CN110334865A - A kind of electrical equipment fault rate prediction technique and system based on convolutional neural networks - Google Patents
A kind of electrical equipment fault rate prediction technique and system based on convolutional neural networks Download PDFInfo
- Publication number
- CN110334865A CN110334865A CN201910602682.6A CN201910602682A CN110334865A CN 110334865 A CN110334865 A CN 110334865A CN 201910602682 A CN201910602682 A CN 201910602682A CN 110334865 A CN110334865 A CN 110334865A
- Authority
- CN
- China
- Prior art keywords
- prps
- spectrum data
- neural networks
- convolutional neural
- electrical equipment
- 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.)
- Granted
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 47
- 102100025142 Beta-microseminoprotein Human genes 0.000 claims abstract description 59
- 101000576812 Homo sapiens Beta-microseminoprotein Proteins 0.000 claims abstract description 59
- 238000001228 spectrum Methods 0.000 claims abstract description 54
- 230000007547 defect Effects 0.000 claims abstract description 36
- 238000013528 artificial neural network Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 18
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000009413 insulation Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 238000013277 forecasting method Methods 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 4
- 230000004888 barrier function Effects 0.000 claims description 3
- 210000004218 nerve net Anatomy 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 9
- 238000001514 detection method Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 7
- 238000012360 testing method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 238000012502 risk assessment Methods 0.000 description 4
- 230000005611 electricity Effects 0.000 description 3
- 238000009394 selective breeding Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Operations Research (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Development Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The electrical equipment fault rate prediction technique based on convolutional neural networks that the invention discloses a kind of comprising training step and prediction steps, wherein training step includes: the case PRPS map that (1) collects power equipment;(2) the case PRPS spectrum data of collection is pre-processed;(3) the first convolution neural network module is constructed, and the first convolution neural network module is trained, so that its output is the corresponding defect type of case PRPS spectrum data;(4) data set of each defect type is constructed based on defect type;(5) corresponding each defect type constructs the respective classification of failure two submodule respectively, and wherein each failure two classification submodule is based on the second convolution neural network module and constructs;The second convolutional neural networks of training, so that each failure two is classified, submodule is based on the probability value to break down obtained by case PRPS spectrum data, and the judgement whether output power equipment breaks down.
Description
Technical field
The present invention relates to the failure prediction methods and system more particularly to a kind of electrical equipment fault probability in electric system
Prediction technique and system.
Background technique
Gas insulated combined electrical equipment (GIS) is used as key equipment in the power system, and degree of risk influences whole system
Safe operation.The characteristics of structure is complicated due to GIS, if it exists defect can develop as after failure, and then cause great damage
It loses, therefore, it is necessary to understand the operating condition of GIS, finds defect in time.Since there are when insulation defect by GIS, it may occur that put part
Electricity, therefore, it is possible to according to the failure rate of the pre- measurement equipment of Partial Discharge Detection data, for further risk assessment.Wind
Dangerous assessment result is probability of equipment failure and the consequence product that failure generates, since consequence may be set according to actual conditions, because
This, probability of equipment failure is to calculate key.
As shelf depreciation live detection technology continues to develop, electrification can be passed through when insulation defect occurs in GIS device
Whether the defect of detection data analytical equipment is serious.In shelf depreciation live detection technology, superfrequency detection method is widely applied,
GIS equipment partial discharge data can be collected by superfrequency detection method.When shelf depreciation occurs for GIS, different is exhausted
Failure rate caused by edge defect is not identical.
Based on this, it is expected that obtaining a kind of electrical equipment fault rate prediction technique, can use including Partial Discharge Detection
The data of data and device history information, by assuming that probability of malfunction meet between certain distribution characteristics or quantity of state it is existing
Failure rate is calculated in association transfer, to obtain accurate prediction result, risk assessment to power equipment and sets
Standby maintenance has positive effect.
Summary of the invention
The electrical equipment fault rate prediction technique based on convolutional neural networks that one of the objects of the present invention is to provide a kind of,
The electrical equipment fault rate prediction technique can predict equipment failure rate based on convolutional neural networks according to different defect classifications.
According to foregoing invention purpose, the present invention proposes a kind of electrical equipment fault rate prediction side based on convolutional neural networks
Method comprising training step and prediction steps, in which:
Training step includes:
(1) power equipment is collected away from the case PRPS map number in the first designated time period before the setting scheduled date
According to the case PRPS spectrum data in the second designated time period;Second designated time period is than the first specified time segment length;
(2) the case PRPS spectrum data of collection is pre-processed;
(3) construct the first convolution neural network module, by the first designated time period by pretreated case PRPS
Spectrum data is trained the first convolution neural network module as input, so that its output is case PRPS spectrum data
Corresponding defect type;
(4) data set based on the corresponding defect type building N kind defect type of each case PRPS spectrum data;
(5) corresponding N kind defect type constructs N number of failure two respectively and classifies submodule, and wherein each failure two is classified son
Module is based on the second convolution neural network module and constructs;By the process in the second designated time period corresponding with each data set
Pretreated case PRPS spectrum data is trained each failure two classification submodule as input, so that each failure two is classified
The third specified time after setting the scheduled date that submodule is characterized based on input case PRPS spectrum data therein
The probability value that breaks down in section, and whether output power equipment is sent out setting in the third designated time period after the scheduled date
The judgement of raw failure;
Prediction steps include:
(a) acquisition power equipment away from the first designated time period of the date foregoing description to be predicted PRPS spectrum data and
PRPS spectrum data in second designated time period;
(b) PRPS spectrum data is pre-processed;
(c) the first convolutional neural networks, the first convolution nerve net will be inputted by pretreated PRPS spectrum data
Network exports the corresponding defect type of PRPS spectrum data;
(d) by being inputted and the defect type by pretreated PRPS spectrum data in corresponding second designated time period
Corresponding failure two is classified in submodule, then the failure two classification submodule exports the third specified time after the date to be predicted
The probability value that power equipment in section breaks down.
In technical solutions according to the invention, using the case PRPS spectrum data of collection to the first convolutional neural networks
Module and the second convolution neural network module carry out deep learning, obtain defect classification, and finally obtain the hair of power equipment
The electrical equipment fault rate prediction technique of the probability value (i.e. prediction rate) of raw failure, this case can be to avoid artificial selection probability point
Cloth, and accurate prediction result can be obtained, so as to the risk assessment for further power equipment.
It should be noted that in above scheme, in step (5), as probability value > 50%, it is believed that event occurs
Barrier.
Further, in the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks,
In step (2) and step (b), pretreatment, which is included at least, carries out linear normalization to PRPS spectrum data.
Further, in the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks, electricity
Power equipment includes at least GIS device, and the insulation defect number of types N=4 of GIS device.
Further, in the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks, the
One convolution neural network module includes 1 input layer, 5 convolutional layers, 5 pond layers, 1 full articulamentum and 1 output category
Layer.
Further, in electrical equipment fault probability forecasting method of the present invention, to the first convolutional neural networks
It includes: using cross entropy cost function that module, which is trained, and using Adam method, (Adam method is proposed by Kingma and Lei Ba, is made
For a kind of random targets function optimization algorithm, the ART network based on low-order moment) update model parameter;Wherein activation primitive is adopted
With unsaturated nonlinear function;Output category layer uses Softmax classifier.
Further, in the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks, the
Two convolutional neural networks modules include 1 input layer, 4 convolutional layers, 4 pond layers, 1 full articulamentum and 1 output category
Layer.
Further, in electrical equipment fault probability forecasting method of the present invention, to the second convolutional neural networks
It includes: to update model parameter using stochastic gradient descent method using cross entropy cost function that module, which is trained,;Wherein activate letter
Number is using unsaturated nonlinear function;Output category layer uses Softmax classifier.
Further, in the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks, the
One designated time period is one month;And the second designated time period is at least three months.
Further, the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks, third
Designated time period is one month.
Correspondingly, another object of the present invention is to provide a kind of electrical equipment fault rate based on volume machine neural network is pre-
Examining system, the electrical equipment fault rate prediction technique can be based on the pre- measurement equipment of convolutional neural networks according to different defect classifications
Failure rate.
According to foregoing invention purpose, the present invention proposes a kind of electrical equipment fault rate prediction system based on convolutional neural networks
The step of uniting, executing above-mentioned electrical equipment fault rate prediction technique.
Electrical equipment fault rate prediction technique and system of the present invention based on convolutional neural networks has following institute
The advantages of stating and the utility model has the advantages that
Electrical equipment fault rate prediction technique of the present invention can be to avoid artificial selection probability distribution, and can obtain
Accurate prediction result is obtained, probabilistic forecasting effect is improved, is highly suitable for carrying out further risk to power equipment
Assessment.
In addition, electrical equipment fault probabilistic forecasting system of the present invention also has above advantages and beneficial to effect
Fruit.
Detailed description of the invention
Fig. 1 is that the principle of the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks is illustrated
Figure.
Fig. 2 is the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks in some embodiment party
The model schematic of the first convolution neural network module in formula.
Fig. 3 is the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks in some embodiment party
The model schematic of the classification submodule of failure two in formula.
Specific embodiment
Below in conjunction with Figure of description and specific embodiment to the electricity of the present invention based on convolutional neural networks
Power equipment failure rate prediction technique and system make further explanation, however the explanation and illustration is not to of the invention
Technical solution constitutes improper restriction.
Event of the electrical equipment fault rate forecasting system based on convolutional neural networks to power equipment in present embodiment
Before barrier situation is predicted, it is necessary first to be trained, training step includes:
(1) power equipment is collected away from the case PRPS map number in the first designated time period before the setting scheduled date
According to the case PRPS spectrum data in the second designated time period;Second designated time period is than the first specified time segment length;
(2) the case PRPS spectrum data of collection is pre-processed;
(3) construct the first convolution neural network module, by the first designated time period by pretreated case PRPS
Spectrum data is trained the first convolution neural network module as input, so that its output is case PRPS spectrum data
Corresponding defect type.When constructing the first convolution neural network module, the first convolution neural network module includes 1 input layer, 5
A convolutional layer, 5 pond layers, 1 full articulamentum and 1 output category layer, wherein the case PRPS in the first designated time period
The corresponding defect type of spectrum data includes suspended discharge, insulation class electric discharge, four kinds of tip corona, particulate electric discharge defect (i.e. N=
4);
(4) data set based on the corresponding defect type building N kind defect type of each case PRPS spectrum data;
(5) corresponding N kind defect type constructs N number of failure two respectively and classifies submodule, and wherein each failure two is classified son
Module is based on the second convolution neural network module and constructs;By the process in the second designated time period corresponding with each data set
Pretreated case PRPS spectrum data is trained each failure two classification submodule as input, so that each failure two is classified
The third specified time after setting the scheduled date that submodule is characterized based on input case PRPS spectrum data therein
The probability value that breaks down in section, and whether output power equipment is sent out setting in the third designated time period after the scheduled date
The judgement of raw failure.
When constructing the second convolution neural network module, the second convolution neural network module includes 1 input layer, 4 convolution
Layer, 4 pond layers, 1 full articulamentum and 1 output category layer.
It should be noted that case PRPS spectrum data may come from substation field collection in training step
GIS device fault case.
And in step (2), since case PRPS spectrum data is the two-dimensional matrix that format is 72 × 50, thus, it is pre- to locate
Reason includes carrying out linear normalization to PRPS spectrum data.Normalized can use following formula:
In formula: initial data in x representing matrix, xminIndicate minimum value in two-dimensional matrix, xmaxIt indicates in two-dimensional matrix most
Big value, x ' indicate the data after linear normalization.
It should be noted that when being trained to the first convolutional neural networks, cross entropy can be used in step (3)
Cost function, updates model parameter using Adam method, updates model parameter using stochastic gradient descent method;Wherein, first layer is rolled up
Lamination convolution kernel is having a size of 5 × 5, and for remaining convolutional layer convolution kernel having a size of 3 × 3, activation primitive uses unsaturated nonlinear function,
The size of each pond layer is 1 × 2, and using maximum down-sampled method, the neuron number of full articulamentum is 1024, output category
Layer uses Softmax classifier.
In step (5), when being trained to the second convolutional neural networks, cross entropy cost function can be used, is utilized
Adam method updates model parameter, updates model parameter using stochastic gradient descent method;Wherein, first layer and second layer convolutional layer
Convolution kernel is having a size of 5 × 5, and for remaining convolutional layer convolution kernel having a size of 3 × 3, activation primitive is each using unsaturated nonlinear function
The size of pond layer is 1 × 2, and using maximum down-sampled method, the neuron number of full articulamentum is 512, and output category layer is adopted
With Softmax classifier.
The first convolutional neural networks and the second convolution nerve net of training completion can be obtained by above-mentioned training step
Network.Based on the first convolutional neural networks and nervus opticus network that training is completed, by collected power equipment away to be predicted
The PRPS spectrum data in PRPS spectrum data and the second designated time period before date in the first designated time period measures in advance
The probability value that the power equipment in third designated time period after to the date to be predicted breaks down.In the present embodiment,
Scheduled date and date to be predicted can be the different dates, and the first designated time period can be one month, second it is specified when
Between section be at least three months, third designated time period can be one month.
Its prediction principle is as shown in Figure 1.Fig. 1 is the electrical equipment fault rate of the present invention based on convolutional neural networks
The schematic illustration of prediction technique.
As shown in Figure 1, when predicting electrical equipment fault rate, prediction steps include:
(a) PRPS spectrum data and second of the acquisition power equipment before away from the date to be predicted in first designated time period
PRPS spectrum data in designated time period;
(b) PRPS spectrum data is pre-processed;
(c) the first convolutional neural networks will be inputted by pretreated PRPS spectrum data, the first convolutional neural networks are defeated
The corresponding defect type of PRPS spectrum data out;
(d) by being inputted and the defect type by pretreated PRPS spectrum data in corresponding second designated time period
Corresponding failure two is classified in submodule, then it is specified to export the third after the date to be predicted for the failure two classification submodule
The probability value that power equipment in period breaks down.
It should be noted that in the prediction step, PRPS map can be obtained directly by live live detection data.
And in step (b), pretreatment includes carrying out linear normalization to PRPS spectrum data.Normalized can adopt
With following formula:
In formula: initial data in x representing matrix, xminIndicate minimum value in two-dimensional matrix, xmaxIt indicates in two-dimensional matrix most
Big value, x ' indicate the data after linear normalization.
Fig. 2 is the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks in some embodiment party
The model schematic of the first convolution neural network module in formula.
As shown in Fig. 2, in the first convolution neural network module instruction will be divided by pretreatment case PRPS spectrum data
Practice sample set data and test sample collection data, wherein training sample set data account for the 80% of total amount, test sample collection data
Account for the 20% of total amount.Training sample set data are trained the first convolution neural network module, and micro- using Adam progress
It adjusts, optimizes the parameter of the first convolution neural network module.And using the data in test sample collection data as test data, with complete
At the training of the first convolution neural network module.
Fig. 3 is the electrical equipment fault rate prediction technique of the present invention based on convolutional neural networks in some embodiment party
The model schematic of the classification submodule of failure two in formula.
As shown in figure 3, corresponding four kinds of defect types construct four failures two respectively and divide in the classification submodule of failure two
Class submodule, each failure two classification submodule are based on the second convolution neural network module and construct.It will be by pretreatment
Case PRPS spectrum data is divided into training sample set data and test sample collection data, wherein training sample set data Zhan is total
The 80% of amount, test sample collection data account for the 20% of total amount.Using training sample set data to the second convolution neural network module
It is trained, the second convolution neural network module is optimized by minimizing error, optimizes the second convolutional neural networks mould
The parameter of block.
It should be noted that since breakdown judge is two points of problems, it is therefore contemplated that when a failure occurs it, output point
The judgment value of class layer is 1, i.e., the data fault label is 1, and when not breaking down, the judgment value of output category layer is 0, the number
It is 0 according to faulty tag.
In summary as can be seen that electrical equipment fault rate prediction technique of the present invention can be general to avoid artificial selection
Rate distribution, and accurate prediction result can be obtained, probabilistic forecasting effect is improved, is highly suitable for power equipment
Carry out further risk assessment.
In addition, electrical equipment fault probabilistic forecasting system of the present invention also has above advantages and beneficial to effect
Fruit.
It should be noted that prior art part is not limited to given by present specification in protection scope of the present invention
Embodiment, all prior arts not contradicted with the solution of the present invention, including but not limited to first patent document, formerly
Public publication, formerly openly use etc., it can all be included in protection scope of the present invention.
In addition, in this case in the combination of each technical characteristic and unlimited this case claim documented combination or
It is combination documented by specific embodiment, all technical characteristics that this case is recorded can be freely combined in any way
Or combine, unless generating contradiction between each other.
It is also to be noted that embodiment enumerated above is only specific embodiments of the present invention.The obvious present invention is not
Above embodiments are confined to, the similar variation or deformation made therewith are that those skilled in the art can be from present disclosure
It immediately arrives at or is easy to just to associate, be within the scope of protection of the invention.
Claims (10)
1. a kind of electrical equipment fault rate prediction technique based on convolutional neural networks comprising training step and prediction steps,
It is characterized by:
The training step includes:
(1) collect power equipment away from setting the scheduled date before the first designated time period in case PRPS spectrum data and
Case PRPS spectrum data in second designated time period;Second designated time period is than the first specified time segment length;
(2) the case PRPS spectrum data of collection is pre-processed;
(3) construct the first convolution neural network module, by the first designated time period by pretreated case PRPS map
Data are trained the first convolution neural network module as input, so that its output is that case PRPS spectrum data is corresponding
Defect type;
(4) data set based on the corresponding defect type building N kind defect type of each case PRPS spectrum data;
(5) corresponding N kind defect type constructs the N number of classification of failure two submodule respectively, wherein each failure two classification submodule
It is based on the second convolution neural network module and constructs;By the pre- place of process in the second designated time period corresponding with each data set
The case PRPS spectrum data of reason is trained each failure two classification submodule as input, the submodule so that each failure two is classified
Block is being set in the third designated time period after the scheduled date based on what input case PRPS spectrum data therein was characterized
The probability value to break down, and output power equipment is setting event whether occurs in the third designated time period after the scheduled date
The judgement of barrier;
The prediction steps include:
(a) acquisition power equipment is away from PRPS spectrum data in the first designated time period of the date foregoing description to be predicted and described
PRPS spectrum data in second designated time period;
(b) PRPS spectrum data is pre-processed;
(c) first convolutional neural networks, the first convolution nerve net will be inputted by pretreated PRPS spectrum data
Network exports the corresponding defect type of PRPS spectrum data;
(d) by being inputted and the defect type by pretreated PRPS spectrum data in corresponding second designated time period
Corresponding failure two is classified in submodule, then it is specified to export the third after the date to be predicted for the failure two classification submodule
The probability value that power equipment in period breaks down.
2. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that
In step (2) and step (b), the pretreatment, which is included at least, carries out linear normalization to PRPS spectrum data.
3. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that institute
It states power equipment and includes at least GIS device, and the insulation defect number of types N=4 of GIS device.
4. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that institute
Stating the first convolution neural network module includes 1 input layer, 5 convolutional layers, 5 pond layers, 1 full articulamentum and 1 output
Classification layer.
5. electrical equipment fault probability forecasting method as claimed in claim 4, which is characterized in that the first convolutional neural networks
It includes: to update model parameter using Adam method using cross entropy cost function that module, which is trained,;Wherein activation primitive is not using
Saturation nonlinearity function;The output category layer uses Softmax classifier.
6. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that institute
Stating the second convolution neural network module includes 1 input layer, 4 convolutional layers, 4 pond layers, 1 full articulamentum and 1 output
Classification layer.
7. electrical equipment fault probability forecasting method as claimed in claim 6, which is characterized in that the second convolutional neural networks
It includes: to update model parameter using stochastic gradient descent method using cross entropy cost function that module, which is trained,;Wherein activate letter
Number is using unsaturated nonlinear function;The output category layer uses Softmax classifier.
8. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that institute
Stating the first designated time period is one month;And the second designated time period is at least three months.
9. the electrical equipment fault rate prediction technique based on convolutional neural networks as described in claim 1, which is characterized in that institute
Stating third designated time period is one month.
10. a kind of electrical equipment fault rate forecasting system based on convolutional neural networks, which is characterized in that it is executed as right is wanted
The step of seeking electrical equipment fault rate prediction technique described in any one of 1-9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910602682.6A CN110334865B (en) | 2019-07-05 | 2019-07-05 | Power equipment fault rate prediction method and system based on convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910602682.6A CN110334865B (en) | 2019-07-05 | 2019-07-05 | Power equipment fault rate prediction method and system based on convolutional neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110334865A true CN110334865A (en) | 2019-10-15 |
CN110334865B CN110334865B (en) | 2023-04-18 |
Family
ID=68144227
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910602682.6A Active CN110334865B (en) | 2019-07-05 | 2019-07-05 | Power equipment fault rate prediction method and system based on convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110334865B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110703057A (en) * | 2019-11-04 | 2020-01-17 | 国网山东省电力公司电力科学研究院 | Power equipment partial discharge diagnosis method based on data enhancement and neural network |
CN111413937A (en) * | 2020-04-07 | 2020-07-14 | 浙江工业大学 | Remote monitoring and predictive maintenance system for broaching equipment and fault prediction method |
CN111965476A (en) * | 2020-06-24 | 2020-11-20 | 国网江苏省电力有限公司淮安供电分公司 | Low-voltage diagnosis method based on graph convolution neural network |
CN112200790A (en) * | 2020-10-16 | 2021-01-08 | 鲸斛(上海)智能科技有限公司 | Cloth defect detection method, device and medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103745119A (en) * | 2014-01-22 | 2014-04-23 | 浙江大学 | Oil-immersed transformer fault diagnosis method based on fault probability distribution model |
CN104573740A (en) * | 2014-12-22 | 2015-04-29 | 山东鲁能软件技术有限公司 | SVM classification model-based equipment fault diagnosing method |
CN109145961A (en) * | 2018-07-30 | 2019-01-04 | 上海交通大学 | A kind of mode identification method and system of unstructured Partial Discharge Data |
CN109212392A (en) * | 2018-09-25 | 2019-01-15 | 上海交通大学 | Recognition methods, system and the Caffe convolutional neural networks of direct current cables shelf depreciation defect failure |
CN109669087A (en) * | 2019-01-31 | 2019-04-23 | 国网河南省电力公司 | A kind of method for diagnosing fault of power transformer based on Multi-source Information Fusion |
-
2019
- 2019-07-05 CN CN201910602682.6A patent/CN110334865B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103745119A (en) * | 2014-01-22 | 2014-04-23 | 浙江大学 | Oil-immersed transformer fault diagnosis method based on fault probability distribution model |
CN104573740A (en) * | 2014-12-22 | 2015-04-29 | 山东鲁能软件技术有限公司 | SVM classification model-based equipment fault diagnosing method |
CN109145961A (en) * | 2018-07-30 | 2019-01-04 | 上海交通大学 | A kind of mode identification method and system of unstructured Partial Discharge Data |
CN109212392A (en) * | 2018-09-25 | 2019-01-15 | 上海交通大学 | Recognition methods, system and the Caffe convolutional neural networks of direct current cables shelf depreciation defect failure |
CN109669087A (en) * | 2019-01-31 | 2019-04-23 | 国网河南省电力公司 | A kind of method for diagnosing fault of power transformer based on Multi-source Information Fusion |
Non-Patent Citations (4)
Title |
---|
QINQIN ZHANG等: ""Fault Identification Based on PD Ultrasonic Signal Using RNN, DNN and CNN"", 《2018 CONDITION MONITORING AND DIAGNOSIS》 * |
ZHU YUFENG等: ""Partial discharge pattern recognition of DC XLPE cables based on convolutional neural network"", 《2018 CONDITION MONITORING AND DIAGNOSIS(CMD)》 * |
唐炬 等: ""基于分解组分分析的SF6设备绝缘故障诊断方法与技术的研究现状"", 《电工技术学报》 * |
宋辉 等: ""运行条件下GIS局部放电严重程度评估方法"", 《中国电机工程学报》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110703057A (en) * | 2019-11-04 | 2020-01-17 | 国网山东省电力公司电力科学研究院 | Power equipment partial discharge diagnosis method based on data enhancement and neural network |
CN110703057B (en) * | 2019-11-04 | 2022-02-25 | 国网山东省电力公司电力科学研究院 | Power equipment partial discharge diagnosis method based on data enhancement and neural network |
CN111413937A (en) * | 2020-04-07 | 2020-07-14 | 浙江工业大学 | Remote monitoring and predictive maintenance system for broaching equipment and fault prediction method |
CN111965476A (en) * | 2020-06-24 | 2020-11-20 | 国网江苏省电力有限公司淮安供电分公司 | Low-voltage diagnosis method based on graph convolution neural network |
CN112200790A (en) * | 2020-10-16 | 2021-01-08 | 鲸斛(上海)智能科技有限公司 | Cloth defect detection method, device and medium |
Also Published As
Publication number | Publication date |
---|---|
CN110334865B (en) | 2023-04-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110334865A (en) | A kind of electrical equipment fault rate prediction technique and system based on convolutional neural networks | |
Zio | Challenges in the vulnerability and risk analysis of critical infrastructures | |
CN110807550B (en) | Distribution transformer overload recognition and early warning method based on neural network and terminal equipment | |
Xu et al. | A classification approach for power distribution systems fault cause identification | |
CN104700321B (en) | A kind of power transmission and transformation equipment state operation trend analysis method | |
CN102779230B (en) | State analysis and maintenance decision judging method of power transformer system | |
Ghiasi et al. | Resiliency/cost-based optimal design of distribution network to maintain power system stability against physical attacks: A practical study case | |
CN110334866A (en) | Consider the electrical equipment fault probability forecasting method and system of insulation defect classification and fault correlation | |
CN109492857A (en) | A kind of distribution network failure risk class prediction technique and device | |
Arif et al. | Distribution network outage data analysis and repair time prediction using deep learning | |
CN104200288A (en) | Equipment fault prediction method based on factor-event correlation recognition | |
CN110334948A (en) | Power equipment shelf depreciation Severity method and system based on characteristic quantity prediction | |
CN106447227A (en) | Urban power grid abnormal state analyzing method and system | |
CN116231871B (en) | Power grid situation supervision method, system and storage medium based on digital twinning | |
Sun et al. | An ensemble system to predict the spatiotemporal distribution of energy security weaknesses in transmission networks | |
CN115730749B (en) | Power dispatching risk early warning method and device based on fusion power data | |
Wang et al. | Data-driven prediction method for characteristics of voltage sag based on fuzzy time series | |
CN116523294A (en) | Knowledge graph-based power grid risk early warning and decision system, method and equipment | |
CN109685344A (en) | A kind of power equipment O&M strategy determines method, apparatus and storage medium | |
CN115409264A (en) | Power distribution network emergency repair stagnation point position optimization method based on feeder line fault prediction | |
CN104680239A (en) | Distribution network maintenance scheme optimization method and device based on element failure model | |
CN113469457B (en) | Power transmission line fault probability prediction method integrating attention mechanism | |
Sun et al. | A multi-model-integration-based prediction methodology for the spatiotemporal distribution of vulnerabilities in integrated energy systems under the multi-type, imbalanced, and dependent input data scenarios | |
CN104732107B (en) | Using medium parameter as the residual life of sleeve of transformer Forecasting Methodology for assessing parameter | |
Li et al. | A probabilistic data-driven method for response-based load shedding against fault-induced delayed voltage recovery in power systems |
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 |