CN109086913A - A kind of transient stability evaluation in power system method and system based on deep learning - Google Patents
A kind of transient stability evaluation in power system method and system based on deep learning Download PDFInfo
- Publication number
- CN109086913A CN109086913A CN201810759783.XA CN201810759783A CN109086913A CN 109086913 A CN109086913 A CN 109086913A CN 201810759783 A CN201810759783 A CN 201810759783A CN 109086913 A CN109086913 A CN 109086913A
- Authority
- CN
- China
- Prior art keywords
- transient stability
- model
- stability evaluation
- noise reduction
- transient
- 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
- 230000001052 transient effect Effects 0.000 title claims abstract description 152
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000011156 evaluation Methods 0.000 title claims abstract description 44
- 238000013135 deep learning Methods 0.000 title claims abstract description 28
- 230000009467 reduction Effects 0.000 claims abstract description 49
- 238000012706 support-vector machine Methods 0.000 claims abstract description 20
- 239000000284 extract Substances 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 30
- 238000013210 evaluation model Methods 0.000 claims description 29
- 239000013598 vector Substances 0.000 claims description 23
- 238000000605 extraction Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 18
- 238000007726 management method Methods 0.000 claims description 11
- 239000012141 concentrate Substances 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 6
- 238000013097 stability assessment Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 2
- 238000013136 deep learning model Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000012854 evaluation process Methods 0.000 description 6
- 230000003044 adaptive effect Effects 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 108020005210 Integrons Proteins 0.000 description 1
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 244000131316 Panax pseudoginseng Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 206010016256 fatigue Diseases 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000007787 solid Substances 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The Transient Stability Evaluation method and system based on deep learning that the invention discloses a kind of, comprising: building Transient Stability Evaluation primitive character collection；Establish offline contingency set；It is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder Feature Selection Model, realize that the multilayer feature being originally inputted extracts；Support vector machine ensembles regression model is established, judges the transient stability margin of a variety of methods of operation under fixed forecast accident；Generate multiple following methods of operation and online contingency set；Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each forecast accident, Severity gradation is carried out to the transient stability degree of system under different running method.The present invention is combined with existing Distributed Parallel Computing technology, it is final to realize the optimum network structure that adaptively determine the deep learning model with a large amount of network nodes and multiple hidden layers, the transient stability degree that electric system can fast and accurately be assessed again, meets application on site requirement.
Description
Technical field
The invention belongs to Electrical Power System Dynamic security evaluation field more particularly to a kind of electric system based on deep learning
Transient Stability Evaluation method and system.
Background technique
In recent years, China's extra-high voltage alternating current-direct current serial-parallel power grid scale constantly expands, high permeability intermittence generation of electricity by new energy
The uncertainty of grid sources lotus bilateral is exacerbated with the response of magnanimity flexible load, the bulk power grid method of operation and dynamic behaviour are increasingly multiple
Miscellaneous, taking place frequently for natural calamity keeps issuable forecast accident scene more complicated, to the pool level of decision-making of management and running and
Collaborative Control ability proposes requirements at the higher level.At the same time, deepening continuously with smart grid construction, regulation central products at different levels
A large amount of management and running data are tired out, carrying out Electrical Power System Dynamic security evaluation based on data has online calculating speed fast, easy
The advantages that generating the heuristic rule of decision can be constituted good mutual with traditional security and stability analysis method based on model
It mends.Safety on line Stability Assessment, which is carried out, based on data mining technology provides new thinking for the intelligent scheduling control of bulk power grid.
Data digging method currently used for Transient Stability Evaluation is shallow-layer learning model mostly, such as support vector machines, certainly
Plan tree and extreme learning machine etc. are limited in that, computation complexity height limited to the characterization ability of complicated function, generalization ability
Difference.Therefore, deep learning model is introduced into Transient Stability Evaluation problem, such as deepness belief network, stacking autocoder.
Deep learning approaches complicated function using multilayered nonlinear network structure, and the distributed nature for learning input data indicates have
The ability of substantive characteristics is extracted from a small amount of sample set.However, the existing Transient Stability Evaluation model based on deep learning is deposited
In following deficiency: first, how adaptively to determine the deep learning model with a large amount of network nodes and multiple hidden layers most
Good network structure；Second, the degree of stability of system is not assessed.
In conclusion needing a kind of deep learning that can adaptively determine with a large amount of network nodes and multiple hidden layers
The optimum network structure of model, and the method and system of electric power system transient stability degree can be assessed.
Summary of the invention
In order to solve the deficiencies in the prior art, it is steady that the first object of the present invention is to provide a kind of transient state based on deep learning
Determine appraisal procedure, can adaptively determine the best net of the deep learning model with a large amount of network nodes and multiple hidden layers
Network structure, and electric power system transient stability degree can be assessed.
To achieve the goals above, the present invention adopts the following technical scheme:
The first object of the present invention is to disclose a kind of Transient Stability Evaluation method based on deep learning, comprising:
According to Operation of Electric Systems characteristic, Transient Stability Evaluation primitive character collection is constructed；
Offline contingency set is established, corresponding sample set is generated at random to each forecast accident；
Each forecast accident is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder feature extraction mould
Type realizes that the multilayer feature being originally inputted extracts；
To stack each layer feature of noise reduction autocoder extraction as the input of supporting vector loom learner, branch is established
It holds vector machine and integrates regression model, predict the transient stability margin index of a variety of methods of operation under fixed forecast accident；
Based on on-line operation mode, using planning data and prediction data, multiple following methods of operation and online pre- are generated
Think accident set；
Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each fixed forecast accident,
Severity gradation is carried out based on transient stability degree of the utility theory to system under different running method.
Further, the Transient Stability Evaluation primitive character collection includes: system features, fault point and its neighbour before failure
The position of single machine feature and forecast accident near field and duration.
Further, described adaptively established based on Mutual Information Theory to each forecast accident stacks noise reduction autocoding
Device Feature Selection Model.
The hidden layer output for stacking noise reduction autocoder is the abstract expression for being originally inputted different levels, and the number of hidden nodes is determined
The dimension of extraction feature is determined, hidden layers numbers determine the level of abstraction for extracting feature, and the weight vectors between hidden layer will input
Feature Conversion plays the role of " filtering " at more abstract expression, to input feature vector, and multilayer feature may be implemented and automatically extract, and guarantees
The objectivity of characteristic extraction procedure.It is measured using MIFS measure information standard and stacks noise reduction autocoder hidden node power
Weight be originally inputted between correlation and hidden node weight between redundancy, simplify network structure to reach to establish
Purpose.
Further, the transient stability margin index is the difference of critical clearing time and fault clearing time.
Further, severity point is carried out based on transient stability degree of the utility theory to system under different running method
Grade, specifically:
Wherein, T is the threshold value of setting, and M is transient stability margin index；
When transient stability margin index is greater than T, it is believed that system absolutely not Transient Instability risk；Work as transient stability margin
When index is less than 0, it is believed that Transient Instability can occur for system；When transient stability margin index belongs to section [0, T], index is used
Function is as the severity function in the section.
Further, the severity of sample is divided into 5 grades, enables SrSeverity grade is represented, Severity gradation rule is as follows:
Wherein, the sample that severity grade is 3 is near security domain boundaries, and temporarily steady nargin is lower；Severity grade is 4
Sample include unstability sample and neutrality sample.
Further, further includes: constantly obtain on-line operation data from Energy Management System, update Transient Stability Evaluation
The structure and parameter of model, including stack noise reduction autocoder Feature Selection Model structure and parameter update and support to
The update of amount machine integrated model structure and parameter；
When online forecast accident concentrates the forecast accident for occurring not training, the transient state under the on-line training accident is steady
Determine assessment models.
The second object of the present invention is to disclose a kind of transient stability evaluation in power system system based on deep learning, packet
It includes:
Primitive character collection constructs module, is configured as: according to Operation of Electric Systems characteristic, it is former to construct Transient Stability Evaluation
Beginning feature set；
Sample set generation module, is configured as: establishing offline contingency set, generates at random to each forecast accident
Corresponding sample set；
It stacks noise reduction autocoder Feature Selection Model and adaptively establishes module, be configured as: being managed based on mutual information
Noise reduction autocoder Feature Selection Model is stacked by adaptively establishing to each forecast accident, realizes the multilayer being originally inputted
Feature extraction；
Transient Stability Evaluation model building module, is configured as: special with each layer for stacking the extraction of noise reduction autocoder
The input as supporting vector loom learner is levied, support vector machine ensembles regression model is established, is predicted under fixed forecast accident
The transient stability margin of a variety of methods of operation；
The following Run-time scenario generation module, is configured as: being based on on-line operation mode, utilizes planning data and prediction number
According to the multiple following methods of operation of generation and online contingency set；
Transient stability margin prediction and Severity gradation module, are configured as: utilizing the corresponding branch of each forecast accident
It holds vector machine and integrates forecast of regression model transient stability margin, it is steady based on transient state of the utility theory to system under different running method
Determine degree and carries out Severity gradation.
Further, further includes: model online updating module is configured as: constantly obtaining from Energy Management System
On-line operation data update the structure and parameter of Transient Stability Evaluation model, concentrate when online forecast accident and occur not training
Transient Stability Evaluation model when the forecast accident crossed, under the on-line training accident.
The third object of the present invention be a kind of transient stability evaluation in power system system based on deep learning is disclosed, including
Server, the server include memory, processor and storage on a memory and the computer that can run on a processor
Program, the processor perform the steps of when executing described program
According to Operation of Electric Systems characteristic, Transient Stability Evaluation primitive character collection is constructed；
Offline contingency set is established, corresponding sample set is generated at random to each forecast accident；
Each forecast accident is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder feature extraction mould
Type realizes that the multilayer feature being originally inputted extracts；
To stack each layer feature of noise reduction autocoder extraction as the input of supporting vector loom learner, branch is established
It holds vector machine and integrates regression model, predict the transient stability margin index of a variety of methods of operation under fixed forecast accident；
Based on on-line operation mode, using planning data and prediction data, multiple following methods of operation and online pre- are generated
Think accident set；
Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each fixed forecast accident,
Severity gradation is carried out based on transient stability degree of the utility theory to system under different running method.
The invention has the advantages that:
The present invention is based on the Transient Stability Evaluation processes of deep learning to be divided into 3 links.Off-line training link generates at random
Sample set and corresponding offline forecast failure collection, carry out model training to the sample set under each failure respectively, adaptive to establish
Feature Selection Model and Transient Stability Evaluation model under each failure.It is in real time periodically using link Starting mode
(15min) triggering, is based on on-line operation mode, generates multiple following methods of operation and online contingency set, using each pre-
Think that the corresponding model of accident carries out Transient Stability Evaluation.Online updating link obtains on-line operation number from Energy Management System
According to the structure and parameter of update Transient Stability Evaluation model, when online forecast accident concentrates the anticipation thing for occurring not training
Therefore when, the Transient Stability Evaluation model under the on-line training accident.
The present invention stacks adaptively establishing for noise reduction autocoder Feature Selection Model since each failure is corresponding
Journey and Transient Stability Evaluation process are mutually indepedent, and above 3 links can be combined with existing Distributed Parallel Computing technology,
It is final to realize the optimum network knot that adaptively determine the deep learning model with a large amount of network nodes and multiple hidden layers
Structure, and the transient stability degree of electric system can be fast and accurately assessed, meet application on site requirement.
Detailed description of the invention
The accompanying drawings constituting a part of this application is used to provide further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation are not constituted an undue limitation on the present application for explaining the application.
Fig. 1 is the Transient Stability Evaluation method flow diagram of the invention based on deep learning；
Fig. 2 is stacking noise reduction autocoder Feature Selection Model structural schematic diagram of the invention；
Fig. 3 is support vector machine ensembles regression model structural schematic diagram of the invention；
Fig. 4 is the Transient Stability Evaluation system schematic of the invention based on deep learning.
Specific embodiment
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
The present invention includes three links, carries out power system transient stability margin prediction.Off-line training link realizes transient state
The off-line training of Stability Assessment model, including step 1~step 4 are commented using link using the transient stability that training is completed in real time
Estimate model to be assessed in real time, include step 5~step 6, online updating link realize Transient Stability Evaluation model it is online more
Newly to enhance generalization ability, including step 7, as shown in Fig. 1.
Off-line training link: the off-line training of Transient Stability Evaluation model is realized
Step 1: according to Operation of Electric Systems characteristic, constructing one group of Transient Stability Evaluation primitive character collection.Transient stability is commented
Estimate primitive character collection to be made of three parts feature.
Since behavioral characteristics are to the more demanding of dynamic phasor measurement system, and it is steady in order to avoid carrying out transient state when sentencing steady
Devise a stratagem is calculated, and primitive character collection selects static nature.The a part of system features as primitive character collection before selecting failure, can be with
Characterize the operation conditions of electric system entirety；If selecting whole single machine features that can will appear " dimension disaster " problem, by
" relaxation with one heart " theoretical inspiration, the fault message that closer region includes with fault point electrical distance is more, therefore selects event
The a part of single machine feature as primitive character collection in barrier point and its adjacent domain；Due to only considering failure effect most serious
Three phase short circuit fault, so the position of forecast accident and duration to be selected as to a part of primitive character collection.
Step 2: consider the factors such as topological structure of electric, generator output and load variations, establishes offline forecast failure collection,
Corresponding sample set is generated at random to each forecast accident.
Step 3: the stacking noise reduction autocoder feature adaptively established under security domain concept based on Mutual Information Theory is mentioned
Modulus type realizes that the multilayer feature being originally inputted extracts, as shown in Figure 2 automatically.
Mutual information measures two complementary degree of variable, indicates the content that information is co-owned between two variables.It is false
If sample set S is by with n dimensional feature (f1,f2,…,fn) N number of sample composition, P (fi) it is feature fiFor the general of different probable values
Rate.fiThe uncertainty of value is measured by following comentropy:
On this basis, two feature fiWith ftBetween mutual information can be defined as
I(fi；ft)=H (fi)+H(ft)-H(fi；ft)
Vacation lets d be each feature f in sample set SiThe mean value of mutual information between class label C indicates feature set and corresponding
The correlation of classification；R is the size of mutual information between feature in sample set S, indicates the redundancy between feature:
Using MIFS measure information standard, evaluation function J (f) is
Max J (f)=D- β R
Wherein, β is adjustment factor, and when β ∈ [0.5,1], algorithm performance is preferable.
The hidden layer output for stacking noise reduction autocoder is the abstract expression for being originally inputted different levels, and the number of hidden nodes is determined
The dimension of extraction feature is determined, hidden layers numbers determine the level of abstraction for extracting feature, and the weight vectors between hidden layer will input
Feature Conversion plays the role of " filtering " at more abstract expression, to input feature vector, and multilayer feature may be implemented and automatically extract, and guarantees
The objectivity of characteristic extraction procedure.It is measured using MIFS measure information standard and stacks noise reduction autocoder hidden node power
Weight be originally inputted between correlation and hidden node weight between redundancy, simplify network structure to reach to establish
Purpose.
The detailed process for stacking noise reduction autocoder model is established based on mutual information are as follows:
Step 1: defining the lesser initial stacking noise reduction autocoder network structure of a scale, input layer number
Equal to input feature vector dimension；Concurrently set J (f) threshold value.
Step 2: determining the number of nodes of i-th of hidden layer.Increase m node newly, it is whole to carry out weights initialisation, it calculates existing
The J (f) of node weight vector.
Step 3: judging whether J (f) is greater than threshold value.If it is less than threshold value, second step is returned；Otherwise newly-increased node is deleted,
Carry out the 4th step.
Step 4: judging whether m is equal to 1.If m is not equal to 1, node is increased using step length changing method, For
It is rounded downwards, returns to second step；If m is equal to 1, it is the number of hidden nodes that 1 is subtracted from present node number.
Step 5: running the stacking noise reduction autocoder of established i hidden layer, judge temporarily whether steady Evaluated effect has
Improve, i.e. whether the precision of transient stability margin prediction improves.If there is improving, i=i+1 returns to step 2；If do not changed
Kind, i-1 hidden layer is to stack the final network structure of noise reduction autocoder.
Step 6: being exported using i-1 hidden layer for stacking noise reduction autocoder, i.e., the spy of i-1 different level of abstractions
Input of the sign expression as subsequent Transient Stability Evaluation, is used for subsequent bulk power grid security risk Situation Awareness.
Step 4: the support vector machine ensembles regression model under security domain concept, such as Fig. 3 are established based on integrated learning approach
It is shown.
Given sample set { xi, i=1,2 ..., n use and stack noise reduction autocoder as trainable feature extraction
Tool, it is assumed that stacking noise reduction autocoder has N number of hidden layer, obtains level characteristics hj(j=1,2 ..., N), hjIntrinsic dimensionality
It is identical with the number of hidden nodes；Use feature set hjThe corresponding supporting vector machine model of training, that is, use different input feature vector collection
It constructs the base with otherness and returns device；Using the output of " method of average " integron learner.
The output y of support vector machine ensembles regression modelI:
In formula: yiThe transient stability margin predicted value of device is returned for i-th of son.
Link is applied in real time: being assessed in real time using the Transient Stability Evaluation model that training is completed.
Step 5: being based on on-line operation mode, utilize planning data (such as section power plan, maintenance plan) and prediction
Data (such as ultra-short term) generate multiple following methods of operation and online contingency set.
Step 6: utilizing the corresponding support vector machine ensembles forecast of regression model transient stability margin of each forecast accident, base
The transient stability degree of system carries out Severity gradation under utility theory is to different running method.
Using the transient stability margin index M based on critical clearing time:
M=tCCT-tcl
In formula: tclIt is fault clearing time, tCCTIt is critical clearing time.
As M > 0, system is stablized, otherwise unstability.The temporary of sample in three subregions is respectively obtained using method shown in Fig. 4
Steady nargin constructs severity function based on effect theory: setting threshold value T, when M is greater than T, it is believed that system absolutely not transient state is lost
Steady risk, severity functional value are 0；When M is less than 0, it is believed that Transient Instability can occur for system, and severity functional value is 3；Work as M
When belonging to section [0, T], use exponential function as the severity function in the section:
Sm=a1e-M+a2
In formula: a1And a2It is coefficient.
Since severity function is continuous function, so function crosses (T, 0) and (0,3) two o'clock, two o'clock coordinate is brought into, is obtained
To exponential function are as follows:
In conclusion obtaining the severity function of temporarily steady nargin:
The severity of sample is divided into 5 grades, enables SrSeverity grade is represented, then obtains the Severity gradation rule of this paper:
The sample that severity grade is 3 is near security domain boundaries, and temporarily steady nargin is lower；The sample that severity grade is 4
This includes unstability sample and neutrality sample.In actual schedule operation, the method for operation for being 3 and 4 by severity grade is excellent
It is first presented to dispatcher, Control Measure is formulated for the high-risk method of operation for dispatcher and reference information is provided.
Online updating link: the online updating of Transient Stability Evaluation model is realized.
Step 7: obtaining on-line operation data from Energy Management System, update the structure and ginseng of Transient Stability Evaluation model
Number.When online forecast accident concentrates the forecast accident for occurring not training, the transient stability under the on-line training accident is commented
Estimate model.
When the structure of update Transient Stability Evaluation model refers to on-line operation, new online fortune is constantly obtained from EMS
The network structure of row data update transient stability model；The structure and parameter for updating Transient Stability Evaluation model includes: including heap
The update of folded noise reduction autocoder Feature Selection Model structure and parameter and support vector machine ensembles model structure and parameter
Update.
Transient Stability Evaluation process based on deep learning of the invention is divided into 3 links.Off-line training link is given birth at random
At sample set and corresponding offline forecast failure collection, model training is carried out to the sample set under each failure respectively, is obtained each
Feature Selection Model and Transient Stability Evaluation model under failure.It is in real time periodically (15min) touching using link Starting mode
Hair is based on on-line operation mode, generates multiple following methods of operation and online contingency set, corresponding using each forecast accident
Model carry out Transient Stability Evaluation.Online updating link obtains on-line operation data from Energy Management System, updates transient state
The structure and parameter of Stability Assessment model, it is online to instruct when online forecast accident concentrates the forecast accident for occurring not training
Practice the Transient Stability Evaluation model under the accident.The present invention is mentioned due to the corresponding stacking noise reduction autocoder feature of each failure
The adaptive establishment process and Transient Stability Evaluation process of modulus type are mutually indepedent, and above 3 links can be with existing distribution
Formula parallel computing combines, and final realize can adaptively determine the depth with a large amount of network nodes and multiple hidden layers
The optimum network structure of learning model, and the transient stability degree of electric system can be fast and accurately assessed, meet application on site
It is required that.
A kind of transient stability evaluation in power system system based on deep learning of the invention, as shown in Figure 4, comprising:
(1) primitive character collection constructs module, is configured as: according to Operation of Electric Systems characteristic, it is steady to construct one group of transient state
Accepted opinion estimates primitive character collection；
In primitive character collection building module, Transient Stability Evaluation primitive character collection is made of three parts feature.
Since behavioral characteristics are to the more demanding of dynamic phasor measurement system, and it is steady in order to avoid carrying out transient state when sentencing steady
Devise a stratagem is calculated, and primitive character collection selects static nature.The a part of system features as primitive character collection before selecting failure, can be with
Characterize the operation conditions of electric system entirety；If selecting whole single machine features that can will appear " dimension disaster " problem, by
" relaxation with one heart " theoretical inspiration, the fault message that closer region includes with fault point electrical distance is more, therefore selects event
The a part of single machine feature as primitive character collection in barrier point and its adjacent domain；Due to only considering failure effect most serious
Three phase short circuit fault, so the position of forecast accident and duration to be selected as to a part of primitive character collection.
(2) sample set generation module is configured as: considering topological structure of electric, generator output and load variations etc.
Factor establishes offline forecast failure collection, generates corresponding sample set at random to each forecast accident；
(3) it stacks noise reduction autocoder Feature Selection Model and adaptively establishes module, be configured as: based on mutual information
Theory to each forecast accident adaptively establish stack noise reduction autocoder Feature Selection Model, realize be originally inputted it is more
Layer feature extraction；
Established in module in the stacking noise reduction autocoder Feature Selection Model, using MIFS measure information standard come
It measures between the correlation and hidden node weight between stacking noise reduction autocoder hidden node weight and being originally inputted
Redundancy, with achieve the purpose that establish simplify network structure.Feature Selection Model under the security domain concept refers to be
Tide flow before failure of uniting is established a stacking noise reduction autocoder feature to each forecast accident and is mentioned as input feature vector
Modulus type.
(4) Transient Stability Evaluation model building module is configured as: establishing security domain concept based on integrated learning approach
Under support vector machine ensembles regression model；
In the Transient Stability Evaluation model building module, the support vector machine ensembles under the security domain concept are returned
Model refers to that the input of each layer feature to stack the extraction of noise reduction autocoder as supporting vector loom learner, judgement are solid
Determine the transient stability of a variety of methods of operation under forecast accident, if there are unstability risks for system, pre- prevention and control can be taken in time
Measure processed.
(5) the following Run-time scenario generation module, is configured as: based on on-line operation mode, using planning data (as broken
Face power planning, maintenance plan etc.) and prediction data (such as ultra-short term), generate multiple following methods of operation and
Line contingency set；
(6) transient stability margin prediction and Severity gradation module, are configured as: corresponding using each forecast accident
Support vector machine ensembles forecast of regression model transient stability margin, based on utility theory to the transient state of system under different running method
Degree of stability carries out Severity gradation.
In transient stability margin prediction and Severity gradation module, when transient stability margin index is limit excision
Between and fault clearing time difference；Based on effect theory structural index function as severity function, system operation people is embodied
Member meets the actual conditions of electric system to the psychological bearing capability of temporary steady nargin variation.
(7) model online updating module, is configured as: obtaining on-line operation data from Energy Management System, updates
The structure and parameter of Transient Stability Evaluation model.When online forecast accident concentrates the forecast accident for occurring not training,
Transient Stability Evaluation model under the line training accident.
When the structure of update Transient Stability Evaluation model refers to on-line operation, new online fortune is constantly obtained from EMS
The network structure of row data update transient stability model；The structure and parameter for updating Transient Stability Evaluation model includes: including heap
The update of folded noise reduction autocoder Feature Selection Model structure and parameter and support vector machine ensembles model structure and parameter
Update.
The present invention is based on the Transient Stability Evaluation processes of deep learning to be divided into 3 links.Off-line training link generates at random
Sample set and corresponding offline forecast failure collection, carry out model training to the sample set under each failure respectively, adaptive to establish
Feature Selection Model and Transient Stability Evaluation model under each failure.It is in real time periodically using link Starting mode
(15min) triggering, is based on on-line operation mode, generates multiple following methods of operation and online contingency set, using each pre-
Think that the corresponding model of accident carries out Transient Stability Evaluation.Online updating link obtains on-line operation number from Energy Management System
According to the structure and parameter of update Transient Stability Evaluation model, when online forecast accident concentrates the anticipation thing for occurring not training
Therefore when, the Transient Stability Evaluation model under the on-line training accident.The present invention is automatic due to the corresponding stacking noise reduction of each failure
The adaptive establishment process and Transient Stability Evaluation process of encoder feature extraction model are mutually indepedent, above 3 links
It is combined with existing Distributed Parallel Computing technology, final realize can adaptively determine there is a large amount of network nodes and multiple
The optimum network structure of the deep learning model of hidden layer, and the transient stability degree of electric system can be fast and accurately assessed,
Meet application on site requirement.
The transient stability evaluation in power system system based on deep learning that the present invention further discloses a kind of, including service
Device, the server include memory, processor and storage on a memory and the computer program that can run on a processor,
The processor performs the steps of when executing described program
According to Operation of Electric Systems characteristic, Transient Stability Evaluation primitive character collection is constructed；
Offline contingency set is established, corresponding sample set is generated at random to each forecast accident；
Each forecast accident is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder feature extraction mould
Type realizes that the multilayer feature being originally inputted extracts；
To stack each layer feature of noise reduction autocoder extraction as the input of supporting vector loom learner, branch is established
It holds vector machine and integrates regression model, judge the transient stability margin of a variety of methods of operation under fixed forecast accident；
Based on on-line operation mode, using planning data and prediction data, multiple following methods of operation and online pre- are generated
Think accident set；
Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each forecast accident, it is based on
Utility theory carries out Severity gradation to the transient stability degree of system under different running method.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.
Claims (10)
1. a kind of Transient Stability Evaluation method based on deep learning characterized by comprising
According to Operation of Electric Systems characteristic, Transient Stability Evaluation primitive character collection is constructed；
Offline contingency set is established, corresponding sample set is generated at random to each forecast accident；
Each forecast accident is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder Feature Selection Model, it is real
The multilayer feature being now originally inputted extracts；
Using stack noise reduction autocoder extraction each layer feature be used as supporting vector loom learner input, establish support to
Amount machine integrates regression model, predicts the transient stability margin index of a variety of methods of operation under fixed forecast accident；
It generates multiple following methods of operation using planning data and prediction data based on on-line operation mode and envisions thing online
Gu Ji；
Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each fixed forecast accident, it is based on
Utility theory carries out Severity gradation to the transient stability degree of system under different running method.
2. a kind of Transient Stability Evaluation method based on deep learning as described in claim 1, which is characterized in that the transient state
Stability Assessment primitive character collection includes: system features before failure, fault point and its single machine feature in adjacent domain and pre-
Think position and the duration of accident.
3. a kind of Transient Stability Evaluation method based on deep learning as described in claim 1, which is characterized in that described to be based on
Mutual Information Theory adaptively establishes each forecast accident and stacks noise reduction autocoder Feature Selection Model, specifically:
Step 1: defining the lesser initial stacking noise reduction autocoder network structure of a scale, input layer number is equal to
Input feature vector dimension；Concurrently set the threshold value of the evaluation function J (f) based on MIFS measure information standard.
Step 2: determining the number of nodes of i-th of hidden layer.Increase m node newly, it is whole to carry out weights initialisation, calculate existing node
The J (f) of weight vectors.
Step 3: judging whether J (f) is greater than threshold value.If it is less than threshold value, second step is returned；Otherwise newly-increased node is deleted, is carried out
4th step.
Step 4: judging whether m is equal to 1.If m is not equal to 1, node is increased using step length changing method, It is downward
It is rounded, returns to second step；If m is equal to 1, it is the number of hidden nodes that 1 is subtracted from present node number.
Step 5: running the stacking noise reduction autocoder of established i hidden layer, judges temporarily whether steady Evaluated effect has and change
Kind, i.e. whether the precision of transient stability margin prediction improves.If there is improving, i=i+1 returns to step 2；If do not improved,
I-1 hidden layer is to stack the final network structure of noise reduction autocoder.
Step 6: being exported using i-1 hidden layer for stacking noise reduction autocoder, i.e., the mark sheet of i-1 different level of abstractions
Up to the input as subsequent Transient Stability Evaluation, it to be used for subsequent bulk power grid security risk Situation Awareness.
4. a kind of Transient Stability Evaluation method based on deep learning as described in claim 1, which is characterized in that the transient state
Stability margin index is the difference of critical clearing time and fault clearing time.
5. a kind of Transient Stability Evaluation method based on deep learning as described in claim 1, which is characterized in that be based on effectiveness
Theory carries out Severity gradation to the transient stability degree of system under different running method, specifically:
Wherein, T is the threshold value of setting, and M is transient stability margin index；
When transient stability margin index is greater than T, it is believed that system absolutely not Transient Instability risk；Work as transient stability margin index
When less than 0, it is believed that Transient Instability can occur for system；When transient stability margin index belongs to section [0, T], exponential function is used
As the severity function in the section.
6. a kind of Transient Stability Evaluation method based on deep learning as claimed in claim 5, which is characterized in that by sample
Severity is divided into 5 grades, enables SrSeverity grade is represented, Severity gradation rule is as follows:
Wherein, the sample that severity grade is 3 is near security domain boundaries, and temporarily steady nargin is lower；The sample that severity grade is 4
This includes unstability sample and neutrality sample.
7. a kind of Transient Stability Evaluation method based on deep learning as claimed in claim 5, which is characterized in that further include:
On-line operation data are constantly obtained from Energy Management System, update the structure and parameter of Transient Stability Evaluation model, including stack
The update of noise reduction autocoder Feature Selection Model structure and parameter and support vector machine ensembles model structure and parameter
It updates；
When online forecast accident concentrates the forecast accident for occurring not training, the transient stability under the on-line training accident is commented
Estimate model.
8. a kind of transient stability evaluation in power system system based on deep learning characterized by comprising
Primitive character collection constructs module, is configured as: according to Operation of Electric Systems characteristic, constructing the original spy of Transient Stability Evaluation
Collection；
Sample set generation module, is configured as: establishing offline contingency set, generates at random to each forecast accident corresponding
Sample set；
It stacks noise reduction autocoder Feature Selection Model and adaptively establishes module, be configured as: based on Mutual Information Theory pair
Each forecast accident, which is adaptively established, stacks noise reduction autocoder Feature Selection Model, realizes the multilayer feature being originally inputted
It extracts；
Transient Stability Evaluation model building module, is configured as: being made with stacking each layer feature of noise reduction autocoder extraction
For the input of supporting vector loom learner, support vector machine ensembles regression model is established, is predicted a variety of under fixed forecast accident
The transient stability margin of the method for operation；
The following Run-time scenario generation module, is configured as: it is based on on-line operation mode, using planning data and prediction data,
Generate multiple following methods of operation and online contingency set；
Transient stability margin prediction and Severity gradation module, be configured as: using each forecast accident it is corresponding support to
Amount machine integrates forecast of regression model transient stability margin, based on utility theory to the transient stability journey of system under different running method
Degree carries out Severity gradation.
9. a kind of transient stability evaluation in power system system based on deep learning as claimed in claim 8, which is characterized in that
Further include: model online updating module is configured as: constantly being obtained on-line operation data from Energy Management System, is updated
The structure and parameter of Transient Stability Evaluation model, when online forecast accident concentrates the forecast accident for occurring not training,
Transient Stability Evaluation model under the line training accident.
10. a kind of transient stability evaluation in power system system based on deep learning, which is characterized in that described including server
Server include memory, processor and storage on a memory and the computer program that can run on a processor, the place
Reason device performs the steps of when executing described program
According to Operation of Electric Systems characteristic, Transient Stability Evaluation primitive character collection is constructed；
Offline contingency set is established, corresponding sample set is generated at random to each forecast accident；
Each forecast accident is adaptively established based on Mutual Information Theory and stacks noise reduction autocoder Feature Selection Model, it is real
The multilayer feature being now originally inputted extracts；
Using stack noise reduction autocoder extraction each layer feature be used as supporting vector loom learner input, establish support to
Amount machine integrates regression model, predicts the transient stability margin index of a variety of methods of operation under fixed forecast accident；
It generates multiple following methods of operation using planning data and prediction data based on on-line operation mode and envisions thing online
Gu Ji；
Using the corresponding support vector machine ensembles forecast of regression model transient stability margin index of each fixed forecast accident, it is based on
Utility theory carries out Severity gradation to the transient stability degree of system under different running method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810759783.XA CN109086913B (en) | 2018-07-11 | 2018-07-11 | Power system transient stability assessment method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810759783.XA CN109086913B (en) | 2018-07-11 | 2018-07-11 | Power system transient stability assessment method and system based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109086913A true CN109086913A (en) | 2018-12-25 |
| CN109086913B CN109086913B (en) | 2020-06-09 |
Family
ID=64837461
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810759783.XA Active CN109086913B (en) | 2018-07-11 | 2018-07-11 | Power system transient stability assessment method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109086913B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109787222A (en) * | 2019-01-04 | 2019-05-21 | 三峡大学 | Distribution based on support vector machines is transported to electric system transient stability discrimination method |
| CN109829627A (en) * | 2019-01-04 | 2019-05-31 | 三峡大学 | A kind of safe confidence appraisal procedure of Electrical Power System Dynamic based on integrated study scheme |
| CN110298398A (en) * | 2019-06-25 | 2019-10-01 | 大连大学 | Wireless protocols frame feature selection approach based on improved mutual imformation |
| CN110728457A (en) * | 2019-10-17 | 2020-01-24 | 广西电网有限责任公司电力科学研究院 | Operation risk situation perception method considering multi-level weak links of power distribution network |
| CN111049159A (en) * | 2019-12-20 | 2020-04-21 | 四川大学 | Power system transient stability prevention control method embedded into deep belief network |
| CN111079805A (en) * | 2019-12-03 | 2020-04-28 | 浙江工业大学 | Abnormal image detection method combining attention mechanism and information entropy minimization |
| CN111311008A (en) * | 2020-02-21 | 2020-06-19 | 山东大学 | AC/DC large power grid dynamic security risk situation rolling prospective early warning method and system |
| CN111628501A (en) * | 2020-06-18 | 2020-09-04 | 国网山东省电力公司济南供电公司 | AC/DC large power grid transient voltage stability assessment method and system |
| CN111652479A (en) * | 2020-05-19 | 2020-09-11 | 三峡大学 | Data driving method for dynamic security assessment of power system |
| CN111753963A (en) * | 2020-06-28 | 2020-10-09 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for outputting information |
| CN111950765A (en) * | 2020-07-06 | 2020-11-17 | 四川大川云能科技有限公司 | Probabilistic transient stability prediction method based on stacked noise reduction self-encoder |
| CN112232427A (en) * | 2020-10-13 | 2021-01-15 | 宁波大学 | Wind driven generator fault detection method based on distributed regression model |
| CN112865118A (en) * | 2021-01-19 | 2021-05-28 | 河海大学 | Deep learning model generation method for power grid dynamic reactive power reserve demand calculation |
| CN114330884A (en) * | 2021-12-29 | 2022-04-12 | 山东大学 | Multi-time scale coordinated alternating current and direct current power grid dynamic safety early warning method and system |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107846012A (en) * | 2017-09-27 | 2018-03-27 | 国网湖北省电力公司 | A kind of transient stability evaluation in power system method based on depth learning technology |
-
2018
- 2018-07-11 CN CN201810759783.XA patent/CN109086913B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107846012A (en) * | 2017-09-27 | 2018-03-27 | 国网湖北省电力公司 | A kind of transient stability evaluation in power system method based on depth learning technology |
Non-Patent Citations (4)
| Title |
|---|
| CHANGJIANG JIANG,ETC.: "A new numerical simulation for stochastic transient stability analysis of power systems integrated wind power", 《IEEE》 * |
| 尹雪燕: "基于深度学习的暂态稳定评估与严重度分级", 《电力自动化设备》 * |
| 李杨: "基于广域动态信息的电力***暂态稳定评估研究", 《中国博士学位论文全文数据库》 * |
| 管霖: "基于模式匹配的智能稳定评估方法", 《电力自动化设备》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109787222A (en) * | 2019-01-04 | 2019-05-21 | 三峡大学 | Distribution based on support vector machines is transported to electric system transient stability discrimination method |
| CN109829627A (en) * | 2019-01-04 | 2019-05-31 | 三峡大学 | A kind of safe confidence appraisal procedure of Electrical Power System Dynamic based on integrated study scheme |
| CN110298398A (en) * | 2019-06-25 | 2019-10-01 | 大连大学 | Wireless protocols frame feature selection approach based on improved mutual imformation |
| CN110728457A (en) * | 2019-10-17 | 2020-01-24 | 广西电网有限责任公司电力科学研究院 | Operation risk situation perception method considering multi-level weak links of power distribution network |
| CN110728457B (en) * | 2019-10-17 | 2022-11-22 | 广西电网有限责任公司电力科学研究院 | Operation risk situation perception method considering multi-level weak links of power distribution network |
| CN111079805A (en) * | 2019-12-03 | 2020-04-28 | 浙江工业大学 | Abnormal image detection method combining attention mechanism and information entropy minimization |
| CN111049159A (en) * | 2019-12-20 | 2020-04-21 | 四川大学 | Power system transient stability prevention control method embedded into deep belief network |
| CN111049159B (en) * | 2019-12-20 | 2023-09-29 | 四川大学 | Power system transient stability prevention control method embedded in deep belief network |
| CN111311008A (en) * | 2020-02-21 | 2020-06-19 | 山东大学 | AC/DC large power grid dynamic security risk situation rolling prospective early warning method and system |
| CN111311008B (en) * | 2020-02-21 | 2021-04-30 | 山东大学 | AC/DC large power grid dynamic security risk situation rolling prospective early warning method and system |
| CN111652479A (en) * | 2020-05-19 | 2020-09-11 | 三峡大学 | Data driving method for dynamic security assessment of power system |
| CN111628501A (en) * | 2020-06-18 | 2020-09-04 | 国网山东省电力公司济南供电公司 | AC/DC large power grid transient voltage stability assessment method and system |
| CN111753963A (en) * | 2020-06-28 | 2020-10-09 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for outputting information |
| CN111753963B (en) * | 2020-06-28 | 2024-04-02 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for outputting information |
| CN111950765A (en) * | 2020-07-06 | 2020-11-17 | 四川大川云能科技有限公司 | Probabilistic transient stability prediction method based on stacked noise reduction self-encoder |
| CN111950765B (en) * | 2020-07-06 | 2024-04-19 | 四川大川云能科技有限公司 | Probabilistic transient stability prediction method based on stacked noise reduction self-encoder |
| CN112232427B (en) * | 2020-10-13 | 2023-10-03 | 宁波大学 | Wind driven generator fault detection method based on distributed regression model |
| CN112232427A (en) * | 2020-10-13 | 2021-01-15 | 宁波大学 | Wind driven generator fault detection method based on distributed regression model |
| CN112865118B (en) * | 2021-01-19 | 2022-09-02 | 河海大学 | Deep learning model generation method for power grid dynamic reactive power reserve demand calculation |
| CN112865118A (en) * | 2021-01-19 | 2021-05-28 | 河海大学 | Deep learning model generation method for power grid dynamic reactive power reserve demand calculation |
| CN114330884A (en) * | 2021-12-29 | 2022-04-12 | 山东大学 | Multi-time scale coordinated alternating current and direct current power grid dynamic safety early warning method and system |
| CN114330884B (en) * | 2021-12-29 | 2024-04-16 | 山东大学 | Multi-time scale coordinated AC/DC power grid dynamic safety early warning method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109086913B (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109086913A (en) | A kind of transient stability evaluation in power system method and system based on deep learning | |
| Zhu et al. | An exact microgrid formation model for load restoration in resilient distribution system | |
| Panda et al. | Security constrained optimal power flow solution of wind-thermal generation system using modified bacteria foraging algorithm | |
| Trivedi et al. | Implementation of artificial intelligence techniques in microgrid control environment: Current progress and future scopes | |
| EP3889855A1 (en) | Ac/dc bulk power grid dynamic security risk situation rolling prospective early warning method and system | |
| CN110311376A (en) | A kind of Electrical Power System Dynamic security evaluation collective model and space-time method for visualizing | |
| Amraee et al. | An enhanced under-voltage load-shedding scheme to provide voltage stability | |
| Khorramdel et al. | Optimal stochastic reactive power scheduling in a microgrid considering voltage droop scheme of DGs and uncertainty of wind farms | |
| Jayasankar et al. | Estimation of voltage stability index for power system employing artificial neural network technique and TCSC placement | |
| Wang et al. | Emergency load shedding strategy for microgrids based on dueling deep Q-learning | |
| CN107769191A (en) | key transmission cross-section automatic identifying method based on network topology and flow data | |
| Tamilselvan et al. | A hybrid method for optimal load shedding and improving voltage stability | |
| Ma et al. | Multi-objective dynamic optimal power flow of wind integrated power systems considering demand response | |
| Jikeng et al. | Two-stage method for optimal island partition of distribution system with distributed generations | |
| Oboudi et al. | A feasible method for controlled intentional islanding in microgrids based on PSO algorithm | |
| Adinolfi et al. | Pseudo-measurements modeling using neural network and Fourier decomposition for distribution state estimation | |
| CN110676839A (en) | Method and system for evaluating reliability of alternating current-direct current hybrid power distribution network | |
| CN109102132A (en) | A kind of micro-capacitance sensor vulnerability assessment method and device based on probability theory | |
| CN107025524A (en) | Consider the apparatus overload aid decision-making method of load power supply reliability requirement | |
| Afraz et al. | Optimal sizing and sitting in radial standard system using pso | |
| Arabzadeh et al. | A new mechanism for remedial action schemes design in a multi-area power system considering competitive participation of multiple electricity market players | |
| CN103886514B (en) | Power grid development assessment method with economical efficiency and security considered | |
| Pavella | Power system transient stability assessment–traditional vs modern methods | |
| CN106329515B (en) | A kind of horizontal determination method of electric network reliability based on static reliability probability level | |
| Karapidakis | Machine learning for frequency estimation of power systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |