CN111259953B - Equipment defect time prediction method based on capacitive equipment defect data - Google Patents
Equipment defect time prediction method based on capacitive equipment defect data Download PDFInfo
- Publication number
- CN111259953B CN111259953B CN202010039425.9A CN202010039425A CN111259953B CN 111259953 B CN111259953 B CN 111259953B CN 202010039425 A CN202010039425 A CN 202010039425A CN 111259953 B CN111259953 B CN 111259953B
- Authority
- CN
- China
- Prior art keywords
- data
- capacitive
- defect
- equipment
- equipment defect
- 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
Classifications
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
According to the equipment defect time prediction method based on the capacitive equipment defect data, firstly, the capacitive equipment defect data is obtained, the obtained abnormal, redundant and missing data are better processed through a series of feature engineering methods, the occurrence time of the defects is better predicted through establishing a capacitive equipment defect occurrence time model, the model can extract effective features from big data, and the defect occurrence time of the capacitive equipment is accurately predicted through the features. The equipment defect time prediction method based on the capacitive equipment defect data has the advantages of being simple to implement, high in calculation speed, high in prediction accuracy, good in prediction robustness and systematic in prediction flow, and solves the problems that in the prior art, the experimental data are simply considered to give up the capacitor equipment work and maintenance data collected by a power grid company for years to conduct research and prediction, a conclusion on one side can be obtained, and the equipment defect time cannot be accurately predicted.
Description
Technical Field
The application relates to the technical field of electrical equipment and information, in particular to an equipment defect time prediction method based on capacitive equipment defect data.
Background
The capacitive device is a device employing a capacitive shielding insulating structure. The transformer mainly comprises a current transformer, a voltage transformer, a capacitive pipe sleeve, a coupling capacitor and the like, and the capacitor is about 40 to 50 percent of the total amount of power transmission and transformation equipment, and is the equipment with the largest amount in a transformer substation. Healthy operation of capacitive devices and safety of electrical devices are critical to the substation, and any unexpected failure can lead to major accidents and very large economic losses. Therefore, the realization of the on-line detection and prediction of the capacitive device has very important research significance.
At present, on-line monitoring research on capacitive equipment is mainly focused on development of digital measurement transmitters and on-line monitoring systems, the influence of research environments on the capacitive equipment is required to be tested in a climate chamber, and the research is relatively complex and is relatively few at home and abroad.
The experimental method has the advantages that experiments on the influence of environmental factors on capacitive equipment are conducted by a large-scale artificial climate chamber, comprehensive and accurate experimental data are obtained, a correction model for the influence based on main link factors of a Support Vector Machine (SVM) is provided, and model parameters are optimized by a genetic algorithm. However, this approach requires a separate laboratory and is obviously impractical to do with all types of equipment because the performance of the individual manufacturer equipment parameters is not very consistent. Meanwhile, the actual environment of the capacitor equipment work is more complex than the experimental environment, and the study and prediction can be carried out by simply giving up the capacitor equipment work and maintenance data collected by the power grid company in the past year by taking the experimental data into consideration, so that a conclusion on one side can be obtained.
Disclosure of Invention
The application provides an equipment defect time prediction method based on capacitive equipment defect data, which aims to solve the problems that in the prior art, the experimental data is simply considered to discard the capacitor equipment work and maintenance data collected by a power grid company for years to conduct research prediction, a conclusion on one side can be obtained, and the equipment defect time cannot be accurately predicted.
The technical scheme adopted by the application for solving the technical problems is as follows:
a device defect time prediction method based on capacitive device defect data, the method comprising:
performing data cleaning treatment on the defect data set of the capacitive equipment;
carrying out characteristic transformation and coding on the cleaned data to obtain characteristic data;
performing reduction and denoising on the characteristic data by using a self-encoder recoding method to obtain recoded characteristic data;
training a plurality of machine learning models by utilizing the obtained recoding data characteristics, evaluating the model quality degree by using training time and root mean square error, and selecting an optimal model;
and storing the optimal model, and predicting the defect time of the capacitive equipment by using the optimal model.
Optionally, the data cleaning process includes: and filling the missing values in the data cleaning, wherein the filling method is to fill by using a K nearest neighbor algorithm and a random forest algorithm.
Optionally, the feature transformation and encoding includes feature decomposition and interleaving.
Optionally, the method for recoding by the self-encoder includes: sparse self-encoder, noise reduction self-encoder and variation self-encoder;
the sparse encoder selects an encoder comprising 4 fully connected layers and a decoder comprising 4 fully connected layers.
Optionally, the training a plurality of machine learning models using the obtained recoded data features includes:
taking the marked and cleaned data characteristics as a label sample, and carrying out a normalization pretreatment process on the label sample;
five machine learning models were developed using the label samples: k nearest neighbor regression, support vector regression, random forest, gradient lifting tree and deep learning are respectively trained;
calculating a mean square loss function of the machine learning model, and if the loss function meets the conditions and all the hyper-parameters of the grid search are trained, obtaining the model by using the hyper-parameters with the best effect; otherwise, the label sample is reused to train the five machine learning models continuously.
Through evaluating the losses of the five machine learning models, the model with the minimum loss is selected.
Optionally, in the five machine learning models, a K value in the K nearest neighbor regression is selected to be 3, and manhattan distance is used as a distance measure; the Gaussian kernel is selected to be used in the support vector regression; the number of decision trees in the random forest is 100, and the maximum depth of the trees is 3; the number of decision trees in the gradient lifting number is 100, the maximum depth of the trees is 3, and the learning rate is 0.1; four groups of convolution and batch normalization layers are included in the deep learning, two layers of full-connection layers are connected, a ReLU is selected as an activation function, an MSE is selected as a loss function, and Adam is used as an optimization method.
Optionally, the machine learning models are trained by adopting a grid hyper-parameter searching method of five-fold cross validation, so that optimal hyper-parameters are ensured to be searched, a set mean square error threshold value is reached, and if the mean square error is larger than the threshold value, the training is continued.
The technical scheme provided by the application has the following beneficial technical effects:
according to the equipment defect time prediction method based on the capacitive equipment defect data, firstly, the capacitive equipment defect data is obtained, the obtained abnormal, redundant and missing data are better processed through a series of feature engineering methods, the occurrence time of the defects is better predicted through establishing a capacitive equipment defect occurrence time model, the model can extract effective features from big data, and the defect occurrence time of the capacitive equipment is accurately predicted through the features. The equipment defect time prediction method based on the capacitive equipment defect data has the advantages of being simple to implement, high in calculation speed, high in prediction accuracy, good in prediction robustness and systematic in prediction flow, and solves the problems that in the prior art, the experimental data are simply considered to give up the capacitor equipment work and maintenance data collected by a power grid company for years to conduct research and prediction, a conclusion on one side can be obtained, and the equipment defect time cannot be accurately predicted.
Drawings
In order to more clearly illustrate the technical solution of the present application, the drawings that are needed in the embodiments will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.
FIG. 1 is a flowchart of a method for predicting device defect time based on capacitive device defect data according to an embodiment of the present application;
fig. 2 is a schematic diagram of an embodiment of the present application.
Detailed Description
In order to enable those skilled in the art to better understand the technical solutions of the present application, the technical solutions of the application embodiments will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application; it will be apparent that the described embodiments are only some, but not all, embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
Referring to fig. 1, fig. 1 is a flowchart of a method for predicting device defect time based on capacitive device defect data according to an embodiment of the present application, as shown in fig. 1, the method for predicting device defect time based on capacitive device defect data according to the present application includes the following steps:
s1: performing data cleaning treatment on the defect data set of the capacitive equipment;
s2: carrying out characteristic transformation and coding on the cleaned data to obtain characteristic data;
s3: performing reduction and denoising on the characteristic data by using a self-encoder recoding method to obtain recoded characteristic data;
s4: training a plurality of machine learning models by utilizing the obtained recoding data characteristics, evaluating the model quality degree by using training time and root mean square error, and selecting an optimal model;
s5: and storing an optimal model, and predicting the defect time of the capacitive equipment by using the optimal model.
According to the equipment defect time prediction method based on the capacitive equipment defect data, firstly, the capacitive equipment defect data are obtained, the obtained abnormal, redundant and missing data are better processed through a series of feature engineering methods, the occurrence time of the defects is better predicted through establishing a capacitive equipment defect occurrence time model, the model can extract effective features from big data, and the defect occurrence time of the capacitive equipment is accurately predicted through the features. The equipment defect time prediction method based on the capacitive equipment defect data has the advantages of being simple to implement, high in calculation speed, high in prediction accuracy, good in prediction robustness and systematic in prediction flow.
Optionally, the data cleaning process includes: and filling the missing values in the data cleaning, wherein the filling method is to fill by using a K nearest neighbor algorithm and a random forest algorithm.
By adopting the K nearest neighbor algorithm and the random forest algorithm, the result of the missing value filling is more robust, other characteristic information is combined, more available information is introduced, the data is more close to real data, and the following modeling step is facilitated.
Optionally, the feature transformation and encoding includes feature decomposition and interleaving.
Optionally, the method of recoding from the encoder includes: sparse self-encoder, noise reduction self-encoder and variation self-encoder; the sparse encoder selects an encoder comprising 4 fully connected layers and a decoder comprising 4 fully connected layers.
By adopting the technical scheme, in the dimension reduction of the characteristic data, the function space is larger than that of the function space by using a method model recoded by a self-encoder, and the loss is smaller while the dimension reduction is performed; in the noise reduction of the characteristic data, the self-coding method can ensure the sparsity of the data and remove most of the noise in the data, so that the specific establishment process of the self-coder recoding method model is obtained, and the method is combined into a specific application scene to set specific input and output parameters.
Optionally, training a plurality of machine learning models using the resulting recoded data features, comprising:
taking the marked and cleaned data characteristics as a label sample, and carrying out a normalization pretreatment process on the label sample;
five machine learning models were developed using label samples: k nearest neighbor regression, support vector regression, random forest, gradient lifting tree and deep learning are respectively trained;
calculating a mean square loss function of the machine learning model, and if the loss function meets the conditions and all the hyper-parameters of the grid search are trained, obtaining the model by using the hyper-parameters with the best effect; otherwise, the label sample is reused to train the five machine learning models continuously.
By evaluating the losses of the five machine learning models, the model with the smallest loss is selected, so that the best machine learning model is obtained.
Optionally, in the five machine learning models, the K value in the K nearest neighbor regression is selected to be 3, and manhattan distance is used as a distance measure; the Gaussian kernel is selected to be used in the support vector regression; the number of decision trees in the random forest is 100, and the maximum depth of the trees is 3; the number of decision trees in the gradient lifting number is 100, the maximum depth of the trees is 3, and the learning rate is 0.1; four groups of convolution and batch normalization layers are included in the deep learning, two layers of full-connection layers are connected, a ReLU is selected as an activation function, an MSE is selected as a loss function, and Adam is used as an optimization method.
By adopting the technical scheme, a specific machine learning model establishment process is obtained, and the specific input and output parameters are set in combination with a specific application scene.
Optionally, the machine learning model is trained by adopting a grid super-parameter searching method of five-fold cross validation, so that the optimal super-parameter is ensured to be searched, a set mean square error threshold value is reached, and if the mean square error is larger than the threshold value, the training is continued.
By adopting the technical scheme, the determination standard of the machine learning model is obtained.
In particular, the embodiment of the application also provides a specific manner of facts, as follows:
step 1: performing data cleaning treatment on the defect data set of the capacitive device: removing the value with more than 70% missing, and filling the missing value of the value with more than 30% missing by using a K nearest neighbor algorithm and a random forest algorithm; drawing a box graph of each feature according to the data features, removing abnormal values of the data, and deleting all redundant data and null data;
step 2: performing feature transformation and coding on the cleaned data: all character type features were subjected to feature decomposition as shown in table 1 below. The decomposed character type features are tag-coded, i.e. the value of each feature is assigned to a number. For continuous numerical features, feature binning techniques, such as latitude and longitude, are performed, giving a different code every 10 °. Finally, all the coded features are subjected to feature crossing, and the features are multiplied by each other to form new features.
TABLE 1 character profile comparison Table
Step 3: and (3) recoding the characteristic data obtained in the step (2) by a sparse self-encoder: and selecting sparse self-coding according to the quality of the characteristic data, and performing dimension reduction and denoising recoding on the characteristic data. The sparse encoder selects an encoder comprising 4 fully connected layers and a decoder comprising 4 fully connected layers. The encoder input is the features obtained in step 2, and the first layer output is the 32 features after dimension reduction. The encoder second, third and fourth layers are shaped as (64, 32), (32, 32) and (32, 16), respectively. The decoder first, second and third layers are shaped as (16, 32), (32, 32) and (32, 64), respectively. The fourth layer of the decoder is input with 64 features, and the output is the feature quantity obtained in the step 2. The activation function selected is the Tanh function. The self-encoder incorporates an L1 regularization term and the loss function is chosen to be the mean square error. After five-fold cross validation, 200 rounds of training are performed respectively, a trained sparse self-encoder is obtained. The resulting data characteristic of step 2 is input from the encoder and the output of the encoder is taken as the new recoding characteristic. The sparse self-encoder structure is shown in fig. 2.
Step 4: constructing a model by using the characteristic data obtained in the step 3: and predicting the occurrence time of the defects of the capacitive equipment by using K nearest neighbor regression, support vector regression, random forest, gradient lifting tree and deep learning methods respectively. And taking the feature data with defects as a training set with labels, taking the feature data without defects as a test set, and normalizing the feature data set. Training the five models by using a training set through five-fold cross validation and a grid search method respectively, and selecting the optimal super parameters. The resulting model and its mean squared error are shown in table 2 below, where the K-nearest neighbor regression K value is selected to be 3, using manhattan distance as the distance measure; support vector regression selects the use of gaussian kernels; the number of decision trees in the random forest is 100, and the maximum depth of the tree is 3; the number of decision trees of the gradient lifting number is 100, the maximum depth of the tree is 3, and the learning rate is 0.1; the deep learning model includes four sets of convolution and batch normalization layers followed by two full connection layers. The activation function selects ReLU, the loss function selects MSE, and Adam is used as the optimization method. And evaluating the five models to finally obtain the model with the best effect: gradient lifting tree model.
Table 2 five models and mean square error comparison table
Step 5: and (3) saving the model obtained in the step (4), and predicting test set data: and (3) saving parameters of the gradient lifting decision tree model, extracting feature vectors of the test set in the step (4), and inputting the feature vectors into the model for prediction. Finally, the prediction result of the defect time of the capacitive equipment is obtained, and the result is rewritten into the data table to finish.
According to the equipment defect time prediction method based on the capacitive equipment defect data, firstly, the capacitive equipment defect data are obtained, the obtained abnormal, redundant and missing data are better processed through a series of feature engineering methods, the occurrence time of the defects is better predicted through establishing a capacitive equipment defect occurrence time model, the model can extract effective features from big data, and the defect occurrence time of the capacitive equipment is accurately predicted through the features. The equipment defect time prediction method based on the capacitive equipment defect data has the advantages of being simple to implement, high in calculation speed, high in prediction accuracy, good in prediction robustness and systematic in prediction flow, and solves the problems that in the prior art, the experimental data are simply considered to give up the capacitor equipment work and maintenance data collected by a power grid company for years to conduct research and prediction, a conclusion on one side can be obtained, and the equipment defect time cannot be accurately predicted.
It is noted that relational terms such as "first" and "second", and the like, are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that an article or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The foregoing is only a specific embodiment of the application to enable those skilled in the art to understand or practice the application. 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 application. Thus, the present application 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.
It will be understood that the application is not limited to what has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.
Claims (6)
1. A method for predicting device defect time based on capacitive device defect data, the method comprising:
performing data cleaning treatment on the defect data set of the capacitive equipment;
carrying out characteristic transformation and coding on the cleaned data to obtain characteristic data;
performing reduction and denoising on the characteristic data by using a self-encoder recoding method to obtain recoded characteristic data;
training a plurality of machine learning models by utilizing the obtained recoding data characteristics, evaluating the model quality degree by using training time and root mean square error, and selecting an optimal model;
storing the optimal model, and predicting the defect time of the capacitive equipment by using the optimal model;
the training of a plurality of machine learning models using the resulting recoded data features includes:
taking the marked and cleaned data characteristics as a label sample, and carrying out a normalization pretreatment process on the label sample;
five machine learning models were developed using the label samples: k nearest neighbor regression, support vector regression, random forest, gradient lifting tree and deep learning are respectively trained;
calculating a mean square loss function of the machine learning model, and if the loss function meets the conditions and all the hyper-parameters of the grid search are trained, obtaining the model by using the hyper-parameters with the best effect; otherwise, the label sample is reused to train the five machine learning models continuously;
through evaluating the losses of the five machine learning models, the model with the minimum loss is selected.
2. The method for predicting equipment defect time based on capacitive equipment defect data according to claim 1, wherein the data cleaning process comprises: and filling the missing values in the data cleaning, wherein the filling method is to fill by using a K nearest neighbor algorithm and a random forest algorithm.
3. The method of device defect time prediction based on capacitive device defect data of claim 1, wherein the feature transformation and encoding comprises feature decomposition and interleaving.
4. The method for device defect time prediction based on capacitive device defect data of claim 1, wherein the method for recoding from the encoder comprises: sparse self-encoder, noise reduction self-encoder and variation self-encoder;
the sparse self encoder selects an encoder comprising 4 fully connected layers and a decoder comprising 4 fully connected layers.
5. The method for predicting equipment defect time based on capacitive equipment defect data according to claim 1, wherein in the five machine learning models, a K value in a K nearest neighbor regression is selected to be 3, and manhattan distance is used as a distance measure; the Gaussian kernel is selected to be used in the support vector regression; the number of decision trees in the random forest is 100, and the maximum depth of the trees is 3; the number of decision trees in the gradient lifting number is 100, the maximum depth of the trees is 3, and the learning rate is 0.1; four groups of convolution and batch normalization layers are included in the deep learning, two layers of full-connection layers are connected, a ReLU is selected as an activation function, an MSE is selected as a loss function, and Adam is used as an optimization method.
6. The method for predicting equipment defect time based on capacitive equipment defect data according to claim 1, wherein the machine learning models are trained by adopting a grid hyper-parameter searching method of five-fold cross validation, so that optimal hyper-parameters are ensured to be searched, a set mean square error threshold is reached, and if the mean square error is larger than the threshold, training is continued.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010039425.9A CN111259953B (en) | 2020-01-15 | 2020-01-15 | Equipment defect time prediction method based on capacitive equipment defect data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010039425.9A CN111259953B (en) | 2020-01-15 | 2020-01-15 | Equipment defect time prediction method based on capacitive equipment defect data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111259953A CN111259953A (en) | 2020-06-09 |
CN111259953B true CN111259953B (en) | 2023-10-20 |
Family
ID=70948786
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010039425.9A Active CN111259953B (en) | 2020-01-15 | 2020-01-15 | Equipment defect time prediction method based on capacitive equipment defect data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111259953B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2022117218A1 (en) * | 2020-12-02 | 2022-06-09 | Hitachi Energy Switzerland Ag | Prognosis of high voltage equipment |
CN112926627A (en) * | 2021-01-28 | 2021-06-08 | 电子科技大学 | Equipment defect time prediction method based on capacitive equipment defect data |
CN113378908B (en) * | 2021-06-04 | 2023-04-18 | 浙江大学 | Heating ventilation air conditioning system fault diagnosis method based on LightGBM algorithm and grid search algorithm |
CN114004515A (en) * | 2021-11-04 | 2022-02-01 | 云南电网有限责任公司电力科学研究院 | Singular value decomposition-based transformer substation equipment health monitoring sensor arrangement method |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014205231A1 (en) * | 2013-06-19 | 2014-12-24 | The Regents Of The University Of Michigan | Deep learning framework for generic object detection |
CN108255656A (en) * | 2018-02-28 | 2018-07-06 | 湖州师范学院 | A kind of fault detection method applied to batch process |
CN108459955A (en) * | 2017-09-29 | 2018-08-28 | 重庆大学 | Software Defects Predict Methods based on depth autoencoder network |
CN109165664A (en) * | 2018-07-04 | 2019-01-08 | 华南理工大学 | A kind of attribute missing data collection completion and prediction technique based on generation confrontation network |
CN109271374A (en) * | 2018-10-19 | 2019-01-25 | 国网江苏省电力有限公司信息通信分公司 | A kind of database health scoring method and scoring system based on machine learning |
CN110174499A (en) * | 2019-07-10 | 2019-08-27 | 云南电网有限责任公司电力科学研究院 | A kind of prediction technique and device of sulfur hexafluoride electrical equipment Air Leakage Defect |
CN110348633A (en) * | 2019-07-11 | 2019-10-18 | 电子科技大学 | A kind of linear classification model defect Occurrence forecast method |
CN110399685A (en) * | 2019-07-29 | 2019-11-01 | 云南电网有限责任公司电力科学研究院 | Capacitance type equipment defect rank prediction technique and device |
CN110515931A (en) * | 2019-07-02 | 2019-11-29 | 电子科技大学 | A kind of capacitance type equipment failure prediction method based on random forests algorithm |
CN110659937A (en) * | 2019-09-20 | 2020-01-07 | 鞍钢集团矿业有限公司 | Gradient-lifting-tree-based improved supplier quantitative scoring prediction algorithm |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7958068B2 (en) * | 2007-12-12 | 2011-06-07 | International Business Machines Corporation | Method and apparatus for model-shared subspace boosting for multi-label classification |
CN104021264B (en) * | 2013-02-28 | 2017-06-20 | 华为技术有限公司 | A kind of failure prediction method and device |
-
2020
- 2020-01-15 CN CN202010039425.9A patent/CN111259953B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014205231A1 (en) * | 2013-06-19 | 2014-12-24 | The Regents Of The University Of Michigan | Deep learning framework for generic object detection |
CN108459955A (en) * | 2017-09-29 | 2018-08-28 | 重庆大学 | Software Defects Predict Methods based on depth autoencoder network |
CN108255656A (en) * | 2018-02-28 | 2018-07-06 | 湖州师范学院 | A kind of fault detection method applied to batch process |
CN109165664A (en) * | 2018-07-04 | 2019-01-08 | 华南理工大学 | A kind of attribute missing data collection completion and prediction technique based on generation confrontation network |
CN109271374A (en) * | 2018-10-19 | 2019-01-25 | 国网江苏省电力有限公司信息通信分公司 | A kind of database health scoring method and scoring system based on machine learning |
CN110515931A (en) * | 2019-07-02 | 2019-11-29 | 电子科技大学 | A kind of capacitance type equipment failure prediction method based on random forests algorithm |
CN110174499A (en) * | 2019-07-10 | 2019-08-27 | 云南电网有限责任公司电力科学研究院 | A kind of prediction technique and device of sulfur hexafluoride electrical equipment Air Leakage Defect |
CN110348633A (en) * | 2019-07-11 | 2019-10-18 | 电子科技大学 | A kind of linear classification model defect Occurrence forecast method |
CN110399685A (en) * | 2019-07-29 | 2019-11-01 | 云南电网有限责任公司电力科学研究院 | Capacitance type equipment defect rank prediction technique and device |
CN110659937A (en) * | 2019-09-20 | 2020-01-07 | 鞍钢集团矿业有限公司 | Gradient-lifting-tree-based improved supplier quantitative scoring prediction algorithm |
Non-Patent Citations (3)
Title |
---|
(美)斯瓦米等."经典学习方法".《无线传感器网络:信号处理与通信》.2015,183-184. * |
"基于决策支持***平台的设备缺陷预测";陈拥军等;《浙江电力》(第6期);1-4 * |
高顿财经研究院."训练机器学习模型的步骤".《CFA 二级中文教材》.2019, * |
Also Published As
Publication number | Publication date |
---|---|
CN111259953A (en) | 2020-06-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111259953B (en) | Equipment defect time prediction method based on capacitive equipment defect data | |
CN109978079A (en) | A kind of data cleaning method of improved storehouse noise reduction self-encoding encoder | |
CN111325403B (en) | Method for predicting residual life of electromechanical equipment of highway tunnel | |
CN107561997B (en) | A kind of power equipment state monitoring method based on big data decision tree | |
CN113642754B (en) | Complex industrial process fault prediction method based on RF noise reduction self-coding information reconstruction and time convolution network | |
CN110334866B (en) | Power equipment fault probability prediction method and system considering insulation defect type and fault correlation | |
CN111966076A (en) | Fault positioning method based on finite-state machine and graph neural network | |
CN115563563A (en) | Fault diagnosis method and device based on transformer oil chromatographic analysis | |
CN111340110B (en) | Fault early warning method based on industrial process running state trend analysis | |
CN112434390B (en) | PCA-LSTM bearing residual life prediction method based on multi-layer grid search | |
CN109472097B (en) | Fault diagnosis method for online monitoring equipment of power transmission line | |
CN112926627A (en) | Equipment defect time prediction method based on capacitive equipment defect data | |
CN113780420B (en) | GRU-GCN-based method for predicting concentration of dissolved gas in transformer oil | |
CN111784061B (en) | Training method, device and equipment for power grid engineering cost prediction model | |
CN116842337A (en) | Transformer fault diagnosis method based on LightGBM (gallium nitride based) optimal characteristics and COA-CNN (chip on board) model | |
CN114397526A (en) | Power transformer fault prediction method and system driven by state holographic sensing data | |
CN116379360A (en) | Knowledge migration-based hydrogen-doped natural gas pipeline damage prediction method and system | |
CN114580934A (en) | Early warning method for food detection data risk based on unsupervised anomaly detection | |
Yang et al. | Teacher–Student Uncertainty Autoencoder for the Process-Relevant and Quality-Relevant Fault Detection in the Industrial Process | |
CN113283546B (en) | Furnace condition abnormity alarm method and system of heating furnace integrity management centralized control device | |
Yang et al. | A Fault Identification Method for Electric Submersible Pumps Based on DAE‐SVM | |
CN117034169A (en) | Power grid main transformer equipment abnormal state prediction method based on time sequence causality network | |
CN117056865A (en) | Method and device for diagnosing operation faults of machine pump equipment based on feature fusion | |
CN112380763A (en) | System and method for analyzing reliability of in-pile component based on data mining | |
CN117332330A (en) | Transformer fault evaluation method and system based on multi-source data graph fusion model |
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 |