CN106296044A - power system risk scheduling method and system - Google Patents
power system risk scheduling method and system Download PDFInfo
- Publication number
- CN106296044A CN106296044A CN201610882652.1A CN201610882652A CN106296044A CN 106296044 A CN106296044 A CN 106296044A CN 201610882652 A CN201610882652 A CN 201610882652A CN 106296044 A CN106296044 A CN 106296044A
- Authority
- CN
- China
- Prior art keywords
- matrix
- knowledge matrix
- knowledge
- risk scheduling
- power system
- 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 25
- 239000011159 matrix material Substances 0.000 claims abstract description 205
- 238000005457 optimization Methods 0.000 claims abstract description 46
- 230000001580 bacterial effect Effects 0.000 claims abstract description 24
- 230000000844 anti-bacterial effect Effects 0.000 claims description 51
- 238000012545 processing Methods 0.000 claims description 24
- 230000010415 tropism Effects 0.000 claims description 17
- 230000005283 ground state Effects 0.000 claims description 13
- 230000009467 reduction Effects 0.000 claims description 13
- 230000001143 conditioned effect Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 4
- 241000894006 Bacteria Species 0.000 claims description 3
- 230000006870 function Effects 0.000 abstract description 34
- 238000004422 calculation algorithm Methods 0.000 abstract description 24
- 238000013508 migration Methods 0.000 abstract description 14
- 230000005012 migration Effects 0.000 abstract description 13
- 230000013016 learning Effects 0.000 abstract description 11
- 238000013526 transfer learning Methods 0.000 abstract description 7
- 230000002431 foraging effect Effects 0.000 abstract description 4
- 230000002787 reinforcement Effects 0.000 abstract 2
- 230000009471 action Effects 0.000 description 18
- 239000000446 fuel Substances 0.000 description 5
- 229940088710 antibiotic agent Drugs 0.000 description 4
- 230000006399 behavior Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000002068 genetic effect Effects 0.000 description 3
- 230000000750 progressive effect Effects 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 230000001174 ascending effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000005303 weighing 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
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (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 invention relates to a risk scheduling method and a system of an electric power system, which are used for acquiring architecture data and new task load section data of the electric power system; and according to the architecture data and the new task load section data, carrying out iterative updating on a preset initial knowledge matrix through a bacterial foraging reinforcement learning algorithm to obtain a corresponding risk scheduling objective function value and an updated knowledge matrix. And performing new task online optimization according to the updated knowledge matrix corresponding to the minimum risk scheduling objective function value to obtain and output a risk scheduling optimization result. And (3) realizing knowledge migration by taking the optimal knowledge matrix in the source task as an initial matrix of the new task, and performing online optimization on the new task by using bacterial foraging reinforcement learning based on the knowledge migration. The online learning speed is greatly improved through the transfer learning, the online dynamic optimization of the risk scheduling problem is realized, the faster solving speed can be still ensured when the problem scale is further enlarged, and the method is suitable for the rapid optimization of large-scale complex risk scheduling.
Description
Technical field
The present invention relates to electric power network technical field, particularly relate to a kind of power system Risk Scheduling method and system.
Background technology
In recent years, with the development of regional power grid interconnection with high pressure distance large capacity transmission, the safety of power system is steady
Determine operation and be faced with more stern challenge.For preferably weighing security of system and economic benefit, strengthen scheduling operation and resist
The level of operation risk, introduces the Risk Theory of power system in generation optimization, and Risk Scheduling has been carried out numerous studies.
Traditional power system Risk Scheduling method is by heredity (genetic algorithm, GA), quantum genetic
(quantum genetic algorithm, QGA), bee colony (artificial bee colony, ABC), population
Intelligent algorithms such as (particle swarm optimization, PSO) is applied to each optimization problem of power system.But, this
The optimization of similar tasks is isolated and is carried out by class intelligent algorithm, it is impossible to effectively preserves the experience and knowledge of task in the past, lacks
Self-learning capability, each new task need to reinitialize when performing, cause Searching efficiency relatively low, it is difficult to adapt to large-scale complex wind
The rapid Optimum of danger scheduling.
Summary of the invention
Based on this, it is necessary to for the problems referred to above, it is provided that a kind of large-scale complex Risk Scheduling rapid Optimum of being suitable for
Power system Risk Scheduling method and system.
A kind of power system Risk Scheduling method, comprises the following steps:
Obtain framework data and the new task load profile data of power system；
According to described framework data and described new task load profile data, look for food nitrification enhancement in advance by antibacterial
If initial knowledge matrix be iterated update, obtain correspondence Risk Scheduling target function value and update after knowledge square
Battle array；Described initial knowledge matrix is the optimum knowledge matrix in originating task；
New task on-line optimization is carried out according to the knowledge matrix after renewal corresponding during Risk Scheduling target function value minimum,
Obtain Risk Scheduling optimum results and export.
A kind of power system Risk Scheduling system, including:
Task data acquisition module, for obtaining framework data and the new task load profile data of power system；
Knowledge matrix more new module, for according to described framework data and described new task load profile data, by carefully
Bacterium look for food nitrification enhancement default initial knowledge matrix is iterated update, obtain correspondence Risk Scheduling object function
Knowledge matrix after value and renewal；Described initial knowledge matrix is the optimum knowledge matrix in originating task；
Risk Scheduling optimizes module, for according to the knowledge square after renewal corresponding during Risk Scheduling target function value minimum
Battle array carries out new task on-line optimization, obtains Risk Scheduling optimum results and exports.
Above-mentioned power system Risk Scheduling method and system, the framework data and the new task load that obtain power system break
Face data；According to framework data and new task load profile data, look for food nitrification enhancement to default initial by antibacterial
Knowledge matrix is iterated updating, the knowledge matrix after obtaining corresponding Risk Scheduling target function value and updating.According to wind
The knowledge matrix after renewal corresponding during the regulation goal functional value minimum of danger carries out new task on-line optimization, obtains Risk Scheduling excellent
Change result and export.Optimum knowledge matrix in originating task is realized knowledge migration as the initial matrix of new task, utilizes base
Antibacterial in knowledge migration intensified learning of looking for food carries out on-line optimization to new task.Online is greatly improved by transfer learning
The speed practised, it is achieved the online dynamic optimization of Risk Scheduling problem, still ensures that ask faster when problem scale expands further
Solve speed, be suitable for large-scale complex Risk Scheduling rapid Optimum.
Accompanying drawing explanation
Fig. 1 is the flow chart of power system Risk Scheduling method in an embodiment；
Fig. 2 is that the antibacterial that in an embodiment, knowledge based migrates is looked for food the knowledge acquisition schematic diagram of nitrification enhancement；
Fig. 3 is the dimension reduction schematic diagram that in an embodiment, knowledge based extends；
Fig. 4 is knowledge migration schematic diagram in an embodiment；
Fig. 5 is test system topological figure in an embodiment；
Fig. 6 is the structure chart of power system Risk Scheduling system in an embodiment.
Detailed description of the invention
In one embodiment, a kind of power system Risk Scheduling method, as it is shown in figure 1, comprise the following steps:
Step S120: obtain framework data and the new task load profile data of power system.
The framework data of power system specifically can include the data such as bus nodes, transmission line, transformator and electromotor.Newly appointed
Business load profile data includes one or more load section, and each load section is as a new task.Obtain power system
Framework data and new task load profile data carry out Risk Scheduling optimization for follow-up.
Step S130: according to framework data and new task load profile data, is looked for food nitrification enhancement pair by antibacterial
The initial knowledge matrix preset is iterated updating, the knowledge square after obtaining corresponding Risk Scheduling target function value and updating
Battle array.
Initial knowledge matrix is the optimum knowledge matrix in originating task.Using the optimum knowledge matrix in originating task as new post
The initial matrix of business realizes knowledge migration, combines look for food stochastic search pattern and the probability of optimized algorithm of antibacterial by bacterial flora empty
Between Action Selection strategy execution Action Selection, it is achieved the antibacterial utilizing knowledge based to migrate is looked for food nitrification enhancement
(Transfer Bacteria Foraging Optimization, TBFO) carries out on-line optimization to new task.
The particular type of initial knowledge matrix is not unique, and in the present embodiment, initial knowledge matrix is Q matrix.Q learns to calculate
In method, (s a) represents the expectation of selection action a gained jackpot prize value under state s to the element Q in Q matrix.Matrix have recorded
Intelligent body is mapped to the knowledge of this process of action state.Using Q matrix as the knowledge matrix of record colony optimization information, lead to
Cross the similarity analyzed between Different Optimization task, utilize the knowledge matrix of originating task to form the initial knowledge matrix of new task, with
The mode of knowledge migration realizes the online dynamic optimization to different time section task, it is ensured that optimize reliability.
In TBFO algorithm, bacterial flora obtains the action policy for specific environment state from initial knowledge matrix, and utilizes
Update original knowledge from the feedback information obtained test is repeated several times, form the intrinsic reaction to particular state, so that antibacterial
The energy value that group accumulates during looking for food reaches maximum.
In one embodiment, step S130 includes that step 131 is to step 136.
Step 131: according to framework data and new task load profile data, controls the antibacterial guidance at initial knowledge matrix
Under carry out tropism operation, migrating property operation and replicability operation.
Antibacterial, under the guidance of initial knowledge matrix, is operated by tropism operation, migrating property and replicability operation obtains
Knowledge.In TBFO algorithm, foraging areas will be scanned for by whole antibacterials according to initial knowledge matrix, and reward feedback by gained
To knowledge matrix.As in figure 2 it is shown, according to the operation being carrying out, antibacterial is drawn and is assigned as tending to and migrating two states by TBFO.
In algorithm single iteration circulates, two states giving a certain proportion of organisms respectively, two groups of antibacterials have performed each behaviour
After work, calculate and the energy value of whole antibacterial of sorting, enter replicability operation, so that the energy that bacterial flora is accumulated during looking for food
Value reaches maximum.During new round iteration is followed, according to energy value height in last iteration, bacterial condition is reallocated, energy value
Bigger antibacterial keeps region constant and carries out tropism operation, antibacterial execution the migrating property operation that energy value is relatively low.
Specifically, sort based on energy value, the advantage individuality in flora is placed in trend state, still undertakes Local Search
Task.Its approach behavior can be expressed from the next:
In formula, θi(j, k are l) that organisms i replicates operation and jth generation trend operation at l for Transfer free energy, kth generation
After position；Δ represent travelling after unit vector in the random direction that determines.
CkI () can be fixing step-length, it is also possible to be the step-length of change.In the present embodiment, CkI () is non-linear successively decreasing
Inertia step-length, CkI () update mode is shown below:
In formula: CkI () is inertia step-length during kth time iteration, C0For initial travelling step-length, CeFor final travelling step-length,
Cly is greatest iteration step number.
To being in the antibacterial of the state of migrating, migrate probability P when it meetsedTime, antibacterial is taken turns according to action probability matrix
Dish selects；Otherwise antibacterial migrates (greedy strategy) according to the action that maximum knowledge element is corresponding:
In formula: subscript i represents i-th controlled variable, knowledge matrix with i-th is corresponding, i ∈ M；M is controlled variable
Set；Subscript j represents that jth antibacterial, j ∈ N, N are flora set；PedFor migrating probability；R is the random number between 0～1；as
It is then probability matrix PiThe action selected in global scope.When meet migrate condition time, antibacterial is according to action probability matrix PiHold
Row pseudorandom wheel disc selects；PiUpdate mode as follows:
In formula: β is coefficient of variation, it is used for amplifying QiThe diversity of matrix element；eiBelong to intermediate computations matrix.
In one embodiment, introducing the process of intersection in replicability operation, its interleaved mode is as follows:
θi+S/2(j, k, l)=r θi(j,k,l)+(1-r)θi+S/2(j,k,l)
In formula: S is organisms number, i ∈ [1, S/2], r are the random number in [0,1].
Step 132: operate according to the tropism of antibacterial, migrating property operates and replicability operation, calculates power system at base
Trend value under state and preset failure.
Tropism on antibacterial operates, migrating property operates and replicability operates after terminating, and calculates according to accordingly result
Power system trend value under ground state and preset failure.Ground state i.e. refers to that system does not occurs the system failure, preset failure concrete
Kind is the most unique.
Step 133: be calculated Risk Scheduling target letter according to power system trend value under ground state and preset failure
Numerical value.
In TBFO algorithm, award value immediately reflects the direction of optimization, and flora is obtained by iteration optimization knowledge matrix
Dominant strategy, obtains maximum progressive award functional value with expectation.In Risk Scheduling mathematical model, object function is that algorithm is rewarded
The inverse of function, it is desirable to make target letter minimize by optimization.In the present embodiment, reward function design is as follows:
Wherein, FCThe fuel cost that nonlinear function describes, IRThe system safety hazards described for non-linear utility function refers to
Mark.CVIt is the violation degree that under ground state, system always retrains, c1、c2The magnitude between fuel cost and risk indicator is coordinated to close respectively
System, ω1、ω2It is respectively used to embody and corresponding target is stressed degree.
Step 134: bacterial condition is reallocated according to Risk Scheduling target function value.It is being calculated Risk Scheduling
After target function value, according to Risk Scheduling target function value, bacterial condition is reallocated.
Step 135: be iterated updating, after being updated to initial knowledge matrix according to the bacterial condition after reallocation
Knowledge matrix.In one embodiment, step 135 includes step 11 and step 12.
Step 11: initial knowledge matrix is carried out dimension reduction, obtains many sub-knowledge matrixes.
As it is shown on figure 3, be " dimension disaster " problem that effectively solves, use the knowledge extending to carry out dimension reduction, will initially know
Know matrix Q and be divided into many sub-knowledge matrix Qi, with each variable one_to_one corresponding.Connected by knowledge matrix between variable, phase
Element in adjacent matrix is relevant knowledge, say, that xiMotion space AiIt is xi+1State space Si+1.The most first determine
Variable xiAction, could based on its select result select xi+1Action, thus between relevant knowledge, define a kind of chain type
Extension, it is achieved that the decomposition dimensionality reduction to knowledge matrix.
Step 12: according to the bacterial condition after reallocation, many sub-knowledge matrixes are updated, knowing after being updated
Know matrix.Many sub-knowledge matrixes are updated, many sub-knowledge matrixes after updating the knowledge after just can being updated
Matrix.
Being updated as multiagent knowledge matrix is collaborative by flora, whole antibacterials share a knowledge matrix, single iteration
In can update multiple knowledge element simultaneously, be greatly accelerated the efficiency of optimizing.Each main body can be encouraged after trial and error is explored every time
Encourage value assessment.After introducing flora is collaborative, sub-knowledge matrix QiUpdate mode is as follows:
In formula: R (sij k, sij k+1, aij k) represent that kth time iteration is in state skLower selection action akTransfer to state sk+1Time
The reward function value obtained；α is Studying factors, and γ is discount factor.
In another embodiment, step 135 includes step 21 and step 23.
Step 21: calculate the meritorious of each originating task and new task in initial knowledge matrix according to the bacterial condition after reallocation
Power deviation.
Active power deviation definition is the similarity between originating task and new task, and is divided into ascending for meritorious demand
Multiple load sections:
[PDs1,PDs2),[PDs2,PDs3),...[PDsi-1,PDsi)...,[PDsn-1,PDsn)
Step 22: be ranked up originating task according to active power deviation is descending, before obtaining, the source of predetermined number is appointed
Business.The concrete value of predetermined number is not unique, and in the present embodiment, predetermined number is two.
Step 23: initial knowledge matrix is updated by the originating task according to obtaining, the knowledge matrix after being updated.
As a example by two originating tasks of acquisition carry out matrix update, first calculate the contribution system of two originating task transfer learnings
Number, is then updated initial knowledge matrix according to coefficient of migration, obtains the knowledge matrix of new task.
Specifically, it is assumed that the meritorious demand of new task x is PDx, PDi、PDkFor in originating task immediate with task x two
Section load, and meet PDi<PDx<PDk, then two originating task PDi、PDkContribution coefficient η to transfer learning1、η2Can be by following formula meter
Calculate:
Utilize linear transport mode, can obtain the knowledge matrix of new task x:
Utilize the knowledge high with new task similarity, use the originating task section information closest to new task workload demand to enter
Row migrates, it is to avoid migrates, by invalid knowledge, new task learning quality and speed is produced negative interference, improves and calculate accuracy.
It is appreciated that in one embodiment, it is also possible to be initial knowledge matrix first to carry out dimension reduction obtain multiple
Sub-knowledge matrix, then utilizes the knowledge high with new task similarity to be updated many sub-knowledge matrixes, after being updated
Knowledge matrix.
Step 136: judge whether iteration renewal meets pre-conditioned.
Pre-conditioned particular type is not unique, in the present embodiment, pre-conditioned for k > kmaxOrIts
In, kmaxRepresent and preset maximum iteration time；For knowledge matrix2-norm, before and after reflection, twice repeatedly
The extent of deviation of knowledge matrix in Dai.
Judge whether iteration renewal meets pre-conditioned, if it is not, the knowledge matrix after then updating is as initial knowledge square
Battle array, and return step 131, again knowledge matrix is updated；The most then iteration updates and terminates, the knowledge that will finally give
Matrix is as the optimization matrix needed for new task optimization.
Step S140: carry out new task according to the knowledge matrix after renewal corresponding during Risk Scheduling target function value minimum
On-line optimization, obtains Risk Scheduling optimum results and exports.
Initial knowledge matrix is being iterated after renewal terminates, by during Risk Scheduling target function value minimum corresponding more
Knowledge matrix after Xin carries out on-line optimization as optimizing matrix to new task, obtains Risk Scheduling optimum results and exports.Defeated
The concrete mode going out risk optimizing scheduling result is unique, can be that output to memorizer stores, it is also possible to be output
Show to display.
Additionally, in one embodiment, before step S130, power system Risk Scheduling method also includes step 110.
Step 110: receive originating task and be trained, obtains optimum knowledge matrix as initial knowledge matrix.
Step 110 can be before step S120, it is also possible to is after step S120.TBFO algorithm is learning rank in advance
The a series of originating task of Duan Zhihang is to obtain optimum knowledge matrix, and therefrom excavates initial knowledge, for the most relevant new post
It is engaged in ready.As shown in Figure 4, the relevant initial knowledge from originating task will be used in on-line optimization, according to originating task with new
Similarity between task, originating task QSInitial knowledge matrix be new task Q by migrationNInitial knowledge matrix.
For the ease of being more fully understood that above-mentioned power system Risk Scheduling method, carry out in detail below in conjunction with specific embodiment
Illustrate.
Using a certain reliability test system as the simulation object of Risk Scheduling.Selecting system reference capacity is 100MVA,
Having 24 bus nodes, 34 transmission lines/transformator and 32 electromotors in system, its topological structure is as shown in Figure 5.Entirely
In 10, portion electromotor node, the electromotor node 21 that single-machine capacity is maximum is set to system-wide balance node, remaining 9 joints
Point is PV (Control of Voltage) node.
The adaptability being optimized different load level for testing algorithm, the present embodiment carries out the risk of 96 sections and adjusts
Degree optimization Simulation.In the present embodiment, have chosen typical day load curve, and break every division in 15 minutes according to sequential
Face, obtains section 1 to section 96.
Based on upper mounting plate, follow the steps below Risk Scheduling optimization.
(1) choose the generated power at PV node to exert oneself PGFor control variable, action variable space A (APG1, APG2...,
APGi) with control variable space be one to one, i be on PV node unit sum.The motion space of previous variable is down
The state space of one variable.The sub-knowledge matrix corresponding with each variable states-motion space is respectively QPG1, QPG2..., QPGi。
Antibacterial, under the guidance of knowledge matrix, is operated by tropism operation, migrating property and replicability operation obtains knowledge.
(2) calculating of the object function of Risk Scheduling depends on non-linear Load flow calculation.If using standard BFO algorithm, if
Ned、NreAnd NcExpression is migrated, is replicated and the operand of approach behavior respectively, and maximum travelling number of times is Ns, it is contemplated that fault set comprises
Fault NpIndividual, then Load flow calculation number of times can up to NedNreNcNs(Np+ 1) secondary so that solution procedure is the slowest.By to algorithm
The improvement of optimizing pattern, eliminates the nested circulation of former algorithm, improves the efficiency of algorithm.Bacterial flora combine BFO algorithm with
Machine search pattern and probability space Action Selection strategy execution Action Selection.
(3) unit of equal fuel cost coefficient will be had on same node to divide a control variable into, 31 units are gained merit
Exert oneself and be divided into 13 variablees for control variable altogether.Using previous unit output size as the state space of a rear unit.
Wherein, the state space of First unit is the active power size of current section, thus reduces the dimension of knowledge matrix.
(4) in TBFO algorithm, award value immediately has reacted the direction optimized, and flora is obtained by iteration optimization knowledge matrix
Obtain optimal strategy, obtain maximum progressive award functional value with expectation.
(5) it is the similarity between originating task and new task by active power deviation definition, and by ascending for meritorious demand
It is divided into multiple load section.
By invalid knowledge, new task learning quality and speed are produced negative interference for avoiding migrating, learning process should be use up
Amount utilizes the knowledge high with new task similarity, and the present embodiment only uses two originating tasks closest to new task workload demand to break
Surface information migrates.The meritorious demand assuming new task x is PDx, PDi、PDkFor in originating task immediate with task x two
Section load, and meet PDi<PDx<PDk, then obtain two originating tasks contribution coefficient to transfer learning, then utilize and linearly move
Shifting mode, can obtain the knowledge matrix of new task x.
Complete the iteration to initial knowledge matrix update after, by during Risk Scheduling target function value minimum corresponding more
New task is carried out online, obtaining Risk Scheduling optimum results and exporting by the knowledge matrix after Xin as optimizing matrix.
Above-mentioned power system Risk Scheduling method, using the optimum knowledge matrix in originating task as the initial matrix of new task
Realizing knowledge migration, the antibacterial utilizing knowledge based to migrate intensified learning of looking for food carries out on-line optimization to new task.By migrating
Study greatly improves the speed of on-line study, it is achieved the online dynamic optimization of Risk Scheduling problem, when problem scale is further
Expand and still ensure that solving speed faster, be suitable for large-scale complex Risk Scheduling rapid Optimum.
In one embodiment, a kind of power system Risk Scheduling system, as shown in Figure 6, obtain mould including task data
Block 120, knowledge matrix more new module 130 and Risk Scheduling optimize module 140.
Task data acquisition module 120 is for obtaining framework data and the new task load profile data of power system.
The framework data of power system specifically can include the data such as bus nodes, transmission line, transformator and electromotor, and new task load breaks
Face data include one or more load section.The framework data and the new task load profile data that obtain power system are used for
Follow-up carry out Risk Scheduling optimization.
Knowledge matrix more new module 130, for according to framework data and new task load profile data, is looked for food by antibacterial
Nitrification enhancement default initial knowledge matrix is iterated update, obtain correspondence Risk Scheduling target function value and
Knowledge matrix after renewal.
Initial knowledge matrix is the optimum knowledge matrix in originating task.Using the optimum knowledge matrix in originating task as new post
The initial matrix of business realizes knowledge migration, combines look for food stochastic search pattern and the probability of optimized algorithm of antibacterial by bacterial flora empty
Between Action Selection strategy execution Action Selection, it is achieved utilize TBFO algorithm that new task is carried out on-line optimization.
The particular type of initial knowledge matrix is not unique, and in the present embodiment, initial knowledge matrix is Q matrix.By Q matrix
As the knowledge matrix of record colony optimization information, by analyzing the similarity between Different Optimization task, utilize knowing of originating task
Know matrix and form the initial knowledge matrix of new task, realize the online of different time section task is moved in the way of knowledge migration
State optimizes, it is ensured that optimize reliability.
In one embodiment, knowledge matrix more new module 130 include the first processing unit, the second processing unit, the 3rd
Processing unit, fourth processing unit, the 5th processing unit and the 6th processing unit.
First processing unit, for according to framework data and new task load profile data, controls antibacterial at initial knowledge square
Tropism operation, the operation of migrating property and replicability operation is carried out under the guidance of battle array.
Antibacterial, under the guidance of initial knowledge matrix, is operated by tropism operation, migrating property and replicability operation obtains
Knowledge.Specifically, sort based on energy value, the advantage individuality in flora is placed in trend state, still undertake appointing of Local Search
Business.Its approach behavior can be expressed from the next:
CkI () can be fixing step-length, it is also possible to be the step-length of change.In the present embodiment, CkI () is non-linear successively decreasing
Inertia step-length, CkI () update mode is shown below:
To being in the antibacterial of the state of migrating, migrate probability P when it meetsedTime, antibacterial is taken turns according to action probability matrix
Dish selects；Otherwise antibacterial migrates (greedy strategy) according to the action that maximum knowledge element is corresponding:
When meet migrate condition time, antibacterial is according to action probability matrix PiPerform pseudorandom wheel disc and select aS；PiRenewal side
Formula is as follows:
In one embodiment, introducing the process of intersection in replicability operation, its interleaved mode is as follows:
θi+S/2(j, k, l)=r θi(j,k,l)+(1-r)θi+S/2(j,k,l)
Second processing unit operates for the tropism according to antibacterial, migrating property operates and replicability operation, calculates electric power
System trend value under ground state and preset failure.
Tropism on antibacterial operates, migrating property operates and replicability operates after terminating, and calculates according to accordingly result
Power system trend value under ground state and preset failure.Ground state i.e. refers to that system does not occurs the system failure, preset failure concrete
Kind is the most unique.
3rd processing unit is adjusted for being calculated risk according to power system trend value under ground state and preset failure
Degree target function value.
In TBFO algorithm, award value immediately has reacted the direction optimized, and flora is obtained by iteration optimization knowledge matrix
Dominant strategy, obtains maximum progressive award functional value with expectation.In Risk Scheduling mathematical model, object function is that algorithm is rewarded
The inverse of function, it is desirable to make target letter minimize by optimization.In the present embodiment, reward function design is as follows:
Wherein, FCThe fuel cost that nonlinear function describes, IRThe system safety hazards described for non-linear utility function refers to
Mark.CVIt is the violation degree that under ground state, system always retrains, c1、c2The magnitude between fuel cost and risk indicator is coordinated to close respectively
System, ω1、ω2It is respectively used to embody and corresponding target is stressed degree.
Fourth processing unit is for reallocating to bacterial condition according to Risk Scheduling target function value.It is being calculated
After Risk Scheduling target function value, according to Risk Scheduling target function value, bacterial condition is reallocated.
5th processing unit, for being iterated updating to initial knowledge matrix according to the bacterial condition after reallocation, obtains
Knowledge matrix after renewal.
In one embodiment, the 5th processing unit includes dimension reduction unit and matrix update unit.
Dimension reduction unit, for initial knowledge matrix is carried out dimension reduction, obtains many sub-knowledge matrixes.Employing is known
Know extension and carry out dimension reduction, initial knowledge matrix Q is divided into many sub-knowledge matrix Qi, with each variable one_to_one corresponding.
Matrix update unit, for being updated many sub-knowledge matrixes according to the bacterial condition after reallocation, obtains more
Knowledge matrix after Xin.Many sub-knowledge matrixes are updated, many sub-knowledge matrixes after updating just can be updated
After knowledge matrix.
Being updated as multiagent knowledge matrix is collaborative by flora, whole antibacterials share a knowledge matrix, single iteration
In can update multiple knowledge element simultaneously, be greatly accelerated the efficiency of optimizing.Each main body can be encouraged after trial and error is explored every time
Encourage value assessment.After introducing flora is collaborative, sub-knowledge matrix QiUpdate mode is as follows:
In another embodiment, the 5th processing unit includes computing unit, extraction unit and updating block.
Computing unit is for calculating each originating task and new task in initial knowledge matrix according to the bacterial condition after reallocation
Active power deviation.Active power deviation definition is the similarity between originating task and new task, and by meritorious demand by little to
It is divided into greatly multiple load section:
[PDs1,PDs2),[PDs2,PDs3),...[PDsi-1,PDsi)...,[PDsn-1,PDsn)
Extraction unit, for being ranked up originating task according to active power deviation is descending, obtains front predetermined number
Originating task.The concrete value of predetermined number is not unique, and in the present embodiment, predetermined number is two.
Initial knowledge matrix is updated by the originating task that updating block is used for according to obtaining, the knowledge square after being updated
Battle array.
As a example by two originating tasks of acquisition carry out matrix update, first calculate the contribution system of two originating task transfer learnings
Number, is then updated initial knowledge matrix according to coefficient of migration, obtains the knowledge matrix of new task.Two originating task PDi、
PDkContribution coefficient η to transfer learning1、η2Can be calculated by following formula:
Utilize linear transport mode, can obtain the knowledge matrix of new task x:
Utilize the knowledge high with new task similarity, use the originating task section information closest to new task workload demand to enter
Row migrates, it is to avoid migrates, by invalid knowledge, new task learning quality and speed is produced negative interference, improves and calculate accuracy.
It is appreciated that in one embodiment, it is also possible to be that the 5th processing unit includes dimension reduction unit and matrix more
New unit, matrix update unit includes computing unit, extraction unit and updating block.First initial knowledge matrix is carried out dimension contracting
Subtract and obtain many sub-knowledge matrixes, then utilize the knowledge high with new task similarity that many sub-knowledge matrixes are updated,
Knowledge matrix after being updated.
6th processing unit is used for judging whether iteration renewal meets pre-conditioned, and presets bar at iteration renewal not met
During part, the knowledge matrix after updating as initial knowledge matrix, and control the first processing unit again according to framework data and
New task load profile data, control antibacterial carry out under the guidance of initial knowledge matrix tropism operation, migrating property operation and
Replicability operates.
Pre-conditioned particular type is not unique, in the present embodiment, pre-conditioned for k > kmaxOrJudge
Whether iteration renewal meets pre-conditioned, if it is not, the knowledge matrix after then updating is carried out repeatedly again as initial knowledge matrix
In generation, updates, the most then iteration updates and terminates, the optimization matrix needed for being optimized as new task by the knowledge matrix finally given.
Risk Scheduling optimizes module 140 for according to the knowledge after renewal corresponding during Risk Scheduling target function value minimum
Matrix carries out new task on-line optimization, obtains Risk Scheduling optimum results and exports.
Initial knowledge matrix is being iterated after renewal terminates, by during Risk Scheduling target function value minimum corresponding more
Knowledge matrix after Xin carries out on-line optimization as optimizing matrix to new task, obtains Risk Scheduling optimum results and exports.Defeated
The concrete mode going out risk optimizing scheduling result is unique, can be that output to memorizer stores, it is also possible to be output
Show to display.
Additionally, in one embodiment, power system Risk Scheduling system also includes matrix training module.
Matrix training module is used in knowledge matrix more new module 130 according to framework data and new task load section number
According to, by antibacterial look for food nitrification enhancement default initial knowledge matrix is iterated update, obtain correspondence risk adjust
Before knowledge matrix after degree target function value and renewal, receive originating task and be trained, obtain optimum knowledge matrix conduct
Initial knowledge matrix.TBFO algorithm performs a series of originating task to obtain optimum knowledge matrix in the pre-study stage, and therefrom
Excavate initial knowledge, ready for the most relevant new task.
Above-mentioned power system Risk Scheduling system, using the optimum knowledge matrix in originating task as the initial matrix of new task
Realizing knowledge migration, the antibacterial utilizing knowledge based to migrate intensified learning of looking for food carries out on-line optimization to new task.By migrating
Study greatly improves the speed of on-line study, it is achieved the online dynamic optimization of Risk Scheduling problem, when problem scale is further
Expand and still ensure that solving speed faster, be suitable for large-scale complex Risk Scheduling rapid Optimum.
Each technical characteristic of embodiment described above can combine arbitrarily, for making description succinct, not to above-mentioned reality
The all possible combination of each technical characteristic executed in example is all described, but, as long as the combination of these technical characteristics is not deposited
In contradiction, all it is considered to be the scope that this specification is recorded.
Embodiment described above only have expressed the several embodiments of the present invention, and it describes more concrete and detailed, but also
Can not therefore be construed as limiting the scope of the patent.It should be pointed out that, come for those of ordinary skill in the art
Saying, without departing from the inventive concept of the premise, it is also possible to make some deformation and improvement, these broadly fall into the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (10)
1. a power system Risk Scheduling method, it is characterised in that comprise the following steps:
Obtain framework data and the new task load profile data of power system；
According to described framework data and described new task load profile data, look for food nitrification enhancement to default by antibacterial
Initial knowledge matrix is iterated updating, the knowledge matrix after obtaining corresponding Risk Scheduling target function value and updating；Institute
Stating initial knowledge matrix is the optimum knowledge matrix in originating task；
Carry out new task on-line optimization according to the knowledge matrix after renewal corresponding during Risk Scheduling target function value minimum, obtain
Risk Scheduling optimum results also exports.
Power system Risk Scheduling method the most according to claim 1, it is characterised in that described initial knowledge matrix is Q
Matrix.
Power system Risk Scheduling method the most according to claim 1, it is characterised in that described according to described framework data
Data are set with described new task, by antibacterial nitrification enhancement of looking for food, default initial knowledge matrix are iterated more
Newly, the step of the knowledge matrix after obtaining corresponding Risk Scheduling target function value and updating, comprise the following steps:
According to described framework data and described new task load profile data, control the antibacterial guidance at described initial knowledge matrix
Under carry out tropism operation, migrating property operation and replicability operation；
Tropism operation according to described antibacterial, the operation of migrating property and replicability operation, calculate power system and in ground state and preset
Trend value under fault；
It is calculated Risk Scheduling target function value according to described power system trend value under ground state and preset failure；
According to described Risk Scheduling target function value, bacterial condition is reallocated；
It is iterated updating to described initial knowledge matrix according to the bacterial condition after reallocation, the knowledge square after being updated
Battle array；
Judge whether iteration renewal meets pre-conditioned；
If it is not, the knowledge matrix after then updating is as described initial knowledge matrix, and return described according to described framework data
With described new task load profile data, control antibacterial and under the guidance of described initial knowledge matrix, carry out tropism operation, move
The operation of moving property and the step of replicability operation.
Power system Risk Scheduling method the most according to claim 3, it is characterised in that described according to reallocation after thin
Described initial knowledge matrix is iterated updating by bacterium state, and the step of the knowledge matrix after being updated comprises the following steps:
Described initial knowledge matrix is carried out dimension reduction, obtains many sub-knowledge matrixes；
According to the bacterial condition after reallocation, the plurality of sub-knowledge matrix is updated, the knowledge matrix after being updated.
Power system Risk Scheduling method the most according to claim 1, it is characterised in that described according to described framework data
With described new task load profile data, by antibacterial nitrification enhancement of looking for food, default initial knowledge matrix is iterated
Update, before the step of the knowledge matrix after obtaining corresponding Risk Scheduling target function value and updating, also include following step
Rapid:
Reception originating task is trained, and obtains optimum knowledge matrix as described initial knowledge matrix.
6. a power system Risk Scheduling system, it is characterised in that including:
Task data acquisition module, for obtaining framework data and the new task load profile data of power system；
Knowledge matrix more new module, for according to described framework data and described new task load profile data, is looked for by antibacterial
Food nitrification enhancement default initial knowledge matrix is iterated update, obtain correspondence Risk Scheduling target function value with
And the knowledge matrix after updating；Described initial knowledge matrix is the optimum knowledge matrix in originating task；
Risk Scheduling optimizes module, for entering according to the knowledge matrix after renewal corresponding during Risk Scheduling target function value minimum
Row new task on-line optimization, obtains Risk Scheduling optimum results and exports.
Power system Risk Scheduling system the most according to claim 6, it is characterised in that described initial knowledge matrix is Q
Matrix.
Power system Risk Scheduling system the most according to claim 6, it is characterised in that described knowledge matrix more new module
Including:
First processing unit, for according to described framework data and described new task load profile data, controls antibacterial described
Tropism operation, the operation of migrating property and replicability operation is carried out under the guidance of initial knowledge matrix；
Second processing unit, for the tropism operation according to described antibacterial, the operation of migrating property and replicability operation, calculates electric power
System trend value under ground state and preset failure；
3rd processing unit, adjusts for being calculated risk according to described power system trend value under ground state and preset failure
Degree target function value；
Fourth processing unit, for reallocating to bacterial condition according to described Risk Scheduling target function value；
5th processing unit, for being iterated updating to described initial knowledge matrix according to the bacterial condition after reallocation,
Knowledge matrix after renewal；
6th processing unit, is used for judging whether iteration renewal meets pre-conditioned, and it is pre-conditioned to update not met in iteration
Time, the knowledge matrix after updating is as described initial knowledge matrix, and controls described first processing unit again according to described
Framework data and described new task load profile data, control antibacterial and carry out tropism under the guidance of described initial knowledge matrix
Operation, the operation of migrating property and replicability operation.
Power system Risk Scheduling system the most according to claim 8, it is characterised in that described 5th processing unit bag
Include:
Dimension reduction unit, for described initial knowledge matrix is carried out dimension reduction, obtains many sub-knowledge matrixes；
Matrix update unit, for being updated the plurality of sub-knowledge matrix according to the bacterial condition after reallocation, obtains
Knowledge matrix after renewal.
Power system Risk Scheduling system the most according to claim 6, it is characterised in that also include matrix training module,
Described matrix training module is used in knowledge matrix more new module according to described framework data and described new task load section number
According to, by antibacterial look for food nitrification enhancement default initial knowledge matrix is iterated update, obtain correspondence risk adjust
Before knowledge matrix after degree target function value and renewal, receive originating task and be trained, obtain optimum knowledge matrix conduct
Described initial knowledge matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610882652.1A CN106296044B (en) | 2016-10-08 | 2016-10-08 | Power system risk scheduling method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610882652.1A CN106296044B (en) | 2016-10-08 | 2016-10-08 | Power system risk scheduling method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106296044A true CN106296044A (en) | 2017-01-04 |
| CN106296044B CN106296044B (en) | 2023-08-25 |
Family
ID=57717240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610882652.1A Active CN106296044B (en) | 2016-10-08 | 2016-10-08 | Power system risk scheduling method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106296044B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106526432A (en) * | 2017-01-06 | 2017-03-22 | 中国南方电网有限责任公司电网技术研究中心 | BFOA-based fault location algorithm and device |
| CN108549907A (en) * | 2018-04-11 | 2018-09-18 | 武汉大学 | A kind of data verification method based on multi-source transfer learning |
| CN108734419A (en) * | 2018-06-15 | 2018-11-02 | 大连理工大学 | A kind of blast furnace gas Modeling of Scheduling method of knowledge based migration |
| CN109460890A (en) * | 2018-09-21 | 2019-03-12 | 浙江大学 | A kind of intelligent self-healing method based on intensified learning and control performance monitoring |
| CN109873406A (en) * | 2019-03-28 | 2019-06-11 | 华中科技大学 | A kind of electric system weakness route discrimination method |
| CN110048461A (en) * | 2019-05-16 | 2019-07-23 | 广东电网有限责任公司 | A kind of more virtual plant dispersion self-discipline optimization methods |
| CN111626539A (en) * | 2020-03-03 | 2020-09-04 | 中国南方电网有限责任公司 | Power grid operation section dynamic generation method based on Q reinforcement learning |
| CN112749785A (en) * | 2019-10-29 | 2021-05-04 | 株式会社东芝 | Information processing apparatus, information processing method, and program |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103199544A (en) * | 2013-03-26 | 2013-07-10 | 上海理工大学 | Reactive power optimization method of electrical power system |
| WO2014090037A1 (en) * | 2012-12-10 | 2014-06-19 | 中兴通讯股份有限公司 | Task scheduling method and system in cloud computing |
| CN105023056A (en) * | 2015-06-26 | 2015-11-04 | 华南理工大学 | Power grid optimal carbon energy composite flow obtaining method based on swarm intelligence reinforcement learning |
| CN105373183A (en) * | 2015-10-20 | 2016-03-02 | 同济大学 | Method for tracking whole-situation maximum power point in photovoltaic array |
-
2016
- 2016-10-08 CN CN201610882652.1A patent/CN106296044B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014090037A1 (en) * | 2012-12-10 | 2014-06-19 | 中兴通讯股份有限公司 | Task scheduling method and system in cloud computing |
| CN103199544A (en) * | 2013-03-26 | 2013-07-10 | 上海理工大学 | Reactive power optimization method of electrical power system |
| CN105023056A (en) * | 2015-06-26 | 2015-11-04 | 华南理工大学 | Power grid optimal carbon energy composite flow obtaining method based on swarm intelligence reinforcement learning |
| CN105373183A (en) * | 2015-10-20 | 2016-03-02 | 同济大学 | Method for tracking whole-situation maximum power point in photovoltaic array |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106526432B (en) * | 2017-01-06 | 2019-05-10 | 中国南方电网有限责任公司电网技术研究中心 | BFOA-based fault location algorithm and device |
| CN106526432A (en) * | 2017-01-06 | 2017-03-22 | 中国南方电网有限责任公司电网技术研究中心 | BFOA-based fault location algorithm and device |
| CN108549907A (en) * | 2018-04-11 | 2018-09-18 | 武汉大学 | A kind of data verification method based on multi-source transfer learning |
| CN108549907B (en) * | 2018-04-11 | 2021-11-16 | 武汉大学 | Data verification method based on multi-source transfer learning |
| CN108734419A (en) * | 2018-06-15 | 2018-11-02 | 大连理工大学 | A kind of blast furnace gas Modeling of Scheduling method of knowledge based migration |
| CN109460890B (en) * | 2018-09-21 | 2021-08-06 | 浙江大学 | Intelligent self-healing method based on reinforcement learning and control performance monitoring |
| CN109460890A (en) * | 2018-09-21 | 2019-03-12 | 浙江大学 | A kind of intelligent self-healing method based on intensified learning and control performance monitoring |
| CN109873406A (en) * | 2019-03-28 | 2019-06-11 | 华中科技大学 | A kind of electric system weakness route discrimination method |
| CN109873406B (en) * | 2019-03-28 | 2019-11-22 | 华中科技大学 | A kind of electric system weakness route discrimination method |
| CN110048461A (en) * | 2019-05-16 | 2019-07-23 | 广东电网有限责任公司 | A kind of more virtual plant dispersion self-discipline optimization methods |
| CN110048461B (en) * | 2019-05-16 | 2021-07-02 | 广东电网有限责任公司 | Multi-virtual power plant decentralized self-discipline optimization method |
| CN112749785A (en) * | 2019-10-29 | 2021-05-04 | 株式会社东芝 | Information processing apparatus, information processing method, and program |
| CN111626539A (en) * | 2020-03-03 | 2020-09-04 | 中国南方电网有限责任公司 | Power grid operation section dynamic generation method based on Q reinforcement learning |
| CN111626539B (en) * | 2020-03-03 | 2023-06-16 | 中国南方电网有限责任公司 | Q reinforcement learning-based power grid operation section dynamic generation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106296044B (en) | 2023-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106296044A (en) | power system risk scheduling method and system | |
| Yang et al. | PV arrays reconfiguration for partial shading mitigation: Recent advances, challenges and perspectives | |
| Ye et al. | Predicting electricity consumption in a building using an optimized back-propagation and Levenberg–Marquardt back-propagation neural network: Case study of a shopping mall in China | |
| Aly | An intelligent hybrid model of neuro Wavelet, time series and Recurrent Kalman Filter for wind speed forecasting | |
| Catalao et al. | An artificial neural network approach for short-term wind power forecasting in Portugal | |
| CN106779177A (en) | Multiresolution wavelet neutral net electricity demand forecasting method based on particle group optimizing | |
| CN114217524A (en) | Power grid real-time self-adaptive decision-making method based on deep reinforcement learning | |
| CN113489015B (en) | Multi-time-scale reactive voltage control method for power distribution network based on reinforcement learning | |
| Barzola-Monteses et al. | Energy consumption of a building by using long short-term memory network: a forecasting study | |
| CN106295857A (en) | A kind of ultrashort-term wind power prediction method | |
| Pandey et al. | Optimal placement & sizing of distributed generation (DG) to minimize active power loss using particle swarm optimization (PSO) | |
| Li et al. | An improved multiobjective estimation of distribution algorithm for environmental economic dispatch of hydrothermal power systems | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| CN114595884A (en) | Genetic intelligent optimization neural network wind power generation equipment temperature prediction method | |
| CN104021315A (en) | Method for calculating station service power consumption rate of power station on basis of BP neutral network | |
| Lekouaghet et al. | Adolescent identity search algorithm for parameter extraction in photovoltaic solar cells and modules | |
| Yu et al. | A robust method based on reinforcement learning and differential evolution for the optimal photovoltaic parameter extraction | |
| Ncir et al. | An advanced intelligent MPPT control strategy based on the imperialist competitive algorithm and artificial neural networks | |
| Jamshidi et al. | Using artificial neural networks and system identification methods for electricity price modeling | |
| Razmi et al. | Neural network based on a genetic algorithm for power system loading margin estimation | |
| CN110674460B (en) | E-Seq2Seq technology-based data driving type unit combination intelligent decision method | |
| Ahmed et al. | Solving combined economic and emission dispatch problem using the slime mould algorithm | |
| CN117057623A (en) | Comprehensive power grid safety optimization scheduling method, device and storage medium | |
| Petrova et al. | Neural network modelling of fermentation processes. Microorganisms cultivation model | |
| Peng et al. | Research on fault diagnosis method for transformer based on fuzzy genetic algorithm and artificial neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210810 Address after: 510663 3 building, 3, 4, 5 and J1 building, 11 building, No. 11, Ke Xiang Road, Luogang District Science City, Guangzhou, Guangdong. Applicant after: China South Power Grid International Co.,Ltd. Applicant after: SOUTH CHINA University OF TECHNOLOGY Address before: 510080 water Donggang 8, Dongfeng East Road, Yuexiu District, Guangzhou, Guangdong. Applicant before: China South Power Grid International Co.,Ltd. Applicant before: POWER GRID TECHNOLOGY RESEARCH CENTER. CHINA SOUTHERN POWER GRID Applicant before: SOUTH CHINA University OF TECHNOLOGY |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |