CN109558979A - Power equipments defect prediction technique and device - Google Patents
Power equipments defect prediction technique and device Download PDFInfo
- Publication number
- CN109558979A CN109558979A CN201811451610.8A CN201811451610A CN109558979A CN 109558979 A CN109558979 A CN 109558979A CN 201811451610 A CN201811451610 A CN 201811451610A CN 109558979 A CN109558979 A CN 109558979A
- Authority
- CN
- China
- Prior art keywords
- equipment
- neural network
- attribute data
- training sample
- power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 230000007547 defect Effects 0.000 title claims abstract description 55
- 238000012549 training Methods 0.000 claims abstract description 78
- 238000003062 neural network model Methods 0.000 claims abstract description 62
- 238000005457 optimization Methods 0.000 claims description 17
- 230000007812 deficiency Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 5
- 230000004888 barrier function Effects 0.000 claims 1
- 230000005540 biological transmission Effects 0.000 abstract description 20
- 230000005611 electricity Effects 0.000 abstract description 19
- 238000007689 inspection Methods 0.000 abstract description 4
- 238000012423 maintenance Methods 0.000 abstract description 4
- 238000013528 artificial neural network Methods 0.000 description 15
- 230000008569 process Effects 0.000 description 13
- 230000006870 function Effects 0.000 description 8
- 230000004913 activation Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000012937 correction Methods 0.000 description 5
- 230000003044 adaptive effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000014509 gene expression Effects 0.000 description 4
- 238000002834 transmittance Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000011478 gradient descent method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- 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
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of power equipments defect prediction technique and device, it is related to electric power equipment management technical field, this method comprises: obtaining the device attribute data of power equipment and estimating operation duration, device attribute data include device type, device location, equipment default service life and equipment voltage class;The prediction result that power equipment breaks down when operation duration is estimated in operation is determined by preset BP neural network model based on device attribute data.It is the model based on many factors training for influencing power equipment due to presetting BP neural network model, therefore, the accuracy rate for the prediction result that the power equipment determined by the preset model breaks down when operation duration is estimated in operation is higher than existing failure prediction method.In addition, the prediction result, to avoid the generation of electric power accident, can promote the safety of transmission line of electricity entirety for Maintenance of Electric Transmission Line unit reference, periodically to carry out inspection to high-risk equipment.
Description
Technical field
The present invention relates to electric power equipment management technical fields, in particular to a kind of power equipments defect prediction technique
And device.
Background technique
Transmission line of electricity is the main path of power transmission, and a large amount of power equipment is distributed on the transmission line.Electric power is set
For the influence due to factors such as self-characteristic, runing time and surrounding enviroment, it may occur that equipment deficiency, to cause electricity
Power accident.Therefore, in order to avoid electric power accident generation, improve the safety of transmission line of electricity, need to defects of power equipment into
Row prediction.
It is in the prior art based on a large amount of historical defect data, with mathematical statistics for defects of power equipment prediction
Method analysis and prediction power equipment future occurrence of equipment defect probability, according to prediction result make corresponding safety measure with
Exclude security risk.
But existing failure prediction method can only be directed to thunder by mathematical statistics methods such as regression analysis or variance analyses
Influence of certain single factors to power equipment in electrical activity intensity or other factors is analyzed, and that is to say existing defect
Prediction technique can only study a kind of influence of defect to power equipment, and cause the often more than one of power equipments defect because
Element, thus it is existing to be improved to the accuracy rate of power equipments defect prediction technique in transmission line of electricity.
Summary of the invention
It is an object of the present invention in view of the deficiency of the prior art, provide a kind of power equipments defect prediction side
Method and device, to improve the accuracy rate predicted defects of power equipment in transmission line of electricity, to improve the safety of transmission line of electricity
Property.
To achieve the above object, technical solution used in the embodiment of the present invention is as follows:
In a first aspect, the embodiment of the invention provides a kind of power equipments defect prediction techniques, comprising: obtain power equipment
Device attribute data and estimate operation duration, device attribute data include device type, device location, equipment preset the service life and
Equipment voltage class;Based on device attribute data, pass through preset BP (Back Propagation, backpropagation) neural network
Model determines the prediction result that power equipment breaks down when operation duration is estimated in operation.
Optionally, device attribute data further include the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, factory
Quotient's registered capital and manufacturer's quality.
Optionally, determining that power equipment is being run by preset BP neural network model based on device attribute data
Before the prediction result to break down when estimating operation duration, this method further include: when to device attribute data and estimating operation
Length is normalized.
Optionally, determining that power equipment is being run by preset BP neural network model based on device attribute data
Before the prediction result to break down when estimating operation duration, this method further include: obtain training sample, training sample includes more
The device history data of a power equipment, device history data include device attribute data, actual motion duration and fail result,
Fail result includes yes/no;Based on training sample, BP neural network model is trained.
Optionally, training sample, before being trained to BP neural network model, this method further include: to instruction are being based on
Practice sample to be sampled, obtain optimization training sample, optimization training sample includes the equipment that equal number of fail result is yes
The device history data that historical data and fail result are no;Correspondingly, it is based on training sample, BP neural network model is carried out
Training, comprising: based on optimization training sample, BP neural network model is trained.
Second aspect, the embodiment of the invention also provides a kind of power equipments defect prediction meanss, which includes: first
Module is obtained, for obtaining the device attribute data of power equipment and estimating operation duration, device attribute data include equipment class
Type, device location, equipment preset service life and equipment voltage class;Determining module, for being based on device attribute data, by default
BP neural network model, determine power equipment prediction result for breaking down when operation duration is estimated in operation.
Optionally, device attribute data further include the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, factory
Quotient's registered capital and manufacturer's quality.
Optionally, the device further include: processing module, for carrying out normalizing with operation duration is estimated to device attribute data
Change processing.
Optionally, device further include: second obtains module, and for obtaining training sample, training sample includes multiple electricity
The device history data of power equipment, device history data include device attribute data, actual motion duration and fail result, failure
It as a result include yes/no;Training module is trained BP neural network model for being based on training sample.
Optionally, the device further include: sampling module, for being sampled to training sample, acquisition optimization training sample,
Optimizing training sample includes the device history data that equal number of fail result is yes and the device history that fail result is no
Data;Correspondingly, the training module is specifically used for: based on optimization training sample, being trained to BP neural network model.
The beneficial effects of the present invention are: providing a kind of power equipments defect prediction technique, comprising: obtain power equipment
Device attribute data include device type, device location, equipment default service life and set with operation duration, device attribute data are estimated
Standby voltage class;Based on device attribute data, by preset BP neural network model, determine that power equipment estimates fortune in operation
The prediction result to break down when row duration.It is based on many factors for influencing power equipment due to presetting BP neural network model
Trained model, therefore, the power equipment determined by the preset model break down pre- when operation duration is estimated in operation
The accuracy rate for surveying result is higher than existing failure prediction method.In addition, the prediction result can be for Maintenance of Electric Transmission Line unit
With reference to, periodically to carry out inspection to high-risk equipment, so that the generation of electric power accident is avoided, the safety of promotion transmission line of electricity entirety
Property.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is a kind of flow chart of power equipments defect prediction technique provided in an embodiment of the present invention;
Fig. 2 is the schematic diagram of another power equipments defect prediction technique provided in an embodiment of the present invention;
Fig. 3 is a kind of BP neural network structural schematic diagram provided in an embodiment of the present invention;
Fig. 4 is that a kind of power equipments defect prediction result provided in an embodiment of the present invention shows report;
Fig. 5 is a kind of module diagram of power equipments defect prediction meanss provided in an embodiment of the present invention;
Fig. 6 is the module diagram of another power equipments defect prediction meanss provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.
Fig. 1 is please referred to, is a kind of flow chart of power equipments defect prediction technique provided in an embodiment of the present invention.
Step 101, the device attribute data of power equipment are obtained and estimate operation duration, device attribute data include equipment
Type, device location, equipment preset service life and equipment voltage class.
In order to avoid the generation of electric power accident, the safety of transmission line of electricity is improved, needs to set the electric power on transmission line of electricity
It is standby to carry out failure prediction.Due to causing many because being known as of power equipments defect on transmission line of electricity, available electric power is set
Standby device attribute data and operation duration is estimated, thus to there is the device attribute data run to whether estimating operation duration
It may break down and be predicted.
Wherein, power equipment is the equipment of distribution on the transmission line, and power equipment is divided into primary equipment and secondary device,
Primary equipment is the main body for constituting electric system, includes transformer, cable and generator for producing, conveying and distributing electric energy
Etc. equipment;Secondary device includes relay, control switch, control for being controlled primary equipment, adjusting, protecting and being detected
Instrument processed etc..
Device attribute data are the data for indicating power equipment attribute, and device attribute data include device type, set
Standby position, equipment preset service life and equipment voltage class.
Device type is the classification of power equipment, can be divided into power equipment based on the type of aforementioned middle power equipment
Different device types, such as device type can be transformer, cable and generator etc..
Device location is power equipment information the location of in transmission line of electricity, which believes comprising equipment longitude
Breath and equipment latitude information.
Equipment presets the service life for going out manufacturer's design that the service life is power equipment.
Equipment voltage class is the voltage rating rank series of electric system and power equipment, and is divided into safe electricity by rank
Pressure, low pressure, high pressure, super-pressure and extra-high voltage.
It estimates operation duration and runs for power equipment to sometime point and start the time point come into operation with power equipment
Between time difference.
It should be noted that estimating operation duration can be previously set by staff, operation is estimated for example, can set
Shi Changwei 1 year, then be after putting into operation 1 year to power equipment to power equipments defect prediction in embodiments of the present invention
Defect is predicted.
Certainly, in practical applications, other parameters relevant to power equipment can also be obtained as failure prediction
Reference index, for example, the meteorological data of power equipment present position can also be obtained.
Step 102, determine that power equipment is being run by preset BP neural network model based on device attribute data
The prediction result to break down when estimating operation duration.
It can only be to single when due to being handled using mathematical statistics methods such as regression analysis or the analysiies of covariance data
Defects of power equipment under the influence of factor is predicted that the accuracy rate of prediction is very low, therefore in order to improve the standard of failure prediction
True rate can determine what power equipment broke down when operation duration is estimated in operation by preset BP neural network model
Prediction result.
BP neural network is a kind of Multi-layered Feedforward Networks by Back Propagation Algorithm training, and BP neural network is when training
The weight and threshold value that network is constantly adjusted by backpropagation keep the error sum of squares of network minimum, wherein weight and threshold value
For two parameters for being used to Optimized model in BP neural network model.Basic BP neural network is divided into two processes: work letter
Number forward direction transmittance process and error signal back-propagation process.
BP neural network includes input layer, hidden layer and output layer, when calculating the output of BP neural network according to from defeated
Enter to the progress of the direction of output, this is the forward direction transmittance process of working signal;And the amendment of weight and threshold value from be output to input
Direction carry out, this be error signal back-propagation process.
It should be noted that default neural network model can be by being determined in advance to obtain, for example, can be by instructing in advance
Get the neural network model predicted for power equipments defect.
It the device attribute data for the power equipment that will acquire and estimates operation duration and inputs default neural network model, it can be with
It is (a kind of to be by what complicated data structure was transmitted to that artificial intelligence nerve net carries out analysis and treatment process by TensorFlow
System) system determines power equipment prediction result for breaking down when operation duration is estimated in operation.
Prediction result is to be run power equipment to estimating knot acquired in operation duration based on prediction neural network model
Fruit, prediction result include two kinds of results of yes/no, wherein are to indicate that defect occurs, defect does not occur for no expression.
Certainly, in practical applications, it can also determine that power equipment estimates operation duration in operation otherwise
When the prediction result that breaks down, for example predicted etc. by other types of machine learning model.
In embodiments of the present invention, a kind of power equipments defect prediction technique is provided, comprising: obtain setting for power equipment
For attribute data and operation duration is estimated, device attribute data include device type, device location, equipment default service life and equipment
Voltage class;Based on device attribute data, by preset BP neural network model, determine that power equipment estimates operation in operation
The prediction result to break down when duration.It is based on many factors instruction for influencing power equipment due to presetting BP neural network model
Experienced model, therefore, the prediction that the power equipment determined by the preset model breaks down when operation duration is estimated in operation
As a result accuracy rate is higher than existing failure prediction method.In addition, the prediction result can join for Maintenance of Electric Transmission Line unit
It examines, periodically to carry out inspection to high-risk equipment, to avoid the generation of electric power accident, promotes the safety of transmission line of electricity entirety.
Defects of power equipment is predicted in addition, the embodiment of the present invention uses BP neural network model, BP neural network model
With certain self study and adaptive ability, the rule of reason of output and outlet chamber can be automatically extracted in training, and will
Adaptive learning Content is stored in BP neural network model, while BP neural network model also has certain fault-tolerant energy
Power is still able to maintain accurate prediction result when network is locally destroyed.
It referring to figure 2., is the flow chart of another power equipments defect prediction technique provided in an embodiment of the present invention.
Step 201, training sample is obtained, training sample includes the device history data of multiple power equipments, device history
Data include device attribute data, actual motion duration and fail result, and fail result includes yes/no.
Since BP neural network model is obtained by inputting training sample in input layer, to BP neural network
Before model is trained, training sample can be first obtained.
Wherein, training sample is the input data of BP neural network mode input layer, in embodiments of the present invention, training sample
This is the device history data of multiple power equipments.
Device history data is the device attribute of the power equipment obtained before being trained to BP neural network model
Data, actual motion duration and fail result.
Optionally, device attribute data further include the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, factory
Quotient's registered capital and manufacturer's quality.
In addition to the device type of power equipment itself, device location, equipment preset four kinds of service life and equipment voltage class because
Plain outer, the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, manufacturer's registered capital and manufacturer's quality are for electric power
The influence whether equipment occurs defect is also very big, and therefore, device attribute data can also include setting for the affiliated manufacturer of power equipment
Standby ratio of defects, manufacturer's registion time, manufacturer's registered capital and manufacturer's quality.
Equipment deficiency rate is the defects of power equipment rate with the associated affiliated manufacturer of power equipment, can pass through defect electricity
The ratio that power equipment accounts for total electricity number of devices determines.
Manufacturer's registion time is the date predicted power equipment and the time difference of manufacturer's registration date, and manufacturer infuses
The volume time is as unit of year.
Capital amount when manufacturer's registered capital is manufacturer's registration as unit of ten thousand yuan of RMB.
Manufacturer's quality is to characterize the parameter of manufacturer's power equipment quality and reliability, can based on Manufacturer's authentication quantity and specially
Sharp quantity determines.In embodiments of the present invention, the basic value for the quotient's mass that can set up factories is 0, and Manufacturer's authentication is added in this basic value
Quantity and patent numbers determine manufacturer's quality.Certainly, in practical applications, manufacturer's quality can also be determined by other means,
If than by determining that manufacturer passes through ISO (International Organization for Standardization, the world
Standardization body) certification then can be in manufacturer's quality base value plus 5.
A length of date of observation and power equipment start the time difference between the date come into operation when actual motion.
Fail result is the operation conditions of power equipment in actual motion duration, and fail result includes yes/no.
Step 202, device history data is normalized.
Since different evaluation indexes has different dimension and dimensional unit, this influences whether the knot analyzed for data
Fruit, therefore in order to eliminate the dimension impact between each data target, it needs that device history data is normalized, with solution
The problem of between certainly each data target without comparativity, to carry out overall merit to power equipment.
Wherein, normalized is a kind of mode of simplified calculating, i.e., the expression formula for having dimension is turned to nondimensional mark
Amount form.
Preferably, it can select deviation standardized method that device history data is normalized.
Linear transformation the initial data to device history data can be carried out respectively using deviation standardized algorithm, make result
Value is mapped between [0,1], the transfer function of deviation standardized algorithm are as follows:Wherein, x is in initial data
Any one data in one, min and max are respectively to belong to minimum value in the initial data of homogeneous data and most with x
Big value, x ' is the data after the conversion of deviation standardized algorithm, and the value of x ' is between [0,1].
Certainly, in practical applications, normalizing can also be carried out using initial data of the other modes to device history data
Change processing.For example, can mean value based on initial data and standard deviation be normalized.
Step 203, training sample is sampled, obtains optimization training sample, optimization training sample includes same number
The fail result device history data that is yes and the fail result device history data that is no;Correspondingly, it is based on training sample,
BP neural network model is trained, comprising: based on optimization training sample, BP neural network model is trained.
Since the sample data in original training sample is unbalanced, the training result that will lead to BP neural network tends to greatly
Data sample as a result, it is therefore possible to use the method for lack sampling samples training sample, obtain optimization training sample,
Optimizing training sample includes the device history data that equal number of fail result is yes and the device history that fail result is no
Data.
Wherein, lack sampling is a kind of means for increasing test equipment bandwidth, in order to reach the instruction that can sample wider scope
Practice sample data, aliasing occurs to avoid sample data.
In embodiments of the present invention, can carry out lack sampling with Cascade algorithms, device history data that fail result is yes and
The investment ratio for the device history data that fail result is no is 1:1.Certainly, in practical applications, its other party can also be used
Formula samples training sample, for example, can carry out sampling to training sample using random lack sampling method obtains optimization training
Sample.
Wherein, Cascade algorithms are a kind of sorting algorithms for unbalanced data.
Step 204, it is based on training sample, BP neural network model is trained.
Due to perfect BP neural network model be by input layer input training sample, then by backpropagation come
The constantly multiple obtained prediction model of training of the weight of adjustment network and threshold value, therefore after getting training sample, it needs to BP
Neural network model is trained.
In BP neural network, the node number of input layer and output layer is all determining.For example, in the embodiment of the present invention
In, the node number of input layer can be 10, including device type, equipment longitude, equipment latitude, equipment preset service life and equipment
Voltage class estimates operation duration, equipment deficiency rate, manufacturer's registion time, manufacturer's registered capital and manufacturer's quality;Output layer
Node number can be 1, including fail result, fail result includes two kinds of yes/no, and in output layer, use 1 indicates defect, 0
Indicate non-defective.And the node number of hidden layer be it is uncertain, the number of the number of hidden layer node is for BP neural network
Performance influence very big, general rule of thumb formula:It can determine node in hidden layer, wherein h is implicit
Node layer number, m are input layer number, and n is output layer interstitial content, regulating constant of a between m and n.In the present invention
In embodiment, by repeatedly training, verifies and attempt repeatedly, it is final to determine, when hidden layer node number is 7, for electric power
The accuracy rate of equipment deficiency prediction of result is higher.
As shown in figure 3, provide a kind of BP neural network structure, the BP neural network structure is from left to right successively are as follows: defeated
Enter layer, hidden layer and output layer.Wherein xjIndicate the input of j-th of node of input layer, and j=1,2 ... M, M are positive integer;
wijIndicate i-th of node of hidden layer to the weight between j-th of node of input layer;θiIndicate the threshold value of i-th of node of hidden layer;
The activation primitive of φ (x) expression hidden layer;wkiIndicate k-th of node of output layer to the weight between i-th of node of hidden layer, i
=1,2 ... q, q are positive integer;akIndicate the threshold value of k-th of node of output layer, k=1,2 ... L, L are positive integer;ψ (x) table
Show the activation primitive of output layer;okIndicate the output of k-th of node of output layer.For example, in the forward direction transmittance process of working signal
In, the input of i-th of node of hidden layer can use formulaIt indicates.
Preferably, activation primitive can select Relu (Rectified Linear Unit, line rectification function), Relu
Function formula are as follows: θ (x)=max (0, x), derivative expressions are as follows:
Due to being missed as activation primitive using Relu function relative to other activation primitives (such as Sigmoid function)
When seeking error gradient in the back-propagation process of difference signal, division arithmetic is related to for the derivation of activation primitive, calculation amount is smaller,
To reduce the training process calculation amount of entire BP neural network model.In addition, for the BP neural network model of deep layer, it can also
The case where gradient disappears is easy to appear in back-propagation process using other activation primitives to reduce, i.e., when close to saturation region
Function variation is too slow, the case where derivative tends to 0, thus reduce such case will cause partial information loss possibility and
It is unable to complete the problem of the training for deep layer network, to improve the reliability of trained BP neural network model.
In embodiments of the present invention, the forward direction transmittance process of working signal is by device type, equipment longitude, equipment latitude
Degree, equipment preset the input layer of the data such as service life and equipment voltage class input BP neural network, the backpropagation of error signal
Process is the error amount successively calculated between defects of power equipment prediction result and actual result since output layer, then
The weight and threshold value that each layer is adjusted according to error gradient descent method, make the final output of modified BP neural network reach pre-
Time value.
In addition, the embodiment of the present invention can also use gradient descent method pair to the process that BP neural network model is trained
BP neural network model optimizes, and the correction amount of output layer weight, output layer threshold are successively corrected according to error gradient descent method
The correction amount of value, the correction amount of hidden layer weight and the correction amount of hidden layer threshold value.It by verifying repeatedly and attempts, determines study
Training effect and training rate are best when rate is 0.1.Wherein, weight pace of change speed is measured when learning rate is backpropagation
Index, with the increase of learning rate, each modified correction amount also be will increase.
Step 205, the device attribute data of power equipment are obtained and estimate operation duration, device attribute data include equipment
Type, device location, equipment preset service life and equipment voltage class.
Wherein, the device attribute data of power equipment are obtained and estimate the mode of operation duration, may refer in aforementioned
Associated description no longer repeats one by one herein.
Optionally, device attribute data further include the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, factory
Quotient's registered capital and manufacturer's quality.
Step 206, it to device attribute data and estimates operation duration and is normalized.
Wherein, to device attribute data and estimate mode that operation duration is normalized with to device history data
The mode being normalized is identical, may refer to the associated description in aforementioned, no longer repeats one by one herein.
Step 207, determine that power equipment is being run by preset BP neural network model based on device attribute data
The prediction result to break down when estimating operation duration.
Wherein, determine that power equipment is estimated in operation by preset BP neural network model based on device attribute data
The mode of the prediction result to break down when operation duration may refer to the associated description in aforementioned, no longer repeat one by one herein.
Furthermore it is possible to will be by preset BP neural network model, determining prediction result is by way of showing report
It is supplied to administrative staff's reference, as shown in figure 4, the displaying report includes ID (Identity, identity mark), the BP of power equipment
The output layer output valve of neural network model, the prediction shortage probability based on output valve, prediction when output layer threshold value is set as 0.5
And practical defects of power equipment result as a result.
In embodiments of the present invention, a kind of power equipments defect prediction technique is provided, comprising: obtain setting for power equipment
For attribute data and operation duration is estimated, device attribute data include device type, device location, equipment default service life and equipment
Voltage class;Based on device attribute data, by preset BP neural network model, determine that power equipment estimates operation in operation
The prediction result to break down when duration.It is based on many factors instruction for influencing power equipment due to presetting BP neural network model
Experienced model, therefore, the prediction that the power equipment determined by the preset model breaks down when operation duration is estimated in operation
As a result accuracy rate is higher than existing failure prediction method.In addition, the prediction result can join for Maintenance of Electric Transmission Line unit
It examines, periodically to carry out inspection to high-risk equipment, to avoid the generation of electric power accident, promotes the safety of transmission line of electricity entirety.
Defects of power equipment is predicted in addition, the embodiment of the present invention uses BP neural network model, BP neural network model
With certain self study and adaptive ability, the rule of reason of output and outlet chamber can be automatically extracted in training, and will
Adaptive learning Content is stored in BP neural network model, while BP neural network model also has certain fault-tolerant energy
Power is still able to maintain accurate prediction result when network is locally destroyed.
It referring to figure 5., is a kind of module diagram of power equipments defect prediction meanss provided in an embodiment of the present invention, it should
Device includes: the first acquisition module 501, for obtaining the device attribute data of power equipment and estimating operation duration, equipment category
Property data include that device type, device location, equipment preset service life and equipment voltage class;Determining module 502 is set for being based on
Standby attribute data determines power equipment failure when operation duration is estimated in operation by preset BP neural network model
Prediction result.
Optionally, device attribute data further include the equipment deficiency rate of the affiliated manufacturer of power equipment, manufacturer's registion time, factory
Quotient's registered capital and manufacturer's quality.
Optionally, the device further include: processing module, for carrying out normalizing with operation duration is estimated to device attribute data
Change processing.
Optionally, device further include: second obtains module, and for obtaining training sample, training sample includes multiple electricity
The device history data of power equipment, device history data include device attribute data, actual motion duration and fail result, failure
It as a result include yes/no;Training module is trained BP neural network model for being based on training sample.
Optionally, the device further include: sampling module, for being sampled to training sample, acquisition optimization training sample,
Optimizing training sample includes the device history data that equal number of fail result is yes and the device history that fail result is no
Data;Correspondingly, the training module is specifically used for: based on optimization training sample, being trained to BP neural network model.
The method that above-mentioned apparatus is used to execute previous embodiment offer, it is similar that the realization principle and technical effect are similar, herein not
It repeats again.
The above module can be arranged to implement one or more integrated circuits of above method, such as: one
Or multiple specific integrated circuits (Application Specific Integrated Circuit, abbreviation ASIC), or, one
Or multi-microprocessor (digital singnal processor, abbreviation DSP), or, one or more field programmable gate
Array (Field Programmable Gate Array, abbreviation FPGA) etc..For another example, when some above module passes through processing elements
When the form of part scheduler program code is realized, which can be general processor, such as central processing unit (Central
Processing Unit, abbreviation CPU) or it is other can be with the processor of caller code.For another example, these modules can integrate
Together, it is realized in the form of system on chip (system-on-a-chip, abbreviation SOC).
Fig. 6 is the schematic diagram of a kind of power equipments defect prediction meanss that one embodiment of the invention provides, which can be with
It is integrated in the chip of terminal device or terminal device, which can be the computer for having power equipments defect forecast function
Equipment.
The device includes: processor 601, memory 602.
Memory 602 is for storing program, the program that processor 601 calls memory 602 to store, to execute the above method
Embodiment.Specific implementation is similar with technical effect, and which is not described herein again.
The present invention also provides a kind of program product, such as computer readable storage medium, including program, which is being located
For executing above method embodiment when reason device executes.
In several embodiments provided by the present invention, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or the network equipment etc.) or processor (English: processor) execute this hair
The part steps of bright each embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory
(English: Read-Only Memory, abbreviation: ROM), random access memory (English: Random Access Memory, letter
Claim: RAM), the various media that can store program code such as magnetic or disk.
Claims (10)
1. a kind of power equipments defect prediction technique, which is characterized in that the described method includes:
It obtains the device attribute data of power equipment and estimates operation duration, the device attribute data include device type, set
Standby position, equipment preset service life and equipment voltage class;
Based on the device attribute data, by preset backpropagation BP neural network model, determine that the power equipment exists
The prediction result to break down when estimating operation duration described in operation.
2. power equipments defect prediction technique as described in claim 1, which is characterized in that the device attribute data further include
Equipment deficiency rate, manufacturer's registion time, manufacturer's registered capital and the manufacturer's quality of the affiliated manufacturer of power equipment.
3. power equipments defect prediction technique as claimed in claim 1 or 2, which is characterized in that be based on the equipment described
Attribute data is determined and is gone out when the power equipment estimates operation duration described in the operation by preset BP neural network model
Before the prediction result of existing failure, the method also includes:
The device attribute data and the operation duration of estimating are normalized.
4. power equipments defect prediction technique as claimed in claim 1 or 2, which is characterized in that be based on the equipment described
Attribute data is determined and is gone out when the power equipment estimates operation duration described in the operation by preset BP neural network model
Before the prediction result of existing failure, the method also includes:
Training sample is obtained, the training sample includes the device history data of multiple power equipments, the device history data
Including device attribute data, actual motion duration and fail result, the fail result includes yes/no;
Based on the training sample, the BP neural network model is trained.
5. power equipments defect prediction technique as claimed in claim 4, which is characterized in that be based on the trained sample described
This, before being trained to the BP neural network model, the method also includes:
The training sample is sampled, optimization training sample is obtained, the optimization training sample includes equal number of event
The device history data that the device history data and fail result that barrier result is yes are no;
Correspondingly, described to be based on the training sample, the BP neural network model is trained, comprising: based on described excellent
Change training sample, the BP neural network model is trained.
6. a kind of power equipments defect prediction meanss, which is characterized in that described device includes:
First obtains module, for obtaining the device attribute data of power equipment and estimating operation duration, the device attribute number
Service life and equipment voltage class are preset according to including device type, device location, equipment;
Determining module, for determining that the power equipment exists by preset BP neural network model based on device attribute data
The prediction result to break down when estimating operation duration described in operation.
7. power equipments defect prediction meanss as claimed in claim 6, which is characterized in that the device attribute data further include
Equipment deficiency rate, manufacturer's registion time, manufacturer's registered capital and the manufacturer's quality of the affiliated manufacturer of power equipment.
8. power equipments defect prediction meanss as claimed in claims 6 or 7, which is characterized in that described device further include:
Processing module, for the device attribute data and the operation duration of estimating to be normalized.
9. power equipments defect prediction meanss as claimed in claims 6 or 7, which is characterized in that described device further include:
Second obtains module, and for obtaining training sample, the training sample includes the device history data of multiple power equipments,
The device history data includes device attribute data, actual motion duration and fail result, the fail result include be or
It is no;
Training module is trained the BP neural network model for being based on the training sample.
10. power equipments defect prediction meanss as claimed in claim 9, which is characterized in that described device further include:
Sampling module obtains optimization training sample, the optimization training sample includes for sampling to the training sample
The device history data that the device history data and fail result that equal number of fail result is yes are no;
Correspondingly, the training module is specifically used for: being based on the optimization training sample, carries out to the BP neural network model
Training.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811451610.8A CN109558979A (en) | 2018-11-29 | 2018-11-29 | Power equipments defect prediction technique and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811451610.8A CN109558979A (en) | 2018-11-29 | 2018-11-29 | Power equipments defect prediction technique and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109558979A true CN109558979A (en) | 2019-04-02 |
Family
ID=65868159
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811451610.8A Pending CN109558979A (en) | 2018-11-29 | 2018-11-29 | Power equipments defect prediction technique and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109558979A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111191400A (en) * | 2019-12-31 | 2020-05-22 | 上海钧正网络科技有限公司 | Vehicle part service life prediction method and system based on user fault reporting data |
CN111754020A (en) * | 2020-05-14 | 2020-10-09 | 大唐七台河发电有限责任公司 | Intelligent power plant equipment inspection system and method based on data analysis |
CN112232370A (en) * | 2020-09-16 | 2021-01-15 | 中国电子产品可靠性与环境试验研究所((工业和信息化部电子第五研究所)(中国赛宝实验室)) | Fault analysis and prediction method for engine |
CN112510699A (en) * | 2020-11-25 | 2021-03-16 | 国网湖北省电力有限公司咸宁供电公司 | Transformer substation secondary equipment state analysis method and device based on big data |
CN113379005A (en) * | 2021-08-12 | 2021-09-10 | 新风光电子科技股份有限公司 | Intelligent energy management system and method for power grid power equipment |
CN113837473A (en) * | 2021-09-27 | 2021-12-24 | 佰聆数据股份有限公司 | Charging equipment fault rate analysis system and method based on BP neural network |
WO2022147684A1 (en) * | 2021-01-06 | 2022-07-14 | 罗伯特·博世有限公司 | Method and apparatus for identifying abnormalities in mechanical apparatus or mechanical component |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103426030A (en) * | 2013-04-24 | 2013-12-04 | 华北电力大学 | Power equipment fault rate prediction method considering aging factors |
CN104573866A (en) * | 2015-01-08 | 2015-04-29 | 深圳供电局有限公司 | Method and system for predicting defects of electrical equipment |
CN105225020A (en) * | 2015-11-11 | 2016-01-06 | 国家电网公司 | A kind of running status Forecasting Methodology based on BP neural network algorithm and system |
US20160034856A1 (en) * | 2014-07-31 | 2016-02-04 | Doosan Heavy Industries & Construction Co., Ltd. | Wind power plant management system and method thereof |
CN107769972A (en) * | 2017-10-25 | 2018-03-06 | 武汉大学 | A kind of power telecom network equipment fault Forecasting Methodology based on improved LSTM |
CN108062608A (en) * | 2018-01-25 | 2018-05-22 | 武汉理工大学 | The distribution switchgear reliability estimation method of meter and environmental factor |
CN108288123A (en) * | 2018-01-25 | 2018-07-17 | 武汉理工大学 | Distribution switchgear reliability estimation method based on BP neural network |
CN108375715A (en) * | 2018-03-08 | 2018-08-07 | 中国电力科学研究院有限公司 | A kind of distribution network line fault risk day prediction technique and system |
-
2018
- 2018-11-29 CN CN201811451610.8A patent/CN109558979A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103426030A (en) * | 2013-04-24 | 2013-12-04 | 华北电力大学 | Power equipment fault rate prediction method considering aging factors |
US20160034856A1 (en) * | 2014-07-31 | 2016-02-04 | Doosan Heavy Industries & Construction Co., Ltd. | Wind power plant management system and method thereof |
CN104573866A (en) * | 2015-01-08 | 2015-04-29 | 深圳供电局有限公司 | Method and system for predicting defects of electrical equipment |
CN105225020A (en) * | 2015-11-11 | 2016-01-06 | 国家电网公司 | A kind of running status Forecasting Methodology based on BP neural network algorithm and system |
CN107769972A (en) * | 2017-10-25 | 2018-03-06 | 武汉大学 | A kind of power telecom network equipment fault Forecasting Methodology based on improved LSTM |
CN108062608A (en) * | 2018-01-25 | 2018-05-22 | 武汉理工大学 | The distribution switchgear reliability estimation method of meter and environmental factor |
CN108288123A (en) * | 2018-01-25 | 2018-07-17 | 武汉理工大学 | Distribution switchgear reliability estimation method based on BP neural network |
CN108375715A (en) * | 2018-03-08 | 2018-08-07 | 中国电力科学研究院有限公司 | A kind of distribution network line fault risk day prediction technique and system |
Non-Patent Citations (1)
Title |
---|
李果 等主编: "《生物多样性监测技术手册》", 31 May 2014, 中国环境出版社.北京 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111191400A (en) * | 2019-12-31 | 2020-05-22 | 上海钧正网络科技有限公司 | Vehicle part service life prediction method and system based on user fault reporting data |
CN111191400B (en) * | 2019-12-31 | 2023-12-29 | 上海钧正网络科技有限公司 | Vehicle part life prediction method and system based on user fault reporting data |
CN111754020A (en) * | 2020-05-14 | 2020-10-09 | 大唐七台河发电有限责任公司 | Intelligent power plant equipment inspection system and method based on data analysis |
CN112232370A (en) * | 2020-09-16 | 2021-01-15 | 中国电子产品可靠性与环境试验研究所((工业和信息化部电子第五研究所)(中国赛宝实验室)) | Fault analysis and prediction method for engine |
CN112510699A (en) * | 2020-11-25 | 2021-03-16 | 国网湖北省电力有限公司咸宁供电公司 | Transformer substation secondary equipment state analysis method and device based on big data |
WO2022147684A1 (en) * | 2021-01-06 | 2022-07-14 | 罗伯特·博世有限公司 | Method and apparatus for identifying abnormalities in mechanical apparatus or mechanical component |
CN113379005A (en) * | 2021-08-12 | 2021-09-10 | 新风光电子科技股份有限公司 | Intelligent energy management system and method for power grid power equipment |
CN113837473A (en) * | 2021-09-27 | 2021-12-24 | 佰聆数据股份有限公司 | Charging equipment fault rate analysis system and method based on BP neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109558979A (en) | Power equipments defect prediction technique and device | |
TWI698808B (en) | Model testing method and device | |
US5625751A (en) | Neural network for contingency ranking dynamic security indices for use under fault conditions in a power distribution system | |
CN109765462A (en) | Fault detection method, device and the terminal device of transmission line of electricity | |
Pindoriya et al. | Composite reliability evaluation using Monte Carlo simulation and least squares support vector classifier | |
CN107742125A (en) | Predict and prevent the depth machine learning of the deleterious situation at structural capital | |
JP2020017952A (en) | Method and device for warning | |
CN109657615A (en) | A kind of training method of target detection, device and terminal device | |
WO2014014470A1 (en) | Problem analysis and priority determination based on fuzzy expert systems | |
CN115269342B (en) | Monitoring cloud platform based on edge computing and monitoring method thereof | |
Hong et al. | Towards accurate and efficient classification of power system contingencies and cyber-attacks using recurrent neural networks | |
CN115081591A (en) | Method for predicting failure probability of electric valve | |
CN111600309B (en) | Voltage control method, device, equipment, computer equipment and storage medium | |
CN114118460A (en) | Low-voltage transformer area line loss rate abnormity detection method and device based on variational self-encoder | |
CN111953531B (en) | Network fault analysis method and device | |
US11989013B2 (en) | Abnormality detection apparatus, abnormality detection system, and learning apparatus, and methods for the same and non-temporary computer-readable medium storing the same | |
Sobajic et al. | On-line monitoring and diagnosis of powersystem operating conditions using artificial neural networks | |
CN115903760A (en) | Fault diagnosis and performance evaluation method for unmanned aerial vehicle system | |
CN115796826A (en) | Management method, system, device and storage medium for ship safety management and control | |
CN109325594A (en) | A kind of power communication scene O&M aided diagnosis method improving CBR | |
CN115310673A (en) | Navigation risk identification, prevention and control method, system, device and storage medium | |
Wang et al. | Optimal size of a feedforward neural network: How much does it matter? | |
CN112163702A (en) | Prediction method and device for stable section load rate, electronic device and storage medium | |
Tamura et al. | Reliability analysis based on deep learning for fault big data on bug tracking system | |
Liu et al. | Evaluation of hurricane impact on failure rate of transmission lines using fuzzy expert system |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190402 |