WO2017113998A1 - 计算机存储介质、计算机程序产品、风电机组故障监测方法和装置 - Google Patents

计算机存储介质、计算机程序产品、风电机组故障监测方法和装置 Download PDF

Info

Publication number
WO2017113998A1
WO2017113998A1 PCT/CN2016/105310 CN2016105310W WO2017113998A1 WO 2017113998 A1 WO2017113998 A1 WO 2017113998A1 CN 2016105310 W CN2016105310 W CN 2016105310W WO 2017113998 A1 WO2017113998 A1 WO 2017113998A1
Authority
WO
WIPO (PCT)
Prior art keywords
component
fault
wind turbine
dimensional model
fault monitoring
Prior art date
Application number
PCT/CN2016/105310
Other languages
English (en)
French (fr)
Inventor
乔志强
唐新安
李康
Original Assignee
北京金风科创风电设备有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京金风科创风电设备有限公司 filed Critical 北京金风科创风电设备有限公司
Priority to EP16880789.9A priority Critical patent/EP3399297B1/en
Priority to US15/567,809 priority patent/US10760551B2/en
Priority to KR1020187000957A priority patent/KR102011227B1/ko
Priority to AU2016380302A priority patent/AU2016380302C1/en
Priority to ES16880789T priority patent/ES2778692T3/es
Publication of WO2017113998A1 publication Critical patent/WO2017113998A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M15/00—Testing of engines
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
    • F03D—WIND MOTORS
    • F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F01—MACHINES OR ENGINES IN GENERAL; ENGINE PLANTS IN GENERAL; STEAM ENGINES
    • F01D—NON-POSITIVE DISPLACEMENT MACHINES OR ENGINES, e.g. STEAM TURBINES
    • F01D15/00—Adaptations of machines or engines for special use; Combinations of engines with devices driven thereby
    • F01D15/10—Adaptations for driving, or combinations with, electric generators
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2250/00—Geometry
    • F05B2250/20—Geometry three-dimensional
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2260/00—Function
    • F05B2260/84—Modelling or simulation

Definitions

  • the invention relates to monitoring technology, in particular to a computer storage medium, a computer program product, a wind turbine fault monitoring method and device.
  • a wind turbine fault monitoring method including:
  • the identified fault is presented using a three-dimensional model of the component.
  • a wind turbine fault monitoring device including:
  • Identification module for fault identification of components in the wind turbine
  • a rendering module for presenting the identified fault using a three-dimensional model of the component.
  • a computer storage medium comprising:
  • a computer program stored in a computer storage medium the computer program causing the computer to perform the wind turbine fault monitoring method described above.
  • a computer program product comprising:
  • a computer program readable by a computer storage medium the program causing a computer to perform the wind turbine fault monitoring method described above.
  • FIG. 1 is a schematic flow chart of a method for monitoring faults of a wind turbine according to an embodiment of the present invention
  • FIG. 2 is a schematic flow chart of a method for monitoring faults of a wind turbine set according to an embodiment of the present invention
  • Figure 3 is a flow chart of data analysis in the AWS cloud
  • Figure 4 is a general analysis flow chart
  • Figure 5 is a flow chart of fault matching and processing
  • FIG. 6 is a schematic structural diagram of a wind turbine fault monitoring device according to an embodiment of the present invention.
  • FIG. 7 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • FIG. 8 is a schematic structural diagram of another wind turbine fault monitoring device according to an embodiment of the present invention.
  • FIG. 9 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • FIG. 10 is a schematic structural diagram of another wind turbine fault monitoring device according to an embodiment of the present invention.
  • FIG. 1 is a schematic flowchart of a fault monitoring method for a wind turbine set according to an embodiment of the present invention. As shown in FIG. 1 , the method includes:
  • Step 101 Perform fault identification on components in the wind turbine.
  • vibration, sound, temperature, and imaging sensors are first installed on the components of the wind turbine, and these sensors are used to collect the sensor data sets of the components.
  • the sensor data set includes vibration parameters, sound parameters, temperature parameters, and image parameters.
  • the sensor data set collected by the sensor is sent to the industrial computer through a programmable logic controller (PLC), and the sensor data sets are further uploaded to the AWS cloud through the industrial computer, wherein AWS is a cloud service.
  • PLC programmable logic controller
  • AWS is a cloud service.
  • the following three acquisition methods may be included, but not limited to:
  • the first acquisition method uses a vibration sensor, an acoustic sensor, a temperature sensor, and an imaging sensor, respectively, so that sound, vibration, temperature, and image parameters are separately collected.
  • the sound and vibration are collected by the same sensor, and the collected signals are inferred according to the function relationship between the sound and the vibration, and the values of the sound parameters and the vibration parameters are respectively obtained.
  • the sound and temperature are collected by the same sensor, and the collected signals are inferred according to a function relationship between sound and temperature, and the values of the sound parameters and the temperature parameters are respectively obtained.
  • the acquisition of the imaging parameters may be implemented by using a video recording method, or by using a three-dimensional modeling technique, the signals collected by the vibration sensor, the sound sensor, and the temperature sensor may be processed to obtain the reproduced image data, and the image will be reproduced.
  • the value of the data as an image parameter.
  • the sensor data set is analyzed and processed in the AWS cloud.
  • the pre-trained machine learning model is used to process the sensor data set to obtain the feature test matrix.
  • the row vector of the feature check matrix is used to indicate the component
  • the column vector is used to indicate the physical parameter. Then, the fault identification is performed according to the feature check matrix.
  • the machine learning model can be a BP neural network, and the vibration, sound, temperature and/or image parameters of each component collected by the sensor are used as input parameters of the BP neural network, and the inverse propagation (Back) Propagation, BP) neural network performs principal component analysis, correlation analysis and/or cluster analysis to obtain feature test matrix. If the value of the physical parameter of the component in the feature test matrix matches the value of the physical parameter in the design state, it is determined that the component is operating normally; otherwise, the component is faulty.
  • BP inverse propagation
  • the target fault mode matching the value of the physical parameter of the component in the feature check matrix may also be queried in the fault database.
  • the reason why the AWS cloud can analyze and process the sensor data set based on the sensor acquisition component to determine whether the component corresponding to the signal is faulty depends on the physical parameters of the component derived from the sensor data set. Describe the operational status of the component.
  • a wind turbine is a large system consisting of multiple subsystems, each of which consists of several mechanical components that are matched to a certain standard. The standard is pre-determined based on the functional data of the component and the geometric data. Therefore, the functional data and the geometric data can be used to describe the component.
  • the set of functional data and geometric data is the physical parameter of the component.
  • Physical parameters include, but are not limited to, shaft torque T, natural frequency H, friction force F f , reflected light intensity U, shaft transmitted power W, output power P, speed N, pressure F p , wind speed V, impeller diameter D, zero Component quality M and / or pixel O.
  • the shaft torque T, the friction force F f and the pressure F p are mechanical parameters.
  • Step 102 The three-dimensional model of the component is used to present the identified fault.
  • the identified faulty component extract the mechanical parameter from each physical parameter of the faulty component, and determine the value of the mechanical parameter of the faulty component according to the feature test matrix, and the mechanics
  • the value of the parameter is used as a boundary condition, and the three-dimensional model in which the faulty component is in a normal state is subjected to stress analysis to obtain a three-dimensional model exhibiting a stress state.
  • the target three-dimensional model corresponding to the target failure mode obtained by the fault identification in step 101 is called, and the target three-dimensional model is presented.
  • the 3D model library of the component is used to store the 3D model in the normal state, and the 3D model in each failure mode can be further stored.
  • the 3D model library of the component can store the 3D model of each component in each failure mode and the 3D model in the normal state.
  • the three-dimensional model of the corresponding component may be determined in the three-dimensional model library of the component according to the value of the physical parameter in the feature check matrix obtained in the previous step.
  • the three-dimensional model of the corresponding component may also be determined directly according to the component indicated by the feature check matrix row vector.
  • the reason why the three-dimensional model of the corresponding component is determined in the three-dimensional model library of the component according to the value of the physical parameter in the feature check matrix is that the component has some relatively stable physical parameters, such as geometric parameters and material parameters, thereby Based on this, the components corresponding to the three-dimensional model are identified.
  • the natural frequency of an object is a physical characteristic of the object, which is determined by the structure, size, shape, and material characteristics of the object. Therefore, the natural frequency can be calculated according to the value of the vibration parameter obtained by the test, and then calculated. Physical parameters such as the structure, size, shape, and material of the component. The following formula has a frequency calculation:
  • a most suitable three-dimensional model based on the parameters related to the operating state and geometry, such as: torque T, section moment of inertia I R , section torsion angle variation
  • the length variation d x The length variation d x , the area division micro-e d, the shaft-transferred power W, the shaft rotation speed N, the shaft radius R, the shear force ⁇ R at the radius R, the shear ⁇ R at the radius R , the torsional stiffness G, and Torsion section coefficient W R .
  • torque T section moment of inertia I R
  • section torsion angle variation The length variation d x , the area division micro-e d, the shaft-transferred power W, the shaft rotation speed N, the shaft radius R, the shear force ⁇ R at the radius R, the shear ⁇ R at the radius R , the torsional stiffness G, and Torsion section coefficient W R .
  • the relationship between these parameters is as follows:
  • position sensors can be added to the component to collect position parameters, such as GPS sensors, and the corresponding relationship between the three-dimensional model and the position parameters can be established in advance, so that the corresponding three-dimensional model can be determined according to the position of the collected components.
  • the three-dimensional model of the normal state stored in the three-dimensional model library of the component may be obtained by three-dimensional modeling in advance based on the design value of the physical parameter of the component.
  • the design values of the W, T, N, M, K, G, W R , H, and O parameters are known for each component, according to the mapping function f: X (W, T, N, M, K, G , W R , H, O) ' ⁇ Y(x m , x g , x f )' is calculated to obtain the material coefficient x m , the geometric coefficient x g and the functional coefficient x f .
  • the fault is presented by using the three-dimensional model of the component for the identified fault, so that the fault monitor can visually observe the component.
  • the failure is not only beneficial to take effective fault handling measures before the slight faults become serious faults, thereby reducing the fault hazards, and reducing the experience requirements of the fault monitoring personnel, and avoiding the experience of the fault monitoring personnel in the prior art.
  • FIG. 2 is a schematic flowchart of a fault monitoring method for a wind turbine set according to an embodiment of the present invention.
  • the fault monitoring method in this embodiment may be used for all components in a wind turbine, or may be only for one or more wind turbines.
  • the components in the sub-system are not limited in this embodiment, as shown in FIG. 2, including:
  • Step 201 The sensor data set of the sensor collection component is sequentially uploaded to the AWS cloud through the PLC controller and the industrial computer.
  • the sensor data set includes but is not limited to: vibration, sound, temperature, and image parameters.
  • Step 202 The AWS cloud performs data analysis according to the sensor data set.
  • FIG. 3 is a flow chart of data analysis in the AWS cloud, as shown in Figure 3, including:
  • Step 2021 After the AWS cloud receives the sensing data set received through the data interface, perform a data cleaning operation.
  • Step 2022 Select to perform a general analysis process or an advanced analysis process.
  • a threshold range of vibration, sound, temperature, and image parameters is set in advance for each component, and when any parameter in the sensor data set exceeds a threshold range, an advanced analysis process is performed, thereby performing more accurate and detailed fault analysis. Otherwise perform a general analysis.
  • the general analysis process or the advanced analysis process may be selected according to the user's settings.
  • the general analysis process is faster, but the accuracy is slightly lower, while the advanced analysis process is slower, but the accuracy is lower. high.
  • Step 2023 performing a general analysis process.
  • FIG. 4 is a general analysis flow chart, as shown in FIG. 4, according to the vibration of the component.
  • the sound, temperature and image parameters are input to state functions corresponding to different state modes, such as state function 1 to state function n, n is the number of states, and each state function outputs a judgment value, and then according to the judgment value of each state function.
  • Analysis determining the most matching state mode from each state mode. If the matching status mode is the normal status mode, the component has no fault and is operating normally. Otherwise, make sure the component is faulty.
  • Step 2024 performing an advanced analysis process.
  • the vibration, sound, temperature and image parameters of the component are input into the BP neural network for principal component analysis, correlation analysis and/or cluster analysis to obtain a feature test matrix.
  • the parameter vector of the feature test matrix is
  • X1 to x12 represent 12 characteristic parameters, and each characteristic parameter corresponds to one physical parameter, and the corresponding relationship is as follows:
  • n N′, representing N′ components of the wind turbine.
  • Step 2025 Establish a component model according to the analysis result.
  • the analysis result is mainly used to establish a component model, and the model is characterized by data, and the data may be the feature check matrix obtained in step 2024, or may be the value of the state function in step 2023.
  • Step 203 The AWS cloud presents a three-dimensional model of the component.
  • the 12 ⁇ N′ feature test matrix calculated by the BP neural network is input according to the vibration, sound, temperature and image parameters, and the mechanical parameters are extracted from the physical parameters of the faulty component to determine the faulty component.
  • Step 204 If it is determined that there is a fault, the AWS cloud matches the analysis result with the fault database to determine a target failure mode of the component.
  • FIG. 5 is a flowchart of fault matching and processing, as shown in FIG. 5, including:
  • Step 2041 Match the analysis result with the fault database.
  • the target failure mode matching the value of the physical parameters of the component in the feature check matrix can be queried in the fault database.
  • the fault database the range of the physical parameters of the components in the feature check matrix in each fault mode is described, and the target fault mode to which the component belongs is determined.
  • the corresponding target failure mode is determined according to one of the most matching state patterns determined by the analysis.
  • Step 2042 If there is no matching target failure mode in the fault database, the pairing fails. Furthermore, a failure mode can be added to the fault database, and the feature check matrix and/or the sensor data set corresponding to the faulty component can be stored in the corresponding position of the added fault mode.
  • Step 2043 If there is a matching target failure mode in the fault database, the pairing is successful.
  • Step 205 Processing the fault.
  • the fault is processed by calling a solution database, including:
  • Step 2051 Query the solution database to obtain a solution corresponding to the target failure mode of the component.
  • the solution database describes the related solutions to the faults that have occurred.
  • Step 2052 If the solution is queried, the pairing is successful.
  • Step 2053 If the pairing is successful, determine whether the solution can be executed to eliminate the fault of the component.
  • Step 2054 if yes, execute the solution.
  • Step 2055 otherwise, the solution is output for manual resolution.
  • Step 2056 if the solution is not queried, the pairing fails.
  • the solution corresponding to the target failure mode does not exist in the solution database, determine that the target failure mode is a new failure. After the technician needs to further analyze the newly added faults, for example, based on the relevant data of the components, such as the sensor data set and/or the feature check matrix, the fault cause determination and the solution are formulated, and the solution is added to the solution. In the database.
  • the fault is presented by using the three-dimensional model of the component for the identified fault, so that the fault monitor can visually observe the component.
  • the failure is not only beneficial to take effective fault handling measures before the slight faults become serious faults, thereby reducing the fault hazards, and reducing the experience requirements of the fault monitoring personnel, and avoiding the experience of the fault monitoring personnel in the prior art.
  • the BP neural network is used to identify the fault according to the physical parameters in the design state of each component, the accuracy of the fault identification is improved, and when the new fault occurs, the relevant data of the component is stored in the fault database. In order to continuously improve the fault database.
  • FIG. 6 is a schematic structural diagram of a wind turbine fault monitoring device according to an embodiment of the present invention. As shown in FIG. 6, the method includes: an identification module 61 and a presentation module 62.
  • the identification module 61 is configured to perform fault identification on components in the wind turbine.
  • the presentation module 62 is configured to present the identified fault using a three-dimensional model of the component.
  • FIG. 7 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • the identification module 61 includes an input unit 601 and a determining unit 602.
  • the input unit 601 is configured to input a sensor data set of the components collected by the sensor into a state mode function corresponding to each state, and obtain each judgment value.
  • the determining unit 602 is configured to determine, according to the determination value of each state function, the most matching state mode from each state mode, and if the matching state mode is the normal state mode, the component does not exist. Fault, otherwise, determine the matching state mode as the target failure mode for the component.
  • the wind turbine fault monitoring device further includes: a query module 63 and a processing module 64.
  • the query module 63 is configured to query the solution database to obtain a solution corresponding to the target failure mode of the component.
  • the processing module 64 is configured to troubleshoot components according to the solution, and/or output the solution.
  • the fault is presented by using the three-dimensional model of the component for the identified fault, so that the fault monitor can visually observe the component.
  • the failure is not only beneficial to take effective fault handling measures before the slight faults become serious faults, thereby reducing the fault hazards, and reducing the experience requirements of the fault monitoring personnel, and avoiding the experience of the fault monitoring personnel in the prior art.
  • FIG. 8 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • the identification module 61 includes: an acquisition unit 611, an analysis unit 612, and an identification unit 613. .
  • the collecting unit 611 is configured to collect the sensing data set of the component by using the sensor.
  • the collecting unit 611 is specifically configured to collect the values of the vibration, sound, temperature, and/or image parameters of the component by using the sensor, and take the values of the vibration, sound, temperature, and/or image parameters as the sensing data set.
  • the analyzing unit 612 is configured to perform data analysis on the sensing data set by using a pre-trained machine learning model to obtain a feature checking matrix.
  • the analyzing unit 612 is specifically configured to perform a principal component analysis, a correlation analysis, and/or a cluster analysis on the sensing data set to obtain a feature checking matrix by using a pre-trained BP neural network.
  • the row vector of the feature check matrix is used to indicate the component
  • the column vector is used to indicate the physical parameter
  • physical parameters include, but are not limited to, shaft torque T, natural frequency H, friction force F f , reflected light intensity U, shaft transmitted power W, output power P, speed N, pressure F p , wind speed V, impeller diameter D, Component quality M and / or pixel O.
  • the shaft torque T, the friction force F f and the pressure F p are mechanical parameters.
  • the identifying unit 613 is configured to perform fault identification according to the feature check matrix.
  • the identification unit 613 includes: a determination subunit 6131 and a query subunit 6132.
  • the determining sub-unit 6131 is configured to determine that the component is in normal operation if the value of the physical parameter of the component in the feature checking matrix matches the value of the physical parameter in the design state; otherwise, the component is faulty.
  • the query subunit 6132 is configured to query, in the fault database, a target fault mode that matches the value of the physical parameter of the component in the feature check matrix.
  • the wind turbine fault monitoring device further includes: a query module 63 and a processing module 64.
  • the query module 63 is configured to query the solution database to obtain a solution corresponding to the target failure mode of the component.
  • the processing module 64 is configured to troubleshoot components according to the solution, and/or output the solution.
  • the presentation module 62 presents the fault to the fault identified by the identification module 61, and uses the three-dimensional model of the component to present the fault, so that the fault monitor can intuitively observe Failure to the parts and components not only helps to take effective fault handling measures before the minor faults become serious faults, thereby reducing the fault hazards, and reducing the experience requirements of the fault monitoring personnel, avoiding the fault monitoring personnel in the prior art.
  • the processing module 64 performs fault processing on the component according to the solution corresponding to the fault queried by the query module 63, and/or outputs the solution, so that the wind turbine can automatically handle the fault and reduce the fault monitoring.
  • the workload of personnel improves the efficiency of fault handling.
  • FIG. 9 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • the presentation module 62 further includes: an extracting unit. 621.
  • the extracting unit 621 is configured to extract mechanical parameters from physical parameters of the faulty component for the identified faulty component.
  • the determining unit 622 is configured to determine, according to the feature check matrix, a value of a mechanical parameter of the component that is faulty.
  • the first rendering unit 623 is configured to perform stress analysis on the three-dimensional model in which the faulty component is in a normal state, and obtain a three-dimensional model exhibiting a stress state.
  • FIG. 10 is a schematic structural diagram of another wind turbine fault monitoring apparatus according to an embodiment of the present invention.
  • the presentation module 62 further includes : Invoking unit 624 and second rendering unit 625.
  • the calling unit 624 is configured to invoke a target three-dimensional model corresponding to the target failure mode from the three-dimensional model library of the component;
  • the second rendering unit 625 is configured to present a target three-dimensional model.
  • the fault is presented by using the three-dimensional model of the component for the identified fault, so that the fault monitor can intuitively observe the component. Failures are not only beneficial to take effective fault handling measures before a minor fault evolves into a serious fault, thereby reducing the fault hazard, and reducing the experience requirements of the fault monitor personnel, avoiding the failure of the prior art fault monitor personnel to rely on personnel experience.
  • a two-dimensional curve is analyzed to identify the process by which a component has a fault.
  • the embodiment of the invention further provides a computer storage medium and/or a computer program product, wherein the computer storage medium stores a computer program, the computer program product comprising a computer program readable by a computer storage medium, the computer program The wind turbine fault monitoring method in the above embodiment of the computer.
  • the aforementioned program can be stored in a computer readable storage medium.
  • the program when executed, performs the steps including the foregoing method embodiments; and the foregoing storage medium includes various media that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Mechanical Engineering (AREA)
  • General Engineering & Computer Science (AREA)
  • Sustainable Development (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Sustainable Energy (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Wind Motors (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

一种风电机组故障监测方法和装置，所述方法包括以下步骤：对风电机组中的零部件进行故障识别（101）；采用所述零部件的三维模型对所识别出的故障进行呈现（102）；所述装置包括相应的识别模块（61）和呈现模块（62）；通过该方法和装置使得故障监测人员能够直观地观察到零部件所出现的故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。

Description

计算机存储介质、计算机程序产品、风电机组故障监测方法和装置 技术领域
本发明涉及监测技术，尤其涉及一种计算机存储介质、计算机程序产品、风电机组故障监测方法和装置。
背景技术
随着风力发电的快速发展，风电机组获得了广泛的应用。但在风电机组的应用过程中，不可避免地会产生一些故障，如何有效监测故障从而保障风电机组的正常运转成为亟待解决的问题。
针对风电机组的故障监测，目前往往仅能给出对风电机组中的零部件进行测量之后所获得的二维曲线结果，这种二维曲线所反映的部件故障不够直观，往往还需要有经验的故障监测人员依靠人员经验，对二维曲线进行分析以识别出零部件是否具有故障。
发明内容
为达到上述目的，本发明的实施例采用如下技术方案：
第一方面，提供了一种风电机组故障监测方法，包括：
对风电机组中的零部件进行故障识别；
采用零部件的三维模型对所识别出的故障进行呈现。
第二方面，提供了一种风电机组故障监测装置，包括：
识别模块，用于对风电机组中的零部件进行故障识别；
呈现模块，用于采用零部件的三维模型对所识别出的故障进行呈现。
第三方面，提供一种计算机存储介质，包括：
在计算机存储介质中存储的计算机程序，所述计算机程序使得计算机执行上述风电机组故障监测方法。
第四方面，提供一种计算机程序产品，包括：
计算机存储介质可读的计算机程序，所述程序使得计算机执行上述的风电机组故障监测方法。
附图说明
图1为本发明实施例提供的一种风电机组故障监测方法的流程示意图；
图2为本发明实施例提供的一种风电机组故障监测方法的流程示意图；
图3为AWS云端的数据分析流程图；
图4为一般分析流程图；
图5为故障匹配和处理的流程图；
图6为本发明实施例提供的一种风电机组故障监测装置的结构示意图；
图7为本发明实施例提供的另一种风电机组故障监测装置的结构示意图；
图8为本发明实施例提供的另一种风电机组故障监测装置的结构示意图；
图9为本发明实施例提供的另一种风电机组故障监测装置的结构示意图；
图10为本发明实施例提供的另一种风电机组故障监测装置的结构示意图。
具体实施方式
下面结合附图，对本发明的一些实施方式作详细说明。在不冲突的情况下，下述的实施例及实施例中的特征可以相互组合。
图1为本发明实施例提供的一种风电机组故障监测方法的流程示意图，如图1所示，方法包括：
步骤101、对风电机组中的零部件进行故障识别。
具体的，首先在风电机组的零部件上安装振动、声音、温度和成像传感器，利用这些传感器采集零部件的传感数据集。其中，传感数据集包括振动参数、声音参数、温度参数和图像参数。将传感器所采集到的传感数据集通过可编程逻辑控制器(Programmable Logic Controller，PLC)发送到工控机上，进一步通过工控机将这些传感数据集上传至AWS云端，其中AWS是一种云服务平台，从而在AWS云端对传感数据集进行分析处理，以确定传感数 据集对应的零部件是否存在故障。
例如：在采集这些传感数据集的过程中，可以包括但不限于以下三种采集方法：
第一种采集方法，分别采用振动传感器、声音传感器、温度传感器和成像传感器，从而声音、振动、温度、图像参数分别独立采集。
第二种采集方法，声音和振动采用同一传感器采集，依据声音与振动之间的函数关系对采集到的信号进行推理，分别获得声音参数和振动参数的取值。
第三种采集方法，声音和温度采用同一传感器采集，依据声音与温度之间的函数关系对采集到的信号进行推理，分别获得声音参数和温度参数的取值。
其中，针对成像参数的采集，可以采用视频录制方式实现，也可以借助三维建模技术，对振动传感器、声音传感器和温度传感器等所采集到的信号进行处理，从而获得再现图像数据，将再现图像数据作为图像参数的取值。
然后，在AWS云端对传感数据集进行分析处理具体采用预先训练的机器学习模型，对传感数据集进行处理，获得特征检验矩阵。其中，特征检验矩阵的行向量用于指示零部件，列向量用于指示物理参数。进而根据特征检验矩阵进行故障识别。
作为一种可能的实现方式，机器学习模型可以为BP神经网络，将传感器所采集到的各零部件的振动、声音、温度和/或图像参数作为BP神经网络的输入参数，由逆传播(Back Propagation，BP)神经网络进行主成份分析、关联分析和/或聚类分析获得特征检验矩阵。若特征检验矩阵中零部件物理参数的取值与设计状态下物理参数的取值相匹配，则确定零部件运转正常；否则，识别出零部件存在故障。
进一步，识别出零部件存在故障之后，还可以在故障数据库中，查询与特征检验矩阵中零部件物理参数的取值相匹配的目标故障模式。
本步骤中AWS云端之所以能够基于传感器采集零部件的传感数据集进行分析处理，确定信号对应的零部件是否存在故障，是依赖于传感数据集所推衍出的零部件物理参数能够准确描述该零部件的运行状态。风电机组是一个大的***，这个***由多个分***组成，而每个分***又分别由多个机械零 部件组成，这些零部件是遵循一定的标准配合在一起的。而标准是预先根据零部件的功能数据以及几何数据而制定，因此，可以利用功能数据以及几何数据来描述零部件，功能数据和几何数据所构成的集合便是零部件的物理参数。物理参数包括但不限于：转轴扭矩T、固有频率H、摩擦力Ff、反射光强U、轴传递的功率W、输出功率P、转速N、压力Fp、风速V、叶轮直径D、零部件质量M和/或像素O。其中，转轴扭矩T、摩擦力Ff和压力Fp属于力学参数。
步骤102、采用零部件的三维模型对所识别出的故障进行呈现。
可选的，针对所识别出的存在故障的零部件，从存在故障的零部件各物理参数中提取出力学参数，根据特征检验矩阵，确定存在故障的零部件的力学参数的取值，将力学参数的取值作为边界条件，对存在故障的零部件处于正常状态的三维模型进行应力分析，获得呈现应力状态的三维模型。
或者，可选的，从零部件的三维模型库中，调用与步骤101中故障识别所获得的目标故障模式相对应的目标三维模型，呈现目标三维模型。
其中，零部件的三维模型库用于存储正常状态的三维模型，还可以进一步存储各故障模式下的三维模型。
当零部件为多个时，零部件的三维模型库中可以存储各零部件在各故障模式下的三维模型，以及处于正常状态的三维模型。
作为一种可能的实现方式，可以根据上一步骤中所获得的特征检验矩阵中的物理参数的取值，在零部件的三维模型库中，确定对应零部件的三维模型。
作为另一种可能的实现方式，也可以直接根据特征检验矩阵行向量所指示的零部件，确定对应零部件的三维模型。
之所以能够根据特征检验矩阵中的物理参数的取值，在零部件的三维模型库中确定对应零部件的三维模型是由于零部件存在一些相对稳定的物理参数，比如几何参数和材料参数，从而据此识别出三维模型对应的零部件。
具体来说，物体的固有频率是物体的一种物理特征，由物体的结构、大小、形状以及材料等特性决定，因而，可以根据测试所获得的振动参数的取值计算固有频率，进而计算获得零部件的结构、大小、形状以及材料等物理参数。如下为具有频率计算公式：
Figure PCTCN2016105310-appb-000001
其中M为质量，K为刚度系数，而刚度系数又可以通过下试算得：
K＝Pf/δ
其中Pf为受力，δ为形变量。在已知受力Pf和形变量δ的情况下，通过联立以上两式解得零部件的质量M，而质量M又是密度与几何尺寸的函数，即：
M＝f(ρ,v)
其中ρ是密度，v是体积，由此，可以将各三维模型集所记录的三维模型的密度ρ、体积v，并结合刚度K与特征检验矩阵进行初步的匹配，进而调出匹配中的零部件三维模型。若存在多个匹配结果，还可以进一步根据与运转状态和几何相关的参数进一步确定一个最为匹配的三维模型，例如：扭矩T、截面惯性矩IR、截面扭转角变化量
Figure PCTCN2016105310-appb-000002
长度变化量dx、面积分微元dA、轴传递的功率W、轴的转速N、轴半径R、半径R处的剪切力τR、半径R处的切变γR、扭转刚度G和抗扭截面系数WR。以上这些参数之间的相互关系如下：
Figure PCTCN2016105310-appb-000003
Figure PCTCN2016105310-appb-000004
τ＝Gγ
Figure PCTCN2016105310-appb-000005
Figure PCTCN2016105310-appb-000006
Figure PCTCN2016105310-appb-000007
Figure PCTCN2016105310-appb-000008
Figure PCTCN2016105310-appb-000009
另外，还可以在零部件上增加位置传感器采集位置参数，例如GPS传感器，并预先建立三维模型与位置参数之间的对应关系，从而根据所采集到的零部件的位置确定对应的三维模型。
零部件的三维模型库中所存储的正常状态的三维模型可以是预先根据零部件的物理参数的设计值进行三维建模获得的。首先，针对每一个零部件已知W,T,N,M,K,G,WR,H,O参数的设计值，根据映射函数 f:X(W,T,N,M,K,G,WR,H,O)'→Y(xm,xg,xf)'计算获得材料系数xm、几何系数xg以及功能系数xf。然后，将材料系数xm、几何系数xg以及功能系数xf代入函数f3D＝f(xm,xg,xf)，进行三维建模，获得三维模型。
本发明实施例中，通过对风电机组中的零部件进行故障识别之后，针对所识别出的故障，采用该零部件的三维模型对故障进行呈现，从而使得故障监测人员能够直观的观察到零部件所出现故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。
图2为本发明实施例提供的一种风电机组故障监测方法的流程示意图，本实施例的故障监测方法可以是针对风电机组中的全部零部件，也可以仅针对风电机组中的一个或多个分***中的零部件，本实施例中对此不做限定，如图2所示，包括：
步骤201、传感器采集零部件的传感数据集，依次通过PLC控制器和工控机上传至AWS云端。
其中，传感数据集包括但不限于：振动、声音、温度和图像参数。
步骤202、AWS云端根据传感数据集进行数据分析。
具体的，AWS云端根据传感数据集进行数据分析从而识别出零部件存在的故障，图3为AWS云端的数据分析流程图，如图3所示，包括：
步骤2021、AWS云端在通过数据接口接收到的传感数据集之后，进行数据清洗操作。
步骤2022、选择执行一般分析流程或高级分析流程。
可选的，预先针对各个零部件设置振动、声音、温度和图像参数的阈值范围，当传感数据集中任一个参数超出阈值范围，则执行高级分析流程，从而进行更加准确和细致的故障分析，否则执行一般分析。
或者，可选的，可以根据用户的设置选择执行一般分析流程或者高级分析流程，其中，一般分析流程处理速度较快，但准确性略低，而高级分析流程处理速度较慢，但准确性较高。
步骤2023、执行一般分析流程。
具体的，图4为一般分析流程图，如图4所示，可以根据零部件的振动、 声音、温度和图像参数输入对应不同状态模式的状态函数，如状态函数1至状态函数n，n为状态的个数，每个状态函数均输出一个判断值，进而根据各状态函数的判断值进行分析，从各状态模式中确定一个最为匹配的状态模式。若匹配的状态模式为正常状态模式则该零部件不存在故障，运转正常。否则，确定零部件存在故障。
步骤2024、执行高级分析流程。
具体的，将零部件的振动、声音、温度和图像参数输入BP神经网络进行主成份分析、关联分析和/或聚类分析，获得特征检验矩阵。
其中，特征检验矩阵的参量向量为
Figure PCTCN2016105310-appb-000010
x1至x12代表12个特征参量，每一种特征参量对应一种物理参数，对应关系如下所示：
x1＝T,x2＝H,x3＝Ff,x4＝U,x5＝W,x6＝P,x7＝N,x8＝Fp,x9＝V,x10＝D,x11＝M,x12＝O。
若风电机组所包含的零部件的总数为N'，那么就可以建立12×N'特征检验矩阵：
Figure PCTCN2016105310-appb-000011
其中m＝12，表示12个特征参量，每一种特征参量对应如上所提及的一种物理参数，n＝N'，表示风电机组的N'个零部件。
若特征检验矩阵中零部件物理参数的取值与设计状态下物理参数的取值 相匹配，则确定零部件运转正常。否则，确定零部件存在故障。
步骤2025、根据分析结果，建立零部件模型。
具体的，分析结果主要用于建立零部件模型，该模型是用数据进行表征的，这些数据可以为步骤2024中所获得的特征检验矩阵，也可以是步骤2023中的状态函数的取值。
步骤203、AWS云端呈现该零部件的三维模型。
具体的，根据振动、声音、温度和图像参数输入BP神经网络所计算获得的12×N'特征检验矩阵，从存在故障的零部件各物理参数中提取出力学参数，确定存在故障的零部件的力学参数的取值，进而将力学参数的取值作为边界条件，对存在故障的零部件处于正常状态的三维模型进行应力分析，获得呈现应力状态的三维模型。
步骤204、若确定存在故障，则AWS云端将分析结果与故障数据库进行匹配，以确定零部件的目标故障模式。
具体的，图5为故障匹配和处理的流程图，如图5所示，包括：
步骤2041、将分析结果与故障数据库进行匹配。
若上一步骤中执行了高级分析流程，则可以在故障数据库中，查询与特征检验矩阵中零部件物理参数的取值相匹配的目标故障模式。在故障数据库中记载了各故障模式下特征检验矩阵中零部件物理参数的取值范围，据此判断零部件所属的目标故障模式。
或者，若上一步骤中执行了一般分析流程，则根据分析所确定的一个最为匹配的状态模式确定对应的目标故障模式。
步骤2042、若故障数据库中不存在匹配的目标故障模式，则配对失败。进而，可在故障数据库新增一故障模式，并将存在故障的零部件对应的特征检验矩阵和/或传感数据集存储到所新增故障模式的对应位置。
步骤2043、若故障数据库中存在匹配的目标故障模式，则配对成功。
步骤205、对故障进行处理。
具体的，如图5所示，通过调用解决方案数据库对故障进行处理，包括：
步骤2051、查询解决方案数据库，以获得与该零部件的目标故障模式对应的解决方案。
其中，解决方案数据库中记载了已发生过的故障的相关解决方案。
步骤2052、若查询到解决方案，则配对成功。
步骤2053、若配对成功，判断是否可以执行该解决方案，以消除该零部件所存在的故障。
步骤2054、若是，则执行该解决方案。
步骤2055、否则，输出该解决方案，以进行人工解决。
步骤2056、若未查询到解决方案，则配对失败。
其中，若该解决方案数据库中不存在目标故障模式对应的解决方案，则确定该目标故障模式为新增故障。需要技术人员对新增故障进行进一步分析后，例如：依据零部件的相关数据，如传感数据集和/或特征检验矩阵，进行故障原因判定和解决方案的制定，将解决方案添加到解决方案数据库中。
本发明实施例中，通过对风电机组中的零部件进行故障识别之后，针对所识别出的故障，采用该零部件的三维模型对故障进行呈现，从而使得故障监测人员能够直观的观察到零部件所出现故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。另外，由于采用了BP神经网络依据各零部件的设计状态下的物理参数对故障进行识别，提高了故障识别的准确性，并实现在出现新故障时，对零部件的相关数据存储到故障数据库中，以便不断完善故障数据库。
图6为本发明实施例提供的一种风电机组故障监测装置的结构示意图，如图6所示，包括：识别模块61和呈现模块62。
识别模块61，用于对风电机组中的零部件进行故障识别。
呈现模块62，用于采用零部件的三维模型对所识别出的故障进行呈现。
进一步，图7为本发明实施例提供的另一种风电机组故障监测装置的结构示意图，如图7所示，在图6的基础上，识别模块61包括：输入单元601和确定单元602。
输入单元601，用于将传感器所采集到的零部件的传感数据集输入对应各状态的状态模式函数，获得各判断值。
确定单元602，用于根据各状态函数的判断值，从各状态模式中确定最为匹配的状态模式，若匹配的状态模式为正常状态模式，则该零部件不存在 故障，否则，确定匹配的状态模式为零部件的目标故障模式。
更进一步，风电机组故障监测装置还包括：查询模块63和处理模块64。
查询模块63，用于查询解决方案数据库，获得与零部件的目标故障模式对应的解决方案。
处理模块64，用于根据解决方案对零部件进行故障处理，和/或，输出解决方案。
本发明实施例中，通过对风电机组中的零部件进行故障识别之后，针对所识别出的故障，采用该零部件的三维模型对故障进行呈现，从而使得故障监测人员能够直观的观察到零部件所出现故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。
图8为本发明实施例提供的另一种风电机组故障监测装置的结构示意图，如图8所示，在图6的基础上，识别模块61包括：采集单元611，分析单元612和识别单元613。
采集单元611，用于利用传感器采集零部件的传感数据集。
具体的，采集单元611具体用于利用传感器采集零部件的振动、声音、温度和/或图像参数的取值，并将振动、声音、温度和/或图像参数的取值作为传感数据集。
分析单元612，用于采用预先训练的机器学习模型，对传感数据集进行数据分析，获得特征检验矩阵。
具体的，分析单元612具体用于采用预先训练的BP神经网络，对传感数据集进行主成份分析、关联分析和/或聚类分析获得特征检验矩阵。
其中，特征检验矩阵的行向量用于指示零部件，列向量用于指示物理参数。
例如：物理参数包括但不限于转轴扭矩T、固有频率H、摩擦力Ff、反射光强U、轴传递的功率W、输出功率P、转速N、压力Fp、风速V、叶轮直径D、零部件质量M和/或像素O。其中，转轴扭矩T、摩擦力Ff和压力Fp属于力学参数。
识别单元613，用于根据特征检验矩阵进行故障识别。
进一步，识别单元613，包括：判断子单元6131和查询子单元6132。
判断子单元6131，用于若特征检验矩阵中零部件物理参数的取值与设计状态下物理参数的取值相匹配，则确定零部件运转正常；否则，识别出零部件存在故障。
查询子单元6132，用于在故障数据库中，查询与特征检验矩阵中零部件物理参数的取值相匹配的目标故障模式。
更进一步，风电机组故障监测装置还包括：查询模块63和处理模块64。
查询模块63，用于查询解决方案数据库，获得与零部件的目标故障模式对应的解决方案。
处理模块64，用于根据解决方案对零部件进行故障处理，和/或，输出解决方案。
通过识别模块61对风电机组中的零部件进行故障识别之后，呈现模块62针对识别模块61所识别出的故障，采用该零部件的三维模型对故障进行呈现，从而使得故障监测人员能够直观的观察到零部件所出现故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。另外，通过处理模块64根据查询模块63所查询到的对应该故障的解决方案对零部件进行故障处理，和/或，输出该解决方案，使得风电机组能够自动对故障进行处理，减轻了故障监测人员的工作量，提高了故障处理的效率。
作为一种可能的实现形式，图9为本发明实施例提供的另一种风电机组故障监测装置的结构示意图，在图8的基础上，如图9所示，呈现模块62进一步包括：提取单元621、确定单元622和第一呈现单元623。
提取单元621，用于针对所识别出的存在故障的零部件，从存在故障的零部件各物理参数中提取出力学参数。
确定单元622，用于根据特征检验矩阵，确定存在故障的零部件的力学参数的取值。
第一呈现单元623，用于将力学参数的取值作为边界条件，对存在故障的零部件处于正常状态的三维模型进行应力分析，获得呈现应力状态的三维模型。
或者，作为另一种可能的实现形式，图10为本发明实施例提供的另一种风电机组故障监测装置的结构示意图，在图8的基础上，如图10所示，呈现模块62进一步包括：调用单元624和第二呈现单元625。
调用单元624，用于从零部件的三维模型库中，调用与目标故障模式相对应的目标三维模型；
第二呈现单元625，用于呈现目标三维模型。
本实施例中，通过对风电机组中的零部件进行故障识别之后，针对所识别出的故障，采用该零部件的三维模型对故障进行呈现，从而使得故障监测人员能够直观的观察到零部件所出现故障，不仅有利于在轻微故障演变成严重故障之前采取有效的故障处理措施，从而减轻故障危害，而且降低了对故障监测人员的经验要求，避免了现有技术中故障监测人员依靠人员经验对二维曲线进行分析以识别出零部件是否具有故障的过程。
本发明实施例还提供了一种计算机存储介质和/或计算机程序产品，所述计算机存储介质中存储有计算机程序，所述计算机程序产品包括计算机存储介质可读的计算机程序，所述计算机程序使得计算机上述实施例中的风电机组故障监测方法。
本领域普通技术人员可以理解：实现上述各方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成。前述的程序可以存储于一计算机可读取存储介质中。该程序在执行时，执行包括上述各方法实施例的步骤；而前述的存储介质包括：ROM、RAM、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims (18)

  1. 一种风电机组故障监测方法，其特征在于，包括：
    对风电机组中的零部件进行故障识别；
    采用所述零部件的三维模型对所识别出的故障进行呈现。
  2. 根据权利要求1所述的风电机组故障监测方法，其特征在于，所述对风电机组中的零部件进行故障识别，包括：
    利用传感器采集所述零部件的传感数据集；
    采用预先训练的机器学习模型，对所述传感数据集进行数据分析，获得特征检验矩阵；所述特征检验矩阵的行向量用于指示零部件，列向量用于指示物理参数；
    根据所述特征检验矩阵进行故障识别。
  3. 根据权利要求2所述的风电机组故障监测方法，其特征在于，所述采用所述零部件的三维模型对所识别出的故障进行呈现，包括：
    针对所识别出的存在故障的零部件，从所述存在故障的零部件各物理参数中提取出力学参数；
    根据所述特征检验矩阵，确定所述存在故障的零部件的力学参数的取值；
    将所述力学参数的取值作为边界条件，对所述存在故障的零部件处于正常状态的三维模型进行应力分析，获得呈现应力状态的三维模型。
  4. 根据权利要求2所述的风电机组故障监测方法，其特征在于，所述根据所述特征检验矩阵进行故障识别，包括：
    若所述特征检验矩阵中所述零部件物理参数的取值与设计状态下物理参数的取值相匹配，则确定所述零部件运转正常；
    否则，识别出所述零部件存在故障；
    在故障数据库中，查询与所述特征检验矩阵中所述零部件物理参数的取值相匹配的目标故障模式。
  5. 根据权利要求1所述的风电机组故障监测方法，其特征在于，所述对风电机组中的零部件进行故障识别，包括：
    将传感器所采集到的零部件的传感数据集输入对应各状态模式的状态函数，获得各判断值；
    根据各状态函数的判断值，从各状态模式中确定最为匹配的状态模式，若所述匹配的状态模式为正常状态模式，则所述零部件不存在故障，否则，确定所述匹配的状态模式为所述零部件的目标故障模式。
  6. 根据权利要求4或5所述的风电机组故障监测方法，其特征在于，所述采用所述零部件的三维模型对所识别出的故障进行呈现，包括：
    从所述零部件的三维模型库中，调用与所述目标故障模式相对应的目标三维模型；
    呈现所述目标三维模型。
  7. 根据权利要求4或5所述的风电机组故障监测方法，其特征在于，所述采用所述零部件的三维模型对所识别出的故障进行呈现之后，还包括：
    查询解决方案数据库，获得与所述零部件的目标故障模式对应的解决方案；
    根据所述解决方案对所述零部件进行故障处理，和/或，输出所述解决方案。
  8. 根据权利要求2-4任一项所述的风电机组故障监测方法，其特征在于，所述物理参数包括：转轴扭矩T、固有频率H、摩擦力Ff、反射光强U、轴传递的功率W、输出功率P、转速N、压力Fp、风速V、叶轮直径D、零部件质量M和/或像素O；
    其中，所述转轴扭矩T、摩擦力Ff和压力Fp属于力学参数。
  9. 一种风电机组故障监测装置，其特征在于，包括：
    识别模块，用于对风电机组中的零部件进行故障识别；
    呈现模块，用于采用所述零部件的三维模型对所识别出的故障进行呈现。
  10. 根据权利要求9所述的风电机组故障监测装置，其特征在于，所述识别模块，包括：
    采集单元，用于利用传感器采集所述零部件的传感数据集；
    分析单元，用于采用预先训练的机器学习模型，对所述传感数据集进行数据分析，获得特征检验矩阵；所述特征检验矩阵的行向量用于指示零部件，列向量用于指示物理参数；
    识别单元，用于根据所述特征检验矩阵进行故障识别。
  11. 根据权利要求9所述的风电机组故障监测装置，其特征在于，所述 呈现模块，包括：
    提取单元，用于针对所识别出的存在故障的零部件，从所述存在故障的零部件各物理参数中提取出力学参数；
    确定单元，用于根据所述特征检验矩阵，确定所述存在故障的零部件的力学参数的取值；
    第一呈现单元，用于将所述力学参数的取值作为边界条件，对所述存在故障的零部件处于正常状态的三维模型进行应力分析，获得呈现应力状态的三维模型。
  12. 根据权利要求10所述的风电机组故障监测装置，其特征在于，所述识别单元，包括：
    判断子单元，用于若所述特征检验矩阵中所述零部件物理参数的取值与设计状态下物理参数的取值相匹配，则确定所述零部件运转正常；否则，识别出所述零部件存在故障；
    查询子单元，用于在故障数据库中，查询与所述特征检验矩阵中所述零部件物理参数的取值相匹配的目标故障模式。
  13. 根据权利要求9所述的风电机组故障监测装置，其特征在于，所述识别模块，包括：
    输入单元，用于将传感器所采集到的零部件的传感数据集输入对应各状态模式的状态函数，获得各判断值；
    确定单元，用于根据各状态函数的判断值，从各状态模式中确定最为匹配的状态模式，若所述匹配的状态模式为正常状态模式，则所述零部件不存在故障，否则，确定所述匹配的状态模式为所述零部件的目标故障模式。
  14. 根据权利要求12或13所述的风电机组故障监测装置，其特征在于，所述呈现模块，包括：
    调用单元，用于从所述零部件的三维模型库中，调用与所述目标故障模式相对应的目标三维模型；
    第二呈现单元，用于呈现所述目标三维模型。
  15. 根据权利要求12或13所述的风电机组故障监测装置，其特征在于，所述风电机组故障监测装置还包括：
    查询模块，用于查询解决方案数据库，获得与所述零部件的目标故障模 式对应的解决方案；
    处理模块，用于根据所述解决方案对所述零部件进行故障处理，和/或，输出所述解决方案。
  16. 根据权利要求10-12任一项所述的风电机组故障监测装置，其特征在于，所述物理参数包括：转轴扭矩T、固有频率H、摩擦力Ff、反射光强U、轴传递的功率W、输出功率P、转速N、压力Fp、风速V、叶轮直径D、零部件质量M和/或像素O；
    其中，所述转轴扭矩T、摩擦力Ff和压力Fp属于力学参数。
  17. 一种计算机存储介质，其特征在于，所述计算机存储介质中存储有计算机程序，所述计算机程序使得计算机执行权利要求1-8任一项所述的风电机组故障监测方法。
  18. 一种计算机程序产品，其特征在于，所述计算机程序产品包括计算机存储介质可读的计算机程序，所述程序使得计算机执行权利要求1-8任一项所述的风电机组故障监测方法。
PCT/CN2016/105310 2015-12-31 2016-11-10 计算机存储介质、计算机程序产品、风电机组故障监测方法和装置 WO2017113998A1 (zh)

Priority Applications (5)

Application Number Priority Date Filing Date Title
EP16880789.9A EP3399297B1 (en) 2015-12-31 2016-11-10 Computer storage medium, computer program product, and method and device for monitoring for malfunction of a wind turbine
US15/567,809 US10760551B2 (en) 2015-12-31 2016-11-10 Computer storage medium, computer program product, method for monitoring fault of wind power generator set, and device
KR1020187000957A KR102011227B1 (ko) 2015-12-31 2016-11-10 컴퓨터 저장 매체, 컴퓨터 프로그램 제품, 및 풍력 발전기 세트의 장애를 모니터링하기 위한 방법, 및 장치
AU2016380302A AU2016380302C1 (en) 2015-12-31 2016-11-10 Computer storage medium, computer program product, method for monitoring fault of wind power generator set, and device
ES16880789T ES2778692T3 (es) 2015-12-31 2016-11-10 Medio de almacenamiento informático, producto de programa informático, y método y dispositivo para monitorizar el mal funcionamiento de una turbina eólica

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201511032667.0 2015-12-31
CN201511032667.0A CN105510038B (zh) 2015-12-31 2015-12-31 风电机组故障监测方法和装置

Publications (1)

Publication Number Publication Date
WO2017113998A1 true WO2017113998A1 (zh) 2017-07-06

Family

ID=55718180

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/105310 WO2017113998A1 (zh) 2015-12-31 2016-11-10 计算机存储介质、计算机程序产品、风电机组故障监测方法和装置

Country Status (7)

Country Link
US (1) US10760551B2 (zh)
EP (1) EP3399297B1 (zh)
KR (1) KR102011227B1 (zh)
CN (1) CN105510038B (zh)
AU (1) AU2016380302C1 (zh)
ES (1) ES2778692T3 (zh)
WO (1) WO2017113998A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112504671A (zh) * 2020-11-27 2021-03-16 世科工业设计沧州有限公司 风力发电用动能传递轴强度检测设备
CN114320773A (zh) * 2021-12-22 2022-04-12 中国大唐集团新能源科学技术研究院有限公司 一种基于功率曲线分析与神经网络的风电机组故障预警方法

Families Citing this family (29)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105510038B (zh) 2015-12-31 2018-07-27 北京金风科创风电设备有限公司 风电机组故障监测方法和装置
CN105952588A (zh) * 2016-05-19 2016-09-21 沈阳工程学院 一种基于云计算及大数据的风机监测方法
CN106288152B (zh) * 2016-08-01 2019-04-09 珠海格力电器股份有限公司 设备监控方法和装置
CN107024331B (zh) * 2017-03-31 2019-07-12 中车工业研究院有限公司 一种神经网络对列车电机振动在线检测方法
CN108957315A (zh) * 2017-05-22 2018-12-07 北京金风科创风电设备有限公司 风力发电机组的故障诊断方法和设备
CN109947786A (zh) * 2017-10-17 2019-06-28 中车株洲电力机车研究所有限公司 风电场中故障自动排查方法、装置及计算机可读存储介质
CN108317664B (zh) * 2017-12-21 2020-08-11 珠海格力电器股份有限公司 过滤网状况的提示方法和装置
CN108520270B (zh) * 2018-03-12 2021-06-04 明觉科技（北京）有限公司 零件匹配方法、***及终端
CN108386325A (zh) * 2018-05-28 2018-08-10 华润电力风能（汕头潮南）有限公司 一种风力发电机组智能诊断的方法及***
CN108629432A (zh) * 2018-06-12 2018-10-09 中国三峡新能源有限公司 故障处理规划方法、***及装置
CN109543210A (zh) * 2018-09-28 2019-03-29 国电电力宁夏新能源开发有限公司 一种基于机器学***台的风电机组故障预测***
CN111400959B (zh) * 2018-12-27 2024-01-30 北京金风科创风电设备有限公司 风力发电机组的叶片故障诊断方法及装置
CN109872323A (zh) * 2019-02-28 2019-06-11 北京国网富达科技发展有限责任公司 输电线路的绝缘子缺陷检测方法及装置
CN110070603A (zh) * 2019-03-21 2019-07-30 江苏东方国信工业互联网有限公司 一种基于数字双胞胎的炉体仿真***及其方法
EP3748154A1 (en) * 2019-06-06 2020-12-09 Siemens Gamesa Renewable Energy A/S A method for computer-implemented monitoring of a compo-nent of a wind turbine
EP4063875A4 (en) * 2019-11-18 2023-08-09 Zhuzhou CRRC Times Electric Co., Ltd. INFORMATION COMBINATION METHOD AND DEVICE FOR EARLY WARNING OF FAULTS FOR AN INVERTER
DE102019135628A1 (de) * 2019-12-20 2021-06-24 fos4X GmbH Verfahren zur Bestimmung einer Restnutzungsdauer einer Windenergieanlage und Windenergieanlage
DE102019135622A1 (de) * 2019-12-20 2021-06-24 fos4X GmbH Verfahren zum Bestimmen von Betriebsdaten einer Windenergieanlage
CN111382673A (zh) * 2020-01-09 2020-07-07 南京艾拓维讯信息技术有限公司 一种用于风车发电状态监测的kvm***及方法
US11428212B2 (en) 2020-02-11 2022-08-30 Inventus Holdings, Llc Wind turbine drivetrain wear detection using azimuth variation clustering
DE102020111142A1 (de) 2020-04-23 2021-10-28 fos4X GmbH Verfahren zum Überwachen einer Windenergieanlage, System zum Überwachen einer Windenergieanlage, Windenergieanlagen und Computerprogrammprodukt
CN111639110A (zh) * 2020-06-02 2020-09-08 国电联合动力技术有限公司 风电机组故障预警方法及装置
CN112283050A (zh) * 2020-11-17 2021-01-29 天津科技大学 一种基于机器视觉的风力机叶片在线监测方法
CN112502909B (zh) * 2020-11-26 2023-09-08 中车永济电机有限公司 风力发电机故障检测方法、数采装置、服务器及***
CN112834224B (zh) * 2021-01-05 2023-05-23 广东核电合营有限公司 一种核电汽轮发电机健康状态评估方法及***
DE102021104826A1 (de) 2021-03-01 2022-09-01 KSB SE & Co. KGaA Verfahren zur Evaluierung einer notwendigen Instandhaltungsmaßnahme für eine Maschine, insbesondere für eine Pumpe
CN113436194B (zh) * 2021-08-26 2021-11-23 北京工业大数据创新中心有限公司 一种风电机组的异常检测方法、装置及设备
CN114704438B (zh) * 2022-06-02 2022-09-20 深圳众城卓越科技有限公司 风电机组故障监控方法及装置
CN117786586B (zh) * 2024-02-27 2024-06-11 深圳亚太航空技术股份有限公司 一种自锁螺母检测***及其方法

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6750776B2 (en) * 2002-02-27 2004-06-15 Nec Machinery Corporation Machines having drive member and method for diagnosing the same
CN101299004A (zh) * 2008-06-24 2008-11-05 华南理工大学 一种基于确定学习理论的振动故障诊断方法
CN101907088A (zh) * 2010-05-27 2010-12-08 中国人民解放军国防科学技术大学 基于单类支持向量机的故障诊断方法
CN102033984A (zh) * 2010-11-12 2011-04-27 清华大学 一种基于区间型证据融合的旋转机械设备故障诊断方法
CN102799619A (zh) * 2012-06-19 2012-11-28 北京航空航天大学 一种基于产品三维模型的故障模式可视化方法
CN102854015A (zh) * 2012-10-15 2013-01-02 哈尔滨理工大学 一种滚动轴承故障位置及性能退化程度诊断方法
CN103471841A (zh) * 2013-09-30 2013-12-25 国家电网公司 一种旋转机械振动故障诊断方法
CN104392020A (zh) * 2014-10-22 2015-03-04 北京必可测科技股份有限公司 一种旋转机械振动故障的显示方法及装置
CN105510038A (zh) * 2015-12-31 2016-04-20 北京金风科创风电设备有限公司 风电机组故障监测方法和装置

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8155943B2 (en) * 2007-10-12 2012-04-10 Power Analytics Corporation Systems and methods for automatically converting CAD drawing files into intelligent objects with database connectivity for the design, analysis, and simulation of electrical power systems
DE102008006370A1 (de) * 2008-01-28 2009-07-30 Prüftechnik Dieter Busch AG Verfahren und Vorrichtung zum Überwachen einer Maschine
US20110313726A1 (en) * 2009-03-05 2011-12-22 Honeywell International Inc. Condition-based maintenance system for wind turbines
ES2562253T3 (es) * 2009-06-24 2016-03-03 Vestas Wind Systems A/S Un procedimiento y un sistema para controlar el funcionamiento de una turbina eólica
US8538735B2 (en) * 2009-09-08 2013-09-17 Wm. Stevens Taber, Jr. Use of devices for measuring fluid movement conditions at a distance to reduce the design and manufacturing cost of moving-fluid-driven working devices
WO2011095519A2 (en) * 2010-02-05 2011-08-11 Vestas Wind Systems A/S Method of operating a wind power plant
CN101872165A (zh) * 2010-06-13 2010-10-27 西安交通大学 一种基于遗传神经网络的风电机组故障诊断方法
CN102175449B (zh) * 2011-03-18 2012-07-18 天津工业大学 基于风力机应变能响应的叶片故障诊断方法
CN102809660B (zh) * 2012-08-22 2013-09-18 合肥工业大学 汽车桥壳再制造的疲劳损伤检测方法
CN102980612B (zh) * 2012-12-05 2016-01-13 徐州市产品质量监督检验中心 一种钢结构网架的三维虚拟现实健康监测方法
US8853877B1 (en) * 2013-05-29 2014-10-07 General Electric Company System and method for controlling a wind farm
WO2015005663A1 (ko) * 2013-07-10 2015-01-15 주식회사 글로비즈 신호 계측 진단 모니터링 시스템과 그 방법 및 이를 개별 장치에 적용한 방법 및 시스템
KR101355888B1 (ko) * 2013-08-13 2014-01-27 공주대학교 산학협력단 차량용 전자식 솔레노이드 구조의 진동특성 평가방법
US20150077155A1 (en) * 2013-09-18 2015-03-19 Palle Clausen Monitoring of wind turbine performance
US9835135B2 (en) * 2013-10-31 2017-12-05 General Electric Company System and method for controlling a wind turbine
CN107211287A (zh) * 2014-08-29 2017-09-26 峰鸟航空科技公司 使用混合电动飞机实现区域性空中运输网络的***和方法
CA2961002C (en) * 2014-09-12 2023-12-05 Quantum Energy Storage Corporation Energy storage management system
CN104459388B (zh) * 2014-11-26 2017-02-22 国家电网公司 一种永磁直驱风力发电***一体化故障诊断方法
CN104697767B (zh) * 2014-12-17 2017-03-22 天津大学 一种基于振动分析的转子***故障诊断方法及装置
CN104614179B (zh) * 2015-01-28 2017-08-29 中国大唐集团新能源股份有限公司 一种风电机组齿轮箱状态监测方法
CN104865269A (zh) * 2015-04-13 2015-08-26 华北理工大学 风力机叶片故障诊断方法

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6750776B2 (en) * 2002-02-27 2004-06-15 Nec Machinery Corporation Machines having drive member and method for diagnosing the same
CN101299004A (zh) * 2008-06-24 2008-11-05 华南理工大学 一种基于确定学习理论的振动故障诊断方法
CN101907088A (zh) * 2010-05-27 2010-12-08 中国人民解放军国防科学技术大学 基于单类支持向量机的故障诊断方法
CN102033984A (zh) * 2010-11-12 2011-04-27 清华大学 一种基于区间型证据融合的旋转机械设备故障诊断方法
CN102799619A (zh) * 2012-06-19 2012-11-28 北京航空航天大学 一种基于产品三维模型的故障模式可视化方法
CN102854015A (zh) * 2012-10-15 2013-01-02 哈尔滨理工大学 一种滚动轴承故障位置及性能退化程度诊断方法
CN103471841A (zh) * 2013-09-30 2013-12-25 国家电网公司 一种旋转机械振动故障诊断方法
CN104392020A (zh) * 2014-10-22 2015-03-04 北京必可测科技股份有限公司 一种旋转机械振动故障的显示方法及装置
CN105510038A (zh) * 2015-12-31 2016-04-20 北京金风科创风电设备有限公司 风电机组故障监测方法和装置

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112504671A (zh) * 2020-11-27 2021-03-16 世科工业设计沧州有限公司 风力发电用动能传递轴强度检测设备
CN112504671B (zh) * 2020-11-27 2022-03-25 北京合信锐风新能源发展有限公司 风力发电用动能传递轴强度检测设备
CN114320773A (zh) * 2021-12-22 2022-04-12 中国大唐集团新能源科学技术研究院有限公司 一种基于功率曲线分析与神经网络的风电机组故障预警方法
CN114320773B (zh) * 2021-12-22 2023-09-22 大唐可再生能源试验研究院有限公司 一种基于功率曲线分析与神经网络的风电机组故障预警方法

Also Published As

Publication number Publication date
AU2016380302C1 (en) 2019-01-03
EP3399297A1 (en) 2018-11-07
US10760551B2 (en) 2020-09-01
KR20180016582A (ko) 2018-02-14
US20180119677A1 (en) 2018-05-03
ES2778692T3 (es) 2020-08-11
EP3399297B1 (en) 2020-01-01
CN105510038A (zh) 2016-04-20
AU2016380302A1 (en) 2017-11-23
CN105510038B (zh) 2018-07-27
KR102011227B1 (ko) 2019-08-14
AU2016380302B2 (en) 2018-09-27
EP3399297A4 (en) 2019-01-09

Similar Documents

Publication Publication Date Title
WO2017113998A1 (zh) 计算机存储介质、计算机程序产品、风电机组故障监测方法和装置
US20190301979A1 (en) Abnormality detection system, support device, and abnormality detection method
US9251582B2 (en) Methods and systems for enhanced automated visual inspection of a physical asset
JPWO2018061842A1 (ja) 異常検知プログラム、異常検知方法および異常検知装置
RU2686252C2 (ru) Способ оценки нормальности или ненормальности измеренного значения физического параметра двигателя летательного аппарата
CN102588214A (zh) 自动监视风力涡轮机的方法和设备及创建线性模型的方法
JP2016045946A5 (zh)
JP2012014698A (ja) 相関計量に基づいて、シミュレーションモデルを物理的デバイスと相関させる方法および装置
JP2010088576A5 (zh)
KR102241804B1 (ko) 상담 대상자의 드로잉 과정을 통해 심리 상태를 판단하는 방법 및 컴퓨터 프로그램
TWI518469B (zh) 加工監控系統及方法
WO2019071438A1 (zh) 对流程行业中的设备进行状态监测的方法、装置及介质
JPWO2021079792A5 (zh)
CN113836782A (zh) 计算***、计算方法和气体输送装置
JP2016061293A (ja) 自動故障予測システム及び方法
JP2017006970A5 (zh)
EP2847676A2 (en) System and method of distributed procesing for machine-vision analysis
JP6223897B2 (ja) 異常検知装置及び異常検知システム
US20180268036A1 (en) Communication information generating apparatus, communication information generating method, recording medium, and communication management system
TWI820612B (zh) 衝壓設備的異常檢測方法、裝置、設備及存儲媒體
JP6302523B1 (ja) 製品診断装置、製品診断方法、および、製品診断プログラム
TWI692249B (zh) 監控終端及監控方法
JP6005555B2 (ja) プラントの運転監視装置、その運転監視方法、及びその運転監視プログラム
TWI481811B (zh) 機台狀態偵測系統及方法
CN117872996A (zh) 一种基于ar眼镜的生产设备监测方法、装置以及电子设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16880789

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15567809

Country of ref document: US

ENP Entry into the national phase

Ref document number: 2016380302

Country of ref document: AU

Date of ref document: 20161110

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 20187000957

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE