A wind turbine health assessment and repair method, system, device and medium

Abstract

The application discloses a kind of wind turbine health degree evaluation and repair method, system, equipment and medium, by centralized acquisition and effective utilization data center's operating data and multidimensional monitoring data, existing information resources are fully integrated, information silos are broken, improve work efficiency;After outlier cleaning and missing value reconstruction processing, again in combination with SCADA and CMS, improve the accuracy of fault diagnosis, and then improve the accuracy of wind turbine health degree evaluation, not only reduce the monitoring difficulty, but also can be established according to fault diagnosis data Fault case library, generate health degree evaluation result, and complete fault positioning, optimize the function of data center, improve the function of wind turbine health degree state monitoring and fault diagnosis;Through the combination of single combat system and fault case library, the key equipment hidden danger investigation focus and implementation steps are clear, and the inspection personnel are guided to complete the wind turbine to complete condition-based maintenance, so as to improve the reliability and operation efficiency of wind turbine.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of wind power generation, and particularly relates to a wind turbine health degree evaluation and maintenance method, system, device and medium. BACKGROUND

[0002] Wind turbines operate in extreme and variable weather conditions, and need to monitor the operating state of the unit at all times, diagnose faults in a timely manner, and reasonably arrange maintenance to reduce maintenance costs. Common state monitoring systems for wind turbines include vibration monitoring systems (CMS) and supervisory control and data acquisition systems (SCADA). Among them, the vibration monitoring system (CMS) reflects quickly and accurately locates faults, and is mainly used for fault diagnosis of the transmission chain; the data acquisition system (SCADA) can cover multiple components of the wind turbine and has a wide monitoring range.

[0003] The wind turbine state monitoring system accumulates a large amount of operating data, but these data have not been further mined at present, the data resource utilization rate is low, the maintenance cost is high, and the operating efficiency of the wind farm is greatly reduced. The functions of the wind turbine data center for state monitoring and fault diagnosis are relatively single at present, and further improvement is needed in the application of actual fault diagnosis, positioning and condition-based maintenance; the data quality evaluation system is still not perfect, and the whole-life management and control of data assets are lacking; the seasonal and random characteristics of wind energy increase the difficulty of condition-based maintenance.

[0004] Therefore, the existing wind turbine health degree monitoring measures have relatively single functions of state monitoring and fault diagnosis, and have great monitoring difficulty and low efficiency, which cannot guarantee the reliability and operating efficiency of the wind turbine. SUMMARY

[0005] To solve the above technical problems, the present application provides a wind turbine health degree evaluation and maintenance method, system, device and medium, which can improve the functions of wind turbine health degree state monitoring and fault diagnosis, not only reduce the monitoring difficulty, but also improve the maintenance efficiency, and can guarantee the reliability and operating efficiency of the wind turbine.

[0006] In order to achieve the above purpose, the technical scheme of the present application is as follows:

[0007] A wind turbine health degree evaluation and maintenance method, comprising:

[0008] Obtaining operating data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction processing on the operating data and multi-dimensional monitoring data to obtain cleaned and reconstructed data;

[0009] According to the SCADA data and the CMS data in the cleaning reconstruction data, SCADA fault diagnosis and CMS fault diagnosis are performed to obtain fault diagnosis data, a fault case library is established according to the fault diagnosis data, a health degree evaluation result is generated, and fault positioning is completed;

[0010] The wind turbine is completed state maintenance in combination with the individual combat system and the fault case library.

[0011] Further, the specific steps of cleaning the abnormal values in the operation data and the multi-dimensional monitoring data are as follows:

[0012] The abnormal values deviating from the normal power curve in the operation data and the multi-dimensional monitoring data are removed, and the abnormal values include the abandoned wind parameter and the incorrect wind speed parameter.

[0013] Further, the specific steps of reconstructing the missing values in the operation data and the multi-dimensional monitoring data are as follows:

[0014] The existing data in the operation data and the multi-dimensional monitoring data are used to complete the training of the conditional generative adversarial network by taking the working conditions corresponding to the existing data as conditions, and the data of unknown working conditions are generated by using the conditional generative adversarial network.

[0015] Further, the specific steps of SCADA fault diagnosis are as follows:

[0016] Based on the Pearson correlation coefficient, the correlation of all SCADA data and a certain fault parameter is solved to obtain correlation data; three models of deep neural network, adversarial variational autoencoder and spatio-temporal graph neural network are trained by using the correlation data, and SCADA fault diagnosis is completed by using the three trained models.

[0017] Further, the fault case library includes variable pitch system, gearbox system, generator system, blade and main bearing parameter data.

[0018] The fault case library further includes a fault diagnosis rule library, a device anomaly database and an algorithm library.

[0019] The fault diagnosis rule library is used to establish the fault diagnosis rules of SCADA and CMS, and provide diagnosis standards.

[0020] The device anomaly database is used to store the verified fault diagnosis data, and specifically includes a SCADA database, a CMS database and an operation load database.

[0021] The algorithm library is used to store the verified effective algorithms, and specifically includes a data cleaning algorithm and a diagnosis and early warning algorithm.

[0022] Further, a multi-dimensional standard health index curve is generated according to the fault diagnosis data, and the health degree of the wind turbine and its components is analyzed.

[0023] Further, the specific steps of completing the state maintenance of the wind turbine by combining the single-soldier combat system and the fault case library include:

[0024] According to the fault case library, a hidden danger investigation system is established, the hidden danger investigation system is used for collecting and deeply analyzing typical faults and representative cases of key components, dividing key system components and corresponding key hidden danger investigation units, and clearly defining hidden danger investigation focuses and implementation steps of key equipment;

[0025] According to the hidden danger investigation system and the single-soldier combat system, a state maintenance intelligent inspection system is established, and the state maintenance intelligent inspection system completes the state maintenance of the wind turbine according to the hidden danger investigation focuses and the implementation steps;

[0026] The single-soldier combat system includes a 5G+AR binocular intelligent helmet and an Internet of Things device.

[0027] A wind turbine health degree evaluation and maintenance system includes:

[0028] A cleaning and reconstruction module is configured to obtain operation data and multi-dimensional monitoring data of a wind turbine data center, perform outlier cleaning and missing value reconstruction processing on the operation data and the multi-dimensional monitoring data, and obtain cleaned and reconstructed data.

[0029] A health evaluation module is configured to perform SCADA fault diagnosis and CMS fault diagnosis according to SCADA data and CMS data in the cleaned and reconstructed data, obtain fault diagnosis data, establish a fault case library according to the fault diagnosis data, generate a health degree evaluation result, and complete fault positioning.

[0030] A state maintenance module is configured to complete state maintenance of the wind turbine by combining a single-soldier combat system and a fault case library.

[0031] An apparatus includes:

[0032] A memory is configured to store a computer program.

[0033] A processor is configured to implement steps of the above wind turbine health degree evaluation and maintenance method when executing the computer program.

[0034] A computer readable storage medium stores a computer program, and the computer program is configured to implement steps of the above wind turbine health degree evaluation and maintenance method when executed by a processor.

[0035] Compared with the prior art, the present application has the following beneficial effects:

[0036] The application provides a wind turbine health degree evaluation and maintenance method, which fully integrates existing information resources, breaks information islands and improves work efficiency by centrally acquiring and effectively utilizing operation data and multi-dimensional monitoring data of a data center, improves the accuracy of fault diagnosis by combining SCADA and CMS after abnormal value cleaning and missing value reconstruction, and further improves the accuracy of wind turbine health degree evaluation, thereby reducing monitoring difficulty, establishing a fault case library according to fault diagnosis data, generating health degree evaluation results, completing fault positioning, optimizing data center functions, and improving the functions of wind turbine health degree state monitoring and fault diagnosis.

[0037] The application also provides a wind turbine health degree evaluation and maintenance system, which can improve the functions of wind turbine health degree state monitoring and fault diagnosis, reduce monitoring difficulty, improve maintenance efficiency, and guarantee the reliability and operation efficiency of the wind turbine. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A flowchart of a wind turbine health degree evaluation and maintenance method provided by the embodiment of the application is shown in the figure.

[0039] Figure 2 A schematic diagram of a wind turbine health degree evaluation and maintenance system provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0040] The application provides a wind turbine health degree evaluation and maintenance method, as shown in the figure, which comprises: Figure 1 The application provides a wind turbine health degree evaluation and maintenance method, as shown in the figure, which comprises:

[0041] Operation data and multi-dimensional monitoring data of a wind turbine data center are acquired, and the operation data and multi-dimensional monitoring data are subjected to abnormal value cleaning and missing value reconstruction to obtain cleaned and reconstructed data.

[0042] The specific steps of abnormal value cleaning are as follows:

[0043] Abnormal values deviating from normal power curves in the operation data and multi-dimensional monitoring data are removed, and the abnormal values include abandoned wind parameters and incorrect wind speed parameters.

[0044] The specific steps of missing value reconstruction are as follows:

[0045] The existing data in the operation data and multi-dimensional monitoring data are used to train the conditional generative adversarial network by taking the corresponding working conditions as conditions, and the conditional generative adversarial network is used to generate data of unknown working conditions.

[0046] The SCADA fault diagnosis and the CMS fault diagnosis are performed according to the SCADA data and the CMS data in the cleaning reconstructed data, fault diagnosis data are obtained, a fault case library is established according to the fault diagnosis data, a health degree evaluation result is generated, and fault positioning is completed; or a multi-dimensional standard health index curve can be generated according to the fault diagnosis data to analyze the health degree of the wind turbine and components thereof.

[0047] The specific steps of the SCADA fault diagnosis are as follows:

[0048] Based on the Pearson correlation coefficient, the correlation of all SCADA data and a certain fault parameter is solved to obtain correlation data; three models of deep neural network, adversarial variational autoencoder and spatio-temporal graph neural network are trained using the correlation data, and the three trained models are used to complete the SCADA fault diagnosis.

[0049] The above fault case library includes parameter data of a variable pitch system, a gearbox system, a generator system, a blade and a main bearing;

[0050] It also includes a fault diagnosis rule library, an equipment anomaly database and an algorithm library.

[0051] The fault diagnosis rule library is used to establish the fault diagnosis rules of SCADA and CMS, and to provide diagnosis standards.

[0052] The equipment anomaly database is used to store fault diagnosis data that have been verified, and specifically includes a SCADA database, a CMS database and a running load database.

[0053] The algorithm library is used to store algorithms that have been verified to be effective, and specifically includes data cleaning algorithms and diagnosis and early warning algorithms.

[0054] The single-soldier combat system and the fault case library are combined to complete the condition-based maintenance of the wind turbine, and the specific steps include:

[0055] The hidden danger investigation system is established according to the fault case library, and is used to collect and deeply analyze typical faults and representative cases of key components, divide key system components and corresponding key hidden danger investigation units, and clarify the hidden danger investigation focus and implementation steps of key equipment;

[0056] The condition-based maintenance intelligent inspection system is established according to the hidden danger investigation system and the single-soldier combat system, and completes the condition-based maintenance of the wind turbine according to the hidden danger investigation focus and implementation steps.

[0057] The single soldier combat system comprises a 5G+AR binocular intelligent helmet and an Internet of Things device.

[0058] The application further provides a wind turbine health degree evaluation and maintenance system, comprising a cleaning and reconstruction module, a health evaluation module and a condition maintenance module; the cleaning and reconstruction module is used for obtaining operation data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction processing on the operation data and the multi-dimensional monitoring data, and obtaining cleaned and reconstructed data; the health evaluation module is used for performing SCADA fault diagnosis and CMS fault diagnosis according to SCADA data and CMS data in the cleaned and reconstructed data, obtaining fault diagnosis data, establishing a fault case library according to the fault diagnosis data, generating a health degree evaluation result, and completing fault positioning; and the condition maintenance module is used for completing condition maintenance of the wind turbine in combination with the single soldier combat system and the fault case library.

[0059] The application further provides a device, comprising a memory for storing a computer program and a processor for executing the computer program to realize the steps of the wind turbine health degree evaluation and maintenance method.

[0060] The processor executes the computer program to realize the steps of the wind turbine health degree evaluation and maintenance method, for example, obtaining operation data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction processing on the operation data and the multi-dimensional monitoring data, and obtaining cleaned and reconstructed data; performing SCADA fault diagnosis and CMS fault diagnosis according to SCADA data and CMS data in the cleaned and reconstructed data, obtaining fault diagnosis data, establishing a fault case library according to the fault diagnosis data, generating a health degree evaluation result, and completing fault positioning; and completing condition maintenance of the wind turbine in combination with the single soldier combat system and the fault case library.

[0061] Alternatively, the processor executes the computer program to realize the functions of the modules in the system, for example, the cleaning and reconstruction module, the health evaluation module and the condition maintenance module; the cleaning and reconstruction module is used for obtaining operation data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction processing on the operation data and the multi-dimensional monitoring data, and obtaining cleaned and reconstructed data; the health evaluation module is used for performing SCADA fault diagnosis and CMS fault diagnosis according to SCADA data and CMS data in the cleaned and reconstructed data, obtaining fault diagnosis data, establishing a fault case library according to the fault diagnosis data, generating a health degree evaluation result, and completing fault positioning; and the condition maintenance module is used for completing condition maintenance of the wind turbine in combination with the single soldier combat system and the fault case library.

[0062] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a preset function, which are used to describe the execution process of the computer program in the wind turbine health degree evaluation and maintenance device. For example, the computer program can be divided into a cleaning reconstruction module, a health evaluation module and a condition maintenance module; the specific functions of each module are as follows: the cleaning reconstruction module is used to obtain the operation data and multi-dimensional monitoring data of the wind turbine data center, and perform outlier cleaning and missing value reconstruction processing on the operation data and multi-dimensional monitoring data to obtain cleaned and reconstructed data; the health evaluation module is used to perform SCADA fault diagnosis and CMS fault diagnosis according to the SCADA data and the CMS data in the cleaned and reconstructed data, obtain fault diagnosis data, establish a fault case library according to the fault diagnosis data, generate a health degree evaluation result, and complete fault positioning; and the condition maintenance module is used to complete condition maintenance of the wind turbine in combination with a single soldier combat system and the fault case library.

[0063] The wind turbine health degree evaluation and maintenance device can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The wind turbine health degree evaluation and maintenance device can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above is an example of the wind turbine health degree evaluation and maintenance device, and does not constitute a limitation on the wind turbine health degree evaluation and maintenance device, and can include more components than the above, or combine certain components, or different components, for example, the wind turbine health degree evaluation and maintenance device can also include an input / output device, a network access device, a bus and the like.

[0064] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like, which is a control center of the wind turbine health degree evaluation and maintenance device, and connects each part of the wind turbine health degree evaluation and maintenance device through various interfaces and lines.

[0065] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the wind turbine health degree evaluation and maintenance equipment by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory.

[0066] The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0067] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the wind turbine health degree evaluation and maintenance equipment method.

[0068] The modules / units of the wind turbine health degree evaluation and maintenance equipment system can be stored in a computer readable storage medium if they are realized in the form of software function units and sold or used as independent products.

[0069] Based on such understanding, the application realizes all or part of the processes of the wind turbine health degree evaluation and maintenance equipment method, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer readable storage medium, and the computer program can realize the steps of the ring intersection channelization and signal timing optimization method when executed by a processor. The computer program includes computer program codes, which can be in the form of source codes, object codes, executable files or preset intermediate forms, etc.

[0070] The computer readable storage medium can include any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. that can carry the computer program codes.

[0071] It should be noted that the contents contained in the computer readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable storage medium does not include electrical carrier signals and telecommunication signals.

[0072] Embodiments

[0073] The application will be further described below in conjunction with the embodiments and the accompanying drawings:

[0074] As Figure 2 shown, the embodiment provides a wind turbine health assessment and maintenance system, specifically comprising:

[0075] a data resource cleaning module (i.e. a cleaning and reconstruction module), a health diagnosis module (i.e. a health assessment module) and a condition-based maintenance module.

[0076] 1. The function realized by the data resource cleaning module is specifically:

[0077] (1) An automatic cleaning function of wind turbine multi-source heterogeneous data is added in the data platform, which can automatically clean the SCADA and CMS data, and eliminate some abnormal values deviating from the normal power curve caused by wind abandonment, incorrect wind speed measurement, etc. The data cleaning steps are as follows: first, eliminate the data with power and speed less than or equal to 0; then, based on Gaussian process regression, eliminate the upper and lower quartile data to obtain the cleaned data.

[0078] (2) The conditional generative adversarial network is improved, the working condition corresponding to the existing data is taken as the condition, the network is trained by the existing data, and the trained network is used to generate data under unknown working conditions, so as to realize the reconstruction of missing values of multi-source heterogeneous data.

[0079] 2. The function realized by the health diagnosis module is specifically:

[0080] (1) SCADA fault diagnosis: For different faults of wind turbines, select a certain SCADA parameter that best reflects the fault, for example, the parameter that best reflects the "high gearbox oil temperature fault" is "gearbox lubricating oil temperature". Based on the Pearson correlation coefficient, the correlation between all SCADA parameters and the parameter reflecting a certain fault is calculated, and the top 5 parameters with the highest correlation are selected as the training set. The training set is input into the three models of deep neural network, adversarial variational autoencoder, and spatio-temporal graph neural network for training, and the model is optimized through back propagation by gradient descent method. The residual error between the predicted value and the true value is calculated, and the mean value of the residual error is μ and the standard deviation is σ. Set μ+3σ as the upper threshold and μ-3σ as the lower threshold. The test set is input into the trained model to obtain the predicted value, and the residual error between the predicted value and the actual value of the test set is calculated. If the residual error is outside the range of [μ-3σ, μ+3σ], the model alarms for the fault. Two of the three models alarm, and the fault is reported.

[0081] (2) CMS fault diagnosis: The improved data center can display the time domain waveform, frequency spectrum, power spectrum, envelope spectrum, and complex wavelet spectrum of the CMS data of the wind turbine in real time, and can provide fault diagnosis conclusions and state maintenance recommendations based on the analysis of the spectrum.

[0082] (3) Benchmark wind turbine important component fault diagnosis: Establish a wind turbine operation fault case library, which contains fault case data of important components such as pitch system, gearbox system, generator system, blade, and main bearing. The case library can compare and evaluate algorithms internally and provide diagnostic model evaluation services externally. The case library includes a fault diagnosis rule base, an equipment anomaly database, and an algorithm library. The fault diagnosis rule base establishes SCADA and CMS fault diagnosis rules and provides diagnostic standards; the equipment anomaly database contains data of confirmed faults, including SCADA database, CMS database, and operation load database; the algorithm library contains proven effective data cleaning algorithms and diagnostic warning algorithms.

[0083] (4) Health degree query: Based on the long short-term memory neural network, a prediction model of SCADA parameters is established, and the health data is used as the training set to train the neural network model. The test set is input into the trained model to obtain the predicted value, and the residual error between the predicted value and the actual value is calculated. The residual errors of SCADA parameters are weighted and summed (the weight is determined according to the fault maintenance cost, and the weight is high for high fault maintenance cost), integrated into a quantitative multi-dimensional standard health index curve, and the health degree of the unit and components is quantitatively analyzed, and the abnormal results are effectively presented; through the establishment of health quantitative index, the aging degree and fault severity of the unit and different components can be more intuitively displayed.

[0084] (5) Fault location: Add fault location of large components of wind turbine generators in the data platform, count all common faults of gearbox systems, generator systems, blades, variable pitch systems, and main bearings, and establish a fault tree. Use fault tree, ontology technology, Bayesian reasoning, and decision tree algorithm to update the wind power intelligent operation and maintenance system framework with knowledge graph, realize fault diagnosis decision, cognitive diagnosis, targeted dispatch, and perform function testing. Analyze and troubleshoot possible fault locations and provide more accurate fault troubleshooting suggestions according to probability.

[0085] 3. The functions implemented by the condition-based maintenance module are as follows:

[0086] (1) 5G+AR single soldier combat system: Introduce 5G+AR binocular intelligent helmet, build IOT Internet of Things device access system, realize docking of AR platform and on-site Internet of Things device system, including environmental monitoring and device monitoring type device system. Through system docking, the data content is presented and displayed on the AR glasses end to empower the front-line wearers, and the perception ability of the front-line personnel is improved through interconnection. The intelligent single soldier system mainly includes: (1) Daily intelligent inspection. During the inspection process, the intelligent single soldier system can obtain real-time data through video monitoring, Beidou positioning, etc., and facilitate communication with the back-end management personnel; (2) Emergency command and dispatch. The management personnel can communicate with the single soldier, group or all on-site inspection single soldiers at any time, and adjust the inspection task at any time; (3) Expert assisted troubleshooting. When encountering maintenance problems, online communication can be carried out directly through voice and video, and the technical experts are sent the on-site device operation parameters in a timely manner through data sharing, so as to quickly solve the problems, improve the efficiency and reduce the maintenance cost.

[0087] (2) Inspection and investigation key system: Collect and deeply analyze typical faults and representative cases of each system and key components, divide key system components and corresponding key hidden trouble investigation units. Summarize, analyze and sort out the accident types, investigation methods and results of the key hidden trouble investigation units, and use the analysis results as the key points for compiling the key equipment hidden trouble investigation system. Give the key points and implementation steps of each key equipment hidden trouble investigation, and realize the construction of the wind power equipment key accident hidden trouble investigation system.

[0088] (3) Condition-based maintenance intelligent inspection system: Based on the hidden trouble investigation system, combined with the condition-based maintenance single soldier system, establish a 5G+AR intelligent inspection system to guide the inspection personnel to strictly follow the implementation steps to complete the inspection work, realize the full-process record of the inspection work and the management closed loop of the inspection quality evaluation. Through the combination of big data, intelligent fault diagnosis and intelligent 5G+AR single soldier combat system, realize the full-process closed-loop management and quality improvement of the maintenance, and help the inspection personnel to determine the maintenance strategy.

[0089] The embodiment provides a wind turbine health degree evaluation and maintenance system, and the principle is as follows:

[0090] The system adds a data resource cleaning module, a health quantization evaluation module and a condition-based maintenance module to the wind turbine data center, optimizes the function of the data center, and realizes intelligent condition monitoring of the wind turbine;

[0091] The system adds a data resource cleaning module, realizes cleaning and missing value reconstruction of multi-source heterogeneous data of the data center. In combination with the operation mechanism of the wind turbine, mathematical statistics and probability theory knowledge, abnormal values are eliminated; the generative adversarial network is improved to realize missing value reconstruction of multi-source heterogeneous data;

[0092] In addition, the system also adds a case library function of important components of the benchmark wind turbine; develops multi-dimensional data diagnosis models of important components such as the pitch system, gearbox, generator and blade of the benchmark wind turbine, tracks the diagnosis results, and realizes model iteration and upgrading.

[0093] In particular, the system adds a health degree query function. Through correlation analysis, boundary recognition technology, long short-term memory neural network and the like, the operation health level of the parameters is integrated into a quantitative multi-dimensional standard health index curve, the health degree of the unit and components is quantitatively analyzed, and the aging degree and fault severity of the unit and different components are intuitively displayed.

[0094] In addition, the system adds a fault positioning function. By using a knowledge graph, through knowledge representation learning, knowledge fusion, graph analysis calculation, knowledge reasoning and the like, possible fault positions are analyzed and investigated, and more accurate fault investigation suggestions are provided according to the probability.

[0095] By introducing a 5G+AR binocular intelligent helmet, an IOT Internet of Things device access system is constructed, and the AR platform is connected with the on-site Internet of Things device system.

[0096] In combination with the single soldier combat system, a hidden danger investigation unit is added. Key system components and corresponding key hidden danger investigation units are divided, and the accident types, investigation methods and results and common experiences of the key hidden danger investigation units are summarized, counted, analyzed and sorted out, the key and implementation steps of each key equipment hidden danger investigation are given, the wind power equipment key accident hidden danger investigation system is constructed, a 5G+AR intelligent inspection system is added. On the basis of the hidden danger investigation system, the 5G+AR intelligent inspection system is established in combination with the condition-based maintenance single soldier system, the inspection personnel are guided to strictly complete the inspection work according to the implementation steps, the whole process record of the inspection work and the management closed loop of the inspection quality evaluation are realized.

[0097] The embodiment provides a wind turbine health degree evaluation and maintenance method and system, which has the following advantages:

[0098] 1. The embodiment adopts three methods of deep neural network, adversarial variational autoencoder and spatio-temporal graph neural network to realize deep mining of data and improve early warning accuracy.

[0099] 2. The embodiment proposes a health degree determination method, which can quantitatively and intuitively show the fault degree of different wind turbines.

[0100] 3. The system proposed in the embodiment adds the functions of case library, hidden danger investigation and maintenance optimization, and has higher engineering application value.

[0101] 4. The embodiment adds a 5G+AR intelligent inspection system, which combines big data, intelligent fault diagnosis and intelligent 5G+AR single soldier combat system to realize maintenance whole-process closed-loop management and quality improvement.

[0102] The above embodiment is only one of the implementation manners of the technical scheme of the present application, and the scope of protection of the present application is not limited to the above embodiment, but also includes any changes, substitutions and other implementation manners easily thought of by those skilled in the art within the technical scope disclosed by the present application.

Claims

1. A wind turbine health assessment and service method, characterized in that, The method comprises the following steps: obtaining operation data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction on the operation data and multi-dimensional monitoring data to obtain cleaned and reconstructed data; performing SCADA fault diagnosis and CMS fault diagnosis according to the SCADA data and the CMS data in the cleaned and reconstructed data, establishing a fault case library according to the fault diagnosis data, generating a health degree evaluation result, and completing fault positioning; combining a single soldier combat system and the fault case library to complete condition-based maintenance of the wind turbine, the specific steps comprising: establishing a hidden danger investigation system according to the fault case library, the hidden danger investigation system being used for collecting and deeply analyzing typical faults and representative cases of key components, dividing key system components and corresponding key hidden danger investigation units, and clearly defining hidden danger investigation focuses and implementation steps of key equipment; establishing a condition-based maintenance intelligent inspection system according to the hidden danger investigation system and the single soldier combat system, the condition-based maintenance intelligent inspection system completing condition-based maintenance of the wind turbine according to the hidden danger investigation focuses and the implementation steps; the single soldier combat system comprising a 5G+AR binocular intelligent helmet and an Internet of Things device; the specific steps of the SCADA fault diagnosis are: based on a Pearson correlation coefficient, solving the correlation of all SCADA data and a certain fault parameter to obtain correlation data; training three models of a deep neural network, an adversarial variational autoencoder and a spatio-temporal graph neural network using the correlation data, and completing SCADA fault diagnosis using the three trained models.

2. A wind turbine health assessment and service method in accordance with claim 1, wherein, The specific steps of performing outlier cleaning on the operation data and the multi-dimensional monitoring data are: eliminating outliers deviating from a normal power curve in the operation data and the multi-dimensional monitoring data, the outliers including abandoned wind parameters and incorrect wind speed parameters.

3. A wind turbine health assessment and service method in accordance with Claim 1, wherein, The specific steps of performing missing value reconstruction on the operation data and the multi-dimensional monitoring data are: using existing data in the operation data and the multi-dimensional monitoring data to complete training of a conditional generative adversarial network by taking the working conditions corresponding to the data as conditions, and generating data of unknown working conditions using the conditional generative adversarial network.

4. A wind turbine health assessment and service method in accordance with Claim 1, wherein, The fault case library comprises parameter data of a variable pitch system, a gearbox system, a generator system, a blade and a main bearing; the fault case library further comprises a fault diagnosis rule library, an equipment anomaly database and an algorithm library; the fault diagnosis rule library is used for establishing fault diagnosis rules of SCADA and CMS, and providing diagnosis standards; the equipment anomaly database is used for storing verified fault diagnosis data, and specifically comprises an SCADA database, a CMS database and an operation load database; the algorithm library is used for storing verified effective algorithms, and specifically comprises data cleaning algorithms and diagnosis and early warning algorithms.

5. A wind turbine health assessment and service method in accordance with Claim 1, wherein, A multi-dimensional standard health index curve is generated according to the fault diagnosis data to analyze the health degree of the wind turbine and its components.

6. A wind turbine health assessment and service system, characterized in that, The steps for implementing the wind turbine health degree evaluation and maintenance method of any one of claims 1-5 comprise: a cleaning and reconstruction module for obtaining operation data and multi-dimensional monitoring data of a wind turbine data center, performing outlier cleaning and missing value reconstruction on the operation data and multi-dimensional monitoring data to obtain cleaned and reconstructed data; The health assessment module is used for SCADA fault diagnosis and CMS fault diagnosis according to the SCADA data and the CMS data in the cleaning reconstruction data, obtaining fault diagnosis data, establishing a fault case library according to the fault diagnosis data, generating a health degree evaluation result, and completing fault positioning; The condition-based maintenance module is used for completing condition-based maintenance of the wind turbine by combining the single-soldier combat system and the fault case library, and the specific steps include: A hidden danger investigation system is established according to the fault case library, and the hidden danger investigation system is used for collecting and deeply analyzing typical faults and representative cases of key components, dividing key system components and corresponding key hidden danger investigation units, and clearly defining hidden danger investigation focuses and implementation steps of key equipment; A condition-based maintenance intelligent inspection system is established according to the hidden danger investigation system and the single-soldier combat system, and the condition-based maintenance intelligent inspection system completes condition-based maintenance of the wind turbine according to the hidden danger investigation focuses and the implementation steps. The single-soldier combat system includes a 5G+AR binocular intelligent helmet and an Internet of Things device.

7. An apparatus, comprising: It includes: A memory for storing a computer program; A processor for executing the computer program to realize the steps of the wind turbine health degree evaluation and maintenance method in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: The computer program is executed by the processor to realize the steps of the wind turbine health degree evaluation and maintenance method in any one of claims 1-5.

Images