowledge graph, the correlation between the gearbox vibration data, gearbox temperature data, gearbox sound data and the damage of the secondary planetary gear train can be known, so that the gearbox vibration data, gearbox temperature data, and gearbox sound data are extracted from the wind power operation database as the target operation data.

[0040] The process of data preprocessing using the partial knowledge graph of secondary planetary gear train damage can be found in Figure 3A . It can be seen that the knowledge graph can be used to extract gearbox vibration data, gearbox temperature data, and gearbox sound data, that is, data from three different data sources can be extracted as target operating data. Subsequently, different data preprocessing processes can be performed on the data from the three different data sources, and then the damage of the secondary planetary gear train can be diagnosed based on the preprocessed data. It can be seen that compared with fault diagnosis using a single data source, this method has higher accuracy in fault diagnosis.

[0041] Step S210, based on the target knowledge graph, preprocess the target operation data to obtain target preprocessed data for fault detection.

[0042] The preprocessing process includes outlier removal, non-power generation data removal, data removal based on some Boolean variables, removal based on variance, filtering links such as modulation functions, variable window sliding average, time-frequency conversion, linear fitting, etc. Good data preprocessing can improve the convergence speed of the model during model training and improve the prediction accuracy when the model is applied. For example, through data standardization, the numerical range of data features is unified, which helps to optimize the parameter selection and training process of the model.

[0043] The operating data of wind turbines often contain abnormal values, noise and default values, which will affect the accuracy of fault diagnosis models. For example, the data recorded by the SCADA system includes data on the normal operation of wind turbines, data on faults, and data on abnormal operations such as wind speed not reaching the cut-in wind speed and pitch not acting.

[0044] In some embodiments, the type of target operation data can be determined based on the target knowledge graph; according to the type of target operation data, the target operation data is preprocessed to obtain target preprocessed data for fault detection. For example, if the target operation data belongs to slice data, considering the influence of noise, sensors, and communications, data that obviously does not conform to statistical laws can be eliminated, that is, the target operation data is processed by outlier elimination, etc., to ensure that the data input into the fault diagnosis model is accurate and reliable, thereby improving the prediction performance of the model.

[0045] If the target operation data is time series data, it can be processed by smoothing and filtering. For example, for time series data such as vibration, Fourier transform, wavelet transform and other methods can be used for feature extraction. For various time series waveforms, the peak value, amplitude, kurtosis factor and other indicators describing the waveform are extracted, and the information of the time series data is converted into cross-sectional information, which is convenient for fusion processing with other data.

[0046] For invalid data beyond the business range, for example, when discussing faults in the normal power generation process, non-power generation data and start-stop data are invalid data, and data can be eliminated through the reflection of speed, active power, and some Boolean variables. Optionally, the operating status data used for judging the operating status of the wind turbine can be obtained, and the working condition of the wind turbine can be judged according to the operating status data; if the working condition of the wind turbine is different from the fault diagnosis requirement, the target operating data under the working condition different from the fault diagnosis requirement is eliminated. For example, if the fault diagnosis requirement is normal power generation condition, the active power parameter can be used to determine whether it is in normal power generation condition, and the target operating data under abnormal power generation condition is eliminated. On the contrary, if the fault diagnosis requirement is abnormal power generation condition, the active power parameter can be used to determine whether it is in abnormal power generation condition, and the target operating data under normal power generation condition is eliminated. Similarly, if the fault diagnosis requirement is normal operation of the pitch change, the pitch change action Boolean variable is used to determine whether the pitch change is normal, and the target operating data when the pitch change is abnormal is eliminated.

[0047] Figure 3BA schematic diagram of data preprocessing of the temperature of the bottom cabinet of a fan tower using the knowledge graph of the temperature of the bottom cabinet of a fan tower in an embodiment of the present application. Assuming that the fault diagnosis instruction is "abnormal temperature diagnosis of the bottom cabinet of a fan tower", after the wind power professional large language model receives the fault diagnosis instruction, it identifies the fault diagnosis content, calls the knowledge graph of the temperature of the bottom cabinet of the fan tower, and extracts the temperature of the bottom cabinet of the fan tower from the wind power operation database. Since it is affected by the ambient temperature, the ambient temperature can also be obtained. The knowledge graph of the temperature of the bottom cabinet of the fan tower shows that the temperature data of the bottom cabinet of the fan tower belongs to slice data, so NULL values, NA values, constant values ​​and continuous 0 values ​​caused by factors such as communication and sensor failure can be eliminated. Since the active power and wind speed can be used to determine whether it is in normal power generation conditions, the temperature data of the bottom cabinet of the fan tower under normal conditions can be screened out according to the data of active power and wind speed. The difference between the temperature data of the bottom cabinet of the fan tower under normal conditions and the ambient temperature is calculated to obtain the temperature rise data, and the temperature rise data is output to the fault diagnosis model as the target preprocessing data.

[0048] The data preprocessing method based on the AI ​​big model of the embodiment of the present application, through the pre-built wind power professional big language model, identifies the fault diagnosis instruction, accesses the wind power operation database, and calls the target knowledge graph corresponding to the fault diagnosis content. The target operation data related to the fault diagnosis content is mined using the target knowledge graph. The target operation data can belong to one data source or multiple data sources, so as to obtain the cross information between different data sources and realize the all-round perception of a certain component. After extracting the target operation data from the wind power operation database, the target operation data is analyzed using the target knowledge graph, and different data preprocessing processes can be performed for different target operation data. It can be seen that different data preprocessing schemes can be provided for different fault diagnosis instructions through the wind power professional big language model and knowledge graph, so that the processed operation data is more suitable for the fault diagnosis model, thereby effectively improving the accuracy and reliability of subsequent fault diagnosis. In addition, the embodiment of the present application has a relatively low dependence on historical operation data and has a certain generalization ability. In the absence of historical operation data (such as a newly put into use wind turbine), the accuracy of the fault diagnosis model can also be effectively increased.

[0049] It should be noted that although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0050] Corresponding to the above method embodiment, the present application embodiment also provides a data preprocessing device based on an AI large model, see Figure 4 , the data preprocessing device 400 based on the AI ​​big model includes: The instruction acquisition module 402 is used to acquire a fault diagnosis instruction for any component in the wind turbine generator system; The instruction recognition module 404 is used to recognize the fault diagnosis instruction through a pre-built wind power professional large language model to obtain the fault diagnosis content; The knowledge graph calling module 406 is used to access the wind power operation database according to the fault diagnosis content and call the target knowledge graph corresponding to the fault diagnosis content; A data extraction module 408 is used to extract target operation data associated with the fault diagnosis content and belonging to one or more data sources within a preset historical time period from the wind power operation database; The data preprocessing module 410 is used to preprocess the target operation data based on the target knowledge graph to obtain target preprocessed data for fault detection.

[0051] Optionally, the data preprocessing module 410 is specifically used to determine the type of the target operation data based on the target knowledge graph; preprocess the target operation data according to the type of the target operation data to obtain target preprocessed data for fault detection.

[0052] Optionally, the data preprocessing module 410 is specifically used to determine the type of target operation data based on the target knowledge graph; if the target operation data belongs to slice data, outliers are removed from the target operation data; if the target operation data belongs to time series data, smoothing and filtering are performed on the target operation data.

[0053] Optionally, the data preprocessing module 410 is specifically used to obtain operating status data for judging the operating status of the wind turbine; based on the operating status data, judge the operating condition of the wind turbine; if the operating condition of the wind turbine is different from the fault diagnosis requirement, then eliminate the target operating data under the operating condition different from the fault diagnosis requirement.

[0054] Optionally, the data preprocessing device 400 based on the AI ​​big model further includes: The wind power professional large language model construction module is used to obtain wind power professional corpus and convert the wind power professional corpus into question-answer pair data; obtain a general large language model and use the question-answer pair data to fine-tune the general large language model to obtain a wind power professional large language model.

[0055] The specific details of each module or unit in the above device have been described in detail in the corresponding method, so they will not be repeated here.

[0056] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0057] An embodiment of the present application also provides an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the above-mentioned data preprocessing method based on the AI ​​big model in this example implementation.

[0058] Reference Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. The specific embodiment of the present application does not limit the specific implementation of the electronic device.

[0059] like Figure 5 As shown, the electronic device may include: a processor 502 , a communication interface 504 , a memory 506 , and a communication bus 508 .

[0060] The processor 502 , the communication interface 504 , and the memory 506 communicate with each other via a communication bus 508 .

[0061] The communication interface 504 is used to communicate with other electronic devices or servers.

[0062] The processor 502 is used to execute the program 510, and specifically can execute the relevant steps in the above method embodiment.

[0063] Specifically, the program 510 may include program codes, which include computer operation instructions.

[0064] The processor 502 may be a central processing unit, or a specific integrated circuit, or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0065] The memory 506 is used to store the program 510. The memory 506 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0066] Program 510 can be specifically used to enable processor 502 to execute the steps in the above-mentioned data preprocessing method embodiment based on AI large model.

[0067] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and modules can refer to the corresponding process descriptions in the aforementioned method embodiments and will not be repeated here.

[0068] A computer-readable storage medium is also provided in an embodiment of the present application, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned data preprocessing method based on the AI ​​large model is implemented.

[0069] It should be noted that the computer-readable storage medium shown in the present application may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, radio frequency, etc., or any suitable combination of the above.

[0070] In an embodiment of the present application, a computer program product is also provided. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned data preprocessing method based on the AI ​​large model.

[0071] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0072] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A data preprocessing method based on AI big model, characterized in that: include: Obtain fault diagnosis instructions for any component in the wind turbine; The fault diagnosis instruction is identified by using a pre-built large language model for wind power industry to obtain the fault diagnosis content; According to the fault diagnosis content, access the wind power operation database and call the target knowledge graph corresponding to the fault diagnosis content; Based on the target knowledge graph, extracting target operation data associated with the fault diagnosis content within a preset historical time period and belonging to one or more data sources from the wind power operation database; Based on the target knowledge graph, the target operation data is preprocessed to obtain target preprocessed data for fault detection.

2. The data preprocessing method based on AI big model according to claim 1 is characterized in that: The preprocessing of the target operation data based on the target knowledge graph to obtain target preprocessed data for fault detection includes: Based on the target knowledge graph, determining the type of the target operation data; According to the type of the target operating data, the target operating data is preprocessed to obtain target preprocessed data for fault detection.

3. The data preprocessing method based on AI big model according to claim 2 is characterized in that: The preprocessing of the target operation data according to the type of the target operation data includes: If the target operating data belongs to slice data, outlier removal is performed on the target operating data; If the target operation data is time series data, a smoothing filter process is performed on the target operation data.

4. The data preprocessing method based on AI big model according to claim 1 is characterized in that: The preprocessing of the target operation data based on the target knowledge graph includes: Acquiring operating status data for judging the operating status of the fan; Determining the operating condition of the wind turbine generator system according to the operating condition data; If the operating condition of the wind turbine is different from the fault diagnosis requirement, the target operation data under the operating condition different from the fault diagnosis requirement is eliminated.

5. The data preprocessing method based on AI big model according to claim 1 is characterized in that: The method for constructing the wind power professional large language model includes: Acquire wind power professional corpus, and convert the wind power professional corpus into question-answer pair data; A general large language model is obtained, and the general large language model is fine-tuned using the question-answer pair data to obtain the wind power professional large language model.

6. A data preprocessing device based on AI large model, characterized in that: The device comprises: An instruction acquisition module, used to acquire a fault diagnosis instruction for any component in the wind turbine generator set; An instruction recognition module is used to recognize the fault diagnosis instruction through a pre-built large language model of wind power industry to obtain fault diagnosis content; A knowledge graph calling module is used to access the wind power operation database according to the fault diagnosis content and call the target knowledge graph corresponding to the fault diagnosis content; A data extraction module, used to extract target operation data associated with the fault diagnosis content within a preset historical time period and belonging to one or more data sources from the wind power operation database; A data preprocessing module is used to preprocess the target operation data based on the target knowledge graph to obtain target preprocessed data for fault detection.

7. The data preprocessing device based on the AI ​​large model according to claim 6 is characterized in that: The data preprocessing module is specifically used to determine the type of the target operation data based on the target knowledge graph; and preprocess the target operation data according to the type of the target operation data to obtain target preprocessed data for fault detection.

8. The data preprocessing device based on AI large model according to claim 7 is characterized in that: The data preprocessing module is specifically used to determine the type of the target operation data based on the target knowledge graph; if the target operation data belongs to slice data, remove outliers from the target operation data; If the target operation data is time series data, a smoothing filter process is performed on the target operation data.

9. The data preprocessing device based on AI large model according to claim 8 is characterized in that: The data preprocessing module is specifically used to obtain operating status data for judging the operating status of the wind turbine; based on the operating status data, judge the working condition of the wind turbine; if the working condition of the wind turbine is different from the fault diagnosis requirement, then eliminate the target operating data under the working condition different from the fault diagnosis requirement.

10. The data preprocessing device based on AI big model according to claim 6 is characterized in that: The device also includes: The wind power professional large language model construction module is used to obtain wind power professional corpus and convert the wind power professional corpus into question-answer pair data; obtain a general large language model and use the question-answer pair data to fine-tune the general large language model to obtain the wind power professional large language model.

Images