CN114370380A - Wind turbine generator fault diagnosis method considering meteorological factors - Google Patents

Wind turbine generator fault diagnosis method considering meteorological factors Download PDF

Info

Publication number
CN114370380A
CN114370380A CN202111495441.XA CN202111495441A CN114370380A CN 114370380 A CN114370380 A CN 114370380A CN 202111495441 A CN202111495441 A CN 202111495441A CN 114370380 A CN114370380 A CN 114370380A
Authority
CN
China
Prior art keywords
wind turbine
turbine generator
fault
sample
fault diagnosis
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
CN202111495441.XA
Other languages
Chinese (zh)
Inventor
程江洲
冯馨以
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.)
China Three Gorges University CTGU
Original Assignee
China Three Gorges University CTGU
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 China Three Gorges University CTGU filed Critical China Three Gorges University CTGU
Priority to CN202111495441.XA priority Critical patent/CN114370380A/en
Publication of CN114370380A publication Critical patent/CN114370380A/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
    • F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G06F18/24155—Bayesian classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Algebra (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Sustainable Development (AREA)
  • Sustainable Energy (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

A wind turbine generator fault diagnosis method considering meteorological factors collects wind power plant fault information and meteorological historical records when faults occur; analyzing a failure mechanism of the wind turbine generator, determining an initial feature set, extracting features of factors causing fan failure by utilizing a Relieff algorithm, and selecting an optimal feature set; establishing a Bayesian network-based wind turbine generator fault diagnosis model, and adding the extracted features as reasons; training a Bayesian network-based wind turbine generator fault diagnosis model by using the collected training data, finding out prior distribution which accords with the data distribution characteristics, updating the prior state into a posterior state by using the training data, and making an optimal decision for fault prediction according to the continuously learned state model. The method adopts the Bayesian network to diagnose the fault of the wind turbine generator in combination with meteorological factors, thereby avoiding serious consequences caused by the diagnosis of the fan after the fault occurs; the accuracy and speed of the prediction diagnosis are improved.

Description

Wind turbine generator fault diagnosis method considering meteorological factors
Technical Field
The invention relates to the technical field of wind power generation, in particular to a wind turbine generator fault diagnosis method considering meteorological factors.
Background
With the proposal of the double-carbon target, the electric power as clean and efficient secondary energy plays more important roles in supporting the development of social economy, serving the energy demand of residents and constructing a clean low-carbon, safe and efficient energy system. The balance of power supply and demand is the basis of safe and stable operation of a power system, and has great significance for guaranteeing the safety of Chinese energy. However, because the single machine capacity of the wind turbine generator is large, the whole structure of the wind turbine generator is complex, and the wind power generation environment is usually in places with severe natural environments and rare occurrence, such as deserts, mountainous areas and the like, the operation conditions are quite unsatisfactory, and under different meteorological conditions, the operation environment of the wind turbine generator has different influences and the probability of generating faults is also different. Therefore, the prediction and the timely diagnosis and elimination of the possible faults of the wind turbine generator are of great importance to the healthy operation of the wind turbine generator and the efficiency of wind power generation by combining meteorological factors.
Chinese patent application (CN111708798A) discloses a method and a system for diagnosing and processing faults of a wind turbine generator. Modules such as a troubleshooting guide library and a logic diagnosis library are connected with the wind turbine generator, so that the technical problems of untimely fault response, inaccurate fault positioning and insufficient fault elimination experience are solved; the scheme is directly oriented to field fault processing services, accurately positions faults and provides a processing guidance scheme, and the faults are quickly and effectively solved. However, the response still can be performed after the fault occurs, which cannot achieve the effect of precaution in the future, and cannot better eliminate the influence caused by the abnormal operation state.
The bayesian model is a method based on probability theory, which represents the relationship between prior knowledge (signs, symptoms) and posterior knowledge (phenomena, conclusions). Based on Bayesian formula, Bayesian statistical inference and Bayesian network, the random variable existing in the system can be represented by nodes in the Bayesian network, the nodes represent sequential directional relation in a connecting line form, and the posterior probability is obtained by using the prior probability and the sample information, and is mainly used for processing the random information in the uncertainty information. All components and structures in the wind turbine generator can be represented by nodes in a Bayesian network, and the mutual relation among the components is represented by connecting lines, so that a new fault diagnosis idea is provided for a wind power generation system.
Disclosure of Invention
The method aims at the problems of fault prediction and fault timely diagnosis of the wind power generation system under the complex meteorological condition. The invention provides a wind turbine generator fault diagnosis method considering meteorological factors, which adopts a Bayesian network to diagnose the wind turbine generator by combining the meteorological factors, and avoids serious consequences caused by the diagnosis of a fan after the fault occurs; the accuracy and speed of the prediction diagnosis are improved.
The technical scheme adopted by the invention is as follows:
a wind turbine generator fault diagnosis method considering meteorological factors comprises the following steps:
the method comprises the following steps: acquiring fault information of the wind power plant and historical records of weather when the fault occurs;
step two: analyzing a failure mechanism of the wind turbine generator, determining an initial feature set, extracting features of factors causing fan failure by utilizing a Relieff algorithm, and selecting an optimal feature set;
step three: establishing a Bayesian network-based wind turbine generator fault diagnosis model, and adding the characteristics extracted in the step two as reasons into a reason layer of the wind turbine generator fault diagnosis model;
step four: training a Bayesian network-based wind turbine generator fault diagnosis model by using the collected training data, finding out prior distribution which accords with the data distribution characteristics, updating the prior state into a posterior state by using the training data, and repeatedly carrying out the optimal decision on fault prediction according to the continuously learned state model.
Step five: and testing the wind turbine generator fault diagnosis model trained in the fourth step by using the test set.
In the first step, the fault information of the wind power plant and meteorological factors when faults occur are collected together, and each piece of obtained fault data comprises fault reasons, fault symptoms, fault phenomena and meteorological factors.
In the second step, the meteorological factors causing the faults in the wind turbine generator fault initial characteristic set include temperature, air pressure, humidity, precipitation and snowfall and wind speed factors.
In the second step, the feature weight is calculated by utilizing a Relieff algorithm, and a classification method with the highest accuracy is selected to extract the features, wherein the method specifically comprises the following steps:
the characteristic weight calculation formula is as follows:
Figure BDA0003399925880000021
in the formula: w (a) and W' (a) represent the feature weight values after and before the iterative update, respectively. j is 1, 2, k, k is the total number of samples. The characteristics A are respectively expressed by n-dimensional arrays, and A is ═ a1,a2...,anEach sample is a point in n-dimensional space. Class (R) represents the type of R sample, C is a C sample, Mj(C) Represents the class C samples in the samples of different classes nearest to the sample R, P (class (R)) is the proportion of the R sample type to the total number of samples, P (C) is the proportion of the class C sample number to the total number of samples,
Figure BDA0003399925880000022
diff(A,R,Hj) Representative samples R and HjRegarding the difference in characteristic A, HjAnd MjRespectively representing the same type of sample and different types of samples which are closest to the sample R in the training set; m represents the number of sampling times; k represents the number of neighboring samples.
The sample characteristic difference calculation formula is as follows:
Figure BDA0003399925880000031
randomly selecting a sample R from the sample data set1And R2，R1[A]And R2[A]Represents a sample R1And R2The sample feature points that meet feature a.
In the third step, a Bayesian network-based wind turbine generator fault diagnosis model is established, and the Bayesian model formula is as follows:
Figure BDA0003399925880000032
wherein: omegaiP (X) is fault alarm information and is the prior probability of the node X; p (omega)i| X) is the conditional probability given X, P (X | ω)i) Is omegaiThe probability of X occurring in case of occurrence, also called a posterior probability;
the prior information of the model comprises switch information, electric quantity information and protection information, meteorological factors are simultaneously fused into the prior information, and a training set sample omega is obtained1、ω2、ω3.., the method is carried into the established Bayesian network-based wind turbine generator fault diagnosis model, and the conditional probability distribution is estimated.
The invention relates to a wind turbine generator fault diagnosis method considering meteorological factors, which has the following technical effects:
1) the invention utilizes the symptom before the fault happens to judge the fault in advance, controls and eliminates the fault in advance, and avoids the serious consequence caused by diagnosing the fan after the fault happens.
2) The method considers the meteorological factor which has a great influence on the running condition of the wind turbine generator, improves the accuracy and speed of the prediction diagnosis, and reduces the misdiagnosis probability.
3) According to the method, the correlation between meteorological factors and faults is utilized, so that the wind turbine generator fault diagnosis model based on the Bayesian network can find the possible faults under the condition that fault symptoms are more tiny, the influence caused by the faults is further reduced, fault nodes can be diagnosed, potential fault nodes which do not occur can also be diagnosed, and the wind turbine generator fault diagnosis model is suitable for process diagnosis and pre-diagnosis.
4) The MF-Bayesian Network model provided by the invention has 94.92% diagnosis accuracy, has certain effectiveness, and can provide a basis for fault diagnosis of the wind turbine generator system.
Drawings
Fig. 1 is a schematic diagram of the basic structure of a bayesian network.
FIG. 2 is a Bayesian network diagram of wind turbine fault diagnosis in view of meteorological factors.
FIG. 3 is a comparison graph of classification accuracy and weight after weight calculation of fault features.
FIG. 4 is a diagram of an MF-Bayesian Network model.
FIG. 5 is a schematic flow chart of the method of the present invention.
Detailed Description
A wind turbine generator fault diagnosis method considering meteorological factors comprises the following steps:
the method comprises the following steps: acquiring fault information of the wind power plant and historical records of weather when the fault occurs;
the wind turbine generator is used as power generation equipment which depends on the natural phenomenon of wind power as energy, and is influenced by natural conditions more than the traditional power generation mode. Therefore, the method has great significance in considering meteorological factors when fault diagnosis is carried out on the wind turbine generator. Before fault diagnosis is carried out, all faults occurring since the wind power plant is built are collected aiming at the wind power plant in a specific area. And querying the historical records of the meteorological conditions, and recording meteorological factors when each fault occurs. According to the method, the fault information of the wind power plant and the meteorological factors when the fault occurs are collected together, and each piece of obtained fault data comprises a fault reason, a fault symptom, a fault phenomenon and the meteorological factors.
The failure causes include: line aging, large air humidity on the day, and the like;
the fault symptoms include: abnormal signal fluctuations, abnormal switching actions, etc.;
the fault phenomena include: component failure, abnormal shutdown, etc.
Step two: analyzing a failure mechanism of the wind turbine generator, determining an initial feature set, extracting features of factors causing fan failure by utilizing a Relieff algorithm, reducing feature dimensions, and selecting an optimal feature set;
the optimal feature set is selected as follows:
calculating the weight of each feature according to a feature weight calculation formula, arranging the features according to the weight, respectively drawing a weight curve and a classification accuracy curve, selecting the weight value of the fault feature with the feature weight near the weight average value and the highest classification accuracy as a weight threshold, and forming an optimal feature set by all fault features with the weight larger than the weight threshold. And the highest classification accuracy is achieved when classification is carried out according to the optimal feature set.
In the second step, the occurrence reasons of the typical fault types of the wind turbine are analyzed, and the typical fault types of the wind turbine generator are sorted as shown in table 1:
TABLE 1 typical failure types of wind turbines
Figure BDA0003399925880000041
The fault generation is related to the factors such as temperature, air pressure, humidity, precipitation and snowfall, wind speed and the like. The initial set of characteristics of the wind turbine fault thus determined contains all the meteorological factors that may cause the fault. And marking and classifying the same weather and the same fault. After the preliminary classification, all the meteorological factors are not influenced to the fan fault to be worth noting, and the characteristics are too many to cause calculation difficulty, so that all the relevant meteorological factors are subjected to characteristic extraction. Redundant data features can complicate the model structure, adversely affect the identification and diagnosis of faults, and even reduce the diagnosis accuracy. Therefore, feature extraction needs to be performed on the initial feature set to remove redundant features. The characteristic extraction is a variable selection process, and is an important step for realizing effective state monitoring and fault detection of the wind turbine generator. Common feature selection methods can be broadly classified into a filtering method, a packaging method, and an embedded method. Among them, relieff (recent features) is a more common filtering feature selection method. Analyzing a failure mechanism of the fan, selecting all meteorological factors which can cause the fan failure, calculating the feature weight by utilizing a Relieff (redundant features) algorithm, and selecting a classification method with the highest accuracy rate to extract the features.
The method comprises the following specific steps:
the characteristic weight calculation formula is as follows:
Figure BDA0003399925880000051
in the formula: w (a) and W' (a) represent the feature weight values after and before the iterative update, respectively. j is 1, 2, k, k is the total number of samples. The characteristics A are respectively expressed by n-dimensional arrays, and A is ═ a1,a2...,anEach sample is a point in n-dimensional space. Class (R) represents the type of R sample, C is a C sample, Mj(C) Represents the class C samples in the samples of different classes nearest to the sample R, P (class (R)) is the proportion of the R sample type to the total number of samples, P (C) is the proportion of the class C sample number to the total number of samples,
Figure BDA0003399925880000052
diff(A,R,Hj) Representative samples R and HjRegarding the difference in characteristic A, HjAnd MjRespectively representing the same type of sample and different types of samples which are closest to the sample R in the training set; m represents the number of sampling times; k represents the number of neighboring samples.
The sample characteristic difference calculation formula is as follows:
Figure BDA0003399925880000053
randomly selecting a sample R from the sample data set1And R2，R1[A]And R2[A]Represents a sample R1And R2The sample feature points that meet feature a.
Step three: and (4) establishing a Bayesian network-based wind turbine generator fault diagnosis model, and adding the characteristics extracted in the step two as reasons into a reason layer of the wind turbine generator fault diagnosis model. As shown in fig. 1, the cause layer represents the cause of a failure, either directly or indirectly, and is the start of the evidence transfer.
In a wind power generation system, each element, such as a line, a transformer, etc., which correspond to a node in a bayesian network, may be divided into an element node, a system node, and a load node. The bayesian network consists of a Directed Acyclic Graph (DAG) and a Conditional Probability Table (CPT), and establishes a network corresponding to elements according to logical relations among the elements, and the network is respectively defined as a reason node, a symptom node and a fault node. The relationship between the nodes is represented by connecting lines with arrows, the direction of which indicates the transfer of evidence, as shown in fig. 1.
Research shows that besides electrical factors, meteorological factors are also a large influence factor influencing the normal operation state of the wind turbine generator. Information which can be generally utilized when a fault of a power system is predicted based on a traditional Bayesian model includes switching information, electrical quantity information, protection information and the like as prior information. The wind turbine generator fault model provided by the invention fuses meteorological factors into prior information of the model. The model construction schematic diagram is shown in FIG. 2
Establishing a Bayesian network-based wind turbine generator fault diagnosis model,
the Bayesian model formula is as follows:
Figure BDA0003399925880000061
wherein: omegaiP (X) is fault alarm information and is the prior probability of the node X; p (omega)i| X) is the conditional probability given X, P (X | ω)i) Is omegaiThe probability of X occurring in case of occurrence, also called a posterior probability;
the prior information of the model comprises switch information, electric quantity information and protection information, and meteorological factors are fused into the prior information.
The construction of the Bayesian model is divided into three main steps: firstly, determining a causal dependency relationship among variables; estimating prior probability distribution; and thirdly, estimating conditional probability distribution.
The main advantages of using a bayesian model are: any state of the node can be updated, and the decision can be made by using the update probability obtained after belief propagation. It also provides a graphical view of the overall process operation. Integrating meteorological factors into the prior information to complete the step I, and determining the causal relationship among variables, so that the likelihood function of the fault alarm sample information in the training set D is now made to beP(X︱ωi) Let the prior probability P (ω) be P (ω) ═ N (ω; 0, σ2) Wherein: sigma2Is the variance.
Will train the set sample omega1、ω2、ω3.., the wind turbine generator fault diagnosis model is brought into the established Bayesian network-based wind turbine generator fault diagnosis model, and the third step is completed to estimate the conditional probability distribution.
Conditional probability refers to the probability of occurrence of an event a if another event B has occurred. As shown in fig. 1, each table in fig. 1 is a conditional probability table of the corresponding node under the influence of other nodes.
Step four: training a Bayesian network-based wind turbine generator fault diagnosis model by using collected training data, finding out prior distribution which accords with data distribution characteristics, updating a prior state into a posterior state by using the training data, repeating the steps, and making an optimal decision for fault prediction according to a continuously learned state model, wherein the optimal decision is as follows:
and training the fault diagnosis model of the wind turbine generator based on the Bayesian network by using the collected training data set. And adopting N (0,1) distribution as prior distribution of the nodes, inputting the training data set into a model, updating the prior state of each node into a posterior state by using a Bayesian formula, and estimating the conditional probability of each node to form a conditional probability table to finish training. When a certain node is in an abnormal operation state, the trained Bayesian model transmits the evidence to each child node in a forward direction and updates the posterior probability, and the node with the maximum posterior probability is the most likely node with the fault.
The prior distribution refers to the probability obtained from past experience and analysis.
The posterior distribution state is probability estimation which is closer to the actual situation and is obtained by correcting the original prior probability of the node by using a Bayesian formula based on new information.
Step five: and testing the wind turbine generator fault diagnosis model trained in the fourth step by using the test set.
Example (b):
(1): and selecting a wind power plant in a certain area, and collecting all faults occurring since the wind power plant is built. And recording the meteorological conditions of all past faults at the same time. Preprocessing the collected data, wherein each piece of fault information includes a fault description and meteorological factors when a fault occurs, taking table 2 as an example:
TABLE 2 weather factors at fault
Figure BDA0003399925880000071
(2): considering that the temperature rise will cause the voltage resistance level and the insulation property of the equipment to be reduced; the high humidity easily reduces the insulation performance of equipment, and causes the occurrence of current leakage, pollution flashover and the like; lightning may directly cause the equipment to withstand lightning strike overvoltages and the like. Through analysis, the wind turbine generator fault characteristics are preliminarily determined as shown in table 3.
TABLE 3 wind turbine Fault signatures
Figure BDA0003399925880000072
Figure BDA0003399925880000081
The initial fault characteristics are subjected to weight calculation, and the weights are arranged from large to small, so that the obtained characteristic weights and the accuracy are shown in table 4.
TABLE 4 feature weights and accuracies
Figure BDA0003399925880000082
The accuracy and the weight are represented by fig. 3, and it is found that the classification accuracy is the highest when the 7 th feature is calculated, and the classification accuracy is greatly reduced when the 8 th feature is calculated with the increase of the feature quantity. Finally, the threshold value of the feature weight takes a value of 0.023.
(3): dividing the elements into reason nodes, symptom nodes and fault nodes, using directional arrows to represent the correlation between the nodes, and establishing a Bayesian network representing the relationship of each element in the wind turbine as shown in FIG. 4
(4): in the past fault data of the wind turbine generator, a part of total data is selected as a training set in a random sampling mode, the established Bayesian model is trained, other data is used as a testing set, and the established Bayesian model is tested.
(5): and integrating meteorological factors into prior information, and estimating conditional probability distribution by using a training set sample. And predicting the probability of the fault occurrence by using a Bayesian model. Converting all fault data into classification data, wherein: normal indicates normal signals and abnormal indicates abnormal signals. The data set used to train the bayesian model is given in table 5.
TABLE 5 data set used for training the Bayesian model
Figure BDA0003399925880000083
Figure BDA0003399925880000091
(6): using the accuracy verification finding, through k-fold cross-validation, 10000 sets of predicted data are obtained, wherein the correct 8057 set is predicted for 8352 normal data (normal), and the correct 1435 set is predicted for 1648 failure data (abnormal), wherein the failure is caused by weather influences 572 set, and the correct 534 set is predicted. The final result shows that the fault prediction accuracy of the MF-Bayesian Network model is 94.92%.
(7): a receiver operating characteristic curve (ROC) is used for verifying the MF-Bayesian Network model, and the result shows that the AUC value of the model is 0.9049, the effectiveness of the model is verified, and an effective basis can be provided for fault diagnosis and risk prediction of the wind power generating unit.

Claims (6)

1. A wind turbine generator fault diagnosis method considering meteorological factors is characterized by comprising the following steps:
the method comprises the following steps: acquiring fault information of the wind power plant and historical records of weather when the fault occurs;
step two: analyzing a failure mechanism of the wind turbine generator, determining an initial feature set, extracting features of factors causing fan failure by utilizing a Relieff algorithm, and selecting an optimal feature set;
step three: establishing a Bayesian network-based wind turbine generator fault diagnosis model, and adding the characteristics extracted in the step two as reasons into a reason layer of the wind turbine generator fault diagnosis model;
step four: training a Bayesian network-based wind turbine generator fault diagnosis model by using the collected training data, finding out prior distribution which accords with the data distribution characteristics, updating the prior state into a posterior state by using the training data, and repeatedly carrying out the optimal decision on fault prediction according to the continuously learned state model.
2. The wind turbine generator system fault diagnosis method considering meteorological factors according to claim 1, wherein: in the first step, the fault information of the wind power plant and meteorological factors when faults occur are collected together, and each piece of obtained fault data comprises fault reasons, fault symptoms, fault phenomena and meteorological factors.
3. The wind turbine generator system fault diagnosis method considering meteorological factors according to claim 1, wherein: in the second step, the meteorological factors causing the faults in the wind turbine generator fault initial characteristic set include temperature, air pressure, humidity, precipitation and snowfall and wind speed factors.
4. The wind turbine generator system fault diagnosis method considering meteorological factors according to claim 1, wherein: in the second step, the feature weight is calculated by utilizing a Relieff algorithm, and a classification method with the highest accuracy is selected to extract the features, wherein the method specifically comprises the following steps:
the characteristic weight calculation formula is as follows:
Figure FDA0003399925870000011
in the formula: w (a) and W' (a) represent the feature weight values after and before the iterative update, respectively; j 1, 2, k, k is the total number of samples; the characteristics A are respectively expressed by n-dimensional arrays, and A is ═ a1,a2...,an-each sample is a point in n-dimensional space; class (R) represents the type of R sample, C is a C sample, Mj(C) Represents the class C samples in the samples of different classes nearest to the sample R, P (class (R)) is the proportion of the R sample type to the total number of samples, P (C) is the proportion of the class C sample number to the total number of samples,
Figure FDA0003399925870000012
diff(A,R,Hj) Representative samples R and HjRegarding the difference in characteristic A, HjAnd MjRespectively representing the same type of sample and different types of samples which are closest to the sample R in the training set; m represents the number of sampling times; k represents the number of adjacent samples; p (C) is the proportion of the number of class C samples to the total number of samples;
the sample characteristic difference calculation formula is as follows:
Figure FDA0003399925870000021
randomly selecting a sample R from the sample data set1And R2，R1[A]And R2[A]Represents a sample R1And R2The sample feature points that meet feature a.
5. The wind turbine generator system fault diagnosis method considering meteorological factors according to claim 1, wherein: in the third step, a Bayesian network-based wind turbine generator fault diagnosis model is established, and the Bayesian model formula is as follows:
Figure FDA0003399925870000022
wherein: omegaiP (X) is fault alarm information and is the prior probability of the node X; p (omega)i| X) is the conditional probability given X, P (X | ω)i) Is omegaiThe probability of X occurring in case of occurrence, also called a posterior probability;
the prior information of the model comprises switch information, electric quantity information and protection information, meteorological factors are simultaneously fused into the prior information, and a training set sample omega is obtained1、ω2、ω3.., the method is carried into the established Bayesian network-based wind turbine generator fault diagnosis model, and the conditional probability distribution is estimated.
6. The wind turbine generator system fault diagnosis method considering meteorological factors according to claim 1, wherein: in the third step, the construction of the Bayesian model is divided into three main steps: firstly, determining a causal dependency relationship among variables; estimating prior probability distribution; estimating conditional probability distribution;
integrating meteorological factors into the prior information to complete the first step, and determining the causal relationship among variables; the likelihood function of the fault alarm sample information in the training set D is P (X | omega)i) Let the prior probability P (ω) be P (ω) ═ N (ω; 0, σ2) Wherein: sigma2Is the variance; will train the set sample omega1、ω2、ω3.., the wind turbine generator fault diagnosis model is brought into the established Bayesian network-based wind turbine generator fault diagnosis model, and the third step is completed to estimate the conditional probability distribution.
CN202111495441.XA 2021-12-08 2021-12-08 Wind turbine generator fault diagnosis method considering meteorological factors Pending CN114370380A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111495441.XA CN114370380A (en) 2021-12-08 2021-12-08 Wind turbine generator fault diagnosis method considering meteorological factors

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111495441.XA CN114370380A (en) 2021-12-08 2021-12-08 Wind turbine generator fault diagnosis method considering meteorological factors

Publications (1)

Publication Number Publication Date
CN114370380A true CN114370380A (en) 2022-04-19

Family

ID=81140328

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111495441.XA Pending CN114370380A (en) 2021-12-08 2021-12-08 Wind turbine generator fault diagnosis method considering meteorological factors

Country Status (1)

Country Link
CN (1) CN114370380A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115037603A (en) * 2022-05-31 2022-09-09 国网湖南省电力有限公司 Diagnosis evaluation method, device and system of electricity consumption information acquisition equipment
CN115268350A (en) * 2022-09-27 2022-11-01 江苏永鼎股份有限公司 Fault early warning method and system for voltage stabilizing transformer
CN115358639A (en) * 2022-10-20 2022-11-18 国网山东省电力公司烟台供电公司 Offshore wind power operation risk analysis system based on data analysis

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115037603A (en) * 2022-05-31 2022-09-09 国网湖南省电力有限公司 Diagnosis evaluation method, device and system of electricity consumption information acquisition equipment
CN115268350A (en) * 2022-09-27 2022-11-01 江苏永鼎股份有限公司 Fault early warning method and system for voltage stabilizing transformer
CN115358639A (en) * 2022-10-20 2022-11-18 国网山东省电力公司烟台供电公司 Offshore wind power operation risk analysis system based on data analysis
CN115358639B (en) * 2022-10-20 2023-01-24 国网山东省电力公司烟台供电公司 Offshore wind power operation risk analysis system based on data analysis

Similar Documents

Publication Publication Date Title
CN111237134B (en) Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model
CN114370380A (en) Wind turbine generator fault diagnosis method considering meteorological factors
CN106054104B (en) A kind of intelligent electric meter failure real-time predicting method based on decision tree
CN110570122B (en) Offshore wind power plant reliability assessment method considering wind speed seasonal characteristics and current collection system element faults
CN109977624A (en) Photovoltaic plant soft fault monitoring method based on deep neural network
CN110362045B (en) Marine doubly-fed wind turbine generator fault discrimination method considering marine meteorological factors
Yang et al. Fault detection of wind turbine generator bearing using attention-based neural networks and voting-based strategy
CN112215305A (en) Virtual laser radar system and method based on mutual early warning of fans
CN114215706B (en) Early warning method and device for cracking faults of wind turbine generator blades
Shi et al. Study of wind turbine fault diagnosis and early warning based on SCADA data
Wang et al. Data-driven prediction method for characteristics of voltage sag based on fuzzy time series
CN113486291B (en) Petroleum drilling machine micro-grid fault prediction method based on deep learning
CN110794254A (en) Power distribution network fault prediction method and system based on reinforcement learning
Atrigna et al. A machine learning approach to fault prediction of power distribution grids under heatwaves
CN109101659B (en) Method for power data abnormity in small hydropower station data acquisition system
Sun et al. A multi-model-integration-based prediction methodology for the spatiotemporal distribution of vulnerabilities in integrated energy systems under the multi-type, imbalanced, and dependent input data scenarios
Gao et al. Design and application of a fault diagnosis and monitoring system for electric vehicle charging equipment based on improved deep belief network
CN110412417A (en) Micro-grid data fault diagnosis method based on intelligent power monitoring instrument
CN108074045B (en) Wind turbine generator complete machine vulnerability analysis and fault sequencing method and electronic terminal
Dong et al. Fault diagnosis and classification in photovoltaic systems using scada data
Khediri et al. Deep-belief network based prediction model for power outage in smart grid
CN110348489B (en) Transformer partial discharge mode identification method based on self-coding network
Hassani et al. Design of a cost-effective deep convolutional neural network–based scheme for diagnosing faults in smart grids
CN116151799A (en) BP neural network-based distribution line multi-working-condition fault rate rapid assessment method
CN116127354A (en) High-voltage cable partial discharge fault positioning method and system based on deep learning

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