CN115907336A - Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm - Google Patents

Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm Download PDF

Info

Publication number
CN115907336A
CN115907336A CN202211330253.6A CN202211330253A CN115907336A CN 115907336 A CN115907336 A CN 115907336A CN 202211330253 A CN202211330253 A CN 202211330253A CN 115907336 A CN115907336 A CN 115907336A
Authority
CN
China
Prior art keywords
individual
population
scheduling
target object
module
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.)
Withdrawn
Application number
CN202211330253.6A
Other languages
Chinese (zh)
Inventor
龙文
焦建军
徐�明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guizhou University of Finance and Economics
Original Assignee
Guizhou University of Finance and Economics
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guizhou University of Finance and Economics filed Critical Guizhou University of Finance and Economics
Priority to CN202211330253.6A priority Critical patent/CN115907336A/en
Publication of CN115907336A publication Critical patent/CN115907336A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • 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

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the technical field of scheduling, and discloses a group optimization scheduling method based on a hybrid intelligent dimensionality reduction algorithm, which comprises the following steps: the system comprises a scheduling decision generation module, an initial population generation module, an extreme value determination module, a position updating module, an iteration module and an optimal scheduling module. According to the method, an entity extraction rule is generated by a scheduling decision generation module based on a preset target object scheduling procedure, so that a mode layer is constructed; and then extracting entities in the preset target object scheduling accident instance through an entity extraction rule to construct a data layer, so that a target object scheduling decision knowledge graph is generated according to the mode layer and the data layer. When a target object scheduling accident occurs, a target object scheduling decision aiming at the target object scheduling accident can be automatically generated according to the knowledge graph, the decision efficiency is high, and the limitation of manual decision is avoided; meanwhile, the iteration efficiency can be greatly improved through the iteration module, so that the scheduling efficiency is improved.

Description

Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm
Technical Field
The invention belongs to the technical field of scheduling, and particularly relates to a group optimization scheduling method based on a hybrid intelligent dimensionality reduction algorithm.
Background
The optimal scheduling of the hydropower station group is generally based on the maximum generated energy as an objective function, and a mathematical model of the optimal scheduling is described as follows: under the conditions of the warehousing flow process and the initial and final water levels of each hydropower station in the known scheduling period, complex constraint conditions such as water level, output, flow and the like are comprehensively considered, and the maximum target is the total generated energy of the hydropower system in the super-large watershed, so that the output and water level operation processes of each hydropower station are determined; however, the existing group optimization scheduling method based on the hybrid intelligent dimension reduction algorithm is slow in decision efficiency due to the fact that a scheduling decision is made manually, and the manual decision has limitations; meanwhile, the iteration efficiency of the population is low, and the scheduling efficiency is influenced.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) The existing group optimization scheduling method based on the hybrid intelligent dimension reduction algorithm is slow in decision efficiency due to the fact that scheduling decisions are made manually, and the manual decisions have limitations.
(2) The iteration efficiency on the population is low, and the scheduling efficiency is influenced.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a group optimization scheduling method based on a hybrid intelligent dimension reduction algorithm.
The invention is realized in this way, a group optimization scheduling system based on a hybrid intelligent dimensionality reduction algorithm comprises:
the system comprises a scheduling decision generation module, an initial population generation module, an extreme value determination module, a position updating module, an iteration module and an optimal scheduling module;
the scheduling decision generation module is connected with the initial population generation module and is used for generating a target object scheduling decision;
the initial population generating module is connected with the scheduling decision generating module and the extreme value determining module and used for sequentially serially coding the characteristics of the target objects at different time intervals according to the sequence of the target objects participating in calculation to obtain a single individual code value, randomly generating an initial population within a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
an extreme value determining module, connected to the initial population generating module and the location updating module, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual remains unchanged, otherwise, the current location of the individual is used to replace the individual extreme value of the individual, and the maximum value of the individual extreme value is selected from the individual extreme values of all the individuals in the current population as a global extreme value, wherein the individual extreme value represents the best location experienced by the individual, and the global extreme value represents the best location experienced by all the individuals in the current population;
the position updating module is connected with the extreme value determining module and the iteration module and used for updating the current position of each individual in the current population according to the global extreme value of the current population and the individual extreme value of each individual in the current population;
the iteration module is connected with the position updating module and the optimal scheduling module and is used for iterating the current population;
and the optimal scheduling module is connected with the iteration module and used for obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration.
A group optimization scheduling method based on a hybrid intelligent dimensionality reduction algorithm comprises the following steps:
generating a target object scheduling decision through a scheduling decision generating module;
sequentially and serially coding the characteristics of each target object at different time intervals according to the sequence of the target objects participating in calculation through an initial population generating module to obtain a single individual code value, randomly generating an initial population within a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
step three, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual is kept unchanged, otherwise, the current position of the individual is used for replacing the individual extreme value of the individual, and the maximum value of the individual extreme value is selected from the individual extreme values of all the individuals in the current population to serve as a global extreme value, wherein the individual extreme value represents the best position experienced by the individual, and the global extreme value represents the best position experienced by all the individuals in the current population;
updating the current position of each individual in the current population by the global extreme value of the current population and the individual extreme value of each individual in the current population through a position updating module;
step five, iteration is carried out on the current population through an iteration module;
and step six, obtaining the optimal scheduling process of each target object in different time periods by the optimal scheduling module according to the global optimal individual of the current population of the last iteration.
Further, the generation method of the scheduling decision generation module is as follows:
1) Acquiring a plurality of target object scheduling decision concepts according to a preset target object scheduling procedure; deconstructing the preset target object scheduling procedure, and determining the relation among the multiple items of target object scheduling decision concepts;
2) Learning the preset target object scheduling rules to generate entity extraction rules; constructing a mode layer according to the scheduling decision concept of the multiple target objects and the relation;
3) Performing entity learning on a preset target object scheduling accident instance according to the entity extraction rule, extracting entities in the preset target object scheduling accident instance, and constructing a data layer by adopting the entities; constructing a target object scheduling decision knowledge graph by adopting the mode layer and the data layer;
4) And when a target object scheduling accident occurs, generating a target object scheduling decision according to the target object scheduling decision knowledge graph.
Further, the step of obtaining a plurality of target object scheduling decision concepts according to the preset target object scheduling procedures comprises:
obtaining a target object scheduling decision term;
and extracting concept terms from the electric power target object scheduling decision terms based on a preset target object scheduling procedure to form a plurality of target object scheduling decision concepts.
Further, the step of deconstructing the preset target object scheduling procedure and determining the relationship between the plurality of target object scheduling decision concepts comprises:
deconstructing the preset target object scheduling procedure, and determining semantic relations among the multiple items of target object scheduling decision concepts;
and extracting the relation among the scheduling decision concepts of the multiple items of target objects according to the semantic relation.
Further, the step of learning the preset target object scheduling procedure and generating an entity extraction rule includes:
extracting general sentence patterns from the preset target object scheduling rules;
and learning the general sentence pattern to generate an entity extraction rule.
Further, the initial population generation module comprises:
sequentially and serially coding the characteristics of each target object in different time periods according to the sequence of the target objects participating in calculation to obtain a single individual code value, wherein the single individual code value is represented as follows:
Figure BDA0003913063630000041
Figure BDA0003913063630000042
representing the characteristic state of the nth target object in the jth time interval, wherein N is the number of target objects, N =1,2, \ 8230, N, T is the number of time intervals in the scheduling period, j =1,2, \ 8230, T;
setting k to an initial value of 1, and
Figure BDA0003913063630000043
randomly generating an initial population U within a preset feasible characteristic range k Wherein, U k Represents a kth generation of population, is selected>
Figure BDA0003913063630000044
Representing the current position of the individual i at the kth iteration, i =1,2, \ 8230;, m, r is [0, 1; (k;)]Interval allocationIn the random number of (a), in the combination of>
Figure BDA0003913063630000045
Respectively the upper limit and the lower limit of the corresponding variable, and m is the population scale.
Further, the extremum determining module includes:
for any individual in the current population, the population is selected from
Figure BDA0003913063630000046
A fitness value is calculated for each individual, wherein,
Figure BDA0003913063630000047
indicates that an individual is present>
Figure BDA0003913063630000048
Corresponding fitness value, G is the number of constraints, P n,j The output, t, of the nth target object in time period j j For scheduling the hours of the period XX k,g Representing particles +>
Figure BDA0003913063630000049
The value corresponding to the g-th constraint in the scheduling process, lambda, is obtained g For a violation penalty factor of the g-th constraint, <' >>
Figure BDA00039130636300000410
Are respectively XX k,q Upper and lower limits of (d);
by
Figure BDA0003913063630000051
Updating the individual extremum and the global extremum, wherein>
Figure BDA0003913063630000052
Represents the individual extremum of the individual i in the population of the kth generation, < > is selected>
Figure BDA0003913063630000053
Represents the individual extreme value, GB, of the individual i in the population of the k-1 generation k Represents a global extremum for the population of the kth generation, <' > is selected>
Figure BDA0003913063630000054
Representing the fitness value of the individual extreme value of an individual i in the population of the kth-1 generation;
for all individual extreme values in the current population, randomly selecting two different individual extreme values from the current population and subtracting to generate a difference vector, superposing the difference vector to the global extreme value according to a preset proportion to generate a variation vector as a new individual extreme value, replacing the individual extreme value of an individual with the new individual extreme value if the fitness value of the varied individual extreme value is better than the fitness value of the individual extreme value before variation, otherwise, keeping the individual extreme value of the individual unchanged comprises:
Figure BDA0003913063630000055
by
Figure BDA0003913063630000056
Performing a mutation operation on an individual extremum in any one of the individuals in the present population, wherein->
Figure BDA0003913063630000057
Represents the new extreme value of the individual generated after the mutation operation and is [0,1 ]]The random numbers allocated to the intervals, ind1 and Ind2, respectively, represent integers randomly selected from the set {1,2, \8230;, m }, and Ind1 ≠ Ind2.
Further, the location update module includes:
by
Figure BDA0003913063630000058
Updating the current position of each individual in the current population, wherein k is the current iteration number and/or the judgment result is obtained>
Figure BDA0003913063630000059
For individual i at the kth iterationThe current position, mbestk, is the optimal center of position of the population at the kth iteration, </or>
Figure BDA00039130636300000510
Indicating an intervening on the kth iteration>
Figure BDA00039130636300000511
And GB k In the position between, v, u is [0,1 ]]With uniformly distributed random numbers, and β k represents the contraction-expansion coefficient at the k-th iteration.
Further, the iterative module iterative method is as follows:
(1) If delta is greater than or equal to P a Then, randomly migrating several individuals from the current population to form an external file set, wherein delta is [0,1 ]]A random number that is randomly distributed in the interval,
Figure BDA0003913063630000061
k represents the current iteration number and represents the maximum iteration number;
(2) Forming a next generation population;
using a negative value of an original fitness value of each individual in the external file set as a target fitness value of each individual in the external file set, performing expansion or contraction operation on a mapping point of the individual corresponding to the maximum target fitness value according to the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value of all the individuals in the external file set, re-determining the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value, performing expansion or expansion operation on the mapping point of the individual corresponding to the new maximum target fitness value until the preset execution times are met, combining the external file set and the current population, and selecting a plurality of previous individuals with better fitness from the combined population to replace the individuals in the current population so as to form a next generation population;
wherein the dilation operation is represented as: x e ＝X center +β(X r -X center ) The shrink operation is expressed as: x c ＝X cente r+γ(X high -X center ) In the formula, X r For said mapped point, X e To map a point X r Expansion point after expansion operation, X c To map point X r Point of contraction after contraction operation, X center Collectively divide the maximum target fitness value f (X) for the external profile high ) Corresponding individual X high The average position of all other individuals, beta is the expansion coefficient, and gamma is the contraction coefficient;
(3) Increasing population iteration times, if the iteration times of the current population are less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration;
and increasing the iteration times of the population, if the iteration times of the current population is less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration comprises the following steps:
(a) Let J =1, use the external archive set S by F (X) = -F (X) k As the target fitness value of each individual in the external archive set, wherein S k The number of the Chinese individuals is D +1, D = N × T;
(b) Determination of S k Maximum target fitness value f (X) of (1) high ) Corresponding individual X high Individual X corresponding to next largest target fitness value f (Xsec) sec Minimum target fitness value f (X) low ) Corresponding individual X low And calculating S k Removing X in high Mean position X of all but one individuals center And by calculating X high Mapping point X of r α is the mapping coefficient, if f (X) low )≤f(X r )≤f(X sec ) Then X high ＝X r And performing step (e) if f (Xr)<f(X low ) Then step (c) is performed if f (X) r )>f(X sec ) If yes, executing step (d);
(c) From X e ＝X center +β(X r -X center ) Mapping point X r Performing an expansion operation to obtain an expansion point X e Beta is a coefficient of expansion, beta is,
if f (X) e )≤f(X low ) Then let X high = Xe and perform step (e), otherwise let X be high ＝X r And performing step (6.5);
(d) If f (X) r )>f(X sec ) And f (X) r )≤f(X high ) Let X be high ＝X r Re-determining Xr according to step (b), then by Xc = X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c If f (Xr)>f(X high ) Then directly from X c ＝X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c And gamma is a shrinkage coefficient;
if f (X) c )≤f(X high ) Then let X high ＝X c And performing step (e);
(e) Increase the value of J by 1 if
Figure BDA0003913063630000071
Then step (f) is performed, otherwise return to performing step (b), ->
Figure BDA0003913063630000072
Is a preset maximum execution algebra;
(f) Merging external archive sets S k With the current population U k Selecting the first m individuals with better fitness from the merged population to replace the current population U k Of (a).
In combination with the above technical solutions and the technical problems to be solved, please analyze the advantages and positive effects of the technical solutions to be protected in the present invention from the following aspects:
first, aiming at the technical problems existing in the prior art and the difficulty in solving the problems, the technical problems to be solved by the technical scheme of the present invention are closely combined with results, data and the like in the research and development process, and some creative technical effects are brought after the problems are solved. The specific description is as follows:
the method comprises the steps that a scheduling decision generation module determines a plurality of target object scheduling decision concepts and a relation between the plurality of target object scheduling decision concepts according to a preset target object scheduling procedure, and generates an entity extraction rule based on the preset target object scheduling procedure, so that a mode layer is constructed; and then extracting entities in the preset target object scheduling accident instance through an entity extraction rule to construct a data layer, so that a target object scheduling decision knowledge graph is generated according to the mode layer and the data layer. When a target object scheduling accident occurs, a target object scheduling decision aiming at the target object scheduling accident can be automatically generated according to the knowledge graph, the decision efficiency is high, and the limitation of manual decision is avoided; meanwhile, the iteration module can greatly improve the iteration efficiency, so that the scheduling efficiency is improved.
Secondly, considering the technical scheme as a whole or from the perspective of products, the technical effect and advantages of the technical scheme to be protected by the invention are specifically described as follows:
the method comprises the steps that a scheduling decision generation module determines a plurality of items of target object scheduling decision concepts and the relation between the plurality of items of target object scheduling decision concepts according to a preset target object scheduling procedure, and generates an entity extraction rule based on the preset target object scheduling procedure, so that a mode layer is constructed; and then extracting entities in the preset target object scheduling accident instance through an entity extraction rule to construct a data layer, so that a target object scheduling decision knowledge graph is generated according to the mode layer and the data layer. When a target object scheduling accident occurs, a target object scheduling decision aiming at the target object scheduling accident can be automatically generated according to the knowledge graph, the decision efficiency is high, and the limitation of manual decision is avoided; meanwhile, the iteration module can greatly improve the iteration efficiency, so that the scheduling efficiency is improved.
Drawings
Fig. 1 is a flowchart of a group optimization scheduling method based on a hybrid intelligent dimension reduction algorithm according to an embodiment of the present invention.
Fig. 2 is a block diagram of a group optimization scheduling system based on a hybrid intelligent dimension reduction algorithm according to an embodiment of the present invention.
Fig. 3 is a flowchart of a method for generating a scheduling decision generating module according to an embodiment of the present invention.
Fig. 4 is a flowchart of an iteration module iteration method provided in an embodiment of the present invention.
In fig. 2: 1. a scheduling decision generating module; 2. an initial population generation module; 3. an extremum determining module; 4. a location update module; 5. an iteration module; 6. and an optimal scheduling module.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
1. The embodiments are explained. This section is an explanatory embodiment expanding on the claims so as to fully understand how the present invention is embodied by those skilled in the art.
As shown in fig. 1, the group optimization scheduling method based on the hybrid intelligent dimensionality reduction algorithm provided by the present invention includes the following steps:
s101, generating a target object scheduling decision through a scheduling decision generating module;
s102, sequentially and serially coding the characteristics of each target object at different time intervals according to the sequence of the target objects participating in calculation through an initial population generation module to obtain a single individual code value, randomly generating an initial population within a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
s103, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual is kept unchanged, otherwise, the current position of the individual is used for replacing the individual extreme value of the individual, and the maximum value of the individual extreme value is selected from the individual extreme values of all the individuals in the current population to serve as a global extreme value, wherein the individual extreme value represents the best position experienced by the individual, and the global extreme value represents the best position experienced by all the individuals in the current population;
s104, updating the current position of each individual in the current population through the position updating module according to the global extreme value of the current population and the individual extreme value of each individual in the current population;
s105, iterating the current population through an iteration module;
and S106, obtaining the optimal scheduling process of each target object in different time periods by the optimal scheduling module according to the global optimal individual of the current population of the last iteration.
As shown in fig. 2, the group optimization scheduling system based on the hybrid intelligent dimension reduction algorithm provided by the embodiment of the present invention includes: the system comprises a scheduling decision generation module 1, an initial population generation module 2, an extreme value determination module 3, a position updating module 4, an iteration module 5 and an optimal scheduling module 6.
The scheduling decision generation module 1 is connected with the initial population generation module 2 and is used for generating a target object scheduling decision;
the initial population generating module 2 is connected with the scheduling decision generating module 1 and the extreme value determining module 3, and is used for sequentially and serially coding the characteristics of the target objects at different time intervals according to the sequence of the target objects participating in the calculation to obtain a single individual code value, randomly generating an initial population in a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
an extreme value determining module 3, connected to the initial population generating module 2 and the location updating module 4, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual remains unchanged, otherwise, the current location of the individual is used to replace the individual extreme value of the individual, and the maximum value of the individual extreme values is selected from the individual extreme values of all individuals in the current population as a global extreme value, wherein the individual extreme value represents the best location experienced by the individual, and the global extreme value represents the best location experienced by all individuals in the current population;
the position updating module 4 is connected with the extreme value determining module 3 and the iteration module 5 and is used for updating the current position of each individual in the current population according to the global extreme value of the current population and the individual extreme value of each individual in the current population;
the iteration module 5 is connected with the position updating module 4 and the optimal scheduling module 6 and is used for iterating the current population;
and the optimal scheduling module 6 is connected with the iteration module 5 and is used for obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration.
As shown in fig. 3, the scheduling decision generating module 1 provided by the present invention generates the following method:
s201, acquiring a plurality of target object scheduling decision concepts according to a preset target object scheduling rule; deconstructing the preset target object scheduling procedure, and determining the relation among the multiple items of target object scheduling decision concepts;
s202, learning the preset target object scheduling procedure to generate an entity extraction rule; constructing a mode layer according to the scheduling decision concept of the multiple target objects and the relation;
s203, performing entity learning on a preset target object scheduling accident instance according to the entity extraction rule, extracting an entity in the preset target object scheduling accident instance, and constructing a data layer by adopting the entity; constructing a target object scheduling decision knowledge graph by adopting the mode layer and the data layer;
and S204, when a target object scheduling accident occurs, generating a target object scheduling decision according to the target object scheduling decision knowledge graph.
The step of obtaining a plurality of target object scheduling decision concepts according to the preset target object scheduling procedures provided by the invention comprises the following steps:
obtaining a target object scheduling decision term;
and extracting concept terms from the electric power target object scheduling decision terms based on a preset target object scheduling procedure to form a plurality of target object scheduling decision concepts.
The steps provided by the invention for deconstructing the preset target object scheduling procedure and determining the relation among the multiple target object scheduling decision concepts comprise:
deconstructing the preset target object scheduling procedure, and determining semantic relations among the multiple items of target object scheduling decision concepts;
and extracting the relation among the multiple items of target object scheduling decision concepts according to the semantic relation.
The step of learning the preset target object scheduling procedure and generating the entity extraction rule provided by the invention comprises the following steps:
extracting a general sentence pattern from the preset target object scheduling rule;
and learning the general sentence pattern to generate an entity extraction rule.
The initial population generation module 2 provided by the invention comprises:
sequentially and serially coding the characteristics of the target objects at different time intervals according to the sequence of the target objects participating in calculation to obtain a single individual code value, wherein the single individual code value is represented as:
Figure BDA0003913063630000111
Figure BDA0003913063630000121
representing the characteristic state of the nth target object in the jth time interval, wherein N is the number of target objects, N =1,2, \ 8230, N, T is the number of time intervals in the scheduling period, j =1,2, \ 8230, T;
setting the initial value of k to 1, and
Figure BDA0003913063630000122
randomly generating an initial population U within a preset feasible characteristic range k Wherein, U k Represents a kth generation of population, <' > based on>
Figure BDA0003913063630000123
Representing the current position of the individual i at the kth iteration, i =1,2, \ 8230;, m, r is [0, 1; (k;)]Random number for interval allocation>
Figure BDA0003913063630000124
Respectively the upper limit and the lower limit of the corresponding variable, and m is the population scale.
The extremum determining module 3 provided by the invention comprises:
for any individual in the current population, the population is determined by
Figure BDA0003913063630000125
A fitness value is calculated for each individual, wherein,
Figure BDA0003913063630000126
indicates that an individual is present>
Figure BDA0003913063630000127
Corresponding fitness value, G is the number of constraints, P n,j The output, t, of the nth target object in time period j j For scheduling the hours of the period XX k,g Indicates that a particle is->
Figure BDA0003913063630000128
Obtaining the value, lambda, corresponding to the g-th constraint in the scheduling process g For a destruction penalty of the g-th constraint>
Figure BDA0003913063630000129
Are respectively XX k,q Upper and lower limits of (d);
by
Figure BDA00039130636300001210
Updating the individual extremum and the global extremum, wherein>
Figure BDA00039130636300001211
Represents the individual extremum of the individual i in the population of the kth generation, < > is selected>
Figure BDA00039130636300001212
Represents the individual extreme value, GB, of the individual i in the population of the k-1 generation k Represents a global extremum for the population of the kth generation, <' > is selected>
Figure BDA00039130636300001213
Expressing the fitness value of the individual extreme value of the individual i in the population of the (k-1) th generation;
for all individual extreme values in the current population, randomly selecting two different individual extreme values from the current population and subtracting to generate a difference vector, superposing the difference vector to the global extreme value according to a preset proportion to generate a variation vector as a new individual extreme value, replacing the individual extreme value of an individual with the new individual extreme value if the fitness value of the varied individual extreme value is better than the fitness value of the individual extreme value before variation, otherwise, keeping the individual extreme value of the individual unchanged comprises:
Figure BDA0003913063630000131
by
Figure BDA0003913063630000132
Performing a mutation operation on an individual extremum of any individual in the current population, wherein ` H `>
Figure BDA0003913063630000133
Represents the new extreme value of the individual generated after mutation operation, and is [0,1 ]]Random numbers of interval allocation, ind1 and Ind2 respectively represent integers randomly selected from the set {1,2, \8230;, m }, and Ind1 ≠ Ind2.
The location update module 4 provided by the present invention includes:
by
Figure BDA0003913063630000134
Updating the current position of each individual in the current population, wherein k is the current iteration number,
Figure BDA0003913063630000135
is the current position of the individual i at the kth iteration, mbestk is the optimal position center of the population at the kth iteration, and ` is `>
Figure BDA0003913063630000136
Indicating an on->
Figure BDA0003913063630000137
And GB k In the position between, v, u is [0,1 ]]With uniformly distributed random numbers, and β k represents the contraction-expansion coefficient at the kth iteration.
As shown in fig. 4, the iteration module 5 provided by the present invention has the following iteration method:
s301, if delta is larger than or equal to P a Then, randomly emigrating a plurality of individuals from the current population to form an external archive set;
wherein, δ is [0,1 ]]The random number is randomly distributed in the interval,
Figure BDA0003913063630000138
k represents the current iteration number and represents the maximum iteration number;
s302, forming a next generation population;
using a negative value of an original fitness value of each individual in the external file set as a target fitness value of each individual in the external file set, performing expansion or contraction operation on a mapping point of the individual corresponding to the maximum target fitness value according to the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value of all the individuals in the external file set, re-determining the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value, performing expansion or expansion operation on the mapping point of the individual corresponding to the new maximum target fitness value until the preset execution times are met, combining the external file set and the current population, and selecting a plurality of previous individuals with better fitness from the combined population to replace the individuals in the current population so as to form a next generation population;
wherein the dilation operation is represented as: x e ＝X center +β(X r -X center ) The shrink operation is expressed as: x c ＝X cente r+γ(X high -X center ) In the formula, X r For said mapped point, X e To map a point X r Expansion point after expansion operation, X c To map a point X r Point of contraction after contraction operation, X center Collectively divide the maximum target fitness value f (X) for the external profile high ) Corresponding individual X high The average position of all other individuals, beta is the expansion coefficient, and gamma is the contraction coefficient;
s303, increasing the iteration times of the population, if the iteration times of the current population is less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the global optimal individual of the current population iterated for the last time;
and increasing the iteration times of the population, if the iteration times of the current population is less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration comprises the following steps:
(a) Let J =1, use the external archive set S by F (X) = -F (X) k As the target fitness value of each individual in the external archive set, wherein S k The number of the medium individuals is D +1, D= N multiplied by T;
(b) Determination of S k Maximum target fitness value f (X) of (1) high ) Corresponding individual X high Individual X corresponding to next largest target fitness value f (Xsec) sec Minimum target fitness value f (X) low ) Corresponding individual X low And calculate S k Middle removing X high Mean position X of all but one individuals center And by calculating X high Mapping point X of r α isMapping coefficients, if f (X) low )≤f(X r )≤f(X sec ) Then X high ＝X r And performing step (e) if f (Xr)<f(X low ) Then step (c) is performed if f (X) r )>f(X sec ) If yes, executing step (d);
(c) From X e ＝X center +β(X r -X center ) Mapping point X r Performing an expansion operation to obtain an expansion point X e And beta is the coefficient of expansion,
if f (X) e )≤f(X low ) Then let X high = Xe and perform step (e), otherwise let X high ＝X r And performing step (6.5);
(d) If f (X) r )>f(X sec ) And f (X) r )≤f(X high ) Let X high ＝X r Re-determining Xr according to step (b), then by Xc = X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c If f (Xr)>f(X high ) Then directly from X c ＝X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c And gamma is a shrinkage coefficient;
if f (X) c )≤f(X high ) Then let X high ＝X c And performing step (e);
(e) Increase the value of J by 1 if
Figure BDA0003913063630000151
Then step (f) is performed, otherwise return to performing step (b), ->
Figure BDA0003913063630000152
Is a preset maximum execution algebra;
(f) Merging external archive sets S k With the current population U k Selecting the first m individuals with better fitness from the merged population to replace the current population U k Of (a).
2. Application examples. In order to prove the creativity and the technical value of the technical scheme of the invention, the part is the application example of the technical scheme of the claims on specific products or related technologies.
The method comprises the steps that a scheduling decision generation module determines a plurality of items of target object scheduling decision concepts and the relation between the plurality of items of target object scheduling decision concepts according to a preset target object scheduling procedure, and generates an entity extraction rule based on the preset target object scheduling procedure, so that a mode layer is constructed; and extracting entities in the preset target object scheduling accident instance through an entity extraction rule to construct a data layer, so that a target object scheduling decision knowledge graph is generated according to the mode layer and the data layer. When a target object scheduling accident occurs, a target object scheduling decision aiming at the target object scheduling accident can be automatically generated according to the knowledge graph, the decision efficiency is high, and the limitation of manual decision is avoided; meanwhile, the iteration module can greatly improve the iteration efficiency, so that the scheduling efficiency is improved.
It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. It will be appreciated by those skilled in the art that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, for example such code provided on a carrier medium such as a diskette, CD-or DVD-ROM, a programmable memory such as read-only memory (firmware) or a data carrier such as an optical or electronic signal carrier. The apparatus of the present invention and its modules may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, or software executed by various types of processors, or a combination of hardware circuits and software, e.g., firmware.
3. Evidence of the relevant effects of the examples. The embodiment of the invention achieves some positive effects in the process of research and development or use, and has great advantages compared with the prior art, and the following contents are described by combining data, diagrams and the like in the test process.
The method comprises the steps that a scheduling decision generation module determines a plurality of target object scheduling decision concepts and a relation between the plurality of target object scheduling decision concepts according to a preset target object scheduling procedure, and generates an entity extraction rule based on the preset target object scheduling procedure, so that a mode layer is constructed; and extracting entities in the preset target object scheduling accident instance through an entity extraction rule to construct a data layer, so that a target object scheduling decision knowledge graph is generated according to the mode layer and the data layer. When a target object scheduling accident occurs, a target object scheduling decision aiming at the target object scheduling accident can be automatically generated according to the knowledge graph, the decision efficiency is high, and the limitation of manual decision is avoided; meanwhile, the iteration module can greatly improve the iteration efficiency, so that the scheduling efficiency is improved.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.

Claims (10)

1. A group optimization scheduling system based on a hybrid intelligent dimension reduction algorithm is characterized by comprising:
the system comprises a scheduling decision generation module, an initial population generation module, an extreme value determination module, a position updating module, an iteration module and an optimal scheduling module;
the scheduling decision generation module is connected with the initial population generation module and used for generating a target object scheduling decision;
the initial population generating module is connected with the scheduling decision generating module and the extreme value determining module and used for sequentially serially coding the characteristics of the target objects at different time intervals according to the sequence of the target objects participating in calculation to obtain a single individual code value, randomly generating an initial population within a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
an extreme value determining module, connected to the initial population generating module and the location updating module, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual remains unchanged, otherwise, the current location of the individual is used to replace the individual extreme value of the individual, and the maximum value of the individual extreme value is selected from the individual extreme values of all the individuals in the current population as a global extreme value, wherein the individual extreme value represents the best location experienced by the individual, and the global extreme value represents the best location experienced by all the individuals in the current population;
the position updating module is connected with the extreme value determining module and the iteration module and used for updating the current position of each individual in the current population according to the global extreme value of the current population and the individual extreme value of each individual in the current population;
the iteration module is connected with the position updating module and the optimal scheduling module and is used for iterating the current population;
and the optimal scheduling module is connected with the iteration module and used for obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration.
2. The group optimization scheduling method based on the hybrid intelligent dimension reduction algorithm according to claim 1, wherein the group optimization scheduling method based on the hybrid intelligent dimension reduction algorithm comprises the following steps:
generating a target object scheduling decision through a scheduling decision generating module;
sequentially and serially coding the characteristics of each target object at different time intervals according to the sequence of the target objects participating in calculation through an initial population generating module to obtain a single individual code value, randomly generating an initial population within a preset feasible characteristic range according to the single individual code value, and taking the initial population as a current population;
step three, for any individual in the current population, if the fitness value of the individual is smaller than the historical optimal fitness value, the individual extreme value of the individual is kept unchanged, otherwise, the current position of the individual is used for replacing the individual extreme value of the individual, and the maximum value of the individual extreme value is selected from the individual extreme values of all the individuals in the current population to serve as a global extreme value, wherein the individual extreme value represents the best position experienced by the individual, and the global extreme value represents the best position experienced by all the individuals in the current population;
updating the current position of each individual in the current population by the global extreme value of the current population and the individual extreme value of each individual in the current population through a position updating module;
step five, iteration is carried out on the current population through an iteration module;
and step six, obtaining the optimal scheduling process of each target object in different time periods by the optimal scheduling module according to the global optimal individual of the current population of the last iteration.
3. The group optimization scheduling system based on the hybrid intelligent dimensionality reduction algorithm of claim 1, wherein the scheduling decision generation module generates the following method:
1) Acquiring a plurality of target object scheduling decision concepts according to a preset target object scheduling procedure; deconstructing the preset target object scheduling procedure, and determining the relation among the multiple items of target object scheduling decision concepts;
2) Learning the preset target object scheduling rules to generate entity extraction rules; constructing a mode layer according to the scheduling decision concept of the multiple target objects and the relation;
3) Performing entity learning on a preset target object scheduling accident instance according to the entity extraction rule, extracting entities in the preset target object scheduling accident instance, and constructing a data layer by adopting the entities; constructing a target object scheduling decision knowledge graph by adopting the mode layer and the data layer;
4) And when a target object scheduling accident occurs, generating a target object scheduling decision according to the target object scheduling decision knowledge graph.
4. The group optimization scheduling system based on the hybrid intelligent dimensionality reduction algorithm according to claim 3, wherein the step of obtaining a plurality of target object scheduling decision concepts according to a preset target object scheduling procedure comprises:
obtaining a target object scheduling decision term;
and extracting concept terms from the electric power target object scheduling decision terms based on a preset target object scheduling procedure to form a plurality of target object scheduling decision concepts.
5. The hybrid intelligent dimension reduction algorithm based group optimized dispatching system of claim 3, wherein the step of deconstructing the preset target object dispatching procedure, determining the relationship between the plurality of target object dispatching decision concepts, comprises:
deconstructing the preset target object scheduling procedure, and determining semantic relations among the multiple items of target object scheduling decision concepts;
and extracting the relation among the multiple items of target object scheduling decision concepts according to the semantic relation.
6. The group-optimized dispatching system based on the hybrid intelligent dimension-reducing algorithm as claimed in claim 3, wherein the step of learning the preset target object dispatching procedure to generate the entity extraction rule comprises:
extracting a general sentence pattern from the preset target object scheduling rule;
and learning the general sentence pattern to generate an entity extraction rule.
7. The group optimal scheduling system based on the hybrid intelligent dimensionality reduction algorithm according to claim 1, wherein the initial population generating module comprises:
sequentially and serially coding the characteristics of each target object in different time periods according to the sequence of the target objects participating in calculation to obtain a single individual code value, wherein the single individual code value is represented as follows:
Figure FDA0003913063620000031
Figure FDA0003913063620000032
representing the characteristic state of the nth target object in the jth time interval, wherein N is the number of target objects, N =1,2, \ 8230, N, T is the number of time intervals in the scheduling period, j =1,2, \ 8230, T;
setting the initial value of k to 1, and
Figure FDA0003913063620000033
randomly generating an initial population U within a preset feasible characteristic range k Wherein, U k Represents a kth generation of population, <' > based on>
Figure FDA0003913063620000034
Representing the current position of the individual i at the kth iteration, i =1,2, \ 8230;, m, r is [0, 1; (k;)]Random number allocated in a section, based on the number of hours or hours>
Figure FDA0003913063620000035
Respectively the upper limit and the lower limit of the corresponding variable, and m is the population scale.
8. The hybrid intelligent dimension reduction algorithm based group optimized dispatch system of claim 1, wherein the extremum determining module comprises:
for any individual in the current population, the population is determined by
Figure FDA0003913063620000041
A fitness value is calculated for each individual, wherein,
Figure FDA0003913063620000044
indicates that an individual is present>
Figure FDA0003913063620000045
Corresponding fitness value, G is the number of constraints, P n,j The output, t, of the nth target object in time period j j For scheduling the hours of the period, XX k,g Indicates that a particle is->
Figure FDA0003913063620000046
Obtaining the value, lambda, corresponding to the g-th constraint in the scheduling process g For a destruction penalty of the g-th constraint>
Figure FDA0003913063620000047
Are respectively XX k,q The upper and lower limits of (d);
by
Figure FDA0003913063620000042
Updating the individual extremum and the global extremum, wherein>
Figure FDA00039130636200000410
Represents the individual extremum of the individual i in the population of the kth generation, < > is selected>
Figure FDA0003913063620000048
Represents the individual extreme value, GB, of the individual i in the population of the k-1 generation k Represents a global extremum for a population of the kth generation>
Figure FDA0003913063620000049
Representing the fitness value of the individual extreme value of an individual i in the population of the kth-1 generation;
for all individual extreme values in the current population, randomly selecting two different individual extreme values from the current population and subtracting to generate a difference vector, superposing the difference vector to the global extreme value according to a preset proportion to generate a variation vector as a new individual extreme value, replacing the individual extreme value of an individual with the new individual extreme value if the fitness value of the varied individual extreme value is better than the fitness value of the individual extreme value before variation, otherwise, keeping the individual extreme value of the individual unchanged comprises:
by
Figure FDA0003913063620000043
Performing a mutation operation on an individual extremum of any individual in the current population, wherein ` H `>
Figure FDA00039130636200000411
Represents the new extreme value of the individual generated after the mutation operation and is [0,1 ]]Random numbers of interval allocation, ind1 and Ind2 respectively represent integers randomly selected from the set {1,2, \8230;, m }, and Ind1 ≠ Ind2.
9. The hybrid intelligent dimensionality reduction algorithm-based group optimization scheduling system of claim 1, wherein the location update module comprises:
by
Figure FDA0003913063620000051
Updating the current position of each individual in the current population, wherein k is the current iteration number and/or the judgment result is obtained>
Figure FDA0003913063620000053
Is the current position of the individual i at the kth iteration, mbestk is the optimal position center of the population at the kth iteration, and ` is `>
Figure FDA0003913063620000054
Indicating an on->
Figure FDA0003913063620000055
And GB k In the position between, v, u is [0,1 ]]With uniformly distributed random numbers, and β k represents the contraction-expansion coefficient at the k-th iteration. />
10. The hybrid intelligent dimension reduction algorithm-based group optimization scheduling system of claim 1, wherein the iterative module iterative method is as follows:
(1) If delta is not less than P a Then, a number of individuals are randomly migrated from the current population to form an external archive set, where δ is [0,1 ]]The random number is randomly distributed in the interval,
Figure FDA0003913063620000052
k represents the current iteration number and represents the maximum iteration number;
(2) Forming a next generation population;
using a negative value of an original fitness value of each individual in the external file set as a target fitness value of each individual in the external file set, performing expansion or contraction operation on a mapping point of the individual corresponding to the maximum target fitness value according to the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value of all the individuals in the external file set, re-determining the individual corresponding to the maximum target fitness value, the individual corresponding to the second largest target fitness value and the individual corresponding to the minimum target fitness value, performing expansion or expansion operation on the mapping point of the individual corresponding to the new maximum target fitness value until the preset execution times are met, combining the external file set and the current population, and selecting a plurality of previous individuals with better fitness from the combined population to replace the individuals in the current population so as to form a next generation population;
wherein the dilation operation is represented as: x e ＝X center +β(X r -X center ) The shrink operation is expressed as: x c ＝X cente r+γ(X high -X center ) In the formula, X r For said mapped point, X e To map point X r Expansion point after expansion operation, X c To map point X r Point of contraction after contraction operation, X center Collectively divide the maximum target fitness value f (X) for the external profile high ) Corresponding individual X high The average position of all other individuals, beta is the expansion coefficient, and gamma is the contraction coefficient;
(3) Increasing population iteration times, if the iteration times of the current population are less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration;
and increasing the iteration times of the population, if the iteration times of the current population is less than the maximum iteration times, taking the next generation population as the current population, and returning to the step of determining the execution extreme value, otherwise, obtaining the optimal scheduling process of each target object in different time periods by the globally optimal individual of the current population of the last iteration comprises the following steps:
(a) Let J =1, use the external archive set S by F (X) = -F (X) k As the target fitness value of each individual in the external archive set, wherein S k The number of the Chinese individuals is D +1, D = N × T;
(b) Determination of S k Maximum target fitness value f (X) of (1) high ) Corresponding individual X high Individual X corresponding to sub-maximum target fitness value f (Xsec) sec Minimum target fitness value f (X) low ) Corresponding individual X low And calculate S k Middle removing X high Mean position X of all but one individuals center And by calculating X high Is mapped to point X r α is the mapping coefficient, if f (X) low )≤f(X r )≤f(X sec ) Then X high ＝X r And performing step (e) if f (Xr)<f(X low ) Then step (c) is performed if f (X) r )>f(X sec ) If yes, executing step (d);
(c) From X e ＝X center +β(X r -X center ) Mapping point X r Performing an expansion operation to obtain an expansion point X e And beta is the coefficient of expansion,
if f (X) e )≤f(X low ) Then let X high = Xe and perform step (e), otherwise let X high ＝X r And performing step (6.5);
(d) If f (X) r )>f(X sec ) And f (X) r )≤f(X high ) Let X high ＝X r Re-determining Xr according to step (b), then by Xc = X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c If f (Xr)>f(X high ) Then directly from X c ＝X center +γ(X high -X center ) Performing a shrinking operation to obtain a shrinking point X c And gamma is a shrinkage coefficient;
if f (X) c )≤f(X high ) Then let X high ＝X c And performing step (e);
(e) Increase the value of J by 1 if
Figure FDA0003913063620000071
Then step (f) is performed, otherwise return to performing step (b), ->
Figure FDA0003913063620000072
Is a preset maximum execution algebra; />
(f) Merging external archive sets S k With the current population U k Selecting the first m individuals with better fitness from the merged population to replace the current population U k Of (a).
CN202211330253.6A 2022-10-27 2022-10-27 Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm Withdrawn CN115907336A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211330253.6A CN115907336A (en) 2022-10-27 2022-10-27 Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211330253.6A CN115907336A (en) 2022-10-27 2022-10-27 Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm

Publications (1)

Publication Number Publication Date
CN115907336A true CN115907336A (en) 2023-04-04

Family

ID=86482558

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211330253.6A Withdrawn CN115907336A (en) 2022-10-27 2022-10-27 Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm

Country Status (1)

Country Link
CN (1) CN115907336A (en)

Similar Documents

Publication Publication Date Title
Kim et al. Zero-centered fixed-point quantization with iterative retraining for deep convolutional neural network-based object detectors
CN108985515B (en) New energy output prediction method and system based on independent cyclic neural network
Juang et al. Reduced interval type-2 neural fuzzy system using weighted bound-set boundary operation for computation speedup and chip implementation
CN109948833A (en) A kind of Hydropower Unit degradation trend prediction technique based on shot and long term memory network
CN112884236B (en) Short-term load prediction method and system based on VDM decomposition and LSTM improvement
Shin et al. Prediction confidence based low complexity gradient computation for accelerating DNN training
CN111160516B (en) Convolutional layer sparsification method and device for deep neural network
CN114676645A (en) Non-stationary time sequence prediction method and system
CN115391745B (en) Rainfall forecast correction method and system based on probability matching average method
CN109840348B (en) Triple acceleration topology optimization method
CN115907131A (en) Method and system for building electric heating load prediction model in northern area
CN115907336A (en) Group optimization scheduling method based on hybrid intelligent dimension reduction algorithm
CN113625144A (en) IGBT fault prediction method and system
CN114239945A (en) Short-term power load prediction method, device, equipment and storage medium
CN115438590B (en) Precipitation prediction correction method and system based on BP neural network
CN117910334A (en) Enterprise power consumption prediction method based on VMD-GWO-HKELM
CN113779861B (en) Photovoltaic Power Prediction Method and Terminal Equipment
CN115907000A (en) Small sample learning method for optimal power flow prediction of power system
CN116384574A (en) W-LSTM-based non-parameter load interval prediction method
CN111931994A (en) Short-term load and photovoltaic power prediction method, system, equipment and medium thereof
Lückehe et al. Simulated Annealing With Parameter Tuning for Wind Turbine Placement Optimization.
CN112381226B (en) Deep convolutional neural network architecture searching method based on particle swarm optimization
CN117972363B (en) Wind speed prediction method and system for energy storage auxiliary black start based on stability
CN110829434A (en) Method for improving expansibility of deep neural network tidal current model
CN114282614B (en) Medium-long runoff prediction method for optimizing CNN-GRU based on random forest and IFDA

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20230404

WW01 Invention patent application withdrawn after publication