CN115238824A - Method for aligning monitoring data of wind generating set - Google Patents

Method for aligning monitoring data of wind generating set Download PDF

Info

Publication number
CN115238824A
CN115238824A CN202211052083.XA CN202211052083A CN115238824A CN 115238824 A CN115238824 A CN 115238824A CN 202211052083 A CN202211052083 A CN 202211052083A CN 115238824 A CN115238824 A CN 115238824A
Authority
CN
China
Prior art keywords
data
monitoring
alignment
aligned
generating
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202211052083.XA
Other languages
Chinese (zh)
Inventor
汪臻
刘艳贵
邓巍
程辰晨
赵勇
张恩享
沈伟文
费宇涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Thermal Power Research Institute Co Ltd
Clean Energy Branch of Huaneng Zhejiang Energy Development Co Ltd
Original Assignee
Xian Thermal Power Research Institute Co Ltd
Clean Energy Branch of Huaneng Zhejiang Energy Development Co Ltd
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 Xian Thermal Power Research Institute Co Ltd, Clean Energy Branch of Huaneng Zhejiang Energy Development Co Ltd filed Critical Xian Thermal Power Research Institute Co Ltd
Priority to CN202211052083.XA priority Critical patent/CN115238824A/en
Publication of CN115238824A publication Critical patent/CN115238824A/en
Pending legal-status Critical Current

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
    • F03D7/00—Controlling wind motors 
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Operations Research (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Algebra (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Sustainable Development (AREA)
  • Sustainable Energy (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

In the method, the device and the storage medium for aligning the monitoring data of the wind generating set, a monitoring data set of at least one operating state of the wind generating set is constructed, a target operating state to be analyzed and a monitoring variable to be aligned in the target operating state are determined, whether the data of the monitoring variable to be aligned in the monitoring data set meet an alignment condition or not is determined, and the alignment data of the monitoring variable to be aligned is generated in response to the fact that the data in the monitoring data set meet the alignment condition. Therefore, the method for aligning the monitoring data of the wind generating set generates new data capable of reflecting the running state mode of the wind generating set, so that the sample size of the state monitoring data is enlarged, the alignment of the monitoring data is realized, and a data base is provided for the follow-up accurate management and control of the state of the wind generating set.

Description

Method for aligning monitoring data of wind generating set
Technical Field
The application relates to the technical field of wind generating set data analysis, in particular to a method and a device for aligning monitoring data of a wind generating set and a storage medium.
Background
The abnormal state mode of the wind generating set is complex in forming reason, and the abnormal state is a result of the combined action of a plurality of coupling factors in the system and a plurality of influence factors such as the external environment of the system operation. Meanwhile, due to the fact that the probability of the abnormal operation state of each type of subsystem of the wind generating set is different, the probability of the abnormal state modes of different types of the wind generating set is different, and therefore the data monitoring data amount corresponding to different moments in the abnormal state modes is different.
Meanwhile, the wind generating set is a complex electromechanical-hydraulic coupling system, the subsystems exchange substances, information and energy continuously, and the subsystems in different state modes are combined with one another, so that the wind generating set has a plurality of abnormal state modes. Based on this, some rare wind turbine generator set operation state modes may occur, and less monitoring data exists in the operation state modes. Therefore, a method for aligning monitoring data of a wind generating set is needed, so that the data volume of less monitoring data and the monitoring data in a normal state mode are aligned, and the characteristics in the operation state mode can be better analyzed based on the aligned data.
Disclosure of Invention
The application provides a method and a device for aligning monitoring data of a wind generating set and a storage medium, which are used for solving the technical problems in the related technology.
An embodiment of a first aspect of the present application provides a method for aligning monitoring data of a wind turbine generator system, including:
constructing a monitoring data set of at least one running state of the wind turbine generator;
determining a target running state to be analyzed and a monitoring variable to be aligned in the target running state;
determining whether the data of the monitoring variables needing to be aligned in the monitoring data set meets an alignment condition;
and generating alignment data of the monitoring variables needing to be aligned in response to the data in the monitoring data set meeting an alignment condition.
An embodiment of a second aspect of the present application provides a wind generating set monitoring data alignment device, including:
the building module is used for building a monitoring data set of at least one operation state of the wind turbine generator;
the device comprises a first determining module, a second determining module and a monitoring module, wherein the first determining module is used for determining a target running state needing to be analyzed and a monitoring variable needing to be aligned in the target running state;
the second determining module is used for determining whether the data of the monitoring variables needing to be aligned in the monitoring data set meets the alignment condition;
and the generating module is used for generating the alignment data of the monitoring variables needing to be aligned in response to the fact that the data in the monitoring data set meet the alignment condition.
A computer storage medium provided in an embodiment of the third aspect of the present application, where the computer storage medium stores computer-executable instructions; the computer executable instructions, when executed by a processor, are capable of performing the method of the first aspect as described above.
A computer device according to an embodiment of a fourth aspect of the present application includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method according to the first aspect is implemented.
The technical scheme provided by the embodiment of the application at least has the following beneficial effects:
in the method, the device and the storage medium for aligning the monitoring data of the wind generating set, a monitoring data set of at least one operating state of the wind generating set is constructed, a target operating state to be analyzed and a monitoring variable to be aligned in the target operating state are determined, whether the data of the monitoring variable to be aligned in the monitoring data set meet an alignment condition or not is determined, and the alignment data of the monitoring variable to be aligned is generated in response to the fact that the data in the monitoring data set meet the alignment condition. Therefore, when the data in the monitoring data set meets the alignment condition, the alignment data of the monitoring variables needing to be aligned is generated, and the effect of aligning the data is improved. Meanwhile, the method for aligning the monitoring data of the wind generating set generates new data capable of reflecting the running state mode of the wind generating set, so that the sample size of the state monitoring data is enlarged, the alignment of the monitoring data is realized, and a data basis is provided for the follow-up accurate management and control of the state of the wind generating set.
Additional aspects and advantages of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.
Drawings
The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 is a schematic flow chart diagram of a method for aligning monitoring data of a wind turbine generator system according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of a wind generating set monitoring data alignment device according to an embodiment of the present application.
Detailed Description
Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are exemplary and intended to be used for explaining the present application and should not be construed as limiting the present application.
The following describes a wind generating set monitoring data alignment method and device according to an embodiment of the present application with reference to the accompanying drawings.
Example one
Fig. 1 is a schematic flowchart of a method for aligning monitoring data of a wind turbine generator system according to an embodiment of the present application, and as shown in fig. 1, the method may include:
step 101, a monitoring data set of at least one operation state of the wind turbine generator is constructed.
In the embodiment of the disclosure, the monitoring data of the wind generating set can be collected through the software and hardware deployment condition of the wind generating set, and a monitoring data set capable of reflecting the running state of the wind generating set is constructed.
Specifically, in the embodiment of the present disclosure, the monitoring data of the wind turbine generator system may be collected through at least one of a data acquisition and monitoring System (SCADA), a status monitoring system (CMS), and a structural health monitoring System (SHM) in the wind turbine generator system.
Wherein, in the embodiment of the present disclosure, the data collecting and monitoring system: monitoring the wind generating set with low resolution and providing data and alarm channels for the set; the state monitoring system comprises: monitoring key subsystems (important parts) of the wind generating set with high resolution, and performing fault diagnosis or pre-diagnosis; the above-mentioned structure health monitoring system: wind turbine generator set critical structure (tower, etc.) components are monitored with lower resolution.
And in the embodiment of the disclosure, the different types of monitoring data are sorted into different operating states according to the acquisition time, the acquired wind generating set number, the current operating state of the wind generating set and the like, so as to construct a wind generating set state monitoring data set.
Step 102, determining a target operation state to be analyzed and a monitoring variable to be aligned in the target operation state.
In the embodiment of the present disclosure, the operation state in which the number of occurrences is smaller than the first threshold may be determined as the target operation state that needs to be analyzed. In the embodiment of the present disclosure, the monitor variable in the target state, in which the data amount of the monitor variable is less than the second threshold, may be determined as the monitor variable in the target state, which needs to be aligned.
And, in the embodiment of the present disclosure, the target operation state to be analyzed and the monitoring variable to be aligned in the target operation state may also be set manually.
Further, in the embodiment of the present disclosure, after the aligned monitored variable is determined, other variables having a coupling relationship with the monitored variable may be obtained through analysis, and the condition of the target operating state may be fully characterized by these variables.
Specifically, in the embodiment of the present disclosure, the following physical mechanisms may be used according to the operation of the wind turbine generator set: the wind generating set is a complex electromechanical-hydraulic coupling system, continuous material flow, energy flow and information flow exist in each key subsystem (important part), and the relationship among a plurality of monitoring data corresponding to the key subsystems is regarded as a coupling relationship; and determining K monitoring variables with the strongest coupling relation in the final target operation state and the determined monitoring variables needing to be aligned together by using an incidence relation mining algorithm among the monitoring variables.
In the embodiment of the present disclosure, the association mining algorithm may include a clustering algorithm (e.g., a density peak-based clustering algorithm, or a DBSCAN clustering algorithm, or a K-means clustering algorithm), an association rule mining algorithm (e.g., an Apriori algorithm, or an FP-Growth algorithm), and a similarity measurement method (e.g., a manhattan distance, or a euclidean distance, or a cosine similarity).
And 103, determining whether the data of the monitoring variables needing to be aligned in the monitoring data set meets an alignment condition.
In an embodiment of the present disclosure, the method for determining whether data of a monitoring variable to be aligned in a monitoring dataset satisfies an alignment condition may include the following steps:
step a, determining whether the data of the monitoring variables needing to be aligned in the monitoring data set has alignment characteristics.
Wherein, in the disclosed embodiment, the above-mentioned alignment feature may include at least one of:
chaos;
non-stationarity;
long-range correlation.
In an embodiment of the present disclosure, the method for determining whether chaos exists in data of monitoring variables needing to be aligned in a monitoring dataset may include: and judging the chaos of the data of the monitoring variables needing to be aligned in the monitoring data set based on the maximum Lyapunov index.
Wherein, in the disclosed embodiment, the maximum Lyapunov exponent lambda is based on Wolf method max The method of computing may comprise the steps of:
1) Selecting a proper method to calculate the length N as a time sequence { x n The delay time τ and embedding dimension m of the reconstruction, the reconstructed phase space is:
X＝[x n ,x n+τ ,...,x n+(m-1)τ ]∈R M ,n＝1,2,...,N-(m-1)τ
2) Setting an initial point x 1 Is a reference point, and finds a phase dividing point x in the reconstructed phase space 1 Phase point with outer distance nearest to reference point
Figure BDA0003824057760000053
As end points, the distance between two points
Figure BDA0003824057760000052
3) Distance between reference point and end pointThrough T 1 After an evolution step length by L 1 Becomes L' 1 The calculation formula for obtaining the exponential growth rate is as follows:
Figure BDA0003824057760000051
4) Division in reconstructed phase space
Figure BDA0003824057760000061
Selecting one of the out-of-phase points and
Figure BDA0003824057760000062
nearest phase point
Figure BDA0003824057760000063
As a new endpoint and the endpoint is to make a vector
Figure BDA0003824057760000064
And
Figure BDA0003824057760000065
angle theta therebetween 1 At the minimum, the temperature of the mixture is controlled,
Figure BDA0003824057760000066
the distance from the new end point is recorded as
Figure BDA0003824057760000067
5) Selecting a new evolution step size as T 2 Datum point
Figure BDA0003824057760000068
And end point
Figure BDA0003824057760000069
Respectively evolve into
Figure BDA00038240577600000610
And
Figure BDA00038240577600000611
distance between two points is L' 2 At this time, the calculation formula of the exponential growth rate is:
Figure BDA00038240577600000612
6) Continuing to select new evolution step length until the last phase point in the phase space, recording the total step number of evolution as M, and taking the average value of exponential growth rate as lambda max Estimate of (a) ("lambda max The calculation formula is as follows:
Figure BDA00038240577600000613
in an embodiment of the present disclosure, the method for determining whether there is non-stationarity in the data of the monitoring variables needing to be aligned in the monitoring dataset may include: the non-stationarity of the data in the monitored dataset for the monitored variables that need to be aligned is determined based on an ADF (extended Diety-Fuller test) test.
In an embodiment of the present disclosure, a method for determining non-stationarity of data in a monitored data set of monitored variables to be aligned based on ADF inspection may include the steps of:
in the embodiment of the present disclosure, if all elements in a time series X are randomly sampled from the same probability distribution and the following three conditions are simultaneously satisfied, the time series X is considered to be stable according to the statistical analysis theory. And the time series is the monitoring data of the same monitoring variable at different moments.
(1) The mean of the time series X is a time-independent constant;
(2) the variance of the time series X is a time-independent constant;
(3) the autocovariance of the time series X is a time-independent constant that is only time-interval dependent.
The stationarity of the time series is monitored by the ADF decision system. Three models of ADF inspection are represented as follows:
model 1:
Figure BDA00038240577600000614
model 2:
Figure BDA0003824057760000071
model 3:
Figure BDA0003824057760000072
alternative testing H of hypothesis testing 1 :H<0, no unit root is present; null hypothesis H 0 H =0, there is a unit root. Whether the time series has stationarity can be judged by the following two conditions:
i) if the hypothesis test result of at least one model rejects the null hypothesis, the model has stationarity;
II) if the hypothesis test results of all models can not reject the null hypothesis, the model has non-stationarity.
And, in an embodiment of the present disclosure, the method for determining whether there is a long-range correlation between data of monitoring variables needing to be aligned in a monitoring dataset may include: the long-range correlation of the data of the monitored variables to be aligned in the monitored dataset is determined based on DFA (Detrended Fluctuation Analysis).
In the embodiment of the present disclosure, the method for determining the long-range correlation of the data of the monitoring variables needing to be aligned in the monitoring dataset based on the DFA may include the following steps:
step one, the original time sequence { x n Calculating to obtain a new time sequence y n }, wherein:
Figure BDA0003824057760000073
step two, when newThe m-sequence y (n) is decomposed into s by length
Figure BDA0003824057760000074
If the length N of y (N) is not an integral multiple of s, a fragment with the length less than s is generated after decomposition, and in order to not lose the part of information, the fragment is decomposed again from the other end of y (N), and finally 2N is obtained s A fragment;
step three, obtaining y (n) and least square fitting p thereof v The difference between them is as follows:
Y(i)＝y(i)-p v (i)
step four, obtaining the variance of the v-th fragment as follows:
Figure BDA0003824057760000075
step five, calculating the mean value of all the segment variances to obtain a calculation formula of the detrending fluctuation function as follows:
Figure BDA0003824057760000076
step six, if the original time sequence { x n With long-range correlation, its detrended fluctuation function will grow in power-law form, i.e.:
F(s)∝s α
step seven, optimizing the algorithm flow to obtain a correction function and an optimized detrending fluctuation function which are respectively as follows:
Figure BDA0003824057760000081
Figure BDA0003824057760000082
wherein, in the above formula: s' is each of the time sequences obtained by disordering and then dividingPartial Length, typically Length s' ≈ Length (x) n )/20。
And b, if the data of the monitoring variables needing to be aligned in the monitoring data set has alignment characteristics, determining that the data of the monitoring variables needing to be aligned in the monitoring data set meets the alignment condition.
In the embodiment of the present disclosure, if the data of the monitoring variable to be aligned in the monitoring data set has the alignment feature, which indicates that the data of the monitoring variable to be aligned in the monitoring data set has a good alignment effect, it is determined that the data of the monitoring data set of the monitoring variable to be aligned satisfies the alignment condition.
And step 104, responding to the data in the monitoring data set meeting the alignment condition, and generating alignment data of the monitoring variables needing to be aligned.
In an embodiment of the present disclosure, a method for generating alignment data of monitoring variables that need to be aligned may include the following steps:
and 1041, determining the width of a time window and the number of sampling points based on the success rate of reconstruction.
In an embodiment of the present disclosure, the reconstruction success rate is: randomly selecting a plurality of data in a period of time sequence for reconstructing a system equation, if the reconstructed system equation meets the precondition (if all the reconstructed time sequence data are in the specified upper and lower limits), determining that the reconstruction is successful, and acquiring a plurality of random data, wherein the percentage of the times of successful reconstruction to the total times is the reconstruction success rate. The reconstruction success rate reflects the matching degree of the monitoring variable to the dynamic relation model, and can also be used for measuring the reasonable degree of the current wind generating set in time and space segmentation.
And, in an embodiment of the present disclosure, a method of determining a width of a time window and a number of sampling points may include: selecting the width D range of the time window as [1 h/sampling interval time, 24 h/sampling interval time ], selecting the range of the number M of sampling points as [2,4h/sampling interval time ], calculating the reconstruction success rate under the combination of different widths D and the number M of sampling points, and selecting the values of D and M in the combination with the maximum success rate.
For example, in the embodiment of the present disclosure, the fan generator set monitoring data of a point is recorded once in 10 minutes, the range of the width D of the time window is selected as [6, 144], and the range of the number M of sampling points is selected as [2, 24]. And, the width D of the time window and the number M of the sampling points are both taken as integers, the number of the combination of the success rate required to be calculated is kept as 3197, and the calculation formula is as follows:
(144-6+1)×(24-2+1)＝3197。
1042, carrying out multivariate dynamic relation modeling on the target running state by a method based on a compressed sensing theory according to the number of sampling points and the width of a time window to obtain a target system equation of the target running state.
In an embodiment of the present disclosure, the method for performing multivariate dynamic relationship modeling on the target operating state based on a compressed sensing theory may include:
for a signal x ∈ R N The compressed sensing problem can be described as knowing a certain measurement matrix Φ ∈ R M×N (M<<N) and the linear measurement y ∈ R of the signal x under the matrix M On the basis of (1), solving an equation system:
y＝Φx
to obtain the original signal x. Since the dimension of y is much lower than that of x, the above formula is an underdetermined system of equations with infinite solutions, and the original signal x cannot be reconstructed. However, if the original signal x is sparse and y and Φ in the equation set satisfy a certain condition, it can be proved that the reconstruction of the original signal x can be solved by solving the problem of the minimized L0 norm:
Figure BDA0003824057760000091
in the above formula:
Figure BDA0003824057760000092
is the reconstructed value of the original signal x; | x | charging 0 Is the L0 norm of vector x, i.e., the number of non-zero elements in vector x.
And, in the embodiments of the present disclosure, the compressed sensing theory indicates that to achieve accurate reconstruction of the K sparse signal x, the measurement order M (i.e., the dimension of y) and the matrix Φ must satisfy the conditions of M = O [ K · lg (N) ] and RIP (Restricted Isometry) respectively. However, since most natural signals in the time domain do not have sparsity, based on which the reconstruction of natural signals cannot follow the above process, the sparse representation of natural signal x can be realized by some transformation Ψ according to the signal sparse representation theory:
x＝Ψα
wherein, in the above formula: alpha is a sparse representation of the natural signal x in the psi transform domain.
And according to the measurement formula y = Φ x, there are:
y＝Φx＝ΦΨα＝Aα
wherein, in the above formula: a is the perceptual matrix, a = Φ Ψ; y is a measure of the sparse signal a with respect to the perceptual matrix a.
And, in the embodiment of the present disclosure, if the sensing matrix a satisfies the RIP condition, the sparse signal α may be reconstructed by solving the minimum L0 norm problem as follows:
Figure BDA0003824057760000101
wherein, the above
Figure BDA0003824057760000102
The reconstructed value of alpha is sparsely represented for the original signal x.
Therein, in an embodiment of the present disclosure, a mathematical model of a nonlinear system equation of the form is constructed in which it is assumed that the system contains only three monitored variables s 1 、s 2 、s 3 And setting the lowest power of the monitoring variable as 0 and the highest power as 2, wherein the expression of the system equation is as follows:
Figure BDA0003824057760000103
in the above formula: s is 1 、s 2 、s 3 Representing signals, s, generated at monitoring points in a wind power plant system 1 * 、s 2 * 、s 3 * And (4) representing a calculated monitoring data value, s 'obtained by calculating a system equation' 1 、s′ 2 、s′ 3 Are respectively the monitored variable s 1 、s 2 、s 3 The varying intensity value (differential) of the signal.
Specifically, in the embodiment of the present disclosure, the multivariate dynamic relationship modeling is performed on the target operating state by using the above nonlinear system equation and a method based on the compressive sensing theory as follows:
according to the characteristics of the wind generating set, the multivariate dynamic relation model is set to be in the following form:
s * ＝F(s)
in the formula, s is a monitoring variable in a group of complex electromechanical systems, and s belongs to R M (ii) a F is a function reflecting the dynamic relation among the monitoring variables; s * And monitoring data values of each monitoring point position calculated by the model.
In the embodiment of the disclosure, the lowest power of each variable is set to be 0, the highest power is set to be n, and the kth component s of s is set to be n k The equation expression of (k =1,2,.. D., m) is expanded into the following form:
Figure BDA0003824057760000111
in the above formula, a is the coefficient vector of the equation, and has sparseness because of F(s) k ) Does not contain s in the expansion k Based on this, the coefficient vector a only contains (1+n) m-1 + m-1 elements.
And, in the disclosed embodiments, the monitored variable s is not paired at any time t k The measurement is carried out, a plurality of monitoring data are randomly extracted in a certain time interval and used for reconstructing a system equation so as to reduce the influence of noise in the monitoring data of the wind generating set, and further, the influence of part of the monitoring data influenced by the noise is effectively reducedAnd (4) influence of data with larger degree on a reconstruction system equation. From the time interval t i ,t j ]In which w monitoring variables s are randomly extracted each time k The measured value is obtained by using the monitoring data as the measured value, and the measured vector is as follows:
Figure BDA0003824057760000112
in the embodiment of the disclosure, unlike a mathematical model for reconstructing a nonlinear system equation, in the multivariate dynamic relationship modeling of the wind turbine generator system, the reconstructed system equation may not necessarily reflect the dynamic relationship between the monitoring variables due to the influence of noise on the monitoring data, and needs to be checked to prove the capability of reflecting the dynamic relationship between the monitoring variables. And, in the disclosed embodiment, the verification method is for the time interval t e [ t ∈ [ [ t ] i ,t j ](i, j ∈ N, and i<j) All monitoring data s in k (t), all of:
Figure BDA0003824057760000113
wherein, in the above formula: sigma is the monitoring data s k (t) reconstructing a threshold value of the error, when σ is sufficiently small, it can be proved that the system equation satisfying the above formula can reflect the dynamic relationship between the monitored variables.
And 1043, constructing a training data set of the countermeasure network based on the target system equation.
In the embodiment of the present disclosure, the method for constructing the training data set of the countermeasure network based on the target system equation may include the following steps:
step a, obtaining analytic data of a monitoring variable required to correspond at the next moment according to the width of a time window and a target system equation;
b, if the analytic data and the real data of the monitoring variables to be aligned at the next moment are larger than a preset deviation threshold, storing the analytic data at the next moment into a training data set of the countermeasure network; otherwise, the process is repeated when the time window is slid to the next time.
In the embodiment of the present disclosure, the preset deviation threshold may be set as needed, and different preset deviation thresholds correspond to different accuracy requirements. And, can be according to different precision requirements, confirm and adjust the STEP length of slipping, set STEP to 1 in the disclosed embodiment, namely interval 1 moment.
And step 1044, carrying out countermeasure training on the Im-TimeGAN model based on the training data set to obtain the trained Im-TimeGAN model.
In this embodiment of the present disclosure, the Im-TimeGAN model may include: embedding network, restoring network, generating network and discriminator.
And 1045, generating alignment data of the monitoring variables needing to be aligned by using the trained Im-TimeGAN model.
In this embodiment of the present disclosure, the method for generating alignment data of monitoring variables that need to be aligned by using a trained Im-TimeGAN model may include:
(1) Embedded network through e s :S→H s ，e x :H s ×H x ×X→H x The data of the original feature space is mapped to a higher-dimensional vector space.
h s ＝e s (s)
h t ＝e x (h s ,h t-1 ,x t )
Wherein e is s Embedded networks being static features, e x Is an embedded network of dynamic features, S, X represents the vector space of static and dynamic features, respectively, H s 、H x Respectively, which represent the potential vector space to which S, X corresponds.
(2) Restoring network transit s :H s →S，r x :H x → X, the resulting higher dimensional vector space is restored to the original static and dynamic features.
Figure BDA0003824057760000121
Figure BDA0003824057760000122
Wherein r is s Is a recovery network of static character, r x Is a recovery network of dynamic nature. Where the embedded network and the restoration network can be parameterized by any chosen method, the only provision is that they are autoregressive and follow a causal order.
(3) Generating a network pass g s :Z S →ss，g x :H S ×H X ×Z X →xx t The synthesized output is generated directly in the feature space.
Figure BDA0003824057760000131
Figure BDA0003824057760000132
Wherein, g s Is a generating network of static characteristics, g x The method is a dynamic feature generation network, wherein Im-timeGAN directly generates synthesis output in a feature space, data for generating the feature space is embedded into a potential vector space through an embedding function when judging, and z is s And z x Is a known distribution subspace.
(4) Discriminator pass d S 、d X And judging by using the classification function of the output layer.
Figure BDA0003824057760000133
Figure BDA0003824057760000134
Wherein,
Figure BDA0003824057760000135
and
Figure BDA0003824057760000136
representing the forward and reverse hidden state sequences respectively,
Figure BDA0003824057760000137
and
Figure BDA0003824057760000138
is a recursive function. As an invertible mapping between the feature space and the latent space, the embedding and recovery functions should be able to be derived from s, X 1:T Is used to represent h as a potential high-dimensional embedding space vector S ,h 1:T Accurate restoration to representation of original low-dimensional feature space
Figure BDA0003824057760000139
Thus, the reconstruction loss function is expressed as:
Figure BDA00038240577600001310
and, in the disclosed embodiments, two losses are mainly minimized in the generation network. Wherein, in the pure open-loop mode, the autoregressive generator receives the synthetic embedding
Figure BDA00038240577600001311
(i.e., its own previous output) to generate the next resultant vector
Figure BDA00038240577600001312
Figure BDA00038240577600001313
And h t The difference between them; maximizing the rate of accuracy of the discriminator (discrimination)
Figure BDA00038240577600001314
Is synthesized, h S ,h 1:T Is original) And minimizing the recognized probability of the generator: (
Figure BDA00038240577600001315
Probability of being identified as a composite), the loss function is expressed as:
Figure BDA00038240577600001316
further, in embodiments of the present disclosure, the generated data distribution is reduced by closed loop and alternating training
Figure BDA00038240577600001317
Distribution with original data (p (h) t |h S ,h 1:t-1 ) To account for the fact that binary antagonistic feedback by the discriminators may not be sufficient to motivate the generator to capture a gradual conditional distribution in the data. Wherein h is t-1 Is the actual data x t-1 Calculated by embedding in the network. The loss function is expressed as:
Figure BDA0003824057760000141
in the embodiment of the present disclosure, if the trained Im-TimeGAN model generates data corresponding to a new variable other than the monitoring variable that needs to be aligned, the process of generating the alignment data in steps 1042 to 1045 is repeated using the data of the new variable to obtain the alignment data of the new variable until the trained Im-TimeGAN model does not generate data of the new variable any more, and the generation of the alignment data is finished.
And, in the embodiment of the present disclosure, the method for determining whether there is a new variable may include: and judging whether the relative error between the actual value of the monitoring data at the next moment and the analytic value of the monitoring data obtained through the target system equation at the previous moment is smaller than a third threshold, if the relative error is greater than the third threshold for Y times continuously, indicating that a new variable appears, storing the new variable, and carrying out an alignment process on the new variable.
It should be noted that, in the embodiment of the present disclosure, the trained Im-TimeGAN model first embeds monitoring data points that satisfy the system model into a high-dimensional vector space, then extracts data points from a known distribution, generates the data points into a feature space through a generation network, maps the feature space into an embedding space, and determines the type of the two types of data in the embedding space by using a discriminator. By generating the countermeasure idea, new data capable of reflecting the target operation state mode of the wind generating set is generated, so that the sample size of the operation state monitoring data is enlarged, and the problem of alignment of typical monitoring data of the wind generating set under multiple influence factors is solved.
Further, in an embodiment of the present disclosure, the method further includes: and in response to the data in the monitoring data set not meeting the alignment condition, processing the data in the monitoring data set, and not generating alignment data of the monitoring variables needing to be aligned.
In summary, in the method for aligning monitoring data of a wind turbine generator set provided by the present application, a monitoring data set of at least one operating state of the wind turbine generator set is constructed, a target operating state to be analyzed and a monitoring variable to be aligned in the target operating state are determined, whether data of the monitoring variable to be aligned in the monitoring data set meets an alignment condition is determined, and in response to that the data in the monitoring data set meets the alignment condition, alignment data of the monitoring variable to be aligned is generated. Therefore, when the data in the monitoring data set meets the alignment condition, the alignment data of the monitoring variables needing to be aligned is generated, and the effect of aligning the data is improved. Meanwhile, the method for aligning the monitoring data of the wind generating set generates new data capable of reflecting the running state mode of the wind generating set, so that the sample size of the state monitoring data is enlarged, the alignment of the monitoring data is realized, and a data basis is provided for the follow-up accurate management and control of the state of the wind generating set.
Example two
Fig. 2 is a schematic structural diagram of a wind generating set monitoring data alignment apparatus according to an embodiment of the present application, and as shown in fig. 2, the apparatus may include:
the building module 201 is used for building a monitoring data set of at least one operation state of the wind turbine generator;
the first determining module 202 is configured to determine a target operating state to be analyzed and a monitoring variable to be aligned in the target operating state;
a second determining module 203, configured to determine whether data of the monitoring variables that need to be aligned in the monitoring data set satisfies an alignment condition;
the generating module 204 is configured to generate alignment data of the monitoring variables that need to be aligned in response to that the data in the monitoring dataset satisfies the alignment condition.
In summary, in the monitoring data alignment device for the wind turbine generator system provided by the application, a monitoring data set of at least one operating state of the wind turbine generator system is constructed, a target operating state to be analyzed and a monitoring variable to be aligned in the target operating state are determined, whether data of the monitoring variable to be aligned in the monitoring data set meet an alignment condition is determined, and alignment data of the monitoring variable to be aligned are generated in response to the fact that the data in the monitoring data set meet the alignment condition. Therefore, when the data in the monitoring data set meet the alignment condition, the alignment data of the monitoring variables needing to be aligned are generated, and the effect of aligning the data is improved. Meanwhile, the method for aligning the monitoring data of the wind generating set generates new data capable of reflecting the running state mode of the wind generating set, so that the sample size of the state monitoring data is enlarged, the alignment of the monitoring data is realized, and a data base is provided for the follow-up accurate management and control of the state of the wind generating set.
In order to implement the above embodiments, the present disclosure also provides a computer storage medium.
The computer storage medium provided by the embodiment of the disclosure stores an executable program; the executable program, when executed by a processor, enables the method as shown in fig. 1 to be implemented.
In order to implement the above embodiments, the present disclosure also provides a computer device.
The computer equipment provided by the embodiment of the disclosure comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor; the processor, when executing the program, is capable of implementing the method as shown in any of fig. 1.
In the description herein, reference to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present application.
Although embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application, and that variations, modifications, substitutions and alterations may be made to the above embodiments by those of ordinary skill in the art within the scope of the present application.

Claims (10)

1. A method for aligning monitoring data of a wind generating set is characterized by comprising the following steps:
constructing a monitoring data set of at least one running state of the wind turbine generator;
determining a target running state to be analyzed and a monitoring variable to be aligned in the target running state;
determining whether the data of the monitoring variables needing to be aligned in the monitoring data set meets an alignment condition;
and generating alignment data of the monitoring variables needing to be aligned in response to the data in the monitoring data set meeting an alignment condition.
2. The method of claim 1, wherein determining whether the data of the alignment-required monitoring variable in the monitoring dataset meets the alignment condition comprises:
determining whether an alignment feature exists in the data of the monitoring variable needing to be aligned in the monitoring data set, wherein the alignment feature comprises at least one of the following:
chaos;
non-stationarity;
long-range correlation;
and if the data of the monitoring variable needing to be aligned in the monitoring data set has the alignment characteristic, determining that the data of the monitoring variable needing to be aligned in the monitoring data set meets the alignment condition.
3. The method of claim 1, wherein generating alignment data for the alignment-required monitor variables comprises:
determining the width of a time window and the number of sampling points based on the success rate of reconstruction;
performing multivariate dynamic relation modeling on the target running state based on a compressed sensing theory method according to the number of sampling points and the width of the time window to obtain a target system equation of the target running state;
constructing a training data set of a confrontation network based on the target system equation;
performing countermeasure training on the Im-TimeGAN model based on the training data set to obtain a trained Im-TimeGAN model;
and generating alignment data of the monitoring variables needing to be aligned by using the trained Im-TimeGAN model.
4. The method of claim 3, wherein constructing a training data set of an antagonistic network based on the target system equation comprises:
obtaining analytic data of a monitoring variable required to correspond at the next moment according to the width of the time window and the target system equation;
if the analytic data and the real data of the monitoring variables needing to be aligned at the next moment are larger than a preset deviation threshold, storing the analytic data at the next moment to a training data set of the countermeasure network; otherwise, the process is repeated when the time window slides to the next time.
5. The method of claim 3, wherein the Im-TimeGAN model comprises: embedding network, restoring network, generating network and discriminator.
6. The method of claim 5, wherein the generating alignment data for the alignment-required monitor variables using the trained Im-TimeGAN model comprises:
embedded network through e s :S→H s ，e x :H s ×H x ×X→H x Mapping data of the original feature space to a higher-dimensional vector space, where e s Embedded networks being static features, e x Is an embedded network of dynamic features, S, X represents the vector space of static and dynamic features, respectively, H s 、H x Respectively representing potential vector spaces corresponding to S, X;
restoring network transit s :H s →S，r x :H x → X, the resulting higher dimensional vector space is restored to the original static and dynamic features, r s Is a recovery network of static character, r x Is a recovery network of dynamic characteristics;
generating a network pass g s :Z S →ss，g x :H S ×H X ×Z X →xx t Generating the synthetic output, g, directly in the feature space s Is a generating network of static characteristics, g x Is a generating network of dynamic characteristics, z s And z x Is a known distribution subspace;
discriminator pass d S 、d X And judging by using the classification function of the output layer.
7. The method of claim 6, further comprising:
if the trained Im-TimeGAN model generates data corresponding to a new variable except the monitoring variable needing to be aligned, repeating the process of generating the alignment data by using the data of the new variable to obtain the alignment data of the new variable until the trained Im-TimeGAN model does not generate the data of the new variable any more, and finishing generating the alignment data.
8. The method of claim 1, further comprising:
and in response to the data in the monitoring data set not meeting the alignment condition, processing the data in the monitoring data set, and not generating the alignment data of the monitoring variables needing to be aligned.
9. A wind generating set monitoring data alignment device, characterized in that, the device includes:
the building module is used for building a monitoring data set of at least one operation state of the wind turbine generator;
the device comprises a first determining module, a second determining module and a monitoring module, wherein the first determining module is used for determining a target running state needing to be analyzed and a monitoring variable needing to be aligned in the target running state;
the second determining module is used for determining whether the data of the monitoring variables needing to be aligned in the monitoring data set meets the alignment condition;
and the generating module is used for generating the alignment data of the monitoring variables needing to be aligned in response to the fact that the data in the monitoring data set meet the alignment condition.
10. A computer storage medium, wherein the computer storage medium stores computer-executable instructions; the computer-executable instructions, when executed by a processor, are capable of performing the method of any one of claims 1-8.
CN202211052083.XA 2022-08-31 2022-08-31 Method for aligning monitoring data of wind generating set Pending CN115238824A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211052083.XA CN115238824A (en) 2022-08-31 2022-08-31 Method for aligning monitoring data of wind generating set

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211052083.XA CN115238824A (en) 2022-08-31 2022-08-31 Method for aligning monitoring data of wind generating set

Publications (1)

Publication Number Publication Date
CN115238824A true CN115238824A (en) 2022-10-25

Family

ID=83681393

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211052083.XA Pending CN115238824A (en) 2022-08-31 2022-08-31 Method for aligning monitoring data of wind generating set

Country Status (1)

Country Link
CN (1) CN115238824A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116662829A (en) * 2023-07-28 2023-08-29 云南中广核能源服务有限公司 Standard power curve definition rule and deviation verification method for field group fan

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116662829A (en) * 2023-07-28 2023-08-29 云南中广核能源服务有限公司 Standard power curve definition rule and deviation verification method for field group fan
CN116662829B (en) * 2023-07-28 2023-10-17 云南中广核能源服务有限公司 Standard power curve definition rule and deviation verification method for field group fan

Similar Documents

Publication Publication Date Title
CN111091233A (en) Wind power plant short-term wind power prediction modeling method based on wavelet analysis and multi-model AdaBoost depth network
CN114169091B (en) Method for establishing prediction model of residual life of engineering machinery part and prediction method
CN115577248A (en) Fault diagnosis system and method for wind generating set
CN112416662A (en) Multi-time series data anomaly detection method and device
CN115238824A (en) Method for aligning monitoring data of wind generating set
Xu et al. Multiscale BLS-based lightweight prediction model for remaining useful life of aero-engine
Yun et al. Research on gas pressure regulator fault diagnosis based on deep confidence network (DBN) theory
CN118013400A (en) Motor fault diagnosis method, device, electronic equipment and storage medium
Veronig et al. Determination of fractal dimensions of solar radio bursts
CN116720103A (en) Cloud theory and nuclear density-based large-scale circuit fault diagnosis method and system
Maragoudakis et al. Random forests identification of gas turbine faults
Aksan et al. Review of the application of deep learning for fault detection in wind turbine
Gkillas et al. Resource efficient federated learning for deep anomaly detection in industrial IoT applications
CN118696304A (en) Thermal anomaly management
CN118094111B (en) Non-invasive load monitoring and identifying method and system based on machine learning
Kvassay et al. Structure function in analysis of multi-state system availability
Serir et al. Evidential Evolving Gustafson-Kessel Algorithm (E2GK) and its application to PRONOSTIA's data streams partitioning
CN118194028B (en) Low orbit satellite abnormal state identification method based on mixed probability principal component analysis
CN116125922B (en) Complex industrial process monitoring method and system based on parallel dictionary learning
CN118332291B (en) Multi-sensor data fault prediction method for aircraft
CN118761688B (en) Data processing method suitable for training and checking system and electronic equipment
Fang et al. Insulator Image Dataset Generation based on Generative Adversarial Network
Schiller et al. Diagnosis of transient faults in quantised systems
CN118889392A (en) Load prediction method based on time sequence library and time sequence model
Chakrabarti et al. A First-Order Stochastic Prognostic System for the Diagnosis of Helicopter Rotor Systems for the US Navy.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination