WO2022012137A1 - 风力发电机组的监控方法、系统及计算机可读存储介质 - Google Patents

风力发电机组的监控方法、系统及计算机可读存储介质 Download PDF

Info

Publication number
WO2022012137A1
WO2022012137A1 PCT/CN2021/093632 CN2021093632W WO2022012137A1 WO 2022012137 A1 WO2022012137 A1 WO 2022012137A1 CN 2021093632 W CN2021093632 W CN 2021093632W WO 2022012137 A1 WO2022012137 A1 WO 2022012137A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
statistic
temperature data
processed
value
Prior art date
Application number
PCT/CN2021/093632
Other languages
English (en)
French (fr)
Inventor
成骁彬
许移庆
赵大文
Original Assignee
上海电气风电集团股份有限公司
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 上海电气风电集团股份有限公司 filed Critical 上海电气风电集团股份有限公司
Priority to KR1020237005315A priority Critical patent/KR20230038281A/ko
Priority to EP21841719.4A priority patent/EP4184007A4/en
Publication of WO2022012137A1 publication Critical patent/WO2022012137A1/zh

Links

Images

Classifications

    • 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
    • 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
    • F03D17/005—Monitoring or testing of wind motors, e.g. diagnostics using computation methods, e.g. neural networks
    • F03D17/0065—Monitoring or testing of wind motors, e.g. diagnostics using computation methods, e.g. neural networks for diagnostics
    • 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
    • F03D17/009—Monitoring or testing of wind motors, e.g. diagnostics characterised by the purpose
    • F03D17/018—Monitoring or testing of wind motors, e.g. diagnostics characterised by the purpose for monitoring temperature
    • 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
    • F05B2240/00—Components
    • F05B2240/50—Bearings
    • 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
    • F05B2270/00—Control
    • F05B2270/30—Control parameters, e.g. input parameters
    • F05B2270/303—Temperature
    • 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
    • F05B2270/00—Control
    • F05B2270/30—Control parameters, e.g. input parameters
    • F05B2270/325—Air temperature
    • 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
    • F05B2270/00—Control
    • F05B2270/70—Type of control algorithm
    • 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
    • F05B2270/00—Control
    • F05B2270/70—Type of control algorithm
    • F05B2270/707—Type of control algorithm fuzzy logic
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/70—Wind energy
    • Y02E10/72—Wind turbines with rotation axis in wind direction

Definitions

  • the present application relates to the field of wind turbines, and in particular, to a monitoring method, system and computer-readable storage medium for a wind turbine.
  • condition monitoring technology of wind turbines conducts real-time data collection, monitoring and related data analysis on key positions such as the impeller, gearbox, generator, yaw and pitch mechanism of the turbine through temperature, speed, vibration and other sensors, so as to understand the operation health of the turbine. Status, timely detection of fault symptoms, processing related faults in advance and reasonable arrangements for power generation operations, thereby improving operational efficiency, reducing operation and maintenance costs, and maximizing benefits.
  • the present application provides a monitoring method, system and computer-readable storage medium for a wind turbine.
  • a monitoring method for a wind turbine comprising: acquiring internal ambient temperature data and external ambient temperature data of the wind turbine, and temperature data of components of the wind turbine ; Determine a first statistic according to the difference between the component temperature data and the internal ambient temperature data, and determine a second statistic according to the difference between the component temperature data and the external ambient temperature data; According to the The first statistic and the second statistic are used to determine whether the temperature data of the component meets the set requirements.
  • judging whether the temperature data of the parts meets the set requirements according to the first statistic and the second statistic further includes: using a multivariate test method to compare the first statistic and the second statistic. Perform test analysis on the second statistic to determine a test value between the first statistic and the second statistic and an upper limit of the test value; monitor whether the test value exceeds the upper limit value to judge whether the temperature data of the component meets the set requirements.
  • the multivariate test method includes a Hotelling T-squared distribution test method, and the multivariate test method is used to test and analyze the first statistic and the second statistic, further comprising: using the Hotelling T-square distribution test method.
  • the first statistic and the second statistic are tested and analyzed by the Tring T-square distribution test method.
  • the acquiring the internal ambient temperature data and the external ambient temperature data of the wind turbine generator set, as well as the component temperature data of the wind turbine generator set includes: acquiring a set number of the internal ambient temperature data, all the external ambient temperature data and the component temperature data.
  • the multivariate test method is used to perform test analysis on the first statistic and the second statistic, and a test value between the first statistic and the second statistic and the difference between the test value are determined.
  • the upper limit value further includes: according to the first statistic, the second statistic, the mean value of the set number of the first statistic, and the set number of the second statistic the mean value of , determine the test value between the first statistical quantity and the second statistical quantity; The upper limit value is determined through the chi-square distribution.
  • the monitoring of whether the inspection value exceeds the upper limit value to determine whether the temperature data of the parts meets the set requirements further includes: if the inspection of the first set number is continuously monitored. If the value is greater than the upper limit value, it is determined that the temperature data of the component does not meet the set requirements.
  • the monitoring of whether the inspection value exceeds the upper limit value to determine whether the temperature data of the parts meets the set requirements further includes: if the second set number of inspections are continuously monitored. If the value is not greater than the upper limit value, it is determined that the temperature data of the component meets the set requirements.
  • the component temperature data includes bearing temperature.
  • collect the SCADA data set of the wind turbine through a SCADA system and obtain the internal ambient temperature data and external ambient temperature data of the wind turbine and the component temperature data of the wind turbine, including: Obtain at least part of the data in the SCADA data set as a data set to be processed, the data set to be processed includes a plurality of data groups, and each data group includes a variety of data representing different information at the same time, and the various data groups
  • the data includes the data to be processed of the internal environment temperature, the data to be processed of the external environment temperature, and the data to be processed of the temperature of the parts; perform data screening processing on the data set to be processed through at least one data screening step to obtain a screening data set, the screening data
  • the set includes the internal environmental temperature data, the external environmental temperature data and the part temperature data; the internal environmental temperature data, the external environmental temperature data and the part temperature data are obtained from the screening data set .
  • the acquiring at least part of the data in the SCADA data set as the data set to be processed includes: acquiring at least part of the data of the set time dimension in the SCADA data set as the data set to be processed.
  • the data screening step includes: using a clustering algorithm to divide the data in the data set to be processed into a plurality of clusters; and removing singular values in each of the clusters and data including the singular values other data for the group.
  • the clustering algorithm includes a fuzzy C-means clustering algorithm
  • the use of the clustering algorithm to divide the data in the data set to be processed into multiple clusters includes: using the fuzzy C-means clustering algorithm to The data in the data set to be processed is divided into multiple clusters.
  • the removing the singular value in each of the clusters and other data of the data group including the singular value includes: using a fuzzy C-means clustering algorithm to determine the center of each of the clusters; determining at least The Euclidean distance between the data in one of the clusters and the center of the corresponding cluster; and using the Laida criterion to remove the distance outliers of the Euclidean distance in at least one of the clusters, and remove all the singular value corresponding to the distance outlier.
  • the data screening step includes: determining abnormal data in the data set to be processed by using a quartile method; removing the abnormal data and other data of the data group including the abnormal data.
  • the method further includes: standardizing the data in the data set to be processed to obtain standardized data.
  • the using a clustering algorithm to divide the data in the data set to be processed into multiple clusters includes: using a clustering algorithm to divide the standardized data into multiple clusters.
  • the normalized data is denormalized after the other data of the data set.
  • using a clustering algorithm to divide the data in the data set to be processed into multiple clusters includes: using a clustering algorithm to divide each type of data in the data set to be processed into multiple clusters respectively. .
  • the method further includes: if the time series of the data in the screening data set is discontinuous, backfilling the data at the missing moment between the discontinuous time series data. .
  • the backfilling the data at the missing moment between the time series non-consecutive data includes: if the number of consecutive multiple data before the time series non-consecutive data is greater than 2, determining the time series The residual mean and variance of a plurality of consecutive data preceding the discontinuous data, and a random number is generated based on the residual mean and variance as the data at the missing moment.
  • the backfilling the data at the missing moment between the time series non-consecutive data includes: if the number of consecutive multiple data before the time series non-consecutive data is not greater than 2, selecting the time series non-consecutive data.
  • the data corresponding to the time before the continuous data is regarded as the data of the missing time.
  • a monitoring system applied to a wind turbine generator set comprising one or more processors for implementing the monitoring method described in any of the above embodiments.
  • a computer-readable storage medium on which a program is stored, and when the program is executed by a processor, implements the monitoring method described in any of the above embodiments.
  • two variables are established in combination with the ambient temperature and the temperature of the components of the wind turbine, and it is determined whether the temperature of the components meets the set requirements according to the two variables, which can reduce the zero impact of environmental factors on the wind turbine.
  • the accuracy of monitoring is reduced due to the influence of the temperature of the components, so as to improve the accuracy of judging the operating health state of the wind turbine through the temperature of the components.
  • FIG. 1 is a schematic flowchart of a monitoring method for a wind turbine according to an exemplary embodiment of the present application.
  • FIG. 2 and FIG. 3 are schematic diagrams of partial refinements of the monitoring method shown in FIG. 1 .
  • FIG. 4 is a schematic flowchart of a data processing method according to an exemplary embodiment of the present application.
  • FIG. 5 is a schematic flowchart of a data screening step of a data processing method according to an exemplary embodiment of the present application.
  • FIG. 6 is a schematic diagram showing a partial refinement of the data screening step shown in FIG. 5 .
  • FIG. 7 is a schematic flowchart of a data screening step of a data processing method according to another exemplary embodiment of the present application.
  • Fig. 8 is a system block diagram of a monitoring system for a wind turbine according to an exemplary embodiment of the present application.
  • first, second, third, etc. may be used in this application to describe various information, such information should not be limited by these terms. These terms are only used to distinguish the same type of information from each other.
  • first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information without departing from the scope of the present application.
  • word "if” as used herein can be interpreted as "at the time of” or "when” or "in response to determining.”
  • the temperature of components affected by environmental factors reflects different conditions. Taking the bearing temperature as an example, the environment in which the wind turbine is located in summer and winter is inconsistent. Assuming that the bearing temperature alarm line is 80°, in summer, due to the generally high outdoor temperature, the bearing temperature is about 60° all year round, which can be regarded as positive. normal status. However, if the bearing temperature is still 60° in winter, even if there is still a certain gap from the alarm line of 80°, this temperature value still needs to be vigilant. However, in the prior art, the influence of environmental factors on the temperature of the components of the wind turbine cannot be effectively monitored, which will reduce the monitoring accuracy, easily cause false alarms, and increase operation and maintenance costs.
  • the present application provides a monitoring method, system and computer-readable storage medium for a wind turbine, which can reduce the influence of environmental factors on the temperature of parts of the wind turbine and reduce the accuracy of monitoring, thereby improving the judgment of the temperature of the parts.
  • the accuracy of the operational health status of wind turbines will be described in detail below with reference to the accompanying drawings. The features of the embodiments and implementations described below may be combined with each other without conflict.
  • an embodiment of the present application provides a monitoring method for a wind turbine, including the following steps S101-S103:
  • Step S101 Obtain the internal ambient temperature data and the external ambient temperature data of the wind turbine generator set, and the component temperature data of the wind turbine generator set.
  • the component may be a bearing of a wind turbine, and the component temperature data may include bearing temperature.
  • the component may also be other components, such as a gearbox, and the temperature data of the component, such as the oil temperature of the gearbox, is not limited in this application for comparison.
  • the internal ambient temperature data may refer to the internal ambient temperature of the nacelle of the wind turbine generator set
  • the external ambient temperature may refer to the external ambient temperature of the nacelle of the wind generator set.
  • Step S102 Determine a first statistic according to the difference between the component temperature data and the internal environment temperature data, and determine a second statistic according to the difference between the component temperature data and the external environment temperature data.
  • t b is the component temperature, such as the bearing temperature.
  • t out is the external ambient temperature of the wind turbine, and t in is the internal ambient temperature of the wind turbine.
  • Step S103 According to the first statistic and the second statistic, determine whether the component temperature data meets the set requirements.
  • two variables are established by combining the ambient temperature and the temperature of the components of the wind turbine. According to the two variables, it is judged whether the temperature data of the components meets the set requirements, which can reduce the impact of environmental factors on the temperature data of the components under variable working conditions. Reduce the accuracy of monitoring, thereby improving the accuracy of judging the operating health status of the wind turbine through the temperature of components.
  • the step S103 of judging whether the temperature data of the parts meets the setting requirements according to the first statistic and the second statistic may further Including sub-steps S1031-S1032:
  • Step S1031 Use a multivariate test method to perform test analysis on the first statistic and the second statistic, and determine the test value and the test value between the first statistic and the second statistic upper limit of .
  • the first statistic and the second statistic of the two variables monitored are essentially residual fields. According to reasonable assumptions, it can be assumed that the distribution of these two variables conforms to the normal distribution, so it can be used to make use of their excellent and controllable distribution.
  • the multivariate test method is used to test, analyze and monitor the two variables to improve the accuracy of monitoring and analysis.
  • Step S1032 Monitor whether the verification value exceeds the upper limit value to determine whether the temperature data of the component meets the set requirements.
  • the present application establishes two variables for inspection and analysis in combination with the ambient temperature and the temperature of the components of the wind turbine, adopts a multivariate inspection method to inspect and analyze the two variables, and compares the inspection value and upper limit obtained by the inspection and analysis. , judging whether the temperature data of the parts meets the set requirements, which can reduce the impact of environmental factors on the temperature data of the parts under variable working conditions and reduce the accuracy of monitoring, thereby improving the judgment of the operation of the wind turbine by the temperature of the parts. Accuracy of health status.
  • the multivariate test method includes Hotelling's T-squared test method (ie, Hotelling T 2 ).
  • the testing and analysis of the first statistic and the second statistic using a multivariate testing method may be performed by using a Hotelling T-squared distribution testing method on the first statistic and the all statistic.
  • the second statistic is used for test analysis.
  • the multivariate testing method may also include other testing methods, as long as the two data samples can be combined and analyzed, which is not limited in this application.
  • step S101 the acquisition of the internal ambient temperature data and the external ambient temperature data of the wind turbine, as well as the component temperature data of the wind turbine, may acquire a set number of of the internal ambient temperature data, the external ambient temperature data and the component temperature data. Since the quantity of the collected temperature data is divided according to the time zone, the temperature of the parts can be monitored and monitored by dividing the time zone by the above method, so as to reduce the data error caused by the excessive amount of data and improve the Monitoring accuracy.
  • the multivariate test method is used to perform test analysis on the first statistic and the second statistic, and a test value between the first statistic and the second statistic is determined and the step S1031 of the upper limit value of the test value, may include sub-steps S10311-S10312:
  • Step S10311 According to the first statistic, the second statistic, the mean value of the set number of the first statistic, and the mean value of the set number of the second statistic, use Equation (3) determines the test value between the first statistic and the second statistic.
  • T 2 is the test value
  • x is the value of the first statistic and the second statistic
  • S -1 is the inverse of the matrix of the covariances of the first statistic t 1 and the second statistic t 2 .
  • Step S10312 According to the number of statistics and the set number determined according to the internal ambient temperature data, the external ambient temperature data and the temperature of the parts, use formula (4) and formula (5) to pass The chi-square distribution determines the upper limit:
  • UCL is the upper limit value
  • LCL is the lower limit value
  • p is the number of variables used for the test analysis
  • m is the order of magnitude used as the initial value for the health state.
  • F means that this formula belongs to the chi-square distribution.
  • a is the upper locus of the chi-square distribution.
  • the value of a is 0.001, and the values of F a, p, and mp can be calculated by themselves according to the Kabbe distribution.
  • the value of m may be set according to the actual situation, which is not limited in this application.
  • the SCADA data of the wind turbine can be collected through the SCADA system, and the SCADA data can include monitoring data such as temperature, wind speed, and power used to characterize the performance of the wind turbine.
  • the internal ambient temperature data and external ambient temperature data of the wind turbine generator set, and the component temperature data of the wind turbine generator set also belong to a type of data in SCADA data.
  • SCADA system is Supervisory Control And Data Acquisition (Supervisory Control And Data Acquisition).
  • SCADA system can monitor and control on-site fans to realize data acquisition, equipment control, measurement, parameter adjustment and various signal alarms and other functions. It is displayed to users in appropriate forms such as sound, graphics, images, etc., and finally achieves the effect of real-time perception of various parameter states of the device.
  • the above-mentioned multivariate test method is used to test and analyze the two variables, and the test value and upper limit value based on the two variables can be obtained.
  • the data is prone to fluctuations. It can be understood that one or two data values currently monitored exceed the standard indicator line, but many subsequent data values conform to the standard indicator line (ie, the alarm point). It will cause so-called false alarms.
  • the alarm method of the "hysteresis" rule can be used to monitor whether the test value exceeds the upper limit value, so as to judge whether the temperature data of the parts meets the set requirements, If it is continuously monitored that the test value of the first set number is greater than the upper limit value, it is determined that the temperature data of the component does not meet the set requirements, and it can be considered that the alarm point has been exceeded.
  • the first set number is 5.
  • N ⁇ 5 that is, when 5 inspection values are continuously monitored and greater than the upper limit value
  • an alarm signal will be issued through the alarm.
  • step S1032 monitor whether the test value exceeds the upper limit to determine whether the temperature data of the parts meets the set requirements, if the second set number of If the verification value is not greater than the upper limit value, it is determined that the temperature data of the component meets the set requirements, and it can be considered that the alarm point is not exceeded.
  • the second set number is 50.
  • the counter starts to reset. It is understandable that after the alarm sends out an alarm signal, when 50 inspection values are continuously monitored and not greater than the upper limit value, it means that the temperature data of the parts monitored in this part meet the set requirements, and the counter can be started at this time. Reset to restart the next round of monitoring.
  • the above method can effectively improve the accuracy of monitoring, reduce false alarms, improve the accuracy of alarms, and reduce operation and maintenance costs.
  • the present application in order to improve the quality of collected temperature data and the accuracy of inspection and analysis, can provide a data processing method for performing data preprocessing on SCADA data, which can remove the data in SCADA data. Abnormal temperature data as well as bad temperature data.
  • the data processing method is used to process the SCADA data set of the wind turbine collected by the SCADA system, and the method includes the following steps S11-S12:
  • Step S11 Obtain at least part of the data in the SCADA data set as a data set to be processed, the data set to be processed includes a plurality of data groups, and each data group includes a variety of data representing different information at the same time, so
  • the various kinds of data include the data to be processed of the internal environment temperature, the data to be processed of the external environment temperature, and the data to be processed of the temperature of the parts.
  • the SCADA data may include monitoring data such as temperature, wind speed, and power.
  • the internal ambient temperature data and external ambient temperature data of the wind turbine generator set, and the component temperature data of the wind turbine generator set also belong to a type of data in SCADA data.
  • SCADA system is Supervisory Control And Data Acquisition (Supervisory Control And Data Acquisition). SCADA system can monitor and control on-site fans to realize data acquisition, equipment control, measurement, parameter adjustment and various signal alarms and other functions. It is displayed to users in appropriate forms such as sound, graphics, images, etc., and finally achieves the effect of real-time perception of various parameter states of the device.
  • Step S12 Perform data screening processing on the data set to be processed through at least one data screening step to obtain a screened data set (which can be understood as a processed data set).
  • the screening dataset includes the internal ambient temperature data, the external ambient temperature data, and the component temperature data.
  • the internal ambient temperature data, the external ambient temperature data, and the component temperature data are obtained from the screening dataset.
  • the data screening step includes a clustering screening step
  • the clustering screening step includes steps S121-S122:
  • Step S121 Use a clustering algorithm to divide the data in the data set to be processed into multiple clusters.
  • Clustering algorithm also known as group analysis, is a statistical analysis method to study (sample or index) classification problems, and is also an important algorithm of data mining. It has the advantages of fast speed, simple calculation and high analysis accuracy.
  • a clustering algorithm is used to divide each piece of data in the data set to be processed into multiple clusters, which can improve the accuracy of data processing and further improve the quality of SCADA data.
  • Step S122 Remove the singular value in each of the clusters and other data of the data group including the singular value.
  • This data screening step is hereinafter referred to as the cluster screening step.
  • the SCADA data is classified by the clustering algorithm, and then the classified data is subjected to data screening processing to remove the singular value in the SCADA data and other data at the same time including the singular value, which can reduce the poor quality in the SCADA data. data to improve the quality of SCADA data.
  • Obtaining the internal ambient temperature data, the external ambient temperature data and the component temperature data from the screening data obtained after the data screening step can remove abnormal temperature data and bad temperature data in the SCADA data, thereby improving the collection efficiency.
  • the clustering algorithm includes a fuzzy C-means clustering algorithm.
  • the clustering algorithm is used to divide the data in the data set to be processed into multiple clusters, including : using the fuzzy C-means clustering algorithm to divide the data in the data set to be processed into multiple clusters.
  • the fuzzy C-means clustering algorithm is a data clustering method based on the optimization of the objective function.
  • the algorithm is an unsupervised fuzzy clustering method, which does not require human intervention in the process of algorithm implementation.
  • the same SCADA data does not belong to one category alone, and the same SCADA data may belong to several categories at the same time. Therefore, the use of such an algorithm is suitable for the diversity of fan SCADA data.
  • fuzzy C-means clustering algorithm is used to cluster the data in the data set to be processed, and then remove the singular value in each of the clusters and other data of the data group including the singular value, that is, including the singular value.
  • Other data at the same time of singular value can reduce the fluctuation of SCADA data and improve the quality of SCADA data.
  • step S122 the removal of the singular value in each of the clusters and other data of the data group including the singular value includes the following sub-step S1221 -S1223:
  • Step S1221 Use the fuzzy C-means clustering algorithm to determine the center of each of the clusters.
  • b represents the weighting index, also known as the smoothing factor, which controls the sharing degree of the mode among the fuzzy classes. Usually, the value of b is 2.
  • Equation 7 Let the partial derivatives of J f to m j and ⁇ j (X i ) be 0, and obtain the necessary conditions for the minimum value of Equation 6. See Equations 7 and 8 below:
  • Equations 7 and 8 are solved by an iterative method until the convergence conditions are met, and the optimal solution is obtained.
  • a set of cluster center values may be randomly given, and then an iterative method is used to solve Equation 7 and Equation 8 until the convergence conditions are met, and an optimal solution is obtained. Or first randomly give a set of weight matrix values, and use an iterative method to solve Equations 7 and 8 until the convergence conditions are met, and the optimal solution is obtained.
  • J f is the target, that is, the objective function of the algorithm iteration.
  • a set of initial weight matrices are randomly given as an example, and the sum of the weights is 1, which means that the sum of the probabilities that each data belongs to each cluster center is equal to 1.
  • each center point m is obtained by formula 7, and then the weight ⁇ is obtained by formula 8 according to the obtained center point m, and the calculation is iterative until the convergence condition is satisfied.
  • Step S1222 Determine the Euclidean distance between the data in at least one of the clusters and the center of the corresponding cluster.
  • each SCADA data will be grouped according to the rules of the fuzzy C-means clustering algorithm.
  • the Euclidean distance of data in at least one cluster from the center of the corresponding cluster is determined.
  • the Euclidean distance between the data in each cluster and the center of the corresponding cluster is determined.
  • the mean and mean square error of the Euclidean distances between the data in all clusters and the centers of the corresponding clusters can be determined. For example, the mean and mean square error of the Euclidean distances between the data in all 10 clusters and the centers of the corresponding clusters are determined.
  • Step S1223 Use Laida's criterion to remove the distance outliers of the Euclidean distance in at least one of the clusters, and remove the singular values corresponding to the distance outliers.
  • the Laida criterion is also called the 3-sigma criterion. Since the mean and mean square error of the Euclidean distance between the data in each cluster and the center of the corresponding cluster conform to a normal distribution, the Laida criterion can usually be used to classify each cluster.
  • the mean and mean square deviation of the Euclidean distance between the data in the class and the center of the corresponding cluster is divided into three intervals, in which the data in the second interval can be understood as meeting the requirements, and the data in the first interval and the third interval It can be understood as unsatisfactory, wherein the second interval is located between the first interval and the third interval.
  • the SCADA data corresponding to the mean and mean square error in the first interval (which can be understood as a range not exceeding the lower limit) and the mean value in the third interval (which can be understood as a range above the upper limit) and
  • the SCADA data corresponding to the mean square error is used as the distance anomaly, and then the singular value in the SCADA data corresponding to the distance anomaly is removed, and other data at the same time including the singular value can be removed.
  • Data deviation, noise, singularity point, reduce the fluctuation of SCADA data, improve the quality of SCADA data and the accuracy of data analysis.
  • the percentage of the second interval is 68.27%
  • the percentages of the first interval and the third interval are both 15.865%.
  • the data in the data set to be processed may also be standardized to obtain a standardized data.
  • the data is standardized by the following formula 9:
  • step S121 using a clustering algorithm to divide the data in the data set to be processed into multiple clusters includes: using a clustering algorithm to divide the standardized data into multiple clusters.
  • the data in the data set to be processed can be standardized by means of, for example, normalization, to obtain standardized data.
  • the standardized data is processed by the fuzzy C-means clustering algorithm, the SCADA data is classified by the fuzzy C-means clustering algorithm, and then the classified data is subjected to data screening processing to remove the singularities in each cluster.
  • value and other data of the data group including the singular value, the normalized data after removing the singular value in each cluster and other data of the data group including the singular value can also be de-normalized, which can be understood as
  • the standardized data after processing by the fuzzy C-means clustering algorithm is restored to the data in the original format, which is convenient for subsequent statistics, analysis and evaluation of the data.
  • the obtaining at least part of the data in the SCADA data set as the data set to be processed includes: obtaining at least part of the data in the set time dimension in the SCADA data set as the data set to be processed the dataset to be processed.
  • SCADA data can be divided into 10min data and 30s data in two dimensions.
  • the 10min data is the average of multiple 30s data. Considering the slowly changing characteristics of the data monitored by the SCADA system due to factors such as wind speed changes, fan speed changes, etc., the fluctuation of 30s data is relatively large.
  • this application selects 10min data, that is, the set time dimension is 10min. Reduce the data fluctuation caused by the variable speed of the fan, and reduce the impact on data analysis and processing.
  • this application can obtain all types of data for data processing, and can also obtain some types of data for data processing, such as several types of data that have a greater impact on fans, such as wind speed, temperature, power and other data.
  • the data screening step may further include: removing data with default values in the data set to be processed and other data in the data group including the data with default values.
  • This data screening step is hereinafter referred to as the step of removing NA, where NA represents a default value.
  • the communication signal may be poor, so the state of signal interruption often occurs.
  • the data when the signal is interrupted is recorded as the default value.
  • the purpose of this data screening step is to remove the The data with the default value in the data set to be processed and other data at the same time in the data group including the data with the default value can reduce the fluctuation of the SCADA data, improve the quality of the SCADA data and the accuracy of the data analysis.
  • the data set to be processed includes power data representing the output power of the wind turbine
  • the data screening step may further include: removing the data set to be processed that characterize the output power Negative power data and other data of the data set including the power data.
  • This data screening step is hereinafter referred to as the removal of negative numbers step.
  • the cut-in wind speed of the fan is for the grid-connected fan, which refers to the wind speed at which the fan reaches the grid-connected condition, that is, the minimum wind speed that can generate electricity, and the fan will automatically stop below this wind speed.
  • the cut-out wind speed of the fan refers to the maximum wind speed at which the fan is connected to the grid for power generation. If the wind speed exceeds this speed, the fan will be cut out of the power grid, that is, the fan will stop and stop generating electricity.
  • the generator of the wind turbine can generate electricity continuously and stably.
  • this data screening step is to remove the power data representing the negative output power in the data set to be processed and other data at the same time in the data group including the power data, which can reduce the fluctuation of SCADA data and improve Quality of SCADA data and accuracy of data analysis.
  • the data screening step may further include: removing the data exceeding the alarm value from the data set to be processed and other data of the data group including the data exceeding the alarm value.
  • This data screening step is hereinafter referred to as the out-of-tolerance removal step.
  • each corresponding SCADA data point can have an alarm value set.
  • the monitored data exceeds the alarm value, it means that the data in this time period is out of tolerance data, which is not a normal fan state, which is not conducive to the follow-up. data analysis.
  • the alarm value of the bearing temperature is 60°.
  • the fan state is judged to be abnormal and an alarm is issued.
  • the purpose of this data screening step is to remove the data exceeding the alarm value in the data set to be processed and other data at the same time in the data group including the data exceeding the alarm value, which can reduce the fluctuation of SCADA data and improve the quality and data analysis of SCADA data. accuracy.
  • the data screening step may further include a quartile screening step, and the quartile screening step includes steps S131-S132:
  • Step S131 Use the quartile method to determine abnormal data in the data set to be processed.
  • Quartiles also known as quartiles, refer to the numerical values at the three dividing points in which all values are arranged from small to large and divided into four equal parts in statistics. Among them, the 1st quartile, Q1, is the 25th percentile. The second quartile, Q2, is the 50th percentile. The third quartile, Q3, is the 75th percentile. Trends in data variables can be compared and analyzed in conjunction with Q1 and Q3.
  • Step S132 Remove the abnormal data and other data of the data group including the abnormal data.
  • This data screening step is hereinafter referred to as the quartile screening step.
  • Obtaining the internal ambient temperature data, the external ambient temperature data and the component temperature data from the screening data obtained after the data screening step can remove abnormal temperature data and bad temperature data in the SCADA data, thereby improving the collection efficiency.
  • the missing data is backfilled between the non-consecutive time series data. time data. It can be understood that after performing data screening processing on the to-be-processed data set through one or more of the above data screening steps, all or most abnormal and bad data can be filtered and removed, leaving relatively high-quality data. However, the time series of the remaining data may not be continuous. The purpose of this step is to fill in the data to obtain continuous time series data, which is convenient for subsequent statistical analysis and evaluation of the data.
  • the time of the first piece of SCADA data in the processed data set is used as the benchmark to check whether the subsequent SCADA data is continuous. Taking the continuous interval of 10 minutes as an example, if the time of the first piece of SCADA data is 2020 -01-01-14:00:00, the time of the second SCADA data is 2020-01-01-14:10:00, then the data is judged to be continuous in time series. If the time of the second SCADA data is 2020-01-01-14:20:00, that is, more than 10 minutes, the data is judged to be discontinuous in time series.
  • backfilling the data at the missing moment between the non-consecutive time series data can include the following two situations:
  • the number of consecutive pieces of data preceding the time series non-consecutive data is greater than 2, then by determining the residual mean and variance of the consecutive pieces of data preceding the time series discontinuous data, and generating random number, as the data for the missing moment.
  • the data corresponding to the time before the time series non-consecutive data is selected as the data at the missing time. For example, the data whose time is 2020-01-01-14:20:00 is judged to be discontinuous, and only one data whose time is 2020-01-01-14:00:00 is judged to be continuous before the data. Then select the data with the time of 2020-01-01-14:00:00 as the data of the missing time.
  • the data processing method of the present application after performing data screening processing on the data set to be processed in the above one or more data screening steps, at least most abnormal and bad data can be filtered and removed, and data with relatively high quality is left, which can reduce SCADA The fluctuation of data improves the quality of SCADA data.
  • Obtaining the internal environmental temperature data, the external environmental temperature data and the component temperature data from the screening data obtained after the data screening step can remove abnormal temperature data and bad temperature data in the SCADA data, thereby improving the collection efficiency.
  • the quality of the temperature data and the accuracy of the inspection analysis may individually perform data processing on the data sets to be processed to obtain respective processing data sets. Then, the data of all the obtained processed data sets are combined to obtain the processed data.
  • multiple data processing steps may be used to sequentially perform data processing on the data set to be processed to obtain the processed data.
  • an embodiment of the present application further provides a monitoring system 30 applied to a wind power generating set, including one or more processors 31 , for implementing the monitoring method described in any of the above embodiments.
  • Embodiments of the monitoring system 30 may be applied to wind turbines.
  • the apparatus embodiment may be implemented by software, or may be implemented by hardware or a combination of software and hardware.
  • FIG. 8 it is a hardware structure diagram of the wind turbine where the monitoring system 30 of the present application is located, except for the processor 31, the internal bus 32, the memory 34, the network interface 33, In addition to the non-volatile memory 35, the wind turbine in which the device is located in the embodiment generally may also include other hardware according to the actual function of the wind turbine, which will not be repeated here.
  • the processor 31 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • a general purpose processor may be a microprocessor or the processor 31 may be any conventional processor or the like.
  • Embodiments of the present application further provide a computer-readable storage medium, on which a program is stored, and when the program is executed by the processor 31, the monitoring method described in any of the foregoing embodiments is implemented.
  • the computer-readable storage medium may be an internal storage unit of the wind power generator set in any of the foregoing embodiments, such as a hard disk or a memory.
  • the computer-readable storage medium can also be an external storage device of the wind turbine, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, and a flash memory card (Flash Card) equipped on the device. Wait.
  • the computer-readable storage medium may also include both an internal storage unit of the wind turbine and an external storage device.
  • the computer-readable storage medium is used to store the computer program and other programs and data required by the wind turbine, and can also be used to temporarily store data that has been output or will be output.

Landscapes

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

Abstract

一种风力发电机组的监控方法、系统及计算机可读存储介质。风力发电机组的监控方法，包括：S101、获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据；S102、根据所述零部件温度数据与所述内部环境温度数据的差值确定第一统计量，根据所述零部件温度数据与所述外部环境温度数据的差值确定第二统计量；S103、根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求。本监控方法可以减少环境因素对风力发电机组的零部件温度的影响而降低监测的准确性的情况，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。

Description

风力发电机组的监控方法、系统及计算机可读存储介质 技术领域
本申请涉及风力发电机领域，尤其涉及一种风力发电机组的监控方法、系统及计算机可读存储介质。
背景技术
随着时代的进步和人类环保意识的提升，对可再生清洁能源的开发和利用越来越受到国际社会的重视。为世界范围内技术最成熟、最具规模化商业开发潜力的新能源之一，风能具有蕴藏量丰富、可再生、分布广、无污染等特性，具备规模化开发利用价值。风能作为一种储量丰富、可再生、零排放的清洁能源，风力发电技术已经成为各国争相发展的重要领域，并且已经被提升到了国家战略的高度。
风电机组的状态监测技术通过温度、转速、振动等传感器对机组的叶轮、齿轮箱、发电机、偏航与变桨机构等关键位置进行实时数据采集、监控和相关数据分析，了解机组的运行健康状态，及时发现故障征兆，提前进行相关故障的处理及发电运行的合理安排，从而提高运营效率，降低运维成本，使效益最大化。
由于风电机组在不同工作环境中，受到环境影响零部件的温度所反映出的情况是不同的，现有相关技术中不能有效监测到环境因素对风力发电机组的零部件的温度的影响会降低监测的准确性，容易造成虚假警报，增加了运维成本。
发明内容
本申请提供一种风力发电机组的监控方法、系统及计算机可读存储介质。
根据本申请实施例的第一方面，提供一种风力发电机组的监控方法，包括：获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据；根据所述零部件温度数据与所述内部环境温度数据的差值确定第一统计量，根据所述零部件温度数据与所述外部环境温度数据的差值确定第二统计量；根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求。
可选地，所述根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求，进一步包括：采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值；监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求。
可选地，所述多变量检验方法包括霍特林T平方分布检验方法，所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，进一步包括：采用霍特林T平方分布检验方法对所述第一统计量和所述第二统计量进行检验分析。
可选地，所述获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以 及所述风力发电机组的零部件温度数据，包括：获取设定数量的所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值，进一步包括：根据所述第一统计量、所述第二统计量、所述设定数量个所述第一统计量的均值、以及所述设定数量个所述第二统计量的均值，确定所述第一统计量和所述第二统计量之间的检验值；根据所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数所确定的统计量的个数值和所述设定数量，经过卡方分布确定所述上限值。
可选地，所述监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求，进一步包括：若连续监测到第一设定个数的所述检验值大于所述上限值，确定所述零部件温度数据不满足设定要求。
可选地，所述监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求，进一步包括：若连续监测到第二设定个数的所述检验值不大于所述上限值，确定所述零部件温度数据满足设定要求。
可选地，所述零部件温度数据包括轴承温度。
可选地，通过SCADA系统采集所述风力发电机组的SCADA数据集，所述获取所述风力发电机组的内部环境温度数据和外部环境温度数据以及所述风力发电机组的零部件温度数据，包括：获取所述SCADA数据集中的至少部分数据，作为待处理数据集，所述待处理数据集包括多个数据组，每个数据组包括同一时刻下的表征不同信息的多种数据，所述多种数据包括内部环境温度待处理数据、外部环境温度待处理数据以及零部件温度待处理数据；通过至少一个数据筛选步骤对所述待处理数据集进行数据筛选处理，得到筛选数据集，所述筛选数据集包括所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据；自所述筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。
可选地，所述获取所述SCADA数据集中的至少部分数据，作为待处理数据集，包括：获取所述SCADA数据集中设定时间维度的至少部分数据，作为所述待处理数据集。
可选地，所述数据筛选步骤包括：采用聚类算法将所述待处理数据集中的数据划分为多个聚类；及去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据。
可选地，所述聚类算法包括模糊C均值聚类算法，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：采用模糊C均值聚类算法将所述待处理数据集中的数据划分为多个聚类。
可选地，所述去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据，包括：采用模糊C均值聚类算法确定每个所述聚类的中心；确定至少一个所述聚类中的数据与对应的所述聚类的中心的欧氏距离；及采用拉依达准则去除至少一个所述聚类中的所述欧氏距离的距离异常值，并去除所述距离异常值对应的所述奇异值。
可选地，所述数据筛选步骤包括：采用四分位数法确定所述待处理数据集中的异常 数据；去除所述异常数据和包括该异常数据的数据组的其他数据。
可选地，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类之前，还包括：对所述待处理数据集中的数据进行标准化处理，得到标准化数据。所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：采用聚类算法将所述标准化数据划分为多个聚类。
可选地，所述去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据之后，还包括：对去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据后的所述标准化数据进行逆标准化处理。
可选地，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：采用聚类算法将所述待处理数据集中的每种数据分别划分为多个聚类。
可选地，对所述待处理数据集进行数据筛选处理之后，还包括：若所述筛选数据集的数据的时间序列不连续，在时间序列非连续的数据之间，回填所缺失时刻的数据。
可选地，所述在时间序列非连续的数据之间，回填所缺失时刻的数据，包括：若所述时间序列非连续的数据之前的连续多个数据的数量大于2，确定所述时间序列非连续的数据之前的连续多个数据的残差均值及方差，并基于所述残差均值和方差生成随机数，作为所缺失时刻的数据。
可选地，所述在时间序列非连续的数据之间，回填所缺失时刻的数据，包括：若所述时间序列非连续的数据之前的连续多个数据的数量不大于2，选取时间序列非连续的数据之前的时刻所对应的数据，作为所缺失时刻的数据。
根据本申请实施例的第二方面，提供一种应用于风力发电机组的监控系统，包括一个或多个处理器，用于实现如上任一实施例所述的监控方法。
根据本申请实施例的第三方面，提供一种计算机可读存储介质，其上存储有程序，该程序被处理器执行时，实现如上任一实施例所述的监控方法。
根据本申请实施例提供的技术方案，结合环境温度和风力发电机组的零部件温度建立两个变量，根据两个变量判断零部件温度是否满足设定要求，可以减少环境因素对风力发电机组的零部件温度的影响而降低监测的准确性的情况，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，并不能限制本申请。
附图说明
此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本申请的实施例，并与说明书一起用于解释本申请的原理。
图1是本申请一示例性实施例示出的一种风力发电机组的监控方法的流程示意图。
图2和图3是图1所示的监控方法的部分细化流程示意图。
图4是本申请一示例性实施例示出的一种数据处理方法的流程示意图。
图5是本申请一示例性实施例示出的一种数据处理方法的数据筛选步骤的流程示意图。
图6是图5所示的数据筛选步骤的部分细化流程示意图。
图7是本申请另一示例性实施例示出的一种数据处理方法的数据筛选步骤的流程示意图。
图8是本申请一示例性实施例示出的一种风力发电机组的监控系统的系统框图。
具体实施方式
这里将详细地对示例性实施例进行说明，其示例表示在附图中。下面的描述涉及附图时，除非另有表示，不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本申请相一致的所有实施方式。相反，它们仅是与如所附权利要求书中所详述的、本申请的一些方面相一致的装置和方法的例子。
在本申请使用的术语是仅仅出于描述特定实施例的目的，而非旨在限制本申请。在本申请和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式，除非上下文清楚地表示其他含义。还应当理解，本文中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。
应当理解，尽管在本申请可能采用术语第一、第二、第三等来描述各种信息，但这些信息不应限于这些术语。这些术语仅用来将同一类型的信息彼此区分开。例如，在不脱离本申请范围的情况下，第一信息也可以被称为第二信息，类似地，第二信息也可以被称为第一信息。取决于语境，如在此所使用的词语“如果”可以被解释成为“在……时”或“当……时”或“响应于确定”。
由于风力发电机组在不同工作环境中，受到环境因素的影响零部件温度所反映出的情况是不同的。以轴承温度为例，风力发电机组在夏天和冬天所处的环境是不一致，假设轴承温度报警线为80°，在夏天由于室外温度普遍较高，轴承温度常年为60°左右均可视为是正常状态。但在冬天若轴承温度依旧为60°，即使离报警线80°还有一定差距，但这个温度值还是需要警戒注意的。而现有相关技术中不能有效监测到环境因素对风力发电机组的零部件的温度的影响会降低监测的准确性，容易造成虚假警报，增加了运维成本。
本申请提供一种风力发电机组的监控方法、系统及计算机可读存储介质，可以减少环境因素对风力发电机组的零部件温度的影响而降低监测的准确性的情况，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。下面结合附图，对本申请的风力发电机组的监控方法、系统及计算机可读存储介质进行详细说明。在不冲突的情况下，下述的实施例及实施方式中的特征可以相互组合。
参见图1所示，本申请实施例提供一种风力发电机组的监控方法，包括以下步骤S101-S103：
步骤S101：获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及 所述风力发电机组的零部件温度数据。可选地，所述零部件可以是风力发电机组的轴承，所述零部件温度数据可以包括轴承温度。所述零部件也可以是其他部件，例如齿轮箱，所述零部件温度数据例如齿轮箱油温，本申请对比不作限制。
其中，所述内部环境温度数据可以是指风力发电机组的机舱的内部环境温度，所述外部环境温度可以是指风力发电机组的机舱的外部环境温度。可以通过在风机发电机组的机舱内外分别设置温度传感器，以及在风力发电机组的零部件上设置温度传感器，来获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据。
步骤S102：根据所述零部件温度数据与所述内部环境温度数据的差值确定第一统计量，根据所述零部件温度数据与所述外部环境温度数据的差值确定第二统计量。
其中，结合所述风力发电机组的内部环境温度数据、外部环境温度数据以及零部件温度数据，可以采用式(1)和式(2)建立新的变量第一统计量t 1和第二统计量t 2。
t 1＝t b-t out    (1)；
t 2＝t b-t in      (2)；
其中，t b为零部件温度，例如轴承温度。t out为风力发电机组的外部环境温度，t in为风力发电机组的内部环境温度。
步骤S103：根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求。
本申请结合环境温度和风力发电机组的零部件温度建立两个变量，根据两个变量，判断零部件温度数据是否满足设定要求，可以减少变工况下环境因素对零部件温度数据的影响会降低监测的准确性的情况，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。
参见图2所示，在一些可选的实施例中，所述根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求的步骤S103，可以进一步包括子步骤S1031-S1032：
步骤S1031：采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值。
需要说明的是，监测的两个变量第一统计量和第二统计量其本质是残差领域，根据合理的假设可以假设这两个变量分布符合正态分布，因此可以利用其优良且可控的性质，采用多变量检验方法对两个变量进行检验分析及监控，提高监控分析的准确性。
步骤S1032：监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求。
通过上述方法，本申请结合环境温度和风力发电机组的零部件温度建立两个变量进行检验分析，采用多变量检验方法对两个变量进行检验分析，通过比较检验分析得到的检验值和上限值，判断所述零部件温度数据是否满足设定要求，可以减少变工况下环境 因素对零部件温度数据的影响会降低监测的准确性的情况，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。
在一些可选的实施例中，所述多变量检验方法包括霍特林T平方分布检验方法(即Hotelling T 2)。在步骤S1031中，所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，可以是采用霍特林T平方分布检验方法对所述第一统计量和所述第二统计量进行检验分析。需要说明的是，在其他例子中，所述多变量检验方法也可以包括其他检验方法，只要能够实现将两个数据样本结合分析即可，本申请对此不作限制。
可以理解的，结合环境温度和风力发电机组的零部件温度建立两个与轴承温度相关的变量，采用霍特林T平方分布检验方法可以对上述两个变量进行合并为一个变量构造监控图(control chart)并进行检验分析，可以将环境温度对风力发电机组的零部件温度的影响也考虑在内进行分析，从而提高通过零部件温度判断风力发电机组的运行健康状态的准确性。
在一些可选的实施例中，在步骤S101中，所述获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据，可以获取设定数量的所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。由于所采集的温度数据的数量是根据时间区段进行划分的，因此通过上述方法可以对零部件温度以划分时间区段的方式地进行监控及监测，减少数据量过大造成的数据误差，提高监控的准确性。
参见图3所示，所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值的步骤S1031，可以包括子步骤S10311-S10312：
步骤S10311：根据所述第一统计量、所述第二统计量、所述设定数量个所述第一统计量的均值、以及所述设定数量个所述第二统计量的均值，采用式(3)确定所述第一统计量和所述第二统计量之间的检验值。
Figure PCTCN2021093632-appb-000001
其中，T 2为所述检验值，x为第一统计量和第二统计量的数值，
Figure PCTCN2021093632-appb-000002
为设定数量个第一统计量的均值和设定数量个第二统计量的均值，S -1是第一统计量t 1和第二统计量t 2的协方差的矩阵的逆。
步骤S10312：根据所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数所确定的统计量的个数值和所述设定数量，采用式(4)和式(5)经过卡方分布确定所述上限值：
Figure PCTCN2021093632-appb-000003
LCL＝0     (5)；
其中，UCL是上限值，LCL是下限值。p是用于检验分析的变量的个数。m是用于作为健康状态的初始值数量级。F是说明这个式子属于卡方分布(chi-square distribution)。a是卡方分布的上位点。
在本实施例中，用于检验分析的变量为第一统计量和第二统计量，即p＝2。m的取值为1000，即第一统计量和第二统计量的设定个数均为1000，则m-p＝998。a的取值为0.001，则根据卡布分布可以自行计算得出F a,p,m-p的值。当然，在其他例子中，可以根据实际情况设定m的取值，本申请对此不作限制。
可以理解的，假设m的取值为2，第一统计量和第二统计量的数值为[t1,t2]:[1,10]和[3,30]，那么
Figure PCTCN2021093632-appb-000004
Figure PCTCN2021093632-appb-000005
分别是[1,10]-[2,20]＝[-1,-10]以及[3,30]-[2,20]＝[1,10]。
通常可以通过SCADA系统采集得到风力发电机组的SCADA数据，SCADA数据可以包括用于表征风力发电机组性能的温度、风速、功率等监测数据。所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据也属于SCADA数据中的一类数据。SCADA系统即数据采集监控系统(Supervisory Control And Data Acquisition)，SCADA系统可以对现场的风机进行监视和控制，以实现数据采集、设备控制、测量、参数调节以及各类信号报警等各项功能，并以适当的形式如声音、图形、图象等方式显示给用户，最终达到实时感知设备各种参数状态的效果。
采用上述多变量检验方法对两个变量进行检验分析，可以得到基于两个变量的检验值和上限值。但由于SCADA数据本身的缓变特性，数据容易造成波动，可理解为是当前监测的某一两个数据值超出标准指标线，但后续很多数据值均符合标准指标线(即报警点)，就会造成所谓的虚假报警。为了减少这类情况的发生，上述步骤S1032中，可以采用“滞后”规则的报警方式，监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求，若连续监测到第一设定个数的所述检验值大于所述上限值，则确定所述零部件温度数据不满足设定要求，即可认为是超过了报警点。
可选地，所述第一设定个数为5。风力发电机组内可以设置计数器和警报器。可以理解的，当基于当前采集的温度数据，采用上述多变量检验方法检验分析得到的检验值超出上限值，计数器开始计数N＝N+1，N为超出上限值的检验值的数量。当N≥5，即连续监测到5个检验值大于上限值时，通过警报器发出报警信号。通过上述“滞后”规则的报警方式，可以减少虚假报警的情况，提高报警的准确性，降低运维成本。
进一步地，回到图2，步骤S1032中，监控所述检验值是否超出所述上限值，以判 断所述零部件温度数据是否满足设定要求，若连续监测到第二设定个数的所述检验值不大于所述上限值，则确定所述零部件温度数据满足设定要求，即可认为是未超过报警点。
可选地，所述第二设定个数为50。在警报器发出报警信号后，计数器开始重置。可以理解的，警报器发出一次报警信号后，当之后连续监测到50个检验值不大于上限值时，则表示这部分监测的零部件温度数据均满足设定要求，此时可以将计数器开始重置，重新开始下一轮的监测。通过上述方法，可以有效提高监测的准确性，减少虚假报警的情况，提高报警的准确性，降低运维成本。
在一些可选的实施例中，为了提高采集的温度数据的质量及检验分析的准确性，本申请可以提供一种数据处理方法，用于对SCADA数据进行数据前处理，可以去除SCADA数据中的异常温度数据以及不良温度数据。
参见图4所示，所述数据处理方法，用于处理SCADA系统采集的风力发电机组的SCADA数据集，所述方法包括以下步骤S11-S12：
步骤S11：获取所述SCADA数据集中的至少部分数据，作为待处理数据集，所述待处理数据集包括多个数据组，每个数据组包括同一时刻下的表征不同信息的多种数据，所述多种数据包括内部环境温度待处理数据、外部环境温度待处理数据以及零部件温度待处理数据。可以理解的，SCADA数据可以包括温度、风速、功率等监测数据。所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据也属于SCADA数据中的一类数据。同一时刻内SCADA系统采集到的全部上述数据，可以划分在同一个数据组内。SCADA系统即数据采集监控系统(Supervisory Control And Data Acquisition)，SCADA系统可以对现场的风机进行监视和控制，以实现数据采集、设备控制、测量、参数调节以及各类信号报警等各项功能，并以适当的形式如声音、图形、图象等方式显示给用户，最终达到实时感知设备各种参数状态的效果。
步骤S12：通过至少一个数据筛选步骤对所述待处理数据集进行数据筛选处理，得到筛选数据集(可理解为是已处理数据集)。所述筛选数据集包括所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。自所述筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。
参见图5所示，其中，所述数据筛选步骤包括聚类筛选步骤,聚类筛选步骤包括步骤S121-S122：
步骤S121：采用聚类算法将所述待处理数据集中的数据划分为多个聚类。聚类算法又称群分析，它是研究(样品或指标)分类问题的一种统计分析方法，同时也是数据挖掘的一个重要算法，具有速度快，计算简便、分析准确性高等优点。在本实施例中，采用聚类算法将待处理数据集中的每笔数据分别划分为多个聚类，可以提高数据处理的准确性，进一步提高SCADA数据的质量。
步骤S122：去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据。此数据筛选步骤以下简称为聚类筛选步骤。
通过聚类算法对SCADA数据进行分类，再对分类后的数据进行数据筛选处理，将 SCADA数据中的奇异值去除，以及包括该奇异值的同一时刻的其他数据，可以减少SCADA数据中质量较差的数据，提升SCADA数据的质量。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。
在一些可选的实施例中，所述聚类算法包括模糊C均值聚类算法，在步骤S121中，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：采用模糊C均值聚类算法将所述待处理数据集中的数据划分为多个聚类。模糊C均值聚类算法是基于对目标函数的优化基础上的一种数据聚类方法。该算法是一种无监督的模糊聚类方法，在算法实现过程中不需要人为的干预。同时，由于增加模糊技术(fuzzy)，同一笔SCADA数据并不单独属于一个分类，同一个SCADA数据可能同时属于几个分类。因此采用这样的算法，适合风机SCADA数据的多样性的特点。
可以理解的，采用模糊C均值聚类算法将待处理数据集中的数据进行聚类划分，再去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据，即包括该奇异值的同一时刻的其他数据，如此可以减少SCADA数据的波动，提升SCADA数据的质量。
参见图6所示，在一些可选的实施例中，在步骤S122中，所述去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据，包括以下子步骤S1221-S1223：
步骤S1221：采用模糊C均值聚类算法确定每个所述聚类的中心。
其中，上述获取的SCADA数据以给定的数据集X＝{X 1,X 2,…,X n}表示，聚类数目为k，m j(j＝1,2,…,k)为每个聚类的中心，μ j(X i)是第i个样本对应第J类的隶属度函数，表示权重矩阵(可理解为是权重)，则基于隶属度函数的聚类损失函数可以参见下式6：
Figure PCTCN2021093632-appb-000006
其中，
Figure PCTCN2021093632-appb-000007
表示某个数据属于各个聚类的中心的概率，这些概率之和等于1。b表示加权指数，也称为平滑因子，控制模式在模糊类间的分享程度，通常情况下b的取值为2。
令J f对m j和μ j(X i)的偏导为0，求得式6极小值的必要条件。参见下式7和式8：
Figure PCTCN2021093632-appb-000008
Figure PCTCN2021093632-appb-000009
采用迭代的方法求解式7和式8，直至满足收敛条件，得到最优解。
在一些实施例中，可以先随机给出一组聚类中心的值，再采用迭代的方法求解式7和式8，直至满足收敛条件，得到最优解。或是先随机给出一组权重矩阵的值，采用迭代的方法求解式7和式8，直至满足收敛条件，得到最优解。
本实施例中，假设在该算法中有10个分类，即k＝10。则通过上述方法，可以最终得到10个分类的中心点，即m 1,m 2,…,m 10。X 1,X 2,…,X 10表示为训练算法时，不同时间下获取的SCADA数据。μ就是权重，如μ 5(x 3)＝0.8，表明x 3数据属于第5个分类群的权重为0.8。需要说明的是，权重的取值一般在(0，1)范围，权重的值越大，说明该数据属于越贴近于对应分类群。J f为目标，即算法迭代的目标函数。本实施例中，以先随机给出一组初始权重矩阵为例，并满足权重总和为1，代表每个数据属于各个聚类中心的概率之和等于1。根据初始权重矩阵通过式7求出各个中心点m，再根据求出的各个中心点m通过式8求出权重μ，如此迭代计算直到满足收敛条件。
步骤S1222：确定至少一个所述聚类中的数据与对应的所述聚类的中心的欧氏距离。
通过上述方法，可以得到多个(例如10个)模糊C均值聚类算法的中心，并且根据模糊C均值聚类算法的规则每个SCADA数据均会被进行分组。确定至少一个聚类中的数据与对应的聚类的中心的欧氏距离。在本实施例中，确定每个聚类中的数据与对应的聚类的中心的欧氏距离。可以确定所有聚类中的数据与对应的聚类的中心的欧式距离的均值和均方差。例如，确定全部的10个聚类中的数据与对应的聚类的中心的欧式距离的均值和均方差。
步骤S1223：采用拉依达准则去除至少一个所述聚类中的所述欧氏距离的距离异常值，并去除所述距离异常值对应的所述奇异值。
拉依达准则又称3-sigma准则，由于每个聚类中的数据与对应的聚类的中心的欧式距离的均值和均方差符合正态分布，采用拉依达准则通常可以将每个聚类中的数据与对应的聚类的中心的欧式距离的均值和均方差划分为三个区间，其中位于第二区间的数据可以理解为是符合要求的，位于第一区间和第三区间的数据可以理解为是不符合要求的，其中第二区间位于第一区间和第三区间之间。在本实施例中，将位于第一区间(可理解为是不超过下限的范围)的均值和均方差所对应的SCADA数据和位于第三区间(可理解为是上限以上的范围)的均值和均方差所对应的SCADA数据作为所述距离异常值，再将该距离异常值对应的SCADA数据中的奇异值去除，以及包括该奇异值的同一时刻的其他数据，可以去除数据偏差、噪声、奇异点，减少SCADA数据的波动，提升SCADA数据的质量及数据分析的准确性。可选地，第二区间的百分比为68.27％，第一区间和第三区间的百分比均为15.865％。
在一些可选的实施例中，在步骤S121采用聚类算法将所述待处理数据集中的数据划分为多个聚类之前，还可以对所述待处理数据集中的数据进行标准化处理，得到标准化数据。
通过将数据进行标准化处理，能够提高采用模糊C均值聚类算法对数据进行处理的精度。在本实施例中，通过下式9对数据进行标准化处理：
Figure PCTCN2021093632-appb-000010
其中，data表示SCADA数据的原始数据，np.min(data)表示同类SCADA数据中的最小值，np.max(data)表示同类SCADA数据中的最大值。在步骤S121中，采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：采用聚类算法将所述标准化数据划分为多个聚类。可选地，可以采用例如归一化等方式将待处理数据集中的数据进行标准化处理，得到标准化数据。
进一步地，采用模糊C均值聚类算法对标准化处理的数据进行处理，采用模糊C均值聚类算法对SCADA数据进行分类，再对分类后的数据进行数据筛选处理，去除每个聚类中的奇异值及包括该奇异值的数据组的其他数据之后，还可以对去除每个聚类中的奇异值及包括该奇异值的数据组的其他数据后的标准化数据进行逆标准化处理，可以理解为将经过模糊C均值聚类算法等处理后的标准化数据还原到初始格式的数据，便于后续对数据进行统计及分析评估。
在一些可选的实施例中，步骤S11中，所述获取所述SCADA数据集中的至少部分数据，作为待处理数据集，包括：获取所述SCADA数据集中设定时间维度的至少部分数据，作为所述待处理数据集。SCADA数据可以分为10min数据和30s数据两种维度的数据。其中10min数据是多个30s数据的均值。考虑到SCADA系统所监测的数据由于风速变动、风机变转速等因素所造成的缓变特性，30s数据的波动较大，可选地，本申请选用10min数据，即设定时间维度是10min，可以减小风机变转速所带来的数据波动，减小对数据分析处理的影响。此外，由于SCADA数据的种类繁多，本申请可以获取全部种类的全部数据进行数据处理，也可以获取其中一部分种类的数据进行数据处理，例如对风机影响较大的几类数据，如风速、温度、功率等数据。
在一些可选的实施例中，所述数据筛选步骤还可以包括：去除所述待处理数据集中的为缺省值的数据及包括为缺省值的所述数据的数据组的其他数据。此数据筛选步骤以下简称为去除NA步骤，NA表示缺省值。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。
在实际风机运行中，由于风机的地理位置可能通讯信号不佳，因此时常有信号中断的状态发生，在SCADA系统中将信号中断时的数据记录为缺省值，此数据筛选步骤的 目的是去除待处理数据集中的为缺省值的数据及包括为缺省值的数据的数据组中同一时刻的其他数据，可以减少SCADA数据的波动，提升SCADA数据的质量及数据分析的准确性。
在一些可选的实施例中，所述待处理数据集包括表征所述风力发电机组的输出功率的功率数据，所述数据筛选步骤还可以包括：去除所述待处理数据集中表征所述输出功率为负数的功率数据及包括该功率数据的数据组的其他数据。此数据筛选步骤以下简称为去除负数步骤。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。
可以理解的，风机的切入风速是针对并网型风机而言的，是指风机达到并网条件的风速，也就是可以发电的最低风速，低于此风速风机会自动停机。风机的切出风速指风机并网发电的最大风速，超过此风速风机将切出电网，也即风机会停机，停止发电。当风机达到切入风速时，风机的发电机可以持续稳定的发电。
在实际风机运行中，由于风机检修、停机或者风速为小风天等因素，当风机没有达到切出风速时，SCADA系统中记录的SCADA数据对应的功率值即为负数，这些数据不利于后续的数据分析，此数据筛选步骤的目的是去除所述待处理数据集中表征所述输出功率为负数的功率数据及包括该功率数据的数据组中同一时刻的其他数据，可以减少SCADA数据的波动，提升SCADA数据的质量及数据分析的准确性。
在一些可选的实施例中，所述数据筛选步骤还可以包括：去除所述待处理数据集中超出报警值的数据及包括该超出报警值的数据的数据组的其他数据。此数据筛选步骤以下简称为去除超差步骤。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。
在实际风机运行中，每个对应的SCADA数据点位皆可有报警值设定，当监测的数据超过报警值即说明该时间段的数据为超差数据，不是正常的风机状态，不利于后续的数据分析。例如轴承温度的报警值是60°，当监测到的实际轴承温度高于60°时即判断风机状态异常，发出报警。此数据筛选步骤的目的是去除待处理数据集中超出报警值的数据及包括该超出报警值的数据的数据组中同一时刻的其他数据，可以减少SCADA数据的波动，提升SCADA数据的质量及数据分析的准确性。
参见图7所示，在一些可选的实施例中，所述数据筛选步骤还可以包括四分位数筛选步骤，四分位数筛选步骤包括步骤S131-S132：
步骤S131：采用四分位数法确定所述待处理数据集中的异常数据。四分位数(Quartile)也称四分位点，是指在统计学中把所有数值由小到大排列并分成四等份，处于三个分割点位置的数值。其中第1四分位数Q1，即第25百分位数。第2四分位数Q2，即第50百分位数。第3四分位数Q3，即第75百分位数。可以结合Q1和Q3比较 分析数据变量的趋势。
步骤S132：去除所述异常数据和包括该异常数据的数据组的其他数据。采用四分位数法确定所述待处理数据集中的异常数据，可以提高数据分析的准确性。此数据筛选步骤以下简称为四分位数筛选步骤。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。
在一些可选的实施例中，对所述待处理数据集进行数据筛选处理之后，若所述已处理数据集的数据的时间序列非连续，在时间序列非连续的数据之间，回填所缺失时刻的数据。可以理解的，在经过上述一个或多个数据筛选步骤对待处理数据集进行数据筛选处理之后，可以过滤和去除掉全部或大部分异常和不良数据，剩下质量相对较高的数据。但是剩下的数据的时间序列可能不会是连续的，此步骤的目的是对数据进行填充，得到时间序列连续的数据，便于后续对数据进行统计及分析评估。
在本实施例中，以已处理数据集中的第一条SCADA数据的时间作为基准，向后查看后续SCADA数据是否为连续，以连续间隔为10min为例，若第一条SCADA数据的时间为2020-01-01-14:00:00，第二条SCADA数据的时间为2020-01-01-14:10:00，则该条数据判为时间序列连续。若第二条SCADA数据的时间为2020-01-01-14:20:00，即超过10min，则该条数据判为时间序列非连续。
在确定了时间序列非连续的数据后，在时间序列非连续的数据之间，回填所缺失时刻的数据，可以包括以下两种情况：
若所述时间序列非连续的数据之前的连续多个数据的数量大于2，那么通过确定所述时间序列非连续的数据之前的连续多个数据的残差均值及方差，并基于随机方程生成随机数，作为所缺失时刻的数据。
若所述时间序列非连续的数据之前的连续多个数据的数量不大于2，那么选取时间序列非连续的数据之前的时刻所对应的数据，作为所缺失时刻的数据。例如，时间为2020-01-01-14:20:00的数据被判断为非连续的，该数据前只有一个时间为2020-01-01-14:00:00的数据被判断为连续的，那么选取时间为2020-01-01-14:00:00的数据，作为所缺失时刻的数据。
本申请的数据处理方法，上述一个或多个数据筛选步骤对待处理数据集进行数据筛选处理之后，可以过滤和去除掉至少大部分异常和不良数据，剩下质量相对较高的数据，可以减少SCADA数据的波动，提升SCADA数据的质量。自经过数据筛选步骤后得到的筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据，可以去除SCADA数据中的异常温度数据以及不良温度数据，从而提高采集的温度数据的质量及检验分析的准确性。需要说明的是，当数据处理方法包括多个数据筛选步骤时，可以将多个数据处理步骤对待处理数据集单独进行数据处理，得到各自对应的处理数据集。然后再将得到的全部处理数据集的数据合并得到所述已处理数据。 或者，可以用多个数据处理步骤依次对所述待处理数据集进行数据处理，得到所述已处理数据。
参见图8所示，本申请实施例还提供一种应用于风力发电机组的监控系统30，包括一个或多个处理器31，用于实现如上任一实施例所述的监控方法。
监控系统30的实施例可以应用在风力发电机组上。装置实施例可以通过软件实现，也可以通过硬件或者软硬件结合的方式实现。以软件实现为例，作为一个逻辑意义上的装置，是通过其所在风力发电机组的处理器31将非易失性存储器中对应的计算机程序指令读取到内存中运行形成的。从硬件层面而言，如图8所示，为本申请监控系统30所在风力发电机组的一种硬件结构图，除了图8所示的处理器31、内部总线32、内存34、网络接口33、以及非易失性存储器35之外，实施例中装置所在的风力发电机组通常根据该风力发电机的实际功能，还可以包括其他硬件，对此不再赘述。
所述处理器31可以是中央处理单元(Central Processing Unit，CPU)，还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器31也可以是任何常规的处理器等。
本申请实施例还提供一种计算机可读存储介质，其上存储有程序，该程序被处理器31执行时，实现如上任一实施例所述的监控方法。
所述计算机可读存储介质可以是前述任一实施例所述的风力发电机组的内部存储单元，例如硬盘或内存。所述计算机可读存储介质也可以是风力发电机的外部存储设备，例如所述设备上配备的插接式硬盘、智能存储卡(Smart Media Card，SMC)、SD卡、闪存卡(Flash Card)等。进一步的，所述计算机可读存储介质还可以既包括风力发电机组的内部存储单元也包括外部存储设备。所述计算机可读存储介质用于存储所述计算机程序以及所述风力发电机组所需的其他程序和数据，还可以用于暂时地存储已经输出或者将要输出的数据。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (21)

  1. 一种风力发电机组的监控方法，包括：
    获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据；
    根据所述零部件温度数据与所述内部环境温度数据的差值确定第一统计量，根据所述零部件温度数据与所述外部环境温度数据的差值确定第二统计量；
    根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求。
  2. 如权利要求1所述的方法，其特征在于，所述根据所述第一统计量和所述第二统计量，判断所述零部件温度数据是否满足设定要求，进一步包括：
    采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值；
    监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求。
  3. 如权利要求2所述的方法，其特征在于，所述多变量检验方法包括霍特林T平方分布检验方法，所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，进一步包括：采用霍特林T平方分布检验方法对所述第一统计量和所述第二统计量进行检验分析。
  4. 如权利要求3所述的方法，其特征在于，所述获取所述风力发电机组的内部环境温度数据和外部环境温度数据，以及所述风力发电机组的零部件温度数据，包括：
    获取设定数量的所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据；
    所述采用多变量检验方法对所述第一统计量和所述第二统计量进行检验分析，确定所述第一统计量和所述第二统计量之间的检验值和所述检验值的上限值，进一步包括：
    根据所述第一统计量、所述第二统计量、所述设定数量个所述第一统计量的均值、以及所述设定数量个所述第二统计量的均值，确定所述第一统计量和所述第二统计量之间的检验值；
    根据所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数所确定的统计量的个数值和所述设定数量，经过卡方分布确定所述上限值。
  5. 如权利要求2所述的方法，其特征在于，所述监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求，进一步包括：
    若连续监测到第一设定个数的所述检验值大于所述上限值，确定所述零部件温度数据不满足设定要求。
  6. 如权利要求2所述的方法，其特征在于，所述监控所述检验值是否超出所述上限值，以判断所述零部件温度数据是否满足设定要求，进一步包括：
    若连续监测到第二设定个数的所述检验值不大于所述上限值，确定所述零部件温度 数据满足设定要求。
  7. 如权利要求1所述的方法，其特征在于，所述零部件温度数据包括轴承温度。
  8. 如权利要求1所述的方法，其特征在于，通过SCADA系统采集所述风力发电机组的SCADA数据集，所述获取所述风力发电机组的内部环境温度数据和外部环境温度数据以及所述风力发电机组的零部件温度数据，包括：
    获取所述SCADA数据集中的至少部分数据，作为待处理数据集，所述待处理数据集包括多个数据组，每个数据组包括同一时刻下的表征不同信息的多种数据，所述多种数据包括内部环境温度待处理数据、外部环境温度待处理数据以及零部件温度待处理数据；
    通过至少一个数据筛选步骤对所述待处理数据集进行数据筛选处理，得到筛选数据集，所述筛选数据集包括所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据；自所述筛选数据集中获取所述内部环境温度数据、所述外部环境温度数据以及所述零部件温度数据。
  9. 如权利要求8所述的方法，其特征在于，所述获取所述SCADA数据集中的至少部分数据，作为待处理数据集，包括：
    获取所述SCADA数据集中设定时间维度的至少部分数据，作为所述待处理数据集。
  10. 如权利要求8所述的方法，其特征在于，所述数据筛选步骤包括：
    采用聚类算法将所述待处理数据集中的数据划分为多个聚类；及
    去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据。
  11. 如权利要求10所述的方法，其特征在于，所述聚类算法包括模糊C均值聚类算法，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：
    采用模糊C均值聚类算法将所述待处理数据集中的数据划分为多个聚类。
  12. 如权利要求11所述的方法，其特征在于，所述去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据，包括：
    采用模糊C均值聚类算法确定每个所述聚类的中心；
    确定至少一个所述聚类中的数据与对应的所述聚类的中心的欧氏距离；及
    采用拉依达准则去除至少一个所述聚类中的所述欧氏距离的距离异常值，并去除所述距离异常值对应的所述奇异值。
  13. 如权利要求8所述的方法，其特征在于，所述数据筛选步骤包括：
    采用四分位数法确定所述待处理数据集中的异常数据；
    去除所述异常数据和包括该异常数据的数据组的其他数据。
  14. 如权利要求11所述的方法，其特征在于，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类之前，还包括：
    对所述待处理数据集中的数据进行标准化处理，得到标准化数据；
    所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：
    采用聚类算法将所述标准化数据划分为多个聚类。
  15. 如权利要求14所述的方法，其特征在于，所述去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据之后，还包括：
    对去除每个所述聚类中的奇异值及包括该奇异值的数据组的其他数据后的所述标准化数据进行逆标准化处理。
  16. 如权利要求10所述的方法，其特征在于，所述采用聚类算法将所述待处理数据集中的数据划分为多个聚类，包括：
    采用聚类算法将所述待处理数据集中的每种数据分别划分为多个聚类。
  17. 如权利要求8所述的方法，其特征在于，对所述待处理数据集进行数据筛选处理之后，还包括：
    若所述筛选数据集的数据的时间序列不连续，在时间序列非连续的数据之间，回填所缺失时刻的数据。
  18. 如权利要求17所述的方法，其特征在于，所述在时间序列非连续的数据之间，回填所缺失时刻的数据，包括：
    若所述时间序列非连续的数据之前的连续多个数据的数量大于2，确定所述时间序列非连续的数据之前的连续多个数据的残差均值及方差，并基于所述残差均值和方差生成随机数，作为所缺失时刻的数据。
  19. 如权利要求17所述的方法，其特征在于，所述在时间序列非连续的数据之间，回填所缺失时刻的数据，包括：
    若所述时间序列非连续的数据之前的连续多个数据的数量不大于2，选取时间序列非连续的数据之前的时刻所对应的数据，作为所缺失时刻的数据。
  20. 一种应用于风力发电机组的监控系统，包括一个或多个处理器，用于实现如权利要求1-19中任一项所述的监控方法。
  21. 一种计算机可读存储介质，其上存储有程序，该程序被处理器执行时，实现如权利要求1-19中任一项所述的监控方法。
PCT/CN2021/093632 2020-07-15 2021-05-13 风力发电机组的监控方法、系统及计算机可读存储介质 WO2022012137A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
KR1020237005315A KR20230038281A (ko) 2020-07-15 2021-05-13 풍력 터빈 발전기 시스템의 모니터링 방법, 시스템 및 컴퓨터 판독 가능 저장매체
EP21841719.4A EP4184007A4 (en) 2020-07-15 2021-05-13 METHOD AND SYSTEM FOR MONITORING A WIND TURBINE GENERATOR SET AND COMPUTER READABLE STORAGE MEDIUM

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010682155.3A CN111878320B (zh) 2020-07-15 2020-07-15 风力发电机组的监控方法、系统及计算机可读存储介质
CN202010682155.3 2020-07-15

Publications (1)

Publication Number Publication Date
WO2022012137A1 true WO2022012137A1 (zh) 2022-01-20

Family

ID=73155473

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/093632 WO2022012137A1 (zh) 2020-07-15 2021-05-13 风力发电机组的监控方法、系统及计算机可读存储介质

Country Status (4)

Country Link
EP (1) EP4184007A4 (zh)
KR (1) KR20230038281A (zh)
CN (1) CN111878320B (zh)
WO (1) WO2022012137A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115374653A (zh) * 2022-10-21 2022-11-22 宇动源(北京)信息技术有限公司 基于nset模型的风力发电机温度预警方法及相关装置
CN115905980A (zh) * 2022-12-09 2023-04-04 马鞍山玖帆科技有限公司 用于风电能源的设备监测系统及其监测方法

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111881617B (zh) * 2020-07-02 2024-03-26 上海电气风电集团股份有限公司 数据处理方法、风力发电机组的性能评估方法及系统
CN111878320B (zh) * 2020-07-15 2021-07-30 上海电气风电集团股份有限公司 风力发电机组的监控方法、系统及计算机可读存储介质
CN114238033B (zh) * 2021-12-28 2022-09-13 湖南云箭智能科技有限公司 板卡运行状态预警方法、装置、设备及可读存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100138267A1 (en) * 2009-08-31 2010-06-03 Sameer Vittal System and method for wind turbine health management
CN102808743A (zh) * 2012-08-14 2012-12-05 国电联合动力技术有限公司 一种风电机组机舱温度控制方法及系统
CN105804951A (zh) * 2014-12-29 2016-07-27 浙江运达风电股份有限公司 一种温度监测系统及风力机系统
DE102016215614A1 (de) * 2016-08-19 2018-02-22 Siemens Aktiengesellschaft Verfahren zur Bestimmung einer Temperatur und Leistungsmodul
CN109386435A (zh) * 2017-08-04 2019-02-26 阿里巴巴集团控股有限公司 风电机故障监控方法、装置和系统
CN110414155A (zh) * 2019-07-31 2019-11-05 北京天泽智云科技有限公司 一种带有单测点的风机部件温度异常检测和报警方法
CN111878320A (zh) * 2020-07-15 2020-11-03 上海电气风电集团股份有限公司 风力发电机组的监控方法、系统及计算机可读存储介质

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB0515034D0 (en) * 2005-07-21 2005-08-31 Rolls Royce Plc Method and system for operating a multi-stage combustor
WO2014114295A1 (en) * 2013-01-25 2014-07-31 Vestas Wind Systems A/S Control of wind turbines
NO3070262T3 (zh) * 2014-04-02 2018-06-02
JP6356500B2 (ja) * 2014-06-19 2018-07-11 株式会社日立製作所 風力発電装置
CN205154510U (zh) * 2015-12-01 2016-04-13 沈阳华创风能有限公司 一种风机机舱温度采集系统
CN109477464B (zh) * 2016-05-03 2021-10-29 维斯塔斯风力系统集团公司 机械设备，尤其是风力涡轮机的状态监控
CN110874665B (zh) * 2018-08-31 2022-07-08 北京金风科创风电设备有限公司 用于风力发电机组的控制装置和方法
CN110486238B (zh) * 2019-09-05 2020-10-16 国电联合动力技术有限公司 一种低风速风力发电机组智能散热系统及其控制方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100138267A1 (en) * 2009-08-31 2010-06-03 Sameer Vittal System and method for wind turbine health management
CN102808743A (zh) * 2012-08-14 2012-12-05 国电联合动力技术有限公司 一种风电机组机舱温度控制方法及系统
CN105804951A (zh) * 2014-12-29 2016-07-27 浙江运达风电股份有限公司 一种温度监测系统及风力机系统
DE102016215614A1 (de) * 2016-08-19 2018-02-22 Siemens Aktiengesellschaft Verfahren zur Bestimmung einer Temperatur und Leistungsmodul
CN109386435A (zh) * 2017-08-04 2019-02-26 阿里巴巴集团控股有限公司 风电机故障监控方法、装置和系统
CN110414155A (zh) * 2019-07-31 2019-11-05 北京天泽智云科技有限公司 一种带有单测点的风机部件温度异常检测和报警方法
CN111878320A (zh) * 2020-07-15 2020-11-03 上海电气风电集团股份有限公司 风力发电机组的监控方法、系统及计算机可读存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4184007A4 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115374653A (zh) * 2022-10-21 2022-11-22 宇动源(北京)信息技术有限公司 基于nset模型的风力发电机温度预警方法及相关装置
CN115374653B (zh) * 2022-10-21 2022-12-20 宇动源(北京)信息技术有限公司 基于nset模型的风力发电机温度预警方法及相关装置
CN115905980A (zh) * 2022-12-09 2023-04-04 马鞍山玖帆科技有限公司 用于风电能源的设备监测系统及其监测方法

Also Published As

Publication number Publication date
KR20230038281A (ko) 2023-03-17
CN111878320A (zh) 2020-11-03
EP4184007A1 (en) 2023-05-24
CN111878320B (zh) 2021-07-30
EP4184007A4 (en) 2023-12-20

Similar Documents

Publication Publication Date Title
WO2022012137A1 (zh) 风力发电机组的监控方法、系统及计算机可读存储介质
Zhang et al. An improved association rule mining-based method for revealing operational problems of building heating, ventilation and air conditioning (HVAC) systems
WO2022001379A1 (zh) 数据处理方法、风力发电机组的性能评估方法及系统
WO2020140560A1 (zh) 一种生产物流输送装备故障预警方法
CN107701378B (zh) 一种风力发电机故障预警方法
WO2023168950A1 (zh) 一种智能抄表终端的数据采集方法及系统
CN108072524B (zh) 一种风电机组齿轮箱轴承故障预警方法
CN110362045B (zh) 一种考虑海洋气象因素的海上双馈风电机组故障判别方法
Peng et al. Wind turbine failure prediction and health assessment based on adaptive maximum mean discrepancy
CN116976707B (zh) 基于用电信息采集的用户用电数据异常分析方法及系统
CN117234301A (zh) 一种基于人工智能的服务器热管理方法
CN115617784A (zh) 一种信息化配电的数据处理系统及其处理方法
CN110781553A (zh) 一种高速列车多参数智能阈值监测方法
CN110506245A (zh) 诊断装置以及诊断方法
CN117674119A (zh) 电网运行风险评估方法、装置、计算机设备和存储介质
CN116151799A (zh) 一种基于bp神经网络的配电线路多工况故障率快速评估方法
CN114662589A (zh) 一种电表故障研判方法、装置、设备和可读存储介质
CN114429240A (zh) 高速公路设备运行状态监测方法及装置
Ding et al. Artificial intelligence based abnormal detection system and method for wind power equipment
Zhou et al. Abnormal data processing of wind turbine based on combined algorithm and class center imputation
Wang et al. Anomaly monitoring in high-density data centers based on gaussian distribution anomaly detection algorithm
Zhang et al. An unsupervised anomaly detection approach based on industrial big data
Liu et al. Wind turbine spindle operating state recognition and early warning driven by SCADA data
Li et al. Multi-source heterogeneous log fusion technology of power information system based on big data and imprecise reasoning theory
CN115711206B (zh) 一种基于聚类权重的风力发电机叶片覆冰状态监测系统

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: 21841719

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20237005315

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2021841719

Country of ref document: EP

Effective date: 20230215