CN108732528A - A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network - Google Patents
A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network Download PDFInfo
- Publication number
- CN108732528A CN108732528A CN201810521491.2A CN201810521491A CN108732528A CN 108732528 A CN108732528 A CN 108732528A CN 201810521491 A CN201810521491 A CN 201810521491A CN 108732528 A CN108732528 A CN 108732528A
- Authority
- CN
- China
- Prior art keywords
- depth confidence
- confidence network
- electrical energy
- energy meter
- data
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
- G01R35/04—Testing or calibrating of apparatus covered by the other groups of this subclass of instruments for measuring time integral of power or current
Abstract
The present invention relates to a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network, including：（1）Obtain acquired original data；（2）Data prediction；（3）Model hyper parameter is set；（4）With training sample training pattern；（5）With test sample test model；（6）Export digitalized electrical energy meter fault diagnosis result.A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network proposed by the present invention, deep learning theory is applied in electrical energy meter fault diagnosis, data volume is big, easily missing in the case of learn various failures automatically under each gathered data variation characteristic, there is preferable fault-tolerance simultaneously, digitalized electrical energy meter fault diagnosis accuracy and promptness are helped to improve, ensures the safe and stable operation of flexible direct current power transmission system.
Description
Technical field
The present invention relates to fault diagnosis technology field, especially a kind of digitalized electrical energy meter event based on depth confidence network
Hinder diagnostic method.
Background technology
With the further investigation of flexible DC transmission technology and the fast development of power electronic technique, solving at a distance,
Have when the problems that large capacity transmission, the access of new energy distributed generation resource and super-huge alternating current-direct current mixing power grid face
Peculiar advantage.As HVDC Transmission Technology of new generation, flexible DC power transmission is that the change of grid power transmission mode and structure are not sent a telegram here
Net provides effective solution scheme.Flexible DC power transmission engineering enters high-speed development period at present, and flexible direct current power transmission system is made
For novel electrical energy transportation pattern, had differences with traditional AC and DC transmission system.In addition, current metering system is just
In from conventional metered dose system to the transition stage of the digitized measurement system.Only carried out system Construction at present, primary equipment can
It is studied by property and operation and maintenance etc., and in terms of the digitized measurement subsystem operation and maintenance research of flexible DC power transmission
It is but also right not only without the difference of the digitized measurement subsystem and conventional metered dose system of analyzing flexible DC power transmission in blank
Its abnormal operating mode being likely to occur and quick discrimination method also without being studied in detail.In soft straight running
In, it can be abnormal operation due to metering system and electric energy tariffing is caused to go wrong, influence charge calculation.
Currently, Europe, Oceania, America, Asia, the flexible DC transmission engineering of African 16 countries in world wide
It puts into operation or is building.Wherein, the engineering that put into operation experienced the technology that 2 level, modular multilevel are returned to from 2 level to 3 level
Development course is modular multilevel topology building flexible DC power transmission engineering almost all.
The theoretical research of VSC-HVDC technology of transmission of electricitys supports mutually with practical engineering application, domestic and foreign scholars' theoretical research heat
Point is concentrated mainly on two aspect of system modelling and control strategy.For the electrical energy measurement in flexible direct current power transmission system, current edge
It is existing the digitized measurement equipment, but what these equipment were designed both for the electrical energy measurement of traditional transmission system,
Its accuracy run and stability are not verified in existing flexible DC power transmission engineering.Especially currently without filling
Divide the abnormal operation mode and method of real-time of the soft straight the digitized measurement subsystem in the case of priori and Engineering Projects
Research also just start to walk.
Compared to above-mentioned several method, it is based on the digitlization of depth confidence network (Deep Belief Network, DBN)
Electrical energy meter fault diagnostic method can make full use of all types of gathered datas, automatically extract the changing rule of fault moment electrical quantity
And fault signature, improve the independent learning ability and serious forgiveness of model, fault diagnosis speed faster, precision higher.
Invention content
The purpose of the present invention is to provide a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network, with
Overcome defect existing in the prior art.
To achieve the above object, the technical scheme is that：A kind of digitalized electrical energy meter based on depth confidence network
Method for diagnosing faults is realized in accordance with the following steps：
Step S1：Obtain digitalized electrical energy meter Historical Monitoring data, including normal sample and fault sample；
Step S2：To acquired digitalized electrical energy meter Historical Monitoring data carry out data prediction, reject redundant data and
Bad data normalizes original time domain signal 0-1, and divides training set and test set according to preset ratio；
Step S3：The hyper parameter of depth confidence network is determined according to training data, including：Input node, output node, most
The big number of plies, every layer of number of nodes and maximum iteration；
Step S4：Depth confidence network is trained by training data, is iterated up to the cost letter of depth confidence network
Number is less than predetermined threshold value；
Step S5：Test data is imported into trained depth confidence network and is tested, if test essence
Degree is unsatisfactory for predetermined threshold value requirement, then repeatedly step S3 and step S4 trains depth confidence network again；
Step S6：It is defeated by depth confidence network for the digitalized electrical energy meter fault diagnosis in flexible direct current power transmission system
Go out digitalized electrical energy meter fault diagnosis result.
In an embodiment of the present invention, in the step S1, the digitalized electrical energy meter Historical Monitoring data include：Electricity
Pressure, electric current, frequency and generator rotor angle.
In an embodiment of the present invention, in the step S2, the data of missing are inserted by using interpolation method
Median completion, enables data sample be consistent；The normalization is realized as follows；
Wherein, x and x ' indicates the forward and backward original sample point of normalization, x respectivelyminWith xmaxSame collection capacity is indicated respectively
Minimum value and maximum value.
In an embodiment of the present invention, in the step S3, the visual layer unit of RBM is v in registered depth confidence network
={ v1,v2,v3,...,vI∈ { 0,1 }, it is h={ h to imply layer unit1,h2,h3,...,hI∈ { 0,1 }, weight matrix w,
The threshold value of visual layer unit is a and the threshold value of implicit layer unit is b, then all visual elements and implicit unit associations state (v,
H) energy function is：
Wherein, I is the quantity of visual element, and J is the quantity of implicit unit, the energy function E (v, h) obtained according to above formula
The joint probability distribution obtained between hidden layer and visual layers is：
Wherein, Z is the generalized constant of an analog physical system, by the energy between all visual layers and implicit layer unit
Magnitude is added to obtain；By joint probability distribution, obtain visual layers vector v be independently distributed for：
Then under conditions of giving a stochastic inputs visual layers vector v, the probability of hidden layer vector h：
Under conditions of giving a stochastic inputs hidden layer vector h, the probability of visual layers vector v：
Since the structural unit of RBM is two state of value, one logical function sigmoid functions of note areThen obtaining activation probability is：
In an embodiment of the present invention, in the step S4, the training of the depth confidence network includes：Unsupervised
Successively pre-training and two processes of fine tuning for having supervision.
In an embodiment of the present invention, the unsupervised successively pre-training includes：It is generated in the visual layers of first RBM
Value is transmitted to hidden layer by one vector by RBM networks；In turn, reconstruct visual layers are gone with hidden layer, according to reconstruction of layer and
Weight between the difference update hidden layer and visual layers of visual layers is hidden until reaching maximum iterations what is obtained
Layer is used as visual layers；By successively stacking, feature is successively extracted from initial data, obtains high-level expression.
In an embodiment of the present invention, described to there is the fine tuning of supervision to include：After completing unsupervised successively pre-training,
The top of depth confidence network adds label data, is carried out to the parameter of depth confidence network by using back-propagation algorithm
Fine tuning carries out Training to depth confidence network, while being updated to the parameter of all layers of depth confidence network；
In an embodiment of the present invention, test data is imported into trained depth confidence network, output section
The maximum probability value of point indicates corresponding and decompression or electromagnetic interference failure mutually occurs；It is wanted if maximum probability value meets predetermined probabilities threshold value
It asks, then detects the failure of digitalized electrical energy meter in flexible direct current power transmission system by the depth confidence network；If maximum probability value
It is unsatisfactory for predetermined probabilities threshold requirement, then resets the hyper parameter of depth confidence network, repeating said steps S3 and step S4
Until meeting predetermined probabilities threshold requirement.
Compared to the prior art, the invention has the advantages that：It is proposed by the present invention a kind of based on depth confidence net
The digitalized electrical energy meter method for diagnosing faults of network, provide a kind of reasonable design, fault diagnosis comprehensively, analysis be accurately based on depth
The method for diagnosing faults of confidence network, the failure that according to historical failure information training pattern, can obtain digitalized electrical energy meter are special
Sign, to quickly analyze electrical energy meter fault state.Deep learning theory is applied in electrical energy meter fault diagnosis, in data volume
Greatly, easily learn the variation characteristic of each gathered data under various failures automatically in the case of missing, while there is preferable fault-tolerance,
Digitalized electrical energy meter fault diagnosis accuracy and promptness are helped to improve, ensures the safety and stability fortune of flexible direct current power transmission system
Row.
Description of the drawings
Fig. 1 is the flow diagram of the embodiment of the present invention；
Fig. 2 is the basic structure of limited Boltzmann machine (RBM) model；
Fig. 3 is the basic basic structure of depth confidence network；
Attached drawing 4 is DBN training process flow charts；
Attached drawing 5 is unsupervised learning process；
Specific implementation mode
Below in conjunction with the accompanying drawings, technical scheme of the present invention is specifically described.
The present invention provides a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network, as shown in Figure 1, packet
Include following steps：
Step S1：Obtain digitalized electrical energy meter Historical Monitoring data, including voltage, electric current, frequency and generator rotor angle, gathered data
Include normal sample and fault sample；
Step S2：Initial data is pre-processed, rejects redundant data and " bad data " in initial data, while will be in data
The gathered data of different dimensions does 0-1 normalizeds, and according to 7:3 ratio cut partition training datas and test data；
In the present embodiment, for the data of missing, it is typically inserted into median completion, due to Various types of data in initial data
Frequency acquisition is different, so that data sample is consistent with interpolation method；It, need to be under since initial data Various types of data dimension is different
Formula is by data normalization；
Wherein, x and x ' indicates the front and back original sample point of normalization, x respectivelyminWith xmaxSame collection capacity is indicated respectively most
Small value and maximum value；
In the present embodiment, by pretreated sample according to 7:3 ratio is divided into training data and test data, is made
Totally 6500, sample, wherein fault sample accounts for 10%, and it is training sample to randomly select 70% sample, remaining is test specimens
This.
Step S3：Based on training sample amount and input data actual conditions, setting mode input, output node, maximum layer
The hyper parameters such as several, every layer of number of nodes and maximum iteration.
In the present embodiment, used sample is obtained from somewhere flexible direct current power transmission system, and data include three-phase
Voltage, electric current, frequency and generator rotor angle, therefore the depth confidence network model established, input number of nodes are set as 12.
Since the major failure of electric energy meter is decompression and electromagnetic interference, it is contemplated that A, B, C three-phase, therefore the output section established
Points are set as 6.Wherein, the output valve of node 1-3 indicates that the probability of no-voltage fault, the output of 4-6 nodes occur for A, B, C three-phase
Value indicates that the probability of electromagnetic interference failure occurs for A, B, C three-phase.
In the present embodiment, depth confidence network is a generative probabilistic model, by multiple limited Boltzmann machines
(Restricted Boltzmann Machine, RBM) is stacked, the bottom of depth confidence network receive input data to
Amount, and hidden layer is input data by RBM conversions, i.e., the input of high one layer of RBM is from the output of low one layer of RBM.
As shown in Fig. 2, each RBM includes a visual layers and a hidden layer, only visual layers and implicit layer unit it
Between be bi-directionally connected weights, and do not connect between layer unit and visual layer unit and implicit layer unit and implicit layer unit visually
It connects.In given visual layer unit v={ v1,v2,v3,...,vI∈ { 0,1 }, implicit layer unit h={ h1,h2,h3,...,hI}∈
Under conditions of { 0,1 }, weight matrix w, the threshold value a of visual layer unit and hidden layer cell threshode b, all visual elements and implicit
The energy function of unit associations state (v, h) is：
Wherein, I is the quantity of visual element, and J is the quantity of implicit unit.The energy function E (v, h) obtained according to above formula
The joint probability distribution that can be obtained between hidden layer and visual layers is：
Wherein, Z is the generalized constant of an analog physical system, by the energy between all visual layers and implicit layer unit
Magnitude is added to obtain.By the joint probability distribution of above formula, can obtain visual layers vector v be independently distributed for：
Due to all not connected between the same layer any two unit of RBM, so giving a stochastic inputs visual layers
Vector v, all implicit layer units are independent of each other, and according to joint probability distribution, obtain the item in given visual layers vector v
Under part, the probability of hidden layer vector h：
Similar, a stochastic inputs hidden layer vector h is given, can be obtained under conditions of given hidden layer vector h,
The probability of visual layers vector v：
In view of the structural unit of RBM is two state of value, logical function sigmoid functions are being definedUnder the premise of, activation probability can be obtained：
After given visual layers vector v, p (h can be passed throughj=1/v) state for implying layer unit h is calculated, then pass through p
(vi=1/h) state of reconstruct visual layers Cell Reconstruction is obtained, the difference between visual layer unit and the visual layer unit of reconstruct
It is minimum, you can to think that implicit layer unit is another expression of visual layer unit, therefore implicit layer unit can be as visual
The feature extraction of layer input unit is as a result, to achieve the purpose that feature extraction.
In the present embodiment, constructed model training data have 4550, constructed depth confidence network such as Fig. 3 institutes
Show.
Step S4：Depth confidence network model parameter is trained using training data, the training of depth confidence network is by no prison
The successively pre-training superintended and directed and be made of two processes of fine tuning of supervision, entirety training process is as shown in Figure 4；
In the present embodiment, unsupervised successively pre-training is the main distinction of depth confidence network model and other models.
Unsupervised Level by level learning can learn non-linear complicated function, this is also by the way that data are directly mapped to output from input
It has the key point of powerful ability in feature extraction.Detailed process is：First the visual layers of first RBM generate one to
Amount, hidden layer is transmitted to by RBM networks by value, in turn, reconstruct visual layers is gone with hidden layer, according to reconstruction of layer and visual layers
Difference go the weight between update hidden layer and visual layers, until reaching maximum iterations, obtained hidden layer is made
For visual layers, by successively stacking, this depth structure can successively extract feature from initial data, and it is high-level to obtain some
Expression.The method of RBM is successively trained to avoid the complex calculation that whole training depth confidence Netowrk tape comes, by depth confidence net
Network model successively evolves into a shallow-layer neural network, and unsupervised Level by level learning process is as shown in Figure 5.
With reference to the accompanying drawings 5, unsupervised pre-training is carried out to depth confidence network.After the completion of training, by depth confidence
The top of network adds label data, carries out Training to depth confidence network, that is, uses back-propagation algorithm (Back
Propagation, BP) relevant parameter of depth confidence network is finely adjusted.By having carried out supervision to depth confidence network
Training will be further reduced training error and improve depth confidence network class model accuracy rate.With it is every in unsupervised training
One layer of secondary training is compared, and backpropagation has supervision fine tuning to be updated simultaneously to all layers of parameter.
Step S5：Test data is imported into trained model and tests training effect, 6 lists of output node
Maximum probability value represents corresponding mutually generation decompression or electromagnetic interference failure, if precision meets threshold requirement, the model in member
Can be used for detecting the most common failure of digitalized electrical energy meter, if model measurement error is excessive, reset the maximum number of plies, every layer
The hyper parameters such as number of nodes and iterations repeat step S3 and S4 until training precision is met the requirements.
Step S6：It is auxiliary by trained model for the digitalized electrical energy meter fault diagnosis in flexible direct current power transmission system
Operation maintenance personnel is helped to grasp the health status of measuring equipment.
In order to allow those skilled in the art to further appreciate that technical solution proposed by the present invention, with reference to specific example into
Row explanation.
In the present embodiment, somewhere flexible direct current power transmission system digitalized electrical energy meter history number is collected by step S1
According to, three-phase voltage, three-phase current, frequency and generator rotor angle.
Interpolation method completion missing data is used by step S2, while the gathered data of different dimensions in data is done into 0-1
Normalized, and according to 7:3 ratio cut partition training datas and test data.
It is based on training sample amount and input data actual conditions, setting mode input, output node, maximum by step S3
The hyper parameters such as the number of plies, every layer of number of nodes and maximum iteration.Used sample is from a regional flexible direct current power transmission system
It obtains, data include 3 on off states of surrounding, three-phase voltage current data and frequency, therefore the depth confidence network mould established
Type, input number of nodes are set as 12, and constructed model training data have 4550, and constructed depth confidence network is shown in
Attached drawing 3.
In the present embodiment, disturbed emulation signal x (t), y (t), z (t), a (t), the b of noise using six bands
(t) and c (t) simulates the decompression and electromagnetic interference failure of three-phase respectively, and verifies fault signature extraction and diagnosis side based on DBN
The performance of method.
For more actual simulated failure environment, certain random noise, wherein noise are added to each emulation signal
Signal is
W=20*randn (1, n)
Training sample training pattern parameter, while test model training effect in step s 5 are completed by step S4.By
Decide that the number of nodes of input layer, output classification number decide the number of nodes of output layer in input data dimension, therefore will input
The number of nodes of layer and output layer is set to 12 and 6, and node in hidden layer is set as 500, learning rate initial value is set as 0.1,
Initial momentum is set as 0.5, and after iteration 5 times, momentum becomes 0.9, and global reversed fine tuning number is 100.In order to which elimination algorithm is random
Property, it tests all be repeated 10 times every time, take the average value of 10 results.
Six output nodes a1, b1, c1, a2, b2, c2 indicate that decompression and electromagnetic interference failure occur for A, B, C three-phase respectively
Probability, using the highest point of probability as fault diagnosis result.Table 1 gives DBN model under 4,5,6 and 7 four kind of different depth
Classification performance, wherein accuracy rate is the mean value of 10 result of calculation, in order to assess the stability of DBN model, is given 10 times
The standard deviation of result of calculation, while in order to which the classification effectiveness to DBN model is there are one intuitive understanding, also giving different depth
Under training time, constructed depth confidence network model classification performance comparison it is as follows：
1 DBN performance of fault diagnosis of table
As it can be seen from table 1 with the increase of DBN model depth, the time selects one almost in increase at double
Appropriate depth is very important.When corresponding classification accuracy highest when DBN model depth is 5 in table 1 than depth is 4
Corresponding minimum classification accuracy is high by 2.6625%, in addition, when depth is more than 5, classification accuracy does not carry significantly
Height, training time are but increasing considerably always, combined training time and classification accuracy, and 5 layers are that a comparison is reasonably deep
Degree selection.
The above are preferred embodiments of the present invention, all any changes made according to the technical solution of the present invention, and generated function is made
When with range without departing from technical solution of the present invention, all belong to the scope of protection of the present invention.
Claims (8)
1. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network, which is characterized in that in accordance with the following steps
It realizes：
Step S1：Obtain digitalized electrical energy meter Historical Monitoring data, including normal sample and fault sample；
Step S2：Data prediction is carried out to acquired digitalized electrical energy meter Historical Monitoring data, rejects redundant data and bad number
According to original time domain signal 0-1 normalization, and according to preset ratio division training set and test set；
Step S3：The hyper parameter of depth confidence network is determined according to training data, including：Input node, output node, maximum layer
Several, every layer of number of nodes and maximum iteration；
Step S4：Depth confidence network is trained by training data, is iterated until the cost function of depth confidence network is low
In predetermined threshold value；
Step S5：Test data is imported into trained depth confidence network and is tested, if measuring accuracy is not
Meet predetermined threshold value requirement, then repeatedly step S3 and step S4 trains depth confidence network again；
Step S6：By depth confidence network for the digitalized electrical energy meter fault diagnosis in flexible direct current power transmission system, number is exported
Word electrical energy meter fault diagnostic result.
2. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 1, special
Sign is that in the step S1, the digitalized electrical energy meter Historical Monitoring data include：Voltage, electric current, frequency and generator rotor angle.
3. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 1, special
Sign is, in the step S2, the data of missing is inserted into median completion, enables data sample by using interpolation method
It is consistent；The normalization is realized as follows；
Wherein, x and x ' indicates the forward and backward original sample point of normalization, x respectivelyminWith xmaxThe minimum of same collection capacity is indicated respectively
Value and maximum value.
4. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 1, special
Sign is that in the step S3, the visual layer unit of RBM is v={ v in registered depth confidence network1,v2,v3,...,vI}∈
{ 0,1 }, it is h={ h to imply layer unit1,h2,h3,...,hI∈ { 0,1 }, the threshold value of weight matrix w, visual layer unit are a
Threshold value with implicit layer unit is b, then the energy function of all visual elements and implicit unit associations state (v, h) is：
Wherein, I is the quantity of visual element, and J is the quantity of implicit unit, and the energy function E (v, h) obtained according to above formula is obtained
Joint probability distribution between hidden layer and visual layers is：
Wherein, Z is the generalized constant of an analog physical system, by the energy value between all visual layers and implicit layer unit
Addition obtains；By joint probability distribution, obtain visual layers vector v be independently distributed for：
Then under conditions of giving a stochastic inputs visual layers vector v, the probability of hidden layer vector h：
Under conditions of giving a stochastic inputs hidden layer vector h, the probability of visual layers vector v：
Since the structural unit of RBM is two state of value, one logical function sigmoid functions of note areThen
It is to activation probability：
5. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 1, special
Sign is, in the step S4, the training of the depth confidence network includes：Unsupervised successively pre-training and there is supervision
Finely tune two processes.
6. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 5, special
Sign is that the unsupervised successively pre-training includes：A vector is generated in the visual layers of first RBM, passes through RBM networks
Value is transmitted to hidden layer；In turn, reconstruct visual layers are gone with hidden layer, is hidden according to the difference update of reconstruction of layer and visual layers
Weight between layer and visual layers, until reaching maximum iterations, using obtained hidden layer as visual layers；By successively
It stacks, feature is successively extracted from initial data, obtains high-level expression.
7. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 5, special
Sign is, described have the fine tuning of supervision to include：After completing unsupervised successively pre-training, in the top of depth confidence network
Label data is added, the parameter of depth confidence network is finely adjusted by using back-propagation algorithm, to depth confidence network
Training is carried out, while the parameter of all layers of depth confidence network is updated.
8. a kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network according to claim 1, special
Sign is, test data is imported into trained depth confidence network, the maximum probability value expression pair of output node
Decompression or electromagnetic interference failure should mutually occur；If maximum probability value meets predetermined probabilities threshold requirement, pass through the depth confidence
Network detects the failure of digitalized electrical energy meter in flexible direct current power transmission system；It is wanted if maximum probability value is unsatisfactory for predetermined probabilities threshold value
It asks, then resets the hyper parameter of depth confidence network, repeating said steps S3 and step S4 until meeting predetermined probabilities threshold value
It is required that.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810521491.2A CN108732528A (en) | 2018-05-28 | 2018-05-28 | A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810521491.2A CN108732528A (en) | 2018-05-28 | 2018-05-28 | A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108732528A true CN108732528A (en) | 2018-11-02 |
Family
ID=63935568
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810521491.2A Pending CN108732528A (en) | 2018-05-28 | 2018-05-28 | A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108732528A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109946640A (en) * | 2019-04-11 | 2019-06-28 | 广东电网有限责任公司 | Special transformer terminals decompression defluidization judgment method based on metering automation system data |
| CN110132627A (en) * | 2019-05-28 | 2019-08-16 | 上海海事大学 | A kind of method for diagnosing faults of propeller |
| CN110376457A (en) * | 2019-06-28 | 2019-10-25 | 同济大学 | Non-intrusion type load monitoring method and device based on semi-supervised learning algorithm |
| CN110763997A (en) * | 2019-11-04 | 2020-02-07 | 华北电力大学（保定） | Early fault early warning method for synchronous motor stator |
| CN110879377A (en) * | 2019-11-22 | 2020-03-13 | 国网新疆电力有限公司电力科学研究院 | Metering device fault tracing method based on deep belief network |
| CN111142060A (en) * | 2019-12-02 | 2020-05-12 | 国网浙江省电力有限公司 | Self-adaptive threshold adjustment diagnosis method based on improved BP neural network |
| CN111242243A (en) * | 2020-02-28 | 2020-06-05 | 南方电网科学研究院有限责任公司 | Method, system and equipment for detecting electric energy quality of electric energy meter |
| CN111273212A (en) * | 2020-02-24 | 2020-06-12 | 国网湖南省电力有限公司 | Data-driven electric quantity sensor error online evaluation closed-loop improvement method, system and medium |
| CN112287592A (en) * | 2020-10-27 | 2021-01-29 | 北京理工大学 | Industrial equipment fault diagnosis method and system based on deep confidence network |
| CN112766702A (en) * | 2021-01-13 | 2021-05-07 | 广东能源集团科学技术研究院有限公司 | Distributed power station fault analysis method and system based on deep belief network |
| CN113406438A (en) * | 2021-06-23 | 2021-09-17 | 安徽南瑞中天电力电子有限公司 | Intelligent fault diagnosis method suitable for low-voltage transformer area and operation and maintenance system thereof |
| CN113780355A (en) * | 2021-08-12 | 2021-12-10 | 上海理工大学 | Deep convolutional neural network learning method for deep sea submersible propeller fault identification |
| CN115600695A (en) * | 2022-09-06 | 2023-01-13 | 北京航天计量测试技术研究所（Cn） | Fault diagnosis method of metering equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102707256A (en) * | 2012-06-20 | 2012-10-03 | 北京航空航天大学 | Fault diagnosis method based on BP-Ada Boost nerve network for electric energy meter |
| CN103792420A (en) * | 2014-01-26 | 2014-05-14 | 威胜集团有限公司 | Electricity larceny preventing and electricity utilization monitoring method based on load curves |
| CN105158725A (en) * | 2015-08-24 | 2015-12-16 | 中国电力科学研究院 | Multi-dimensional influence quantity-based electric energy meter metering accuracy evaluation method |
| CN106338708A (en) * | 2016-08-30 | 2017-01-18 | 中国电力科学研究院 | Electric energy metering error analysis method by combining deep learning and recursive neural network |
| CN106769048A (en) * | 2017-01-17 | 2017-05-31 | 苏州大学 | Self adaptation depth confidence network Method for Bearing Fault Diagnosis based on Nesterov momentum methods |
| CN107742127A (en) * | 2017-10-19 | 2018-02-27 | 国网辽宁省电力有限公司 | A kind of improved anti-electricity-theft intelligent early-warning system and method |
| CN107831465A (en) * | 2017-10-31 | 2018-03-23 | 国网黑龙江省电力有限公司电力科学研究院 | A kind of intelligent electric energy meter fault judgment method based on BP neural network |
-
2018
- 2018-05-28 CN CN201810521491.2A patent/CN108732528A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102707256A (en) * | 2012-06-20 | 2012-10-03 | 北京航空航天大学 | Fault diagnosis method based on BP-Ada Boost nerve network for electric energy meter |
| CN103792420A (en) * | 2014-01-26 | 2014-05-14 | 威胜集团有限公司 | Electricity larceny preventing and electricity utilization monitoring method based on load curves |
| CN105158725A (en) * | 2015-08-24 | 2015-12-16 | 中国电力科学研究院 | Multi-dimensional influence quantity-based electric energy meter metering accuracy evaluation method |
| CN106338708A (en) * | 2016-08-30 | 2017-01-18 | 中国电力科学研究院 | Electric energy metering error analysis method by combining deep learning and recursive neural network |
| CN106769048A (en) * | 2017-01-17 | 2017-05-31 | 苏州大学 | Self adaptation depth confidence network Method for Bearing Fault Diagnosis based on Nesterov momentum methods |
| CN107742127A (en) * | 2017-10-19 | 2018-02-27 | 国网辽宁省电力有限公司 | A kind of improved anti-electricity-theft intelligent early-warning system and method |
| CN107831465A (en) * | 2017-10-31 | 2018-03-23 | 国网黑龙江省电力有限公司电力科学研究院 | A kind of intelligent electric energy meter fault judgment method based on BP neural network |
Non-Patent Citations (2)
| Title |
|---|
| 王春雨: "多维条件对电子式电能表计量性能影响的建模研究及应用", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
| 葛强强: "基于深度置信网络的数据驱动故障诊断方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109946640A (en) * | 2019-04-11 | 2019-06-28 | 广东电网有限责任公司 | Special transformer terminals decompression defluidization judgment method based on metering automation system data |
| CN110132627A (en) * | 2019-05-28 | 2019-08-16 | 上海海事大学 | A kind of method for diagnosing faults of propeller |
| CN110376457B (en) * | 2019-06-28 | 2020-10-02 | 同济大学 | Non-invasive load monitoring method and device based on semi-supervised learning algorithm |
| CN110376457A (en) * | 2019-06-28 | 2019-10-25 | 同济大学 | Non-intrusion type load monitoring method and device based on semi-supervised learning algorithm |
| CN110763997A (en) * | 2019-11-04 | 2020-02-07 | 华北电力大学（保定） | Early fault early warning method for synchronous motor stator |
| CN110879377A (en) * | 2019-11-22 | 2020-03-13 | 国网新疆电力有限公司电力科学研究院 | Metering device fault tracing method based on deep belief network |
| CN111142060B (en) * | 2019-12-02 | 2023-11-07 | 国网浙江省电力有限公司 | Adaptive threshold adjustment diagnosis method based on improved BP neural network |
| CN111142060A (en) * | 2019-12-02 | 2020-05-12 | 国网浙江省电力有限公司 | Self-adaptive threshold adjustment diagnosis method based on improved BP neural network |
| CN111273212A (en) * | 2020-02-24 | 2020-06-12 | 国网湖南省电力有限公司 | Data-driven electric quantity sensor error online evaluation closed-loop improvement method, system and medium |
| CN111273212B (en) * | 2020-02-24 | 2022-03-11 | 国网湖南省电力有限公司 | Data-driven electric quantity sensor error online evaluation closed-loop improvement method, system and medium |
| CN111242243A (en) * | 2020-02-28 | 2020-06-05 | 南方电网科学研究院有限责任公司 | Method, system and equipment for detecting electric energy quality of electric energy meter |
| CN112287592A (en) * | 2020-10-27 | 2021-01-29 | 北京理工大学 | Industrial equipment fault diagnosis method and system based on deep confidence network |
| CN112766702A (en) * | 2021-01-13 | 2021-05-07 | 广东能源集团科学技术研究院有限公司 | Distributed power station fault analysis method and system based on deep belief network |
| CN113406438A (en) * | 2021-06-23 | 2021-09-17 | 安徽南瑞中天电力电子有限公司 | Intelligent fault diagnosis method suitable for low-voltage transformer area and operation and maintenance system thereof |
| CN113406438B (en) * | 2021-06-23 | 2023-11-24 | 安徽南瑞中天电力电子有限公司 | Intelligent fault diagnosis method suitable for low-voltage transformer area and operation and maintenance system thereof |
| CN113780355A (en) * | 2021-08-12 | 2021-12-10 | 上海理工大学 | Deep convolutional neural network learning method for deep sea submersible propeller fault identification |
| CN113780355B (en) * | 2021-08-12 | 2024-02-09 | 上海理工大学 | Deep convolution neural network learning method for fault identification of deep sea submersible propeller |
| CN115600695B (en) * | 2022-09-06 | 2023-10-17 | 北京航天计量测试技术研究所 | Fault diagnosis method for metering equipment |
| CN115600695A (en) * | 2022-09-06 | 2023-01-13 | 北京航天计量测试技术研究所（Cn） | Fault diagnosis method of metering equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108732528A (en) | A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network | |
| CN108089099A (en) | The diagnostic method of distribution network failure based on depth confidence network | |
| CN106443297B (en) | The decision tree SVM method for diagnosing faults of photovoltaic diode Clamp three-level inverter | |
| CN102707256B (en) | Fault diagnosis method based on BP-Ada Boost nerve network for electric energy meter | |
| CN104155574A (en) | Power distribution network fault classification method based on adaptive neuro-fuzzy inference system | |
| Tong et al. | Detection and classification of transmission line transient faults based on graph convolutional neural network | |
| CN103886405B (en) | Boiler combustion condition identification method based on information entropy characteristics and probability nerve network | |
| CN103995237A (en) | Satellite power supply system online fault diagnosis method | |
| CN109344517A (en) | A kind of high-voltage isulation method for diagnosing faults of new-energy automobile | |
| CN110417011A (en) | A kind of online dynamic secure estimation method based on mutual information Yu iteration random forest | |
| CN109657789A (en) | Gear case of blower failure trend prediction method based on wavelet neural network | |
| CN104899353B (en) | A kind of power quality disturbance localization method based on evidence theory | |
| CN103544542A (en) | Power system transient stability margin predicting method | |
| CN106093678A (en) | A kind of method quick and precisely diagnosing flexible direct current power transmission system converter fault | |
| CN111062569A (en) | Low-current fault discrimination method based on BP neural network | |
| CN104536970A (en) | Fault determining and classifying system and method for remote communication data device | |
| CN112183606A (en) | Power system fault identification and classification method and system based on C4.5 algorithm | |
| CN111814284A (en) | On-line voltage stability evaluation method based on correlation detection and improved random forest | |
| CN103400213B (en) | A kind of bulk transmission grid survivability evaluation method based on LDA Yu PCA | |
| CN113610119B (en) | Method for identifying power transmission line development faults based on convolutional neural network | |
| CN105353270B (en) | Consider the grid-connected fault-tolerant localization method of power quality disturbance of distributed generation resource | |
| Singh et al. | Power system fault diagnosis using fuzzy decision tree | |
| CN108054768A (en) | Transient stability evaluation in power system method based on principal component analysis | |
| CN105741184A (en) | Transformer state evaluation method and apparatus | |
| CN115712871A (en) | Power electronic system fault diagnosis method combining resampling and integrated learning |
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: 20181102 |