CN107179503B - Wind turbine generator fault intelligent diagnosis and early warning method based on random forest - Google Patents
Wind turbine generator fault intelligent diagnosis and early warning method based on random forest Download PDFInfo
- Publication number
- CN107179503B CN107179503B CN201710266789.9A CN201710266789A CN107179503B CN 107179503 B CN107179503 B CN 107179503B CN 201710266789 A CN201710266789 A CN 201710266789A CN 107179503 B CN107179503 B CN 107179503B
- Authority
- CN
- China
- Prior art keywords
- wind turbine
- data
- turbine generator
- model
- sample
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/34—Testing dynamo-electric machines
- G01R31/343—Testing dynamo-electric machines in operation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Wind Motors (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a wind turbine generator fault intelligent diagnosis and early warning method based on random forests, which comprises the following steps: extracting historical data of the state of the wind turbine generator as sample data; carrying out exploratory analysis and pretreatment on the sample data; constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result; and diagnosing the wind turbine equipment in real time by using the analyzed and evaluated model, and sending alarm information by using the model if the diagnosis result is abnormal. The invention adopts a random forest algorithm, considers the overall characteristics of the indexes, can solve the problem that a single index determines the equipment state, and comprehensively considers the hidden knowledge correlation among a plurality of indexes to comprehensively judge the output result.
Description
Technical Field
The invention relates to the technical field of electric power, in particular to a wind turbine generator fault intelligent diagnosis and early warning method based on a random forest.
Background
Wind power generation is one of the world recognized renewable energy technologies closest to commercialization. Under the background of the emphasis on environmental protection and sustainable development, wind power generation without consuming fossil fuel and environmental pollution is considered as the cleanest energy utilization form. Wind power generation has become the fastest growing renewable energy source in the world over the last 10 years due to annual average growth rates approaching 28%.
Along with the rapid development of wind energy and the operation of large-scale wind turbine generators, and because most of the wind turbine generators are installed in remote areas and the load is unstable, many wind turbine generators in China have operation faults, and the safety and the economy of wind power generation are directly influenced. In order to keep the long-term stable development of wind power and enhance the competitiveness of the wind power and traditional energy, the wind power generation efficiency must be continuously improved, the maintenance cost and the operation cost of wind power equipment are reduced, and the economic benefit of enterprises is maximized.
Although a certain scientific and technical means is adopted in the traditional fault diagnosis method, the proportion of the technical and experience of maintenance personnel is large, the fault is basically located by firstly depending on the experience of the maintenance personnel and then accurately locating by the scientific and technical diagnosis method, and the fault problem is found, so that the defects of man-made subjective errors and prolonged unit maintenance time exist to a certain extent;
in addition, as the roads of the areas where the wind power plants are located are far away, the natural environment is severe, maintenance personnel need to watch the wind power plants for a long time, and the waste of the personnel capital can be caused to a certain extent along with the increase of the number of the wind power plants.
Disclosure of Invention
The invention overcomes the problems of long fault finding time, large deviation and personnel waste in fault diagnosis and monitoring of the traditional manual method. Therefore, the invention provides a wind turbine generator fault intelligent diagnosis and early warning method based on a random forest.
In order to achieve the above purpose, the invention provides the following technical scheme:
the intelligent wind turbine fault diagnosis and early warning method based on the random forest comprises the following steps:
extracting historical data of the state of the wind turbine generator as sample data; carrying out exploratory analysis and pretreatment on the sample data;
constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result;
and diagnosing the wind turbine equipment in real time by using the analyzed and evaluated model, and sending alarm information by using the model if the diagnosis result is abnormal.
Preferably, extracting historical data of the wind turbine state as sample data includes:
analyzing common faults of the wind turbine generator to construct a fault index system of the wind turbine generator;
and extracting historical data of the state of the wind turbine generator from the wind turbine generator automation system or the equipment background control system to be used as sample data.
Preferably, exploratory analysis and preprocessing are performed on the sample data, including:
carrying out exploratory analysis on the sample data, namely removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine;
preprocessing the sample data, including: data cleaning, missing value processing, data discretization, attribute reduction and data transformation.
Preferably, the method for constructing the intelligent diagnosis and early warning model of the wind turbine generator fault based on the random forest comprises the following steps:
dividing the preprocessed sample data into training samples and testing samples, and constructing a random forest model expert sample set;
and constructing an intelligent diagnosis and early warning model of the wind turbine generator faults by using the sample set.
Preferably, the intelligent diagnosis and early warning model for analyzing and evaluating the wind turbine generator faults comprises the following steps:
the prediction results of the model on the test set are analyzed and diagnosed,
if all predictions are correct, the diagnosis effect of the model is ideal;
and if the diagnosis result has errors, further optimizing the model.
Preferably, the real-time diagnosis of the wind turbine equipment by using the trained model comprises the following steps:
accessing online data from a wind turbine generator automation system or an equipment background control system;
selecting an attribute in an air outlet generator set fault index system from the online data as an input attribute;
and preprocessing the online data, accessing the preprocessed online data into the model after analysis and evaluation, and monitoring and diagnosing the equipment data in real time.
Preferably, the method further comprises the steps of collecting more historical fault and normal data of the wind turbine generator, training the model periodically, and updating the model in time.
Compared with the prior art, the invention has the beneficial effects that:
aiming at the problem that the fault frequency of the wind turbine generator occurs, the intelligent diagnosis and early warning model for the fault of the wind turbine generator is constructed, the random forest algorithm is adopted to train and learn historical data of the wind turbine generator, characteristic patterns of factors influencing the fault occurrence of the wind turbine generator are excavated, and threshold values of indexes of the fault occurrence are determined, so that the online fault diagnosis and early warning for the wind turbine generator are carried out according to the real-time operation of equipment, the maintenance cost of the wind turbine generator is effectively reduced, and the utilization efficiency of the wind turbine generator is improved;
the invention adopts a random forest algorithm, considers the overall characteristics of the indexes, can solve the problem that a single index determines the equipment state, and comprehensively considers the hidden knowledge correlation among a plurality of indexes to comprehensively judge the output result.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for intelligently diagnosing and warning a fault of a wind turbine generator based on a random forest according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a decision tree provided by a real-time embodiment of the present invention;
FIG. 3 is a schematic diagram of a random forest provided by a real-time embodiment of the present invention;
FIG. 4 is a flow chart of a method for intelligent diagnosis and early warning of faults of a wind turbine generator based on a random forest according to the real-time embodiment of the present invention, wherein the method is used for constructing an intelligent diagnosis and early warning model of faults of a wind turbine generator based on a random forest;
FIG. 5 is a diagram illustrating a relationship between a false positive rate of a random forest and a number of decision trees according to an embodiment of the present invention;
fig. 6 is a flowchart of a method for intelligently diagnosing and warning a fault of a wind turbine generator based on a random forest according to a real-time embodiment of the present invention, wherein the method is implemented by using a wind turbine generator fault intelligent diagnosis and warning model based on a random forest.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1:
the intelligent wind turbine fault diagnosis and early warning method based on the random forest comprises the following steps:
constructing a wind turbine generator fault index system, and extracting historical wind turbine generator state data as sample data;
performing data exploration and pretreatment on the extracted historical sample data of the state of the wind turbine generator;
constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result;
the method comprises the steps of utilizing a built wind turbine generator fault intelligent diagnosis early warning model based on a random forest to realize real-time diagnosis;
and collecting more historical fault and normal data of the wind turbine generator, training the model regularly, and updating the model in time.
Preferably, the constructing a wind turbine generator fault index system, and extracting historical wind turbine generator state data as sample data means:
analyzing common faults of the wind turbine generator to construct a fault index system of the wind turbine generator;
and selectively extracting part of state historical data of the wind turbine generator from the wind turbine generator automation system and the equipment background control system.
Preferably, the data exploration and preprocessing of the extracted historical sample data of the state of the wind turbine generator includes:
carrying out exploratory analysis on the sample, removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine;
and preprocessing the sample, including data cleaning, missing value processing, data discretization, attribute reduction, data transformation and the like.
Preferably, the constructing of the wind turbine generator fault intelligent diagnosis early warning model based on the random forest, the analyzing and evaluating of the model according to the model result includes:
constructing a random forest model expert sample set;
constructing an intelligent fault diagnosis early warning model of the wind turbine generator;
and analyzing and evaluating the model result, and further optimizing the intelligent wind turbine fault diagnosis early warning model according to the analysis and evaluation result.
Preferably, the method for realizing real-time diagnosis by using the established intelligent wind turbine generator fault diagnosis and early warning model based on the random forest comprises the following steps:
according to the established intelligent wind turbine fault diagnosis model (after training is completed), relevant data of wind turbine equipment is accessed into the model, the equipment data is monitored and diagnosed in real time, and once the diagnosis result is abnormal, warning information is sent to the wind turbine.
Preferably, the collecting more historical fault and normal data of the wind turbine generator, training the model periodically, and updating the model in time includes:
more historical fault and normal data of the wind turbine generator are collected, so that the model covers the characteristics of all conditions as far as possible, the model can be trained once a month regularly, the model is updated in time, and the accuracy of the model is continuously improved.
The invention discloses a wind turbine generator fault intelligent diagnosis and early warning method based on a random forest, which comprises the following steps: aiming at the problem that the fault frequency of the wind turbine generator occurs, an intelligent diagnosis and early warning model for the fault of the wind turbine generator is constructed, a random forest algorithm is adopted to train and learn historical data of the wind turbine generator, characteristic patterns of factors influencing the fault occurrence of the wind turbine generator are excavated, and threshold values of indexes of the fault occurrence are determined, so that the online fault diagnosis and early warning for the wind turbine generator are carried out according to real-time operation of equipment, the maintenance cost of the wind turbine generator is effectively reduced, and the utilization efficiency of the wind turbine generator is improved.
The method has certain defects, and the method does not perform independent research on certain key indexes (which often directly determine the equipment state), but adopts a random forest algorithm and considers the overall characteristics of the indexes, so that the method not only can solve the problem that the equipment state is determined by a single index, but also can comprehensively consider the hidden knowledge correlation among a plurality of indexes and make comprehensive judgment on the output result.
Example 2:
referring to fig. 1, a flowchart of a method for intelligently diagnosing and warning a fault of a wind turbine generator based on a random forest according to an embodiment of the present invention is shown, where the method may include:
s101: and constructing a wind turbine generator fault index system, and extracting wind turbine generator state data as sample data.
Analyzing common faults of the wind turbine generator, selecting 13 attributes (gear _ temp, gear _ generator _ vol, …, motor _ temp _ sd, power _ mean, is _ running) as input attributes, and establishing a wind turbine generator fault index system by taking whether the wind turbine generator normally operates as an output attribute;
and selectively extracting part of state historical data of the wind turbine generator from the wind turbine generator automatic system and the equipment background control system to be used as sample data. It should be noted that, in the step, part of the historical data of the state of the wind turbine generator is selectively extracted, because some required index data in the equipment background control system may be incomplete, data with high data integrity should be selected as sample data as far as possible.
S102: and performing data exploration and pretreatment on the extracted historical sample data of the state of the wind turbine generator.
Carrying out exploratory analysis on the sample, removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine;
and preprocessing the sample, including data cleaning, missing value processing, data discretization, attribute reduction, data transformation and the like.
S103: and constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result.
Constructing a random forest model expert sample set;
constructing an intelligent fault diagnosis early warning model of the wind turbine generator;
and analyzing and evaluating the model result, and further optimizing the intelligent wind turbine fault diagnosis early warning model according to the analysis and evaluation result.
S104: and real-time diagnosis is realized by utilizing the constructed wind turbine generator fault intelligent diagnosis early warning model based on the random forest.
According to the established intelligent wind turbine fault diagnosis model (after training is completed), relevant data of wind turbine equipment is accessed into the model, the equipment data is monitored and diagnosed in real time, and once the diagnosis result is abnormal, warning information is sent to the wind turbine.
S105: and collecting more historical fault and normal data of the wind turbine generator, training the model regularly, and updating the model in time.
More historical fault and normal data of the wind turbine generator are collected, so that the model covers the characteristics of all conditions as far as possible, the model can be trained once a month regularly, the model is updated in time, and the accuracy of the model is continuously improved.
In this embodiment, optionally, an intelligent wind turbine generator fault diagnosis and early warning model based on a random forest is constructed. As shown in fig. 6, includes:
firstly, constructing a random forest model expert sample set, comprising the following steps:
s201: setting parameters: the random forest algorithm selects 500 trees, and divides the samples preprocessed by the S102 into training samples and testing samples by using hierarchical sampling, wherein the ratio of the training samples to the testing samples is (0.8, 0.2).
For the layered sampling in S201, it should be noted that:
the layered sampling is to combine scientific grouping method with sampling method, in particular to divide the whole into a plurality of homogeneous layers and randomly sample or mechanically sample in each layer, and the layered sampling ensures that the sampled samples have enough representativeness.
S202: 500 sample subsets are generated from the training sample set Of the S201 by using a replaceable random sampling mode, the number Of samples Of each sample subset is the same as that Of the training sample set Of the S201, theoretically, the 500 sample subsets cover 2/3 data examples in the original sample set, the data which is not contained is called Out-Of-Bag data (OOB), the Out-Of-Bag data can be used as test data, and the classification effect Of the combined classifier can be well evaluated by estimation in a random forest algorithm.
For the example data in the original sample set 2/3 theoretically covered by the sample subset after performing the replaceable random sampling on the data in S202, it should be noted that:
assuming that a sample set D with the number of samples N is subjected to random sampling with putting back, N times of sampling are carried out, and the probability that each sample is not sampled is (1-1/N)NWhen N is sufficiently large (1-1/N)NWill converge to 1/e ≈ 0.368, so i believe that the sample subset covers the sample data of the original sample 2/3.
Then, constructing an intelligent diagnosis and early warning model of the wind turbine generator faults, comprising the following steps:
s203: and growing 500 decision trees by using the generated 500 self-help sample sets. Here, m (m) is randomly selected from the 13 features at each node of each tree<13) features, which are usually taken in actual engineering
One feature is selected to branch from the randomly selected m features according to a certain rule (information gain) each time until the tree grows sufficiently, and pruning operation is not performed in the meantime.
For the information entropy and the information gain recited in S203, it should be noted that:
information entropy: show thatThe larger the uncertainty (degree of disorder) of the information, the larger the entropy, the more disordered and difficult the prediction of the information, the smaller the amount of information provided by the index, and the less important the weight of the index. For a classification system, class C is a variable, which may take the value C1,C2,…,CnAnd the probability of each class occurrence is P (C)1),P(C2),…,P(Cn) And thus n is the total number of categories. The entropy of the classification system can now be expressed as:
the information gain is for a feature, that is, looking at a feature t, what the amount of information is when the system has it and when it does not, and the difference between the two is the amount of information that the feature brings to the system, that is, the gain. The amount of information when the system contains a feature t is the above equation, which represents the amount of information of the system when all features are included.
InfoGain＝H(Y)-H(Y|X)
In the classification system, the selection of attributes and the splitting of the decision tree are selected according to information gain, attribute variables with the maximum information gain are selected aiming at root nodes and child nodes, and then the whole decision tree and a random forest are constructed by adopting a recursive method.
S204: and predicting the test sample set according to the 500 generated decision trees, and determining a final result according to a certain voting mechanism by integrating the test result of each tree. The random forest algorithm utilizes randomness (including randomly generating a sub sample set and randomly selecting sub features), minimizes the correlation among trees, improves the overall classification performance, and has very short generation time of each tree, parallelization of forests can be realized, and the classification speed of the random forests is very high. Assume random forest classifier { hi(x,θiI is 1, …, N), the class label of the classification result is determined by each decision tree hi(x,θi) And probability averaging, for test case x, the predicted class label cpThen, the process of the present invention,
wherein argmaxcThe parameter c representing the search with the largest score, N the number of decision trees in the random forest, I (x) the performance function,
representing the classification result of the decision tree for the class C,
representing a decision tree hiNumber of leaf nodes, WiRepresenting the weight of the ith tree in the random forest.
The model results are as follows:
table 1 model output results
As can be seen from the table above, the out-of-packet data error rate OOB is 2.7% for the normal operation of the wind turbine generator, which indicates that the overall classification effect of the model is very ideal. Fig. 5 shows the random forest OOB false positive rate and the number of decision trees, and it can be seen from fig. 5 that the random forest false positive rate continuously decreases with the increase of the number of decision trees, and finally converges to a smaller fixed value. The broken line labeled 1 in fig. 5 represents a negative sample error rate, the broken line labeled 2 represents a positive sample error rate, and the solid black line represents a total error rate.
Finally, analyzing and evaluating the model result, and determining whether to further optimize the intelligent diagnosis and early warning model of the wind turbine generator system fault according to the analysis and evaluation result, wherein the method comprises the following steps:
s205: analyzing and diagnosing the prediction result of the model on the test set, and analyzing the model diagnosis result:
table 2: test data diagnostic results
By carrying out diagnosis analysis on 8 pieces of data in the test set, the prediction is all correct, the diagnosis effect of the model is relatively ideal, and the intelligent diagnosis and early warning model for the faults of the wind turbine generator does not need to be further optimized. If the diagnosis effect of the model is not ideal, the model can be further optimized by adjusting the number of decision trees, the maximum depth of the trees, the information measurement mode, the feature selection method and other parameters. It can be seen that: according to the test result, the 1 st, 4 th, 6 th and 8 th wind generation sets normally run, the 2 nd, 3 th, 5 th and 7 th wind generation sets send out early warning information, corresponding plan measures should be started immediately, and larger safety accidents and economic losses are prevented from occurring.
In the above embodiment, optionally, real-time diagnosis is realized by using the constructed intelligent wind turbine generator fault diagnosis and early warning model based on the random forest. As shown in fig. 6, includes:
s301: and accessing online data from the wind turbine generator automation system and the equipment background control system.
S302: selecting 13 attributes in the wind turbine generator system fault index system from the online data (
gear _ temp, gear _ rate, rotor _ vol, …, motor _ temp _ sd, power _ mean, is _ running) as input attributes.
S303: and preprocessing the online data, including data cleaning, missing value processing, data discretization, attribute stipulation, data transformation and the like.
S304: and accessing the preprocessed online data into the model, monitoring and diagnosing the equipment data in real time, sending alarm information to the wind turbine generator once the diagnosis result is abnormal, and enabling a worker to immediately take corresponding shutdown measures and alternative schemes according to each parameter.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (7)
1. The method for intelligently diagnosing and early warning the faults of the wind turbine generator based on the random forest is characterized by comprising the following steps of:
extracting historical data of the state of the wind turbine generator as sample data; carrying out exploratory analysis and pretreatment on the sample data;
constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result;
the model after analysis and evaluation is used for diagnosing the wind turbine equipment in real time, and if the diagnosis result is abnormal, the model sends out warning information;
performing exploratory analysis on the sample data, namely removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine; preprocessing the sample data, including: data cleaning, missing value processing, data discretization, attribute stipulation and data transformation;
the method specifically comprises the following steps:
s101: constructing a wind turbine generator fault index system, and extracting wind turbine generator state data as sample data;
analyzing common faults of the wind turbine generator, selecting the following 13 attributes of gear _ temp, gear _ rate _ vol, …, motor _ temp _ sd, power _ mean and is _ running as input attributes, and establishing a wind turbine generator fault index system by taking whether the wind turbine generator normally operates as an output attribute;
selectively extracting part of state historical data of the wind turbine generator from an automatic system of the wind turbine generator and a background control system of the equipment as sample data; the selective extraction of part of the historical data of the state of the wind turbine generator in the step is that some required index data in the equipment background control system are possibly incomplete, so that data with higher data integrity is selected as sample data as much as possible;
s102: performing data exploration and pretreatment on the extracted historical sample data of the state of the wind turbine generator;
carrying out exploratory analysis on the sample, removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine;
preprocessing a sample, including data cleaning, missing value processing, data discretization, attribute stipulation, data transformation and the like;
s103: constructing a wind turbine generator fault intelligent diagnosis early warning model based on a random forest, and analyzing and evaluating the model according to a model result;
constructing a random forest model expert sample set;
constructing an intelligent fault diagnosis early warning model of the wind turbine generator;
analyzing and evaluating the model result, and further optimizing the intelligent wind turbine fault diagnosis early warning model according to the analysis and evaluation result;
s104: the method comprises the steps of utilizing a built wind turbine generator fault intelligent diagnosis early warning model based on a random forest to realize real-time diagnosis;
according to the intelligent wind turbine fault diagnosis model which is built and trained, related data of wind turbine equipment are accessed into the model, the equipment data are monitored and diagnosed in real time, and once the diagnosis result is found to be abnormal, warning information is sent to the wind turbine;
s105: collecting more historical fault and normal data of the wind turbine generator, training the model regularly, and updating the model in time;
more historical fault and normal data of the wind turbine generator are collected, so that the model covers the characteristics of all conditions as far as possible, the model can be trained once a month regularly, the model is updated in time, and the accuracy of the model is continuously improved;
constructing a random forest model expert sample set, comprising the following steps:
s201: setting parameters: selecting 500 trees by a random forest algorithm, and dividing the samples subjected to S102 preprocessing into training samples and testing samples by utilizing hierarchical sampling, wherein the ratio of the training samples to the testing samples is (0.8, 0.2);
for the layered sampling in S201, it should be noted that:
the layered sampling is to combine a scientific grouping method with a sampling method, particularly to divide the whole into a plurality of homogeneous layers and randomly sample or mechanically sample in each layer, wherein the layered sampling ensures that the sampled samples have enough representativeness;
s202: generating 500 sample subsets from the training sample set Of the S201 by using a replaceable random sampling mode, wherein the number Of samples Of each sample subset is the same as that Of the training sample set Of the S201, theoretically, the 500 sample subsets cover 2/3 data examples in the original sample set, the data which are not contained are called Out-Of-Bag data, such as Out-Of-Bag and OOB, the Out-Of-Bag data can be used as test data, and the classification effect Of the combined classifier can be well evaluated in a random forest algorithm;
the method for constructing the intelligent diagnosis and early warning model of the faults of the wind turbine generator comprises the following steps:
s203: growing 500 decision trees by using the generated 500 self-help sample sets; in this case, m is randomly selected from 13 features at each node of each tree, and m is less than or equal to 13 features, which are usually taken in practical engineering
Selecting one feature from the randomly selected m features for branching according to a certain information gain rule each time until the tree grows sufficiently, and not performing pruning operation during the period;
s204: predicting the test sample set according to the 500 generated decision trees, and determining a final result according to a certain voting mechanism by integrating the test result of each tree;
the method comprises the steps of utilizing a built wind turbine generator fault intelligent diagnosis early warning model based on a random forest to realize real-time diagnosis; the method comprises the following steps:
s301: accessing online data from a wind turbine generator automation system and an equipment background control system;
s302: selecting the following 13 attributes gear _ temp, gear _ rate, rotor _ vol, …, motor _ temp _ sd, power _ mean in the wind turbine generator set fault index system from the online data,
is _ running as an input attribute;
s303: preprocessing online data, including data cleaning, missing value processing, data discretization, attribute stipulation, data transformation and the like;
s304: and accessing the preprocessed online data into the model, monitoring and diagnosing the equipment data in real time, sending alarm information to the wind turbine generator once the diagnosis result is abnormal, and enabling a worker to immediately take corresponding shutdown measures and alternative schemes according to each parameter.
2. The intelligent diagnosis and early warning method for the faults of the wind turbine generator based on the random forest as claimed in claim 1, wherein the step of extracting historical data of the states of the wind turbine generator as sample data comprises the following steps:
analyzing common faults of the wind turbine generator to construct a fault index system of the wind turbine generator;
and extracting historical data of the state of the wind turbine generator from the wind turbine generator automation system or the equipment background control system to be used as sample data.
3. The intelligent diagnosis and early warning method for faults of wind turbines based on random forests as claimed in claim 1, characterized in that,
and carrying out exploratory analysis and pretreatment on the sample data, wherein the process comprises the following steps:
carrying out exploratory analysis on the sample data, namely removing aging equipment and just-put-into-production equipment, and preliminarily examining the characteristic conditions of normal and abnormal states of the wind turbine;
preprocessing the sample data, including: data cleaning, missing value processing, data discretization, attribute reduction and data transformation.
4. The method for intelligently diagnosing and warning the faults of the wind turbine generator based on the random forest as claimed in claim 1, wherein the constructing of the intelligent diagnosis and warning model of the faults of the wind turbine generator based on the random forest comprises the following steps:
dividing the preprocessed sample data into training samples and testing samples, and constructing a random forest model expert sample set;
and constructing an intelligent diagnosis and early warning model of the wind turbine generator faults by using the sample set.
5. The method for wind turbine generator fault intelligent diagnosis and early warning based on random forest as claimed in claim 1, wherein analyzing and evaluating the wind turbine generator fault intelligent diagnosis and early warning model comprises:
the prediction results of the model on the test set are analyzed and diagnosed,
if all predictions are correct, the diagnosis effect of the model is ideal;
and if the diagnosis result has errors, further optimizing the model.
6. The intelligent diagnosis and early warning method for the faults of the wind turbine generator based on the random forest as claimed in claim 1, wherein the real-time diagnosis of the wind turbine generator equipment by using the trained model comprises the following steps:
accessing online data from a wind turbine generator automation system or an equipment background control system;
selecting an attribute in an air outlet generator set fault index system from the online data as an input attribute;
and preprocessing the online data, accessing the preprocessed online data into the model after analysis and evaluation, and monitoring and diagnosing the equipment data in real time.
7. The intelligent diagnosis and early warning method for the faults of the wind turbine generator based on the random forest as claimed in claim 1, further comprising the steps of collecting more historical faults and normal data of the wind turbine generator, training the model regularly, and updating the model timely.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710266789.9A CN107179503B (en) | 2017-04-21 | 2017-04-21 | Wind turbine generator fault intelligent diagnosis and early warning method based on random forest |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710266789.9A CN107179503B (en) | 2017-04-21 | 2017-04-21 | Wind turbine generator fault intelligent diagnosis and early warning method based on random forest |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107179503A CN107179503A (en) | 2017-09-19 |
| CN107179503B true CN107179503B (en) | 2020-07-07 |
Family
ID=59832129
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710266789.9A Active CN107179503B (en) | 2017-04-21 | 2017-04-21 | Wind turbine generator fault intelligent diagnosis and early warning method based on random forest |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107179503B (en) |
Families Citing this family (42)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107782442A (en) * | 2017-10-24 | 2018-03-09 | 华北电力大学（保定） | Transformer multiple features parameter selection method based on big data and random forest |
| CN107967485B (en) * | 2017-11-16 | 2024-05-14 | 国网冀北电力有限公司电力科学研究院 | Fault analysis method and device for electricity metering equipment |
| CN109840666B (en) * | 2017-11-29 | 2020-12-18 | 中国电力科学研究院有限公司 | Model construction method and system for predicting future wind turbine generator failure |
| CN108011367B (en) * | 2017-12-04 | 2020-12-18 | 贵州电网有限责任公司电力科学研究院 | Power load characteristic mining method based on depth decision tree algorithm |
| CN107958395B (en) * | 2017-12-13 | 2021-11-26 | 美林数据技术股份有限公司 | Method for identifying abnormal users of power system |
| CN108020785A (en) * | 2017-12-14 | 2018-05-11 | 海安常州大学高新技术研发中心 | A kind of electrical fault forecasting system and data managing method based on micromainframe |
| CN110657072A (en) * | 2018-06-29 | 2020-01-07 | 中车株洲电力机车研究所有限公司 | Wind power fault maintenance method and system based on structured knowledge base |
| CN108535618A (en) * | 2018-07-11 | 2018-09-14 | 云南电网有限责任公司电力科学研究院 | A kind of GIS method for detecting insulation defect |
| CN109283378A (en) * | 2018-08-30 | 2019-01-29 | 番禺珠江钢管（珠海）有限公司 | A kind of rotating arc welding is seamed into shape parameter detection method, system, device and medium |
| CN109297689B (en) * | 2018-09-11 | 2020-11-06 | 中国水利水电科学研究院 | Large-scale hydraulic machinery intelligent diagnosis method introducing weight factors |
| CN110968069B (en) * | 2018-09-28 | 2022-01-25 | 新疆金风科技股份有限公司 | Fault prediction method of wind generating set, corresponding device and electronic equipment |
| CN109543210A (en) * | 2018-09-28 | 2019-03-29 | 国电电力宁夏新能源开发有限公司 | A kind of Wind turbines failure prediction system based on machine learning algorithm platform |
| CN109522599A (en) * | 2018-10-16 | 2019-03-26 | 国电南瑞科技股份有限公司 | Transmission line of electricity catastrophic failure method for early warning caused by a kind of typhoon |
| CN109544035A (en) * | 2018-12-12 | 2019-03-29 | 上海理工大学 | Electric energy efficiency analysis and ranking method based on random forest |
| CN109800127A (en) * | 2019-01-03 | 2019-05-24 | 众安信息技术服务有限公司 | A kind of system fault diagnosis intelligence O&M method and system based on machine learning |
| CN109855879B (en) * | 2019-01-26 | 2020-09-04 | 厦门华夏国际电力发展有限公司 | Online detection and early warning method and system for faults of steam turbine servo mechanism |
| CN109887122A (en) * | 2019-02-19 | 2019-06-14 | 南京洛普股份有限公司 | A kind of ship lock functional failure of electromechanical long-distance intelligent diagnostic method and system |
| CN111724184A (en) * | 2019-03-20 | 2020-09-29 | 北京嘀嘀无限科技发展有限公司 | Transformation probability prediction method and device |
| CN109936113B (en) * | 2019-03-29 | 2021-06-01 | 国网浙江省电力有限公司 | Protection action intelligent diagnosis method and system based on random forest algorithm |
| CN110007232B (en) * | 2019-05-23 | 2021-09-03 | 广东工业大学 | Method and related device for predicting running efficiency of squirrel-cage asynchronous motor |
| CN110188834A (en) * | 2019-06-04 | 2019-08-30 | 广东电网有限责任公司 | A kind of method for diagnosing faults of power telecom network, device and equipment |
| CN110378427A (en) * | 2019-07-23 | 2019-10-25 | 上海电气风电集团有限公司 | Fault detection method, system, equipment and the medium of the blade root bolt of wind electricity blade |
| CN111080466A (en) * | 2019-12-03 | 2020-04-28 | 广东工业大学 | Calender calendering quality early warning system based on big data |
| CN111047215B (en) * | 2019-12-09 | 2023-06-23 | 中国兵器科学研究院 | Method for determining classification of field replaceable units based on random forest |
| CN111260201B (en) * | 2020-01-13 | 2023-04-28 | 北京科技大学 | Variable importance analysis method based on layered random forest |
| CN111310786B (en) * | 2020-01-15 | 2023-07-25 | 青岛海信网络科技股份有限公司 | Traffic detector abnormality diagnosis method and device based on random forest classifier |
| CN111611146B (en) * | 2020-06-18 | 2023-05-16 | 南方电网科学研究院有限责任公司 | Micro-service fault prediction method and device |
| CN112398218A (en) * | 2020-09-28 | 2021-02-23 | 国网山东省电力公司冠县供电公司 | Power grid safety and stability analysis device and method |
| CN112308146A (en) * | 2020-11-02 | 2021-02-02 | 国网福建省电力有限公司 | Distribution transformer fault identification method based on operation characteristics |
| CN112446144A (en) * | 2020-11-17 | 2021-03-05 | 哈工大机器人（合肥）国际创新研究院 | Fault diagnosis method and device for large-scale rotating machine set |
| CN112578794B (en) * | 2020-12-12 | 2023-09-01 | 云南昆船智能装备有限公司 | AGV fault detection method, storage medium and system based on machine learning |
| CN112561176A (en) * | 2020-12-21 | 2021-03-26 | 深圳供电局有限公司 | Early warning method for online running state of electric power metering device |
| CN112765878A (en) * | 2021-01-11 | 2021-05-07 | 龙源（北京）风电工程技术有限公司 | Intelligent diagnosis method and device for abnormity of generator bearing of wind turbine generator |
| US11774324B2 (en) * | 2021-03-12 | 2023-10-03 | General Electric Renovables Espana, S.L. | System and method for detecting actual slip in a coupling of a rotary shaft |
| CN113107785B (en) * | 2021-05-12 | 2022-05-31 | 浙江浙能技术研究院有限公司 | Real-time monitoring method and device for power performance abnormity of wind turbine generator |
| EP4364041A1 (en) * | 2021-08-11 | 2024-05-08 | Siemens Aktiengesellschaft | Method and apparatus for training a model |
| CN113869578A (en) * | 2021-09-26 | 2021-12-31 | 合肥通用机械研究院有限公司 | Intelligent prediction and diagnosis method for salt content of crude oil after removal of electric desalting system of atmospheric and vacuum distillation unit |
| CN114483479B (en) * | 2021-12-23 | 2023-09-12 | 大唐可再生能源试验研究院有限公司 | Fan Gao Wenjiang capacity state assessment method based on random forest |
| WO2023182944A1 (en) * | 2022-03-25 | 2023-09-28 | Nanyang Technological University | Alternator monitoring systems and methods |
| CN115079626B (en) * | 2022-07-21 | 2022-11-22 | 东方电气风电股份有限公司 | Early warning method and system for potential operation risk of wind generating set component |
| CN116664096B (en) * | 2022-08-23 | 2024-02-13 | 国家电投集团科学技术研究院有限公司 | Wind power bolt data processing method and device based on federal learning |
| CN117436569A (en) * | 2023-09-18 | 2024-01-23 | 华能核能技术研究院有限公司 | Nuclear power equipment fault prediction and intelligent calibration method and system based on random forest |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104747368A (en) * | 2015-01-27 | 2015-07-01 | 风脉（武汉）可再生能源技术有限责任公司 | Method and system for optimizing power of wind turbine generator |
| CN105069476A (en) * | 2015-08-10 | 2015-11-18 | 国网宁夏电力公司 | Method for identifying abnormal wind power data based on two-stage integration learning |
-
2017
- 2017-04-21 CN CN201710266789.9A patent/CN107179503B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104747368A (en) * | 2015-01-27 | 2015-07-01 | 风脉（武汉）可再生能源技术有限责任公司 | Method and system for optimizing power of wind turbine generator |
| CN105069476A (en) * | 2015-08-10 | 2015-11-18 | 国网宁夏电力公司 | Method for identifying abnormal wind power data based on two-stage integration learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107179503A (en) | 2017-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107179503B (en) | Wind turbine generator fault intelligent diagnosis and early warning method based on random forest | |
| Hsu et al. | Wind turbine fault diagnosis and predictive maintenance through statistical process control and machine learning | |
| Abdallah et al. | Fault diagnosis of wind turbine structures using decision tree learning algorithms with big data | |
| CN110929918B (en) | 10kV feeder fault prediction method based on CNN and LightGBM | |
| Zaher et al. | A multi-agent fault detection system for wind turbine defect recognition and diagnosis | |
| CN110674189B (en) | Method for monitoring secondary state and positioning fault of intelligent substation | |
| CN110362045B (en) | Marine doubly-fed wind turbine generator fault discrimination method considering marine meteorological factors | |
| CN111582551B (en) | Wind power plant short-term wind speed prediction method and system and electronic equipment | |
| CN114723285B (en) | Power grid equipment safety evaluation prediction method | |
| CN110276372B (en) | Fuel cell engine fault diagnosis method based on cloud platform | |
| Bangalore et al. | An approach for self evolving neural network based algorithm for fault prognosis in wind turbine | |
| CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
| CN112115999A (en) | Wind turbine generator fault diagnosis method of space-time multi-scale neural network | |
| CN111931851A (en) | Fan blade icing fault diagnosis method based on one-dimensional residual error neural network | |
| CN116842337A (en) | Transformer fault diagnosis method based on LightGBM (gallium nitride based) optimal characteristics and COA-CNN (chip on board) model | |
| CN103489041A (en) | Short-term wind power forecast method | |
| CN115758151A (en) | Combined diagnosis model establishing method and photovoltaic module fault diagnosis method | |
| KR20220089853A (en) | Method for Failure prediction and prognostics and health management of renewable energy generation facilities using machine learning technology | |
| CN114320773B (en) | Wind turbine generator system fault early warning method based on power curve analysis and neural network | |
| CN115563885A (en) | Energy conversion system for energy storage battery of wind power plant | |
| Shaaban et al. | A decision tree approach for steam turbine-generator fault diagnosis | |
| Zhenhao et al. | Prediction of wind power ramp events based on deep neural network | |
| CN114819248A (en) | Wind power generation capacity prediction method and device based on random forest | |
| Jankauskas et al. | Short-term wind energy forecasting with advanced recurrent neural network models: a comparative study | |
| CN112465404B (en) | Method and system for evaluating running state of photovoltaic power generation system for multiple time windows |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: 710075 6th floor, national e-commerce demonstration base, No.528, tianguba Road, software new town, high tech Zone, Xi'an City, Shaanxi Province Patentee after: MERITDATA TECHNOLOGY CO.,LTD. Address before: 710000 C501, Hanyun Pavilion, Xi'an Software Park, No.68, Keji 2nd Road, high tech Zone, Xi'an City, Shaanxi Province Patentee before: MERITDATA TECHNOLOGY CO.,LTD. |