CN107993012A - A kind of adaptive electric system on-line transient stability appraisal procedure of time - Google Patents
A kind of adaptive electric system on-line transient stability appraisal procedure of time Download PDFInfo
- Publication number
- CN107993012A CN107993012A CN201711260938.7A CN201711260938A CN107993012A CN 107993012 A CN107993012 A CN 107993012A CN 201711260938 A CN201711260938 A CN 201711260938A CN 107993012 A CN107993012 A CN 107993012A
- Authority
- CN
- China
- Prior art keywords
- rbm
- moment
- transient stability
- hidden layer
- input
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 230000001052 transient effect Effects 0.000 title claims abstract description 48
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000011156 evaluation Methods 0.000 claims abstract description 21
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 238000005259 measurement Methods 0.000 claims abstract description 9
- 238000013507 mapping Methods 0.000 claims abstract description 8
- 230000000750 progressive effect Effects 0.000 claims abstract description 5
- 230000000007 visual effect Effects 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000001537 neural effect Effects 0.000 claims description 12
- 238000013210 evaluation model Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 230000000644 propagated effect Effects 0.000 claims description 6
- 210000005036 nerve Anatomy 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 230000005611 electricity Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000013097 stability assessment Methods 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Educational Administration (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The present invention relates to the electric system on-line transient stability appraisal procedure that a kind of time is adaptive, the measurable dynamic data of power grid Wide Area Measurement System for the time window that length is T is as input feature vector after present invention selection fault clearance, feature extraction, which is carried out, using DBN obtains high-order feature, and then carry out off-line training using GRU and obtain the mapping relations between high-order feature and power system transient stability, during application on site, input feature vector is temporally progressive to be sequentially inputted, until obtaining accurate evaluation result or reaching the stopping of time window length.Therefore transient stability evaluation in power system model proposed by the present invention has the characteristics that to region be directly facing the measurable data of power grid, anti-noise jamming ability is strong, Evaluation accuracy is high, the time is adaptively strong.
Description
Technical field
The invention belongs to power system automatic field, it is related to a kind of adaptive electric system on-line transient stability of time
Appraisal procedure.
Background technology
Transient stability is that electric system keeps stable ability in the case where being disturbed greatly.In past country's outgoing
In raw a lot of large-scale blackouts, Transient Instability is the principal element of such event, therefore keeps power system transient stability
Still it is of great significance in Power System Planning and operation.In order to meet growing electricity needs, load is increasingly
Close to power transmission capacity, the consequence for causing Transient Instability in this case is extremely serious, therefore electric system is online
Transient Stability Evaluation is necessary and very promising method.
Transient stability evaluation in power system method based on machine learning is most potential to be applied to online method.At present
The method that Transient Stability Evaluation is carried out based on machine learning is mainly considered after failure occurs, and spy is used as using the dynamic response of system
Sign, obtains the mapping relations between input feature vector and stable state, emergent control will be taken to be adjusted once unstability occurs.
This method can obtain higher Evaluation accuracy, but it generally requires a period of time after fault clearance that just to carry out transient state steady
Accepted opinion is estimated.This period is longer, its Evaluation accuracy is higher, but the time for keeping for dispatcher to be adjusted may deficiency.
Being also in fault clearance early stage can be with accurate evaluation power system transient stability, therefore the power train that search time is adaptive
On-line transient stability appraisal procedure of uniting is to solve the problems, such as this key.
The content of the invention
To solve the above-mentioned problems, the present invention proposes a kind of adaptive electric system on-line transient stability assessment of time
Method.This method has towards bottom measurement data, anti-noise jamming ability is strong, Evaluation accuracy is high, the assessment time is adaptive
The advantages of, it is adapted to the assessment of electric system on-line transient stability.
A kind of adaptive electric system on-line transient stability appraisal procedure of time proposed by the present invention, it is characterised in that:
High-level feature X is formed to the measurable data of power grid bottom using depth confidence network (DBN), it is latter using fault clearance offline
The data of a time window T obtain between these high-level features and transient stability GRU (Gate Recurrent Unit) training
Mapping relations, application on site is then successively to the high-level feature x at each moment after fault clearancetTransient stability is carried out to comment
Estimate, disclosure satisfy that assessment confidence level or reach time window length and then stop assessing.
The present invention adopts the technical scheme that:
Adaptive electric system on-line transient stability appraisal procedure of a kind of time, it is characterised in that comprise the steps of:
Step 1:The power grid dynamic data under several fault condition is produced using time-domain simulation method, chooses failure removal
The measurable data of power grid Wide Area Measurement System in time window T are as input feature vector collection { x afterwards1,x2,…,xt…xT, wherein,
The measurable data of power grid Wide Area Measurement System refer to each busbar voltage of power grid and its phase angle, xt=[v1,v2,…,vn,θ1,
θ2,…,θn]T, wherein n is busbar number;
Step 2:Feature extraction, which is carried out, using DBN forms high-order feature set { X1,X2,…,Xt…XT, specifically utilize DBN
Stacked and formed by one layer softmax layers and m layers limited Boltzmann machine (RBM) stack, the connection of unsupervised first RBM of training
Weights and biasing so that the RBM can with maximum probability reconstruct input feature vector, using the output of trained first RBM as
The input of next RBM carries out feature reconstruction, until unsupervised training is completed in m-th of RBM training, finally adds softmax layers
Carry out Training and carry out network parameter fine setting, m-th of RBM layers of output is high-order feature { X1,X2,…,Xt…XT};
Step 3:Off-line training is carried out to GRU, obtains the mapping relations between high-order feature and electric power system transient stability,
During off-line training, input as { X1,X2,…,Xt…XT, export as { y1,y2,…,yt,…,yT, training process is by propagated forward
Formed with parameter renewal process, iterations N and error upper limit error is set;
Step 4:Application on site time adaptive Transient Stability Evaluation model, successively to the height at each moment after fault clearance
Level characteristics XtTransient Stability Evaluation is carried out, assessment confidence level is disclosure satisfy that or reaches time window length and then stop assessing, online
The assessment result of the adaptive Transient Stability Evaluation model of application time isα is in formula
Confidence factor, [0, α) and ∪ (1- α, 1] it is confidential interval;Using temporally progressive input sequence, specifically, in fault clearance
Backward mode input x1, stop assessment if assessment result is in confidential interval, otherwise continue to input x2, with reference to last moment
To x1Assessment gained " memory " h1Make the assessment to stability of power system;Until when drawing assessment result or reaching assessment
Between untill window length T.
In the electric system on-line transient stability appraisal procedure that a kind of above-mentioned time is adaptive, step 2 single RBM without
Supervised training process is as follows:
Step 2.1, RBM hidden layer each units are separate, in the state v of given visual layers, j-th of nerve of hidden layer
The activation probability of unit isWhereinbjFor j-th of god of hidden layer
Biasing through unit, ωijFor the connection weight between i-th of visual layer unit of RBM networks and j-th of implicit layer unit;
Step 2.2, RBM hidden layer each units are separate, in the state h of given hidden layer, i-th of nerve of visual layers
The activation probability of unit isWherein aiFor i-th of neural unit biasing of hidden layer;
Step 2.3, the input set { v for quantity for s1,v2,…,vs, by maximizing logarithms of the RBM in input set
Likelihood functionObtain model parameter θ, you can the feature extraction using hidden layer output as visual layers, its
Middle P (vk| h, θ) under the state h and parameter θ of known hidden layer visual layers probability, be expressed as with formula:
Step 2.4, parameter more new formula is:Further,In formula, v(0)To be originally inputted collection, v(l)For to v(0)Carry out obtained by l step gibbs samplers, ρ
For momentum term, η is learning rate, bjFor the biasing of j-th of neural unit of hidden layer, ωijFor i-th of visual layer unit of RBM networks
Connection weight between j-th of implicit layer unit, aiFor i-th of neural unit biasing of hidden layer.
In the electric system on-line transient stability appraisal procedure that a kind of above-mentioned time is adaptive, step 3 specifically includes:
Step 3.1:Perform circulation i=1 to i=N;
Step 3.2:Perform circulation t=1 to t=T;
Step 3.3:Carry out propagated forward process, W(z), U(z)Input and last moment hidden layer are referred respectively to renewal door z
Connection matrix, W(r), U(r)Represent input and last moment hidden layer to the connection matrix for resetting door r, W respectivelyh, UhRepresent respectively
Input and the hidden layer of last moment to the connection matrix of output layer, WoutTo export weight matrix;
Step 3.4:Back-propagation process is carried out, this is a process being updated to weighting parameter, defines single sample
This error isD in formulatFor sample real output value, then its each undated parameter variable quantity is:
WhereinXtIt is defeated for t moment
Enter, ztDoor, z are updated for t momenttDoor, h are reset for t momenttFor t moment output layer, ytFor t moment output valve, st-1For t-1 when
The memory at quarter, rt+1Door is reset for the t+1 moment,The output of door error is reset for the t+1 moment;
Step 3.5:When iterations is more than N or sample overall errorStop circulation during less than error.See
Shown in Fig. 2.
The features of the present invention and beneficial effect:Length is wide for the power grid of the time window of T after the present invention chooses fault clearance
The measurable dynamic data of domain measurement system carries out feature extraction using DBN and obtains high-order feature, and then utilize as input feature vector
GRU carries out off-line training and obtains the mapping relations between high-order feature and power system transient stability, during application on site, input
Feature is temporally progressive to be sequentially inputted, until obtaining accurate evaluation result or reaching the stopping of time window length.Therefore originally
The transient stability evaluation in power system model that invention proposes, which has, region be directly facing the measurable data of power grid, anti-noise jamming ability
By force, the characteristics of Evaluation accuracy is high, the time is adaptively strong, specifically, the invention has the advantages that:(1) DBN can be to input
Data carry out abstract expression and obtain high-order feature, can directly input measurable data and carry out feature extraction and then can filter out electricity
The influence of noise in net, therefore there is excellent antijamming capability;(2) process of DBN feature extractions is there are supervised learning, because
This can form the feature towards transient stability, further improve the Evaluation accuracy that GRU forms Transient Stability Evaluation model, because
This has the characteristics that assessment accuracy rate is high;(3) time window T is bigger, and model evaluation accuracy is higher.During application on site, model
Whether the characteristic of subsequent time can be assessed according to assessment result, until judgement of stability can be made or reached
Untill the time is assessed.In other words, if the judgement of stability can be made in early stage, then can be left to dispatcher
More make a response, therefore model has the characteristics that the time is adaptive.
A kind of adaptive electric system on-line transient stability appraisal procedure of time proposed by the present invention can be applied to electricity
The assessment of Force system on-line transient stability sees that it can early carry out Transient Stability Evaluation in fault clearance, be dispatcher's implementing measure
Valuable time is won.With the continuous development of electrical power system wide-area measuring system, measured data are gradually enriched,
Time adaptive electric system on-line transient stability appraisal procedure can play more next for Electrical Power System Dynamic security evaluation
Bigger effect.
Brief description of the drawings
Fig. 1 is the method flow schematic diagram of the present invention.
Fig. 2 is the application on site of time adaptive transient stability evaluation in power system model in the embodiment of the present invention.
Specific implementation method
The present invention carries out electric system measurable data feature extraction by DBN, and then forms high-order feature using GRU
Mapping relations between transient stability, so as to realize time adaptive electric system using GRU characteristics in the application stage
On-line transient stability is assessed.It is described as follows in conjunction with the accompanying drawings and embodiments:
The present invention adopts the technical scheme that, a kind of adaptive electric system on-line transient stability appraisal procedure of time,
It is characterized in that, comprise the steps of:
Step 1:The power grid dynamic data under several fault condition is produced using time-domain simulation method, chooses failure removal
The measurable data of power grid Wide Area Measurement System in time window T are as input feature vector collection { x afterwards1,x2,…,xt…xT};
Step 2:Feature extraction, which is carried out, using DBN forms high-order feature set { X1,X2,…,Xt…XT};
Step 3:Off-line training is carried out to GRU, obtains the mapping relations between high-order feature and electric power system transient stability;
Step 4:Application on site time adaptive Transient Stability Evaluation model, successively to the height at each moment after fault clearance
Level characteristics XtTransient Stability Evaluation is carried out, assessment confidence level is disclosure satisfy that or reaches time window length and then stop assessing.
The measurable data of power grid Wide Area Measurement System refer to each busbar voltage of power grid and its phase angle, x described in step 1t=
[v1,v2,…,vn,θ1,θ2,…,θn]T, wherein n is busbar number.
Stacked and formed by one layer softmax layers and m layers limited Boltzmann machine (RBM) stack using DBN described in step 2,
The connection weight of unsupervised first RBM of training and biasing so that the RBM can reconstruct input feature vector with maximum probability, will instruct
The output of first RBM perfected carries out feature reconstruction as the input of next RBM, until m-th of RBM training is completed without prison
Supervise and instruct white silk, finally add softmax layers and carry out Training progress network parameter fine setting, m-th of RBM layers of output is high-order
Feature { X1,X2,…,Xt…XT}.The single unsupervised training process of RBM is as follows:
(1) RBM hidden layers each unit is separate, in the state v of given visual layers, j-th of neural unit of hidden layer
Activation probability beWhereinB biases for hidden layer, ω RBM
Network connection weight;
(2) RBM hidden layers each unit is separate, in the state h of given hidden layer, i-th of neural unit of hidden layer
Activation probability beWherein a biases for hidden layer;
(3) input set { v for quantity for s1,v2,…,vs, by maximizing log-likelihoods of the RBM in input set
FunctionObtain model parameter θ, you can using hidden layer output as the feature extraction of visual layers, wherein P
(vk| h, θ) under the state h and parameter θ of known hidden layer visual layers probability, be expressed as with formula:
(4) parameter more new formula is:Further,In formula, v(0)To be originally inputted collection, v(l)For to v(0)Carry out obtained by l step gibbs samplers, ρ
For momentum term, η is learning rate, bjFor the biasing of j-th of neural unit of hidden layer, ωijFor i-th of visual layer unit of RBM networks
Connection weight between j-th of implicit layer unit, aiFor i-th of neural unit biasing of hidden layer;
GRU is trained offline described in step 3, is inputted as { X1,X2,…,Xt…XT, export as { y1,y2,…,
yt,…,yT, its training process is made of propagated forward and parameter renewal process, sets iterations N and error upper limit error,
Detailed process is described as follows:
Step 3.1:Perform circulation i=1 to i=N;
Step 3.2:Perform circulation t=1 to t=T;
Step 3.3:Carry out propagated forward process, W(z), U(z)Input and t-1 moment hidden layer are referred respectively to renewal door z
Connection matrix, W(r), U(r)Represent input and t-1 moment hidden layer to the connection matrix for resetting door r, W respectivelyh, UhRepresent respectively defeated
Enter the connection matrix with the hidden layer at t-1 moment to output layer, WoutTo export weight matrix.
(1) t moment renewal door:zt=σ (W(z)Xt+U(z)st-1), renewal door determines the memory s for leaving the t-1 momentt-1;
(2) t moment resets door:rt=σ (W(r)Xt+U(r)st-1), reset door and determine how to combine new input xtDuring with t-1
The memory s at quartert-1;
(3) t moment output layer:
(4) t moment output valve is:yt=σ (Woutst)。
Step 3.4:Back-propagation process is carried out, this is a process being updated to weighting parameter, defines single sample
This error isD in formulatFor sample real output value, then its each undated parameter variable quantity is:
WhereinWherein ztFor t when
Carve renewal door, ztDoor, h are reset for t momenttFor t moment output layer, ytIt is s for t moment output valvet-1For the memory at t-1 moment,
rt+1Door, r are reset for the t+1 momentt+1Door is reset for the t+1 moment
Step 3.5:When iterations is more than N or sample overall errorIt is less thanerrorWhen stop circulation.
A kind of adaptive electric system on-line transient stability appraisal procedure of time according to claim 1, it is special
Sign is:The adaptive Transient Stability Evaluation model of application on site time described in step 4, assessment result areα is confidence factor in formula, [0, α) and ∪ (1- α, 1] it is confidential interval.Using temporally
Progressive input sequence, specifically, to mode input x after fault clearance1, stop if assessment result is in confidential interval
Assessment, otherwise continues to input x2, with reference to last moment to x1Assessment gained " memory " h1Make and stability of power system is commented
Estimate.And so on, untill drawing assessment result or reaching assessment time window length T.
Claims (3)
1. the electric system on-line transient stability appraisal procedure that a kind of time is adaptive, it is characterised in that comprise the steps of:
Step 1:The power grid dynamic data under several fault condition is produced using time-domain simulation method, when choosing after failure removal
Between measurable data of power grid Wide Area Measurement System in window T as input feature vector collection { x1,x2,…,xt…xT, wherein, power grid
The measurable data of Wide Area Measurement System refer to each busbar voltage of power grid and its phase angle, xt=[v1,v2,…,vn,θ1,θ2,…,
θn]T, wherein n is busbar number;
Step 2:Feature extraction, which is carried out, using DBN forms high-order feature set { X1,X2,…,Xt…XT, specifically using DBN by one
Softmax layers and m layers limited Boltzmann machine (RBM) stack of layer, which stack, to be formed, the connection weight of unsupervised first RBM of training
And biasing so that the RBM can reconstruct input feature vector with maximum probability, using the output of trained first RBM as next
The input of a RBM carries out feature reconstruction, until unsupervised training is completed in m-th of RBM training, finally adds softmax layers of progress
Training carries out network parameter fine setting, and m-th of RBM layers of output is high-order feature { X1,X2,…,Xt…XT};
Step 3:Off-line training is carried out to GRU, obtains the mapping relations between high-order feature and electric power system transient stability, offline
During training, input as { X1,X2,…,Xt…XT, export as { y1,y2,…,yt,…,yT, training process is by propagated forward and ginseng
Number renewal process composition, sets iterations N and error upper limit error;
Step 4:Application on site time adaptive Transient Stability Evaluation model, successively to after fault clearance each moment it is high-level
Feature XtTransient Stability Evaluation is carried out, assessment confidence level is disclosure satisfy that or reaches time window length and then stop assessing, application on site
The time assessment result of adaptive Transient Stability Evaluation model isIn formula α for confidence because
Son, [0, α) and ∪ (1- α, 1] it is confidential interval;Using temporally progressive input sequence, specifically, to mould after fault clearance
Type inputs x1, stop assessment if assessment result is in confidential interval, otherwise continue to input x2, with reference to last moment to x1Comment
Estimate gained " memory " h1Make the assessment to stability of power system;Until draw assessment result or reach assessment time window
Untill spending T.
2. the electric system on-line transient stability appraisal procedure that a kind of time according to claim 1 is adaptive, its feature
It is:The single unsupervised training process of RBM is as follows in step 2:
Step 2.1, RBM hidden layer each units are separate, in the state v of given visual layers, j-th of neural unit of hidden layer
Activation probability beWhereinbjIt is single for j-th of nerve of hidden layer
The biasing of member, ωijFor the connection weight between i-th of visual layer unit of RBM networks and j-th of implicit layer unit;
Step 2.2, RBM hidden layer each units are separate, in the state h of given hidden layer, i-th of neural unit of visual layers
Activation probability beWherein aiFor i-th of neural unit biasing of hidden layer;
Step 2.3, the input set { v for quantity for s1,v2,…,vs, by maximizing log-likelihoods of the RBM in input set
FunctionObtain model parameter θ, you can using hidden layer output as the feature extraction of visual layers, wherein P
(vk| h, θ) under the state h and parameter θ of known hidden layer visual layers probability, be expressed as with formula:
Step 2.4, parameter more new formula is:Further,In formula, v(0)To be originally inputted collection, v(l)For to v(0)Carry out obtained by l step gibbs samplers, ρ
For momentum term, η is learning rate, bjFor the biasing of j-th of neural unit of hidden layer, ωijFor i-th of visual layer unit of RBM networks
Connection weight between j-th of implicit layer unit, aiFor i-th of neural unit biasing of hidden layer.
3. the electric system on-line transient stability appraisal procedure that a kind of time according to claim 1 is adaptive, its feature
It is:Step 3 specifically includes:
Step 3.1:Perform circulation i=1 to i=N;
Step 3.2:Perform circulation t=1 to t=T;
Step 3.3:Carry out propagated forward process, W(z), U(z)Input and last moment hidden layer are referred respectively to the company of renewal door z
Meet matrix, W(r), U(r)Represent input and last moment hidden layer to the connection matrix for resetting door r, W respectivelyh, UhInput is represented respectively
With the hidden layer of last moment to the connection matrix of output layer, WoutTo export weight matrix;
Step 3.4:Back-propagation process is carried out, this is a process being updated to weighting parameter, defines single sample and misses
Difference isD in formulatFor sample real output value, then its each undated parameter variable quantity is:
WhereinXtInputted for t moment, zt
Door, z are updated for t momenttDoor, h are reset for t momenttFor t moment output layer, ytFor t moment output valve, st-1For the note at t-1 moment
Recall, rt+1Door is reset for the t+1 moment,The output of door error is reset for the t+1 moment;
Step 3.5:When iterations is more than N or sample overall errorStop circulation during less than error.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711260938.7A CN107993012B (en) | 2017-12-04 | 2017-12-04 | Time-adaptive online transient stability evaluation method for power system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711260938.7A CN107993012B (en) | 2017-12-04 | 2017-12-04 | Time-adaptive online transient stability evaluation method for power system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107993012A true CN107993012A (en) | 2018-05-04 |
CN107993012B CN107993012B (en) | 2022-09-30 |
Family
ID=62035576
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711260938.7A Expired - Fee Related CN107993012B (en) | 2017-12-04 | 2017-12-04 | Time-adaptive online transient stability evaluation method for power system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107993012B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763360A (en) * | 2018-05-16 | 2018-11-06 | 北京旋极信息技术股份有限公司 | A kind of sorting technique and device, computer readable storage medium |
CN108960077A (en) * | 2018-06-12 | 2018-12-07 | 南京航空航天大学 | A kind of intelligent failure diagnosis method based on Recognition with Recurrent Neural Network |
CN109239527A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | Distribution network failure recognition methods based on depth confidence network |
CN109376939A (en) * | 2018-11-01 | 2019-02-22 | 三峡大学 | A kind of grid stability real-time predicting method based on adaptive neural network |
CN110336277A (en) * | 2019-07-11 | 2019-10-15 | 福州大学 | A kind of transient stability evaluation in power system method based on depth confidence network |
CN112529684A (en) * | 2020-11-27 | 2021-03-19 | 百维金科(上海)信息科技有限公司 | Customer credit assessment method and system based on FWA _ DBN |
CN113435113A (en) * | 2021-06-22 | 2021-09-24 | 清华大学 | Transient stability evaluation method and device for power system |
CN115775114A (en) * | 2022-12-29 | 2023-03-10 | 国网甘肃省电力公司经济技术研究院 | New energy station transient voltage stability evaluation method based on gated cycle unit |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103778467A (en) * | 2014-01-16 | 2014-05-07 | 天津大学 | Power system transient state stability estimation inputted characteristic quantity selection method |
US8898093B1 (en) * | 2012-06-25 | 2014-11-25 | The Boeing Company | Systems and methods for analyzing data using deep belief networks (DBN) and identifying a pattern in a graph |
CN105512808A (en) * | 2015-11-30 | 2016-04-20 | 武汉大学 | Power system transient stability assessment method based on big data |
CN105608512A (en) * | 2016-03-24 | 2016-05-25 | 东南大学 | Short-term load forecasting method |
CN106529476A (en) * | 2016-11-11 | 2017-03-22 | 重庆邮电大学 | Deep stack network-based electroencephalogram signal feature extraction and classification method |
CN106709820A (en) * | 2017-01-11 | 2017-05-24 | 中国南方电网有限责任公司电网技术研究中心 | Electrical power system load prediction method and device based on depth belief network |
CN106991372A (en) * | 2017-03-02 | 2017-07-28 | 北京工业大学 | A kind of dynamic gesture identification method based on interacting depth learning model |
US20170270410A1 (en) * | 2016-03-18 | 2017-09-21 | Michael James Denham | Neuromorphic training algorithm for a Restricted Boltzmann Machine. |
CN107391852A (en) * | 2017-07-26 | 2017-11-24 | 清华大学 | Transient stability real time evaluating method and device based on depth confidence network |
-
2017
- 2017-12-04 CN CN201711260938.7A patent/CN107993012B/en not_active Expired - Fee Related
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8898093B1 (en) * | 2012-06-25 | 2014-11-25 | The Boeing Company | Systems and methods for analyzing data using deep belief networks (DBN) and identifying a pattern in a graph |
CN103778467A (en) * | 2014-01-16 | 2014-05-07 | 天津大学 | Power system transient state stability estimation inputted characteristic quantity selection method |
CN105512808A (en) * | 2015-11-30 | 2016-04-20 | 武汉大学 | Power system transient stability assessment method based on big data |
US20170270410A1 (en) * | 2016-03-18 | 2017-09-21 | Michael James Denham | Neuromorphic training algorithm for a Restricted Boltzmann Machine. |
CN105608512A (en) * | 2016-03-24 | 2016-05-25 | 东南大学 | Short-term load forecasting method |
CN106529476A (en) * | 2016-11-11 | 2017-03-22 | 重庆邮电大学 | Deep stack network-based electroencephalogram signal feature extraction and classification method |
CN106709820A (en) * | 2017-01-11 | 2017-05-24 | 中国南方电网有限责任公司电网技术研究中心 | Electrical power system load prediction method and device based on depth belief network |
CN106991372A (en) * | 2017-03-02 | 2017-07-28 | 北京工业大学 | A kind of dynamic gesture identification method based on interacting depth learning model |
CN107391852A (en) * | 2017-07-26 | 2017-11-24 | 清华大学 | Transient stability real time evaluating method and device based on depth confidence network |
Non-Patent Citations (2)
Title |
---|
CHUN ZHANG 等: "Feature selection of power system transient stability assessment based on random", 《2016 IEEE PES ASIA-PACIFIC POWER AND ENERGY ENGINEERING CONFERENCE (APPEEC)》 * |
叶圣永: "基于机器学习的电力***暂态稳定评估研究", 《中国优秀博士学位论文全文数据库 工程科技Ⅱ辑(月刊)》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763360A (en) * | 2018-05-16 | 2018-11-06 | 北京旋极信息技术股份有限公司 | A kind of sorting technique and device, computer readable storage medium |
CN108960077A (en) * | 2018-06-12 | 2018-12-07 | 南京航空航天大学 | A kind of intelligent failure diagnosis method based on Recognition with Recurrent Neural Network |
CN109239527A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | Distribution network failure recognition methods based on depth confidence network |
CN109376939A (en) * | 2018-11-01 | 2019-02-22 | 三峡大学 | A kind of grid stability real-time predicting method based on adaptive neural network |
CN110336277A (en) * | 2019-07-11 | 2019-10-15 | 福州大学 | A kind of transient stability evaluation in power system method based on depth confidence network |
CN112529684A (en) * | 2020-11-27 | 2021-03-19 | 百维金科(上海)信息科技有限公司 | Customer credit assessment method and system based on FWA _ DBN |
CN113435113A (en) * | 2021-06-22 | 2021-09-24 | 清华大学 | Transient stability evaluation method and device for power system |
CN113435113B (en) * | 2021-06-22 | 2024-06-11 | 清华大学 | Power system transient stability evaluation method and device |
CN115775114A (en) * | 2022-12-29 | 2023-03-10 | 国网甘肃省电力公司经济技术研究院 | New energy station transient voltage stability evaluation method based on gated cycle unit |
Also Published As
Publication number | Publication date |
---|---|
CN107993012B (en) | 2022-09-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107993012A (en) | A kind of adaptive electric system on-line transient stability appraisal procedure of time | |
CN110535146B (en) | Electric power system reactive power optimization method based on depth determination strategy gradient reinforcement learning | |
CN110705743B (en) | New energy consumption electric quantity prediction method based on long-term and short-term memory neural network | |
CN111612651A (en) | Abnormal electric quantity data detection method based on long-term and short-term memory network | |
CN104951836A (en) | Posting predication system based on nerual network technique | |
CN104636985A (en) | Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network | |
CN113659565B (en) | Online prediction method for frequency situation of new energy power system | |
CN105005708B (en) | A kind of broad sense load Specialty aggregation method based on AP clustering algorithms | |
CN112465664A (en) | AVC intelligent control method based on artificial neural network and deep reinforcement learning | |
CN104899608B (en) | The Weighted Fusion Method Using Relevance Vector Machine model of rolling bearing predicting residual useful life | |
CN114006370B (en) | Power system transient stability analysis and evaluation method and system | |
CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
CN114580545A (en) | Wind turbine generator gearbox fault early warning method based on fusion model | |
CN109408896B (en) | Multi-element intelligent real-time monitoring method for anaerobic sewage treatment gas production | |
CN115049115A (en) | RDPG wind speed correction method considering NWP wind speed transverse and longitudinal errors | |
CN117077546B (en) | Power system load modeling method and system based on data driving | |
CN105205572A (en) | Photovoltaic power generation output forecasting method based on EMD (Empirical Mode Decomposition) and Elman algorithm | |
Liang et al. | Dissolved gas analysis of transformer oil based on Deep Belief Networks | |
Mahato et al. | Electric power system transient stability assessment based on Bi-LSTM attention mechanism | |
Liu et al. | Rapid assessment of flood loss based on neural network ensemble | |
CN109490793A (en) | A kind of X-ray high voltage power supply failure prediction method based on wavelet decomposition and LSTM | |
CN115048857A (en) | CNN-based transient stability limit transmission power determination method | |
Gu et al. | Trend analysis of extreme rainfall based on BP neural network | |
CN114049014A (en) | Method, device and system for evaluating operation state of offshore wind turbine generator | |
CN112685900A (en) | Power load simulation method for representing impact load power characteristics |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220930 |