CN112232366B - Electrical equipment fault early warning method and system based on RFID monitoring - Google Patents
Electrical equipment fault early warning method and system based on RFID monitoring Download PDFInfo
- Publication number
- CN112232366B CN112232366B CN202010941484.5A CN202010941484A CN112232366B CN 112232366 B CN112232366 B CN 112232366B CN 202010941484 A CN202010941484 A CN 202010941484A CN 112232366 B CN112232366 B CN 112232366B
- Authority
- CN
- China
- Prior art keywords
- fault early
- warning information
- early warning
- temperature data
- training
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012544 monitoring process Methods 0.000 title claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 79
- 238000013528 artificial neural network Methods 0.000 claims abstract description 46
- 238000007781 pre-processing Methods 0.000 claims abstract description 16
- 230000006870 function Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 8
- 241001334134 Rugopharynx epsilon Species 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000005672 electromagnetic field Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
Classifications
-
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K1/00—Details of thermometers not specially adapted for particular types of thermometer
- G01K1/02—Means for indicating or recording specially adapted for thermometers
-
- 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
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/302—Contactless testing
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention relates to an electrical equipment fault early warning method and system based on RFID monitoring, wherein the method specifically comprises the following steps: acquiring a time sequence temperature data set of the electrical equipment through an RFID temperature acquisition system, preprocessing, inputting a trained denoising self-coding network and a long-short-time memory neural network to respectively obtain first fault early-warning information and first prediction fault early-warning information, and inputting a trained Xgboost model to obtain a fault early-warning grade; the method comprises the steps of acquiring historical time sequence temperature data and corresponding historical fault early warning information of electrical equipment, preprocessing the historical time sequence temperature data, taking the preprocessed historical time sequence temperature data and corresponding historical fault early warning information as a training set by a denoising self-coding network and a long-short-time memory neural network, and taking the historical fault early warning information and corresponding fault early warning level as the training set by an Xgboost model. Compared with the prior art, the method has the advantages of over fitting avoidance, high precision and the like.
Description
Technical Field
The invention relates to an electrical equipment monitoring technology, in particular to an electrical equipment fault early warning method and system based on RFID monitoring.
Background
In the power system, the safe and stable operation of the power equipment is the basis of the stability of the power system. However, in an actual power system, many factors such as loose connection, poor contact, magnetic leakage, overcurrent and the like of electrical equipment can cause equipment overheat to cause equipment failure. Temperature detection is one of the main ways to determine whether an electrical device is abnormal. Aiming at the power equipment temperature detection system based on the ultrahigh frequency radio frequency identification technology, the power equipment temperature time sequence data obtained by the system detection is used for realizing the abnormal trend identification of the time sequence temperature data by fusing the deep learning technology, so that the prediction and the early warning of the power equipment faults are realized.
The temperature acquisition data of the power equipment temperature acquisition system based on the RFID technology has the following problems: abnormal data such as data point missing, data dislocation and the like exists in the data set due to collision between the multi-tag and the multi-reader; the RFID realizes data communication and transmission based on reverse electromagnetic waves, the working environment of power equipment in an actual scene is complex, and electromagnetic fields in an external environment cause interference to data transmission, so that data noise is generated; and when the RFID equipment is in failure, the temperature data is lack in a long time scale, and the change trend is abnormal. The temperature data has obvious time sequence characteristics, the current mainstream in the prediction of time sequence data is to adopt a recurrent neural network RNN, a convolutional neural network CNN or a long and short memory neural network LSTM to realize the prediction of time sequence data, and based on the characteristics, the traditional method of applying RNN, CNN, LSTM technology easily causes the problems of over fitting, low prediction precision and the like.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide the electric equipment fault early warning method and system based on RFID monitoring, which avoid over fitting and have high precision.
The aim of the invention can be achieved by the following technical scheme:
an electrical equipment fault early warning method based on RFID monitoring specifically comprises the following steps:
the time sequence temperature data set of the electrical equipment is acquired in real time through the RFID temperature acquisition system and preprocessed, the preprocessed time sequence temperature data set is input into the trained denoising self-coding network and the long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, the first fault early-warning information and the second fault early-warning information are input into the trained Xgboost model to obtain a fault early-warning level, fault prediction and early warning of the electrical equipment are achieved, larger economic loss of the electrical equipment due to sudden faults is avoided, and prediction accuracy is high.
The denoising self-coding network and the long-short-term memory neural network take the preprocessed historical time sequence temperature data and corresponding historical fault early warning information as training sets to train;
the fault early-warning information corresponding to the historical time sequence temperature data is divided into a plurality of fault early-warning levels according to the severity, the Xgboost model takes the historical fault early-warning information and the corresponding fault early-warning levels as a training set, a meta-learner is obtained after the training set is generalized, the fault early-warning levels are comprehensively judged according to the first fault early-warning information and the second fault early-warning information, and the accuracy is high.
Further, the preprocessing corrects the time sequence temperature data set through a neighbor propagation clustering algorithm, and the process specifically comprises the following steps:
301 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
302 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
303 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
304 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
305 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
Further, the newly obtained time sequence temperature data set, the first fault early warning information and the second fault early warning information are combined into a training set of the denoising self-coding network and the long-short-time memory neural network, and the denoising self-coding network and the long-short-time memory neural network are trained by using the new training set;
and combining the newly obtained first fault early-warning information, second fault early-warning information and fault early-warning level into a training set of the Xgboost model, and training the Xgboost model by utilizing the new training set, so that the denoising self-coding network, the long-short-time memory neural network and the Xgboost model can realize feedback correction, and parameters are continuously corrected to improve prediction accuracy.
Further, the self-coding network, the long-short-time memory neural network and the Xgboost model are trained by adopting a symmetrical embedded measurement learning method, and a formula of a trained loss function J is as follows:
wherein l mn ∈{0,1}，(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by sample x.
An electrical equipment fault early warning system based on RFID monitoring comprises a data acquisition module, a data processing module, a first prediction module, a second prediction module, a fault early warning module and a model training module:
the data acquisition module is used for acquiring a time sequence temperature data set of the electrical equipment through the RFID temperature acquisition system, and simultaneously acquiring historical fault early warning information and corresponding historical fault early warning grades of the electrical equipment, wherein the fault early warning grades are classified into a plurality of grades according to the severity of the fault early warning information;
the data processing module is used for preprocessing the acquired time sequence temperature data set;
the first prediction module is used for inputting the preprocessed time sequence temperature data set into a trained denoising self-coding network to obtain first fault early warning information;
the second prediction module is used for inputting the preprocessed time sequence temperature data set into the trained long-time and short-time memory neural network to obtain second fault early warning information;
the fault early-warning module is used for inputting the first fault early-warning information and the second fault early-warning information into a trained Xgboost model to obtain a fault early-warning grade,
the model training module is used for training the preprocessed historical time sequence temperature data set of the electrical equipment and corresponding fault early warning information as training sets of the denoising self-coding network and the long-short-term memory neural network; the model training module takes historical fault early warning information and corresponding fault early warning grades of the electrical equipment as a training set of the Xgboost model for training.
Further, the data processing module corrects the time sequence temperature data set through a neighbor propagation clustering algorithm, and the process specifically comprises the following steps:
801 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
802 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
803 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
804 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
805 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
Further, the model training module integrates the newly obtained time sequence temperature data set, the first fault early warning information and the second fault early warning information into a training set of the denoising self-coding network and the long-short-time memory neural network, and trains the denoising self-coding network and the long-short-time memory neural network by using the new training set;
the model training module combines the newly obtained first fault early warning information, the second fault early warning information and the fault early warning level into a training set of the Xgboost model, and trains the Xgboost model by using the new training set.
Further, the self-coding network, the long-short-time memory neural network and the Xgboost model are trained by adopting a symmetrical embedded measurement learning method, and a formula of a trained loss function J is as follows:
wherein l mn ∈{0,1}，(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by sample x.
Compared with the prior art, the invention has the following beneficial effects:
(1) The invention collects the time sequence temperature data set of the electrical equipment through the RFID temperature collection system, performs preprocessing, inputs the preprocessed time sequence temperature data set into the trained denoising self-coding network and the long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, and the Xgboost model comprehensively judges the fault early-warning level according to the first fault early-warning information and the second fault early-warning information;
(2) According to the method, the data is preprocessed after the time sequence temperature data set is acquired, the clustering is carried out by adopting a neighbor propagation clustering algorithm, abnormal data in the data are corrected, the overfitting is further avoided, and the prediction precision is improved;
(3) The method comprises the steps of combining a newly obtained time sequence temperature data set, first fault early warning information and second fault early warning information into a training set of a denoising self-coding network and a long-short-time memory neural network, and training the denoising self-coding network and the long-short-time memory neural network by using the new training set; the newly obtained first fault early-warning information, the second fault early-warning information and the fault early-warning level are combined into a training set of the Xgboost model, the Xgboost model is trained by utilizing the new training set, training samples can be continuously expanded, feedback correction is realized, and the prediction precision is high;
(4) According to the invention, a symmetrical embedded measurement learning method is adopted to train the self-coding network, the long-short-time memory neural network and the Xgboost model, so that the training process of the model can be further increased, the hidden time sequence characteristic of the temperature data is captured more carefully, and the fault early warning accuracy is further improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of a denoising self-encoding network;
FIG. 3 is a schematic diagram of a deep learning network;
fig. 4 is a schematic diagram of symmetrical embedding.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples. The present embodiment is implemented on the premise of the technical scheme of the present invention, and a detailed implementation manner and a specific operation process are given, but the protection scope of the present invention is not limited to the following examples.
Example 1
An electrical equipment fault early warning method based on RFID monitoring, as shown in fig. 1, 2 and 3, specifically comprises the following steps:
acquiring a time sequence temperature data set of the electrical equipment through an RFID temperature acquisition system, preprocessing, dividing the preprocessed time sequence temperature data set S into a training set E and a test set R, equally dividing the training set E into 2 subsets which are respectively marked as D 1 And D 2 The aggregate relationship is:
S＝E∪R
E＝D 1 ∪D 2
acquisition and D from history 1 Corresponding fault early warning information and D 2 Corresponding fault early warning information is utilized D 1 And D 1 Corresponding fault early warning information training denoising self-coding network AE, using D 2 And D 2 Training a long-short-time memory neural network LSTM according to the corresponding fault early warning information; dividing fault early-warning information corresponding to historical time sequence temperature data into 4 fault early-warning levels according to severity, wherein the first level is the most serious, acquiring the fault early-warning information of the electrical equipment and the fault early-warning level corresponding to the fault early-warning information according to a historical record, and training an Xgboost model by utilizing the fault early-warning information and the fault early-warning level corresponding to the fault early-warning information;
inputting the test set R into a trained denoising self-coding network and a long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, and inputting the first fault early-warning information and the second fault early-warning information into a trained Xgboost model to obtain a fault early-warning grade;
the new acquired time sequence temperature data set, the first fault early warning information and the second fault early warning information are combined into a training set of the denoising self-coding network and the long-short-time memory neural network, the denoising self-coding network and the long-short-time memory neural network are trained by utilizing the new training set, the new acquired first fault early warning information, the second fault early warning information and the fault early warning level are combined into a training set of the Xgboost model, the Xgboost model is trained by utilizing the new training set, feedback correction of AE, LSTM and the Xgboost model can be realized, and prediction accuracy is improved.
Wherein denoising self-encoding network stacks N e The hidden layer number of the long-short memory neural network is set to be N L 。
Because abnormal data such as data missing and noise interference can occur in sequential temperature data acquired by the RFID temperature acquisition system, correction is needed, and most clustering methods such as K-means clustering, fuzzy clustering and the like are all needed to achieve specified group data, in the embodiment, the preprocessing corrects the sequential temperature data set through a neighbor propagation clustering algorithm, so that abnormal data points can be identified, and the method specifically comprises the following steps:
301 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
302 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
303 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
304 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
305 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
The self-coding network, the long-short-time memory neural network and the Xgboost model are trained by adopting a symmetrical embedded measurement learning method, the sample pairs connected by the solid line in FIG. 4 belong to the same category, the sample pairs connected by the dotted line belong to different categories, and the formula of the trained loss function J is as follows:
wherein l mn ∈{0,1}，(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by the input sample x.
Example 2
An electrical equipment fault early warning system based on RFID monitoring comprises a data acquisition module, a data processing module, a first prediction module, a second prediction module, a fault early warning module and a model training module:
the data acquisition module is used for acquiring a time sequence temperature data set of the electrical equipment through the RFID temperature acquisition system, and simultaneously acquiring historical fault early warning information and corresponding historical fault early warning grades of the electrical equipment, wherein the fault early warning grades are classified into a plurality of grades according to the severity of the fault early warning information;
the data processing module is used for preprocessing the acquired time sequence temperature data set;
the first prediction module is used for inputting the preprocessed time sequence temperature data set into a trained denoising self-coding network to obtain first fault early warning information;
the second prediction module is used for inputting the preprocessed time sequence temperature data set into the trained long-short time memory neural network to obtain second fault early warning information;
the fault early-warning module is used for inputting the first fault early-warning information and the second fault early-warning information into the trained Xgboost model to obtain the fault early-warning level,
the model training module is used for training the preprocessed historical time sequence temperature data set of the electrical equipment and corresponding fault early warning information as training sets of the denoising self-coding network and the long-short-time memory neural network; the model training module takes the historical fault early warning information of the electrical equipment and the corresponding fault early warning level as a training set of the Xgboost model for training.
Wherein denoising self-encoding network stacks N e The hidden layer number of the long-short memory neural network is set to be N L 。
The pretreatment process specifically comprises the following steps: the data processing module corrects the time sequence temperature data set through a clustering algorithm.
The clustering algorithm is a neighbor propagation clustering algorithm, and the preprocessing process specifically comprises the following steps:
801 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
802 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
803 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
804 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
805 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
The model training module is used for combining the newly obtained time sequence temperature data set, the first fault early warning information and the second fault early warning information into training sets of the denoising self-coding network and the long-short-time memory neural network, and training the denoising self-coding network and the long-short-time memory neural network by using the new training sets;
the model training module combines the newly obtained first fault early warning information, the second fault early warning information and the fault early warning level into a training set of the Xgboost model, and trains the Xgboost model by using the new training set.
The self-coding network, the long-short-time memory neural network and the Xgboost model are trained by adopting a symmetrical embedded measurement learning method, and the training loss function J has the formula:
wherein l mn ∈{0,1}，(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by sample x,
embodiment 1 and embodiment 2 propose a method and a system for early warning faults of RFID (radio frequency identification) monitored electrical equipment, because abnormal data, including data noise and data missing, often appear in time sequence temperature data sets collected by an RFID temperature collection system, fault early warning grades are comprehensively analyzed through first fault early warning information and second fault early warning information, overfitting is avoided, and a result is more accurate.
The foregoing describes in detail preferred embodiments of the present invention. It should be understood that numerous modifications and variations can be made in accordance with the concepts of the invention by one of ordinary skill in the art without undue burden. Therefore, all technical solutions which can be obtained by logic analysis, reasoning or limited experiments based on the prior art by the person skilled in the art according to the inventive concept shall be within the scope of protection defined by the claims.
Claims (8)
1. An electrical equipment fault early warning method based on RFID monitoring is characterized by comprising the following steps:
acquiring a time sequence temperature data set of the electrical equipment through an RFID temperature acquisition system, preprocessing, inputting the preprocessed time sequence temperature data set into a trained denoising self-coding network and a long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, and inputting the first fault early-warning information and the second fault early-warning information into a trained Xgboost model to obtain a fault early-warning grade;
the method comprises the steps of acquiring historical time sequence temperature data and corresponding historical fault early warning information of electrical equipment, preprocessing the historical time sequence temperature data, and training a denoising self-coding network and a long-short-time memory neural network by taking the preprocessed historical time sequence temperature data and corresponding historical fault early warning information as a training set, wherein the denoising self-coding network and the long-short-time memory neural network are specifically:
dividing the preprocessed time sequence temperature data set S into a training set E and a test set R, equally dividing the training set E into 2 subsets which are respectively marked as D 1 And D 2 The aggregate relationship is:
S＝E∪R
E＝D 1 ∪D 2
acquisition and D from history 1 Corresponding fault early warning information and D 2 Corresponding fault early warning information is utilized D 1 And D 1 Corresponding fault early warning information training denoising self-coding network AE, using D 2 And D 2 Training a long-short-time memory neural network LSTM according to the corresponding fault early warning information; dividing fault early-warning information corresponding to historical time sequence temperature data into 4 fault early-warning levels according to severity, wherein the first level is the most serious, acquiring the fault early-warning information of the electrical equipment and the fault early-warning level corresponding to the fault early-warning information according to a historical record, and training an Xgboost model by utilizing the fault early-warning information and the fault early-warning level corresponding to the fault early-warning information;
inputting the test set R into a trained denoising self-coding network and a long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, and inputting the first fault early-warning information and the second fault early-warning information into a trained Xgboost model to obtain a fault early-warning grade;
the new time sequence temperature data set, the first fault early warning information and the second fault early warning information are combined into a training set of the denoising self-coding network and the long-short-time memory neural network, the denoising self-coding network and the long-short-time memory neural network are trained by utilizing the new training set, the new first fault early warning information, the second fault early warning information and the fault early warning level are combined into a training set of the Xgboost model, the Xgboost model is trained by utilizing the new training set, and feedback correction of AE, LSTM and the Xgboost model is realized.
2. The method for early warning of faults of electrical equipment based on RFID monitoring according to claim 1, wherein the preprocessing process is specifically as follows: the time series temperature dataset is modified by a clustering algorithm.
3. The electrical equipment fault early warning method based on RFID monitoring according to claim 2, wherein the clustering algorithm is a neighbor propagation clustering algorithm, and the preprocessing process specifically comprises the following steps:
301 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
302 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
303 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
304 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
305 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
4. The method for early warning of faults of electrical equipment based on RFID monitoring according to claim 1, wherein the self-coding network, the long-short-time memory neural network and the Xgboost model are trained by adopting a symmetrical embedded measurement learning method, and a formula of a trained loss function J is as follows:
wherein the method comprises the steps of(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by sample x.
5. An electrical equipment fault early warning system based on RFID monitoring, characterized by comprising:
the data acquisition module is used for acquiring a time sequence temperature data set of the electrical equipment through the RFID temperature acquisition system, and simultaneously acquiring historical fault early warning information and corresponding historical fault early warning grades of the electrical equipment, wherein the fault early warning grades are classified into a plurality of grades according to the severity of the fault early warning information;
the data processing module is used for preprocessing the acquired time sequence temperature data set;
the first prediction module is used for inputting the preprocessed time sequence temperature data set into a trained denoising self-coding network to obtain first fault early warning information;
the second prediction module is used for inputting the preprocessed time sequence temperature data set into the trained long-time and short-time memory neural network to obtain second fault early warning information;
the fault early-warning module is used for inputting the first fault early-warning information and the second fault early-warning information into the trained Xgboost model to obtain the fault early-warning level,
the model training module is used for training the preprocessed historical time sequence temperature data set of the electrical equipment and the corresponding fault early warning information as training sets of the denoising self-coding network and the long-short-time memory neural network, and training the historical fault early warning information of the electrical equipment and the corresponding fault early warning level as training sets of the Xgboost model, and specifically:
dividing the preprocessed time sequence temperature data set S into a training set E and a test set R, equally dividing the training set E into 2 subsets which are respectively marked as D 1 And D 2 The aggregate relationship is:
S＝E∪R
E＝D 1 ∪D 2
acquisition and D from history 1 Corresponding fault early warning information and D 2 Corresponding fault early warning information is utilized D 1 And D 1 Corresponding fault early warning information training denoising self-coding network AE, using D 2 And D 2 Training a long-short-time memory neural network LSTM according to the corresponding fault early warning information; dividing fault early-warning information corresponding to historical time sequence temperature data into 4 fault early-warning levels according to severity, wherein the first level is the most serious, acquiring the fault early-warning information of the electrical equipment and the fault early-warning level corresponding to the fault early-warning information according to a historical record, and training an Xgboost model by utilizing the fault early-warning information and the fault early-warning level corresponding to the fault early-warning information;
inputting the test set R into a trained denoising self-coding network and a long-short-time memory neural network to respectively obtain first fault early-warning information and second fault early-warning information, and inputting the first fault early-warning information and the second fault early-warning information into a trained Xgboost model to obtain a fault early-warning grade;
the new time sequence temperature data set, the first fault early warning information and the second fault early warning information are combined into a training set of the denoising self-coding network and the long-short-time memory neural network, the denoising self-coding network and the long-short-time memory neural network are trained by utilizing the new training set, the new first fault early warning information, the second fault early warning information and the fault early warning level are combined into a training set of the Xgboost model, the Xgboost model is trained by utilizing the new training set, and feedback correction of AE, LSTM and the Xgboost model is realized.
6. The system for early warning of failure of an electrical device based on RFID monitoring of claim 5, wherein the preprocessing process specifically comprises: the data processing module corrects the time sequence temperature data set through a clustering algorithm.
7. The electrical equipment fault early warning system based on RFID monitoring according to claim 6, wherein the clustering algorithm is a neighbor propagation clustering algorithm, and the preprocessing process specifically comprises:
801 For each temperature sequence T in the time-series temperature data set N ＝{L 1 ,L 2 ,…,L N Equal division into X periodic sequencesL is the temperature value, r.epsilon.1, X]Each->All are a dimension sequences, and each +.>Temperature trend sequence>Wherein,
802 (ii) each ofK formation by AP clustering r The clustering center of the ith group is marked as V i ，i∈[1，k r ],i＝N + ；
803 Calculating eachIs->With each V i Similarity sim of (2) n,i ，sim n,i ∈[0,1]；
804 Determining)For the membership degree of each group, the group with the largest membership degree is taken as the +.>A group of membership;
805 A membership threshold is set, when in a populationIs to be said +.>Consider an outlier and correct the outlier.
8. The system for early warning of failure of an electrical device based on RFID monitoring of claim 5, wherein the self-encoding network, the long-short-term memory neural network and the Xgboost model are trained by a symmetric embedded metric learning method, and the trained loss function J has the formula:
wherein the method comprises the steps of(x m ,x n ) For the sample pair, h (x) =max (0, x) is the cross-loss function, α 1 As penalty term, α 1 Is constant, d (x m ,x n ) Is (x) m ,x n ) F (x) is the feature extracted by sample x.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941484.5A CN112232366B (en) | 2020-09-09 | 2020-09-09 | Electrical equipment fault early warning method and system based on RFID monitoring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941484.5A CN112232366B (en) | 2020-09-09 | 2020-09-09 | Electrical equipment fault early warning method and system based on RFID monitoring |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112232366A CN112232366A (en) | 2021-01-15 |
| CN112232366B true CN112232366B (en) | 2024-04-16 |
Family
ID=74115453
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010941484.5A Active CN112232366B (en) | 2020-09-09 | 2020-09-09 | Electrical equipment fault early warning method and system based on RFID monitoring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112232366B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113570862B (en) * | 2021-07-28 | 2022-05-10 | 太原理工大学 | XGboost algorithm-based large traffic jam early warning method |
| CN114550902B (en) * | 2022-01-13 | 2023-06-20 | 重庆医科大学附属第一医院 | Hospital electric equipment management system and method based on RFID |
| CN115178752A (en) * | 2022-09-13 | 2022-10-14 | 广东银纳增材制造技术有限公司 | Fault early warning method and device for 3D printing metal powder production equipment |
| CN115901003B (en) * | 2022-11-23 | 2024-04-09 | 南京乾鑫电器设备有限公司 | Temperature monitoring method and system for environment-friendly gas switch cabinet |
| CN115574963B (en) * | 2022-12-08 | 2023-03-17 | 南京智联达科技有限公司 | Wireless temperature acquisition system and temperature acquisition method based on IOT technology |
| CN115860106A (en) * | 2022-12-23 | 2023-03-28 | 四川物通科技有限公司 | Intelligent transformer substation capacitor fault early warning method based on deep Q learning |
| CN115950557B (en) * | 2023-03-08 | 2023-05-09 | 深圳市特安电子有限公司 | Intelligent temperature compensation method based on pressure transmitter |
| CN117171517B (en) * | 2023-11-02 | 2024-01-26 | 无锡尚航数据有限公司 | Dynamic early warning method for operation fault risk of data center |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104697653A (en) * | 2015-02-10 | 2015-06-10 | 上海交通大学 | Temperature pre-warning system for key equipment of ultrahigh-pressure power distributing station based on web |
| WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
| CN108803576A (en) * | 2018-07-24 | 2018-11-13 | 广东工业大学 | A kind of fault early warning method and relevant apparatus of temperature control system |
| CN109840666A (en) * | 2017-11-29 | 2019-06-04 | 中国电力科学研究院有限公司 | A kind of model building method and system for predicting that the following Wind turbines break down |
| CN110210495A (en) * | 2019-05-21 | 2019-09-06 | 浙江大学 | The XGBoost soft-measuring modeling method extracted based on parallel LSTM self-encoding encoder behavioral characteristics |
| CN110737952A (en) * | 2019-09-17 | 2020-01-31 | 太原理工大学 | prediction method for residual life of key parts of mechanical equipment by combining AE and bi-LSTM |
| CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
| CN110909782A (en) * | 2019-11-15 | 2020-03-24 | 湘潭大学 | Method for diagnosing machine tool spindle fault based on multi-feature combined deep learning |
| CN111237134A (en) * | 2020-01-14 | 2020-06-05 | 上海电力大学 | Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model |
| CN111582298A (en) * | 2020-03-18 | 2020-08-25 | 宁波送变电建设有限公司永耀科技分公司 | Sensing abnormal data real-time detection method based on artificial intelligence |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI662422B (en) * | 2018-04-23 | 2019-06-11 | 國家中山科學研究院 | Air quality prediction method based on machine learning model |
-
2020
- 2020-09-09 CN CN202010941484.5A patent/CN112232366B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
| CN104697653A (en) * | 2015-02-10 | 2015-06-10 | 上海交通大学 | Temperature pre-warning system for key equipment of ultrahigh-pressure power distributing station based on web |
| CN109840666A (en) * | 2017-11-29 | 2019-06-04 | 中国电力科学研究院有限公司 | A kind of model building method and system for predicting that the following Wind turbines break down |
| CN108803576A (en) * | 2018-07-24 | 2018-11-13 | 广东工业大学 | A kind of fault early warning method and relevant apparatus of temperature control system |
| CN110210495A (en) * | 2019-05-21 | 2019-09-06 | 浙江大学 | The XGBoost soft-measuring modeling method extracted based on parallel LSTM self-encoding encoder behavioral characteristics |
| CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
| CN110737952A (en) * | 2019-09-17 | 2020-01-31 | 太原理工大学 | prediction method for residual life of key parts of mechanical equipment by combining AE and bi-LSTM |
| CN110909782A (en) * | 2019-11-15 | 2020-03-24 | 湘潭大学 | Method for diagnosing machine tool spindle fault based on multi-feature combined deep learning |
| CN111237134A (en) * | 2020-01-14 | 2020-06-05 | 上海电力大学 | Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model |
| CN111582298A (en) * | 2020-03-18 | 2020-08-25 | 宁波送变电建设有限公司永耀科技分公司 | Sensing abnormal data real-time detection method based on artificial intelligence |
Non-Patent Citations (3)
| Title |
|---|
| 基于并行的F⁃LSTM 模型及其在电力通信设备故障 预测中的应用;杨济海;武汉大学学报（理学版）;第65卷(第3期);263-268 * |
| 基于状态预测的风力发电机控制策略分析;陈鹏飞;中国优秀硕士学位论文全文数据库（电子期刊）;C042-272 * |
| 基于综合预警算法的开关柜温度预警系统设计;刘耀巍;刘向军;;电器与能效管理技术;20180315(05);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112232366A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112232366B (en) | Electrical equipment fault early warning method and system based on RFID monitoring | |
| Mao et al. | Online detection of bearing incipient fault with semi-supervised architecture and deep feature representation | |
| CN104809722A (en) | Electrical device fault diagnosis method based on infrared thermography | |
| CN116306320B (en) | Method and system for managing production quality of non-vascular lumen guide wire | |
| CN106912067B (en) | WSN wireless communication module fault diagnosis method based on fuzzy neural network | |
| CN111597991A (en) | Rehabilitation detection method based on channel state information and BilSTM-Attention | |
| Tang et al. | Specific emitter identification for IoT devices based on deep residual shrinkage networks | |
| CN117076955A (en) | Fault detection method and system for high-voltage frequency converter | |
| CN117437933A (en) | Feature cluster combination generation type learning-based unsupervised detection method for fault of voiceprint signal of transformer | |
| CN116915582A (en) | Diagnosis and analysis method and device for fault root cause of communication terminal | |
| CN117579101A (en) | Control method and system for carrier communication module | |
| CN114492534B (en) | Construction method and application of cross-size motor bearing fault diagnosis model | |
| CN114611409A (en) | Method and device for establishing power distribution terminal abnormity detection model | |
| CN114581699A (en) | Transformer state evaluation method based on deep learning model in consideration of multi-source information | |
| CN114048546A (en) | Graph convolution network and unsupervised domain self-adaptive prediction method for residual service life of aircraft engine | |
| Raghuvanshi et al. | Machine learning approach for anomaly detection in wireless sensor data | |
| Shia et al. | High rate outlier detection in wireless sensor networks: a comparative study | |
| CN116433049A (en) | Power consumption abnormality detection method based on fuzzy rough entropy | |
| Chen et al. | An Anomaly Detection Scheme with K-means aided Extended Isolation Forest in RSS-based Wireless Positioning System | |
| Lin et al. | DS Evidence Theory Information Fusion Methods and Applications | |
| CN113098640A (en) | Frequency spectrum anomaly detection method based on channel occupancy prediction | |
| Yang et al. | Deep ensemble siamese network for incremental signal classification | |
| Khan et al. | External Magnetic Interference Classification in Magnetostrictive Position Sensors using Neuro-Symbolic AI with Log-Likelihood Ratios | |
| CN111476383A (en) | Pump station unit state maintenance dynamic decision method | |
| CN116231631B (en) | Data-driven voltage sag user production scene identification method |
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 |