CN104809593A - Distributed type parallel computation management method of power system - Google Patents

Distributed type parallel computation management method of power system Download PDF

Info

Publication number
CN104809593A
CN104809593A CN201510239699.1A CN201510239699A CN104809593A CN 104809593 A CN104809593 A CN 104809593A CN 201510239699 A CN201510239699 A CN 201510239699A CN 104809593 A CN104809593 A CN 104809593A
Authority
CN
China
Prior art keywords
user
subsystem
management
computation
calculating
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
Application number
CN201510239699.1A
Other languages
Chinese (zh)
Other versions
CN104809593B (en
Inventor
张海波
朱存浩
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.)
North China Electric Power University
Original Assignee
North China Electric Power University
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 North China Electric Power University filed Critical North China Electric Power University
Priority to CN201510239699.1A priority Critical patent/CN104809593B/en
Publication of CN104809593A publication Critical patent/CN104809593A/en
Application granted granted Critical
Publication of CN104809593B publication Critical patent/CN104809593B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Supply And Distribution Of Alternating Current (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a distributed type parallel computation management method of a power system. The method comprises the following steps: A, when any user of any subsystem carries out any high-grade application computation at any moment, the user sends a computation request to a coordination management object through a user management object; B, the coordination management object sends a computation starting instruction to other subsystems; C, subsystem management objects of the other subsystems develop a new computation progress in respective subsystem, and a shadow user is established in the progress; D, a user management object of the shadow user is connected with the coordination management object and decomposition-coordination computation is prepared; E, a computation object is established by each user and a coordination layer; F, the coordination management object distributes three-dimensional coordinates to all the computation objects, and all the computation objects are mutually positioned and communicated through the three-dimensional coordinates; G, a coordination computation object finishes decomposition coordination between the computation objects and each computation object finishes iteration according to a decomposition-coordination mechanism with a subsystem master-slave mechanism.

Description

Electric system Distributed Parallel Computing management method
Technical field
The present invention relates to Power System Analysis technical field, relate in particular to technical field of electric system energy management.
Background technology
The power production management of China has defined the system of a whole set of " differentiated control, hierarchical control, step-by-step processing ".Dispatching center at different levels all sets up comparatively detailed electric power system model for the electrical network in region within the jurisdiction, but each partition power grid does not understand operational factor and the state of other partition power grid.And physical arrangement, electrical network is interconnected, each electrical network can not ignore isolated calculate local area electrical network and manage of other networks.Physical change on force device is influenced each other by electrical network connection, mutually restricts, makes electric system present model complexity gradually, the feature of data wide area distribution.On the other hand, the appearance of electricity market also been proposed many nontechnical requirements for the operation of electrical network.Contradiction between the interconnected and management system of power system hierarchical subregion of electrical network determines needs the composition decomposition carried out between multi-control center to calculate.
Meanwhile, the safety of electric system not only depends on operation, the analysis and control technology of electric system, also be limited by the validity and reliability of infosystem, and it is closely related with factors such as planning and design, the market operation, game, supervision, input in science and technology and personnel trainings, strong electric network composition still likely shakes under the strike of a series of successive failure and improper operation, the even power failure of system scope.For electric system, realize safety on line qualitative assessment and Ge Dao defence line adaptive optimization and coordination, the task of building the safety and stability defence framework of space-time cooperative is extremely urgent.Self-examination for power outage large several times enlightens, and the energy management system of a new generation will set up intactly space-time cooperative calculation mechanism: from isolated defence to integrated campaign, based on scheduling and the management of space-time cooperative.This just requires that the multi-control central decomposition of spatially distributed existence is coordinated calculating and in time and can be applied upper concurrent realization.
On the other hand, the multiple electric system Distributed Application such as the estimation of distributed dynamic trend, distributions, static security analysis calculates the ripe gradually and practical application in electric system of research, ability energy management system (Energy Management System, EMS) being coordinated to various Distributed Application parallel computation proposes more requirements.Document [1] (Zhang Baiming, Zhang Haibo. composition decomposition model study [J] between multi-control center. Proceedings of the CSEE, 2006,26 (22): 1-5) from the angle of practical application, give the composition decomposition that China realizes multi-control center and calculate the rule of development that should follow, propose respectively at synchronous iteration, under asynchronous iteration and in real time these three kinds of distributed computing models equivalent, interacted system decomposition coordinates specific implementation form and the algorithm of calculating.Document [2] (Zhang Haibo, Jiang Liangmin, inscription on pottery is big. the design and implimentation [J] of practical distributed dynamic load flow calculation system. and Automation of Electric Systems, 2012,36 (9): 67-71) the distributed management mechanism of registration, activation, startup is introduced, solve the conflict resolution problem that multiple subsystem is initiated to calculate simultaneously, and carry out corresponding practical test.Document [3] (Zhang Haibo, Yi Wenfei. based on the electric system distributed state estimation method [J] of asynchronous iteration mode. Automation of Electric Systems, 2014,38 (9): 125-131) propose a kind of bad data recognition method measuring distributions algorithm for estimating and the consideration outer net measurement information revising thought based on equivalence net equivalence, have studied the practical implementation procedure of state estimation composition decomposition calculating and verify.
Up to now, also lack in prior art and practical design is carried out to multi-control central decomposition coordination computation schema, lack and consider that space-time applying three-dimensional coordinate carries out distributed parallel system administrative mechanism, builds the technical scheme that concurrent computational system is applied to distributed energy management solutions system, what this made electric system Distributed Parallel Computing specifically carries out difficulties.
Summary of the invention
In view of this, the object of the invention is to overcome the multi-control central decomposition existed in prior art and coordinate the system management problems that computation schema is applied to practice, design a kind of electric system Distributed Parallel Computing management method, for the distributed many application of energy management system provide suitable access platform, meet the requirement of following electrical network for energy management system of new generation.
In order to realize this object, the technical scheme that the present invention takes is as follows.
A kind of electric system Distributed Parallel Computing management method, described method comprises:
When one user of A, a subsystem initiates senior computation, described user sends computation requests by its user management object to coordinated management object;
B, coordinated management object assign start-up simulation instruction to other subsystems;
The management subsystem object of C, other subsystems described opens up new calculation procedure in respective subsystem, and creates shadow user in process, coordinates initiation to calculate user and carries out Distributed Calculation;
User management object and the described coordinated management object of D, shadow user connect, and prepare composition decomposition and calculate;
The user management object of E, described user and shadow user sets up primary user's calculating object and from user's calculating object, coordinated management object is set up and coordinated calculating object;
F, coordinated management object are that all calculating objects distribute three-dimensional coordinate, and whole calculating object is mutually located by three-dimensional coordinate and communicated;
G, coordination calculating object complete the composition decomposition between each calculating object, and each calculating object completes iterative computation according to composition decomposition mechanism with subsystem principal and subordinate mechanism;
After H, calculating terminate, primary user's calculating object exports result of calculation;
In I, computing system, any one user initiates any one senior computation, completes a computation process according to above-mentioned steps.
In described method, each calculated examples comprises calculating object by three-dimensional coordinate mark formation data hierarchy, belong to identical or the different user of different sub-systems can start identical or different senior computations simultaneously, mutually do not have an impact, parallel between each user's computation process completes, described three-dimensional coordinate is time, space and applying three-dimensional spatial coordinate, wherein
Time dimension coordinate represents the different user participating in calculating,
Space dimension coordinate represents the different sub-systems that calculating object is subordinate to,
Application dimension coordinate represents different senior computations.
In addition, decomposition and cooperation between multi-control center is expanded to multi-user and many applications by described method, when guaranteeing that the many Distributed Application of energy management system access, multi-user starts, the correctness of each subsystem Distributed Calculation, described senior computation to be connected to the distributed senior application of the various electric system of energy management system, comprises state estimation calculating, Load flow calculation and static security analysis.
In addition, also comprise before described steps A:
A01, arrange the cooperation layer of Distributed Calculation at superior control centre, described cooperation layer possesses coordinated management function and coordinates computing function;
A02, utilize subordinate's All Control Center to serve as subsystems, each subsystem connects multiple user, and subsystem possesses management subsystem function, and user possesses subscriber management function and user's computing function.
Management function and the computing function of described cooperation layer, subsystem and user are realized by object, cooperation layer has coordinated management object and coordinates calculating object, and described subsystem has management subsystem object, and described user has user management object and user's calculating object, wherein
Coordinated management object is for managing whole computing system, and comprise sub-system state and User Status manages, all computation process manages, be in core status, the duration is whole computing system;
Coordinate calculating object to be used for serving as each subsystem communication agency, and complete the border coordination process in concrete computation process, the duration is the duration once calculated;
Management subsystem object is for managing this subsystem state, the establishment of leading subscriber process, the foundation of calculating object, descending process, and the duration is the duration of subsystem;
User management object for managing this User Status, set up communication process between user and cooperation layer, the duration is the duration of user;
User's calculating object has been used for actual computation process, and communicated with coordination calculating object by interface, the duration is the duration once calculated.
Especially, step G comprises further:
Before G1, calculating start, cooperation layer reads interconnection information between each subsystem, and according to subsystem principal and subordinate mechanism partition sub-system, each subsystem only has this subsystem network model, is sent to cooperation layer after network structure is carried out Intranet equivalence;
G2, cooperation layer, after each subsystem Intranet Equivalent Model of acquisition, set up the external network equivalent model of each subsystem in conjunction with interconnection information, and are sent to corresponding subsystem;
G3, subsystem accept External network model, form this subsystem electrical network and calculate topology;
After G4, calculating start, in each iterative process of each subsystem, border coordination information corresponding for this computation is sent to cooperation layer, cooperation layer asks for merging parameter according to border coordination information, and boundary node state amount is sent to corresponding subsystem, each subsystem completes iterative computation in conjunction with boundary information.
By current sub-system former interconnection outer boundary node or border plant stand, utilize dominant eigenvalues or boundary node voltage relative to the Decoupling And Linearization distribution factor of outer mesh element, determine buffer network, and count the detailed modeling of this subsystem Intranet.
Wherein the border coordination information of different computing application exchange is different, and wherein dynamic power flow is calculated as border impedance matrix inverse matrix diagonal element, and state estimation is calculated as information matrix inverse matrix diagonal element.
In addition, the communication mode between described cooperation layer and subsystem and user is:
Establish a communications link between cooperation layer and each subsystem;
Cooperation layer and establishing a communications link between each user and shadow user;
Clog-free interference between the multiple communication connection of cooperation layer;
Communication connection between user with cooperation layer is that asynchronous communication is connected.
Described method also comprises user management step, and described user management step comprises:
User starts writing to each other of rear foundation and cooperation layer;
Record current user state, User Status comprises registration and activates, and characterizes authority that active user has;
After receiving User Status handoff request, await a response to cooperation layer Forward-reques;
Accept and recording user state switching result,
Calculating object is set up after accepting to start computations.
By adopting electric system Distributed Parallel Computing management method of the present invention, by multi-control central decomposition coordination, computation schema is generalized to multi-user, many applications are put into practice, solve the parallel computation problem that following distributed energy management solutions system carries out Distributed Calculation, for the distributed many application of ripe electric system provide suitable access platform, solution is provided to three-dimensional coordinated energy management system of new generation.
Accompanying drawing explanation
Fig. 1 is electric system Distributed Parallel Computing management method high-level schematic functional block diagram in the specific embodiment of the invention.
Fig. 2 is the master subsystem interconnection expansion schematic diagram of electric system Distributed Parallel Computing management system in the specific embodiment of the invention.
Fig. 3 is each Subsystem subscribers corresponding relation figure of electric system Distributed Parallel Computing management system in the specific embodiment of the invention.
Fig. 4 is that the distributed parallel system object of electric system Distributed Parallel Computing management method in the specific embodiment of the invention arranges table.
Fig. 5 is each object logics graph of a relation of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Fig. 6 is the three-dimensional coordinate setting schematic diagram of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Fig. 7 is the data hierarchy effect schematic diagram of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Fig. 8 is the deployment way figure of electric system Distributed Parallel Computing management system in the specific embodiment of the invention.
Fig. 9 is that the subsystem principal and subordinate mechanism of electric system Distributed Parallel Computing management method in the specific embodiment of the invention divides schematic diagram.
Figure 10 is the interconnection spreading result table of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Figure 11 is the parallel computation measure of merit table of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Figure 12 is multi-user's parallel computation table with test results of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Figure 13 is many application parallel computation table with test results of electric system Distributed Parallel Computing management method in the specific embodiment of the invention.
Embodiment
Below in conjunction with accompanying drawing, the present invention is elaborated.
The example embodiment that following discloses are detailed.But concrete structure disclosed herein and function detail are only the objects for describing example embodiment.
But should be appreciated that, the present invention is not limited to disclosed concrete example embodiment, but covers all modifications, equivalent and the alternative that fall within the scope of the disclosure.In the description to whole accompanying drawing, identical Reference numeral represents identical element.
Should be appreciated that, term "and/or" as used in this comprises one or morely relevant lists any of item and all combinations simultaneously.Should be appreciated that in addition, when parts or unit are called as " connection " or " coupling " to another parts or unit, it can be directly connected or coupled to miscellaneous part or unit, or also can there is intermediate member or unit.In addition, other words being used for describing relation between parts or unit should be understood according to identical mode (such as, " between " to " directly ", " adjacent " to " direct neighbor " etc.).
The object of the invention is to coordinate for multi-control central decomposition the system management problems that computation schema is applied to practice, design a kind of building mode and management method of electric system distributed parallel system, for the distributed many application of energy management system provide suitable access platform.
Concrete technical scheme of the present invention is as follows.
1. form electric system distributed parallel system control construction, detailed process is:
Coordinating computation schema according to multi-control central decomposition, in order to ensure the correctness of electric system Distributed Parallel Computing result, when the subsystem participating in calculating is more, being coordinated the computation process of subsystems by a cooperation layer.The cooperation layer of a Distributed Calculation is set at superior control centre; Subordinate's All Control Center serves as subsystems, and each subsystem connects multiple user, can initiate control center to this management subsystem; Between cooperation layer and subsystem in correspondence with.Electric system distributed parallel system building mode as shown in Figure 1.
A cooperation layer builds
In practical power systems management, the different user belonging to different sub-systems may initiate the senior computation of difference for this subsystem simultaneously.Therefore, decomposition and cooperation should be generalized to multi-user and many applications, cooperation layer builds just needs to possess coordinated management and the function coordinating calculating two aspect.
(1) function is coordinated and managed
First, based on the requirement of system management for user right, cooperation layer needs to possess condition managing function, comprises the management for subsystem state and User Status.
The state of subsystem is divided into and adds state and exit status: add and represent that current sub-system normally participates in Distributed Calculation; Exit, represent that current sub-system does not participate in Distributed Calculation, usually occur when current sub-system control center collapses.
Calculate authority from initiation to distinguish, User Status has login state and state of activation: login state represents that user joins computing system, can observe this subsystem current state, but haves no right to initiate senior computation; State of activation then represents that user possesses the authority initiating the senior computation of current sub-system.Distinguish from the senior computation pattern of initiation, User Status has real-time status and research state: the user under real-time status is system user, can initiate real-time status and estimate to calculate; User under research state can observe the real-time result of calculation of current sub-system, but lack of competence initiates real-time status estimation calculating, can only initiate the senior computation of research state.
Cooperation layer needs the state status recording current each subsystem and user.Meanwhile, when user needs to change current user state, all need to send request to cooperation layer, finally determine that User Status changes by cooperation layer.
Secondly, for the computation process of Distributed Calculation complexity, cooperation layer also needs to possess management of computing function.
The mode that decomposition and cooperation realizes Distributed Calculation is: keep on the basis of each subsystem independence accuracy in computations, and carry out border coordination by exchanging boundary information mode between subsystem, each subsystem constantly revises the computation process that iteration completes the whole network.Therefore, the management of computing function of cooperation layer shows as: when a user of a certain subsystem initiates to calculate, cooperation layer controls other subsystem synchronous averaging and calculates, and coordinates initiation computing subsystem to complete a Distributed Calculation.Cooperation layer carries out unified coordination for the Distributed Calculation that each subsystem is initiated and calculates.
(2) computing function is coordinated
Specific to the senior each time computation of system, each subsystem is separate, and completing Distributed Calculation needs to exchange border coordination information by cooperation layer.Therefore cooperation layer coordination computing function shows as:
Before calculating, cooperation layer reads interconnection information between each subsystem, and partition sub-system according to this; Each subsystem only has this subsystem network model, is sent to cooperation layer after network structure is carried out Intranet equivalence; Cooperation layer, after each subsystem Intranet Equivalent Model of acquisition, sets up the external network equivalent model of each subsystem in conjunction with interconnection information, and is sent to corresponding subsystem; Subsystem accepts External network model, forms this subsystem electrical network and calculates topology.
Calculate and start, in each iterative process of each subsystem, border coordination information corresponding for this second advanced computation is sent to cooperation layer; Cooperation layer asks for merging parameter according to border coordination information, and boundary node state amount is sent to corresponding subsystem; Each subsystem completes iteration in conjunction with boundary information.Wherein, it is different that different application calculates the border coordination information exchanged, and if dynamic power flow is border impedance matrix inverse matrix diagonal element, state estimation is information matrix inverse matrix diagonal element etc., determines according to concrete senior computation.Specifically: in senior computation, each subsystem iterative process is asynchronous iteration.
B subsystem builds
Corresponding with cooperation layer function setting, subsystem will possess management function and computing function equally.Structure for subsystem architecture carries out from subsystem and user's two aspects: subsystem completes the management to whole subsystem, and this user of user management also completes concrete senior computation.
(1) principal and subordinate's system subdivision
According in cooperation layer management function for the description of management of computing part, the senior computation of any one is all initiated by subsystem, and other subsystem synchronous averaging calculate and coordinated.In once calculating, the result of calculation of institute of this control center compass of competency electrical network Intranet that what promoter was more concerned about must be, but not the state of whole electrical network and result of calculation.Therefore, electric network model divides according to principal and subordinate's mechanism, and as shown in Figure 2, dividing mode is: the subsystem of initiating to calculate is set as master subsystem, coordinates the subsystem calculated to be set as from subsystem; To pessimistic concurrency control in master subsystem, extrapolated by the physical network model boundary node of reality, set up buffer network to expand its Equivalent Model; When any one subsystem initiates to calculate as master subsystem, cooperation layer all divides current each subsystem according to principal and subordinate's mechanism.Obtain buffer network by boundary node and calculate the mode of the Intranet Equivalent Model into master subsystem, effectively can ensure the correctness of master subsystem result of calculation in a second advanced computation.
(2) management subsystem formation function
Many calculation procedure management.According to decomposition and cooperation, when having a Subsystem subscribers to initiate to calculate in computing system, all subsystem synchronous averaging calculate, and jointly complete Distributed Calculation.When composition decomposition calculating expands to multi-user and many applications, the different user likely occurring belonging to multiple subsystem initiates the situation of calculating simultaneously, and the collision problem of Distributed Calculation just may appear repeatedly in computing system.Meanwhile, for keeping the robustness of computing system, same Client-initiated twice Distributed Calculation result, can not have an impact each other.Therefore, need design parallel computation mechanism, complete Distributed Calculation in the mode that multi-process calculates, concrete building mode is: computing system is using user as the concrete Minimal Realization unit calculated; When master subsystem user initiates to calculate, respectively open up calculation procedure from subsystem correspondence, generation shadow user and cooperation layer establish a communications link, and user and shadow user set up one-to-one relationship, form calculated examples, as shown in Figure 3; Cooperation layer assigns instruction, and the user and the shadow user synchronous averaging that belong to same calculated examples calculate, and starts iterative computation.Wherein, the management function of subsystem is mainly reflected in: accept cooperation layer and open up calculation procedure instruction, set up independent calculation procedure and produce shadow user.
(3) user structure builds
Subscriber management function.There is multiple User Status in user, can carry out the switching of User Status, needs to manage User Status.Concrete way to manage is: user starts writing to each other of rear foundation and cooperation layer; Record current user state; After accepting User Status handoff request, await a response to cooperation layer Forward-reques; Accept and recording user state switching result.
User's computing function.User, as the Minimal Realization unit of Distributed Calculation, completes the concrete computation process in actual senior computation.No matter be user or shadow user, all should have concrete senior computation function.
2. objectification realizes system management architecture function, and detailed process is:
Consider the practical of system management architecture, objectification realization is carried out to it.
A functional object
Based on the design of said system control construction, abstract functions: cooperation layer possesses coordinated management function and coordinates computing function; Subsystem possesses management subsystem function; User possesses subscriber management function and user's computing function.Therefore, design five class objects and realize above-mentioned five part of module functions respectively: coordinated management object, coordination calculating object, management subsystem object, user management object and user's calculating object.Concrete five class object functions as shown in Figure 4.
Different object function is different, and status in computing systems and life cycle are not identical yet: management object manages system state and computation process, and durations is longer; Wherein, coordinated management object occupy core status, and manage whole computing system, its durations runs through whole computing system; Management subsystem Object Management group book system-computed process, this User Status of user management Object Management group, two management object matching coordinative management object completion system management, durations is identical with subsystem and user's survival cycle; Calculating object mainly completes once concrete computation process, and durations is identical with a computation process.
Coordinated management object and coordination calculating object are arranged in cooperation layer, management subsystem object placement is in subsystem side, user management object and user's calculating object are arranged in user side, and five kinds of objects cooperatively interact, and jointly complete the management for system and computation process.
B three-dimensional coordinate marks
From data plane, when system is run according to above-mentioned parallel computation mode, the data communication between each calculated examples actual user and shadow user will have been come by cooperation layer, just needs to distinguish the data stream belonging to different calculated examples.Owing to relate to user, subsystem and application three dimensions for the design of distributed parallel system administrative mechanism, the calculating object that the present invention utilizes time, space and applying three-dimensional coordinate solid mark different, realizes the layering isolation belonged between the data stream of different calculated examples.By three-dimensional coordinate notation methods, each calculating object in computing system can have unique three-dimensional coordinate, calculates between each calculating object of data in same calculated examples and flows, realize data hierarchy.
C sets up communication interface
Cooperation layer needs to communicate with between subsystem and user, and connection setup mode is: establish a communications link between cooperation layer and each subsystem; Cooperation layer and establishing a communications link between each user and shadow user; Can not there is occlusion interfere in the multiple communication connection of cooperation layer, communicate to connect as multi-threading correspondence mode; Distributed Calculation iterative process is asynchronous iteration, and the communication that user and cooperation layer are set up possesses asynchronous communication function.
Therefore, electric system Distributed Parallel Computing management method of the present invention comprises step:
When one user of A, a subsystem initiates to calculate, described user sends computation requests by its user management object to coordinated management object;
B, coordinated management object assign start-up simulation instruction to other subsystems;
The management subsystem object of C, other subsystems described opens up new calculation procedure in respective subsystem, and creates shadow user in process;
User management object and the described coordinated management object of D, shadow user connect, and prepare composition decomposition and calculate;
The user management object of E, described user and shadow user sets up primary user's calculating object and from user's calculating object, coordinated management object is set up and coordinated calculating object;
F, coordinated management object are that all calculating objects distribute three-dimensional coordinate, and whole calculating object is mutually located by three-dimensional coordinate and communicated;
G, coordination calculating object complete the composition decomposition between each calculating object, and each calculating object completes iterative computation according to composition decomposition mechanism with subsystem principal and subordinate mechanism;
After H, calculating terminate, primary user's calculating object exports result of calculation;
In I, computing system, any one user initiates any one senior computation, completes a computation process according to above-mentioned steps.
As shown in Figure 6, the three-dimensional coordinate of described calculating object is time, space and applying three-dimensional spatial coordinate, wherein,
Time coordinate represents the user participating in calculating,
Volume coordinate represents the subsystem that calculating object is subordinate to,
Application coordinate represents different senior computing applications.
From data plane, when system is run according to above-mentioned parallel computation mode, the data communication in each calculated examples between actual user and shadow user will have been come by cooperation layer, therefore just needs to distinguish the data stream belonging to different calculated examples.Owing to having related to user, subsystem and application three dimensions for the Management Mechanism Design of Distributed Parallel Computing method, the calculating object that the present invention utilizes time, space and applying three-dimensional coordinate solid mark different, realizes the layering isolation belonged between the data stream of different calculated examples.
Simple row are given one example: suppose that regional power grid has three control centers, electrical network are divided into three subsystems according to distributed computing; Distinguishing label for computing application is: state estimation is calculated as 1, Load flow calculation is 2, static security analysis is calculated as 3 etc.Then under real-time state, start real-time status by the system user 1 of subsystem 1 to estimate to calculate, main master subsystem actual user three-dimensional coordinate should be (1, + 1,1), correspondence is (1 from subsystem shadow user coordinate,-x, 1), cooperation layer calculated examples coordinate is (1,0,1), x is Sub-System Number belonging to calculating object.
Correspondingly, Fig. 7 is the data hierarchy effect schematic diagram of electric system Distributed Parallel Computing management method in the specific embodiment of the invention, by the mode that three-dimensional coordinate marks, each calculating object in computing system can have unique three-dimensional coordinate, located by three-dimensional coordinate, calculate between each calculating object of data in same calculated examples and flow, realize data hierarchy.
Especially, described computing application to be connected to the various electric system Distributed Application of energy management system, comprises state estimation calculating, Load flow calculation and static security analysis.
In addition, described steps A takes a step forward and comprises:
A01, arrange the cooperation layer of Distributed Calculation at superior control centre, described cooperation layer possesses coordinated management function and coordinates computing function;
A02, utilize subordinate's All Control Center to serve as subsystems, each subsystem connects multiple user, and subsystem possesses management subsystem function, and user possesses subscriber management function and user's computing function.
As shown in Figure 5, management function and the computing function of described cooperation layer, subsystem and user are realized by object, have coordinated management object and coordinate calculating object, described subsystem has management subsystem object, described user has user management object and user's calculating object, wherein
Coordinated management object is for managing whole computing system, and comprise sub-system state and User Status manages, all computation process manages, be in core status, the duration is whole computing system;
Coordinate calculating object to be used for serving as each subsystem communication agency, and complete the border coordination process in concrete computation process, the duration is the duration once calculated;
Management subsystem object is for managing this subsystem state, the establishment of leading subscriber process, the foundation of calculating object, descending process, and the duration is the duration of subsystem;
User management object for managing this User Status, set up communication process between user and cooperation layer, the duration is the duration of user;
User's calculating object has been used for actual computation process, and communicated with coordination calculating object by interface, the duration is the duration once calculated.
Wherein, by current sub-system former interconnection outer boundary node or border plant stand, utilize dominant eigenvalues or boundary node voltage relative to the Decoupling And Linearization distribution factor of outer mesh element, determine buffer network, and count the detailed modeling of this subsystem Intranet.
Particularly, step G comprises further:
Before G1, calculating start, cooperation layer reads interconnection information between each subsystem, and according to subsystem principal and subordinate mechanism partition sub-system, each subsystem only has this subsystem network model, is sent to cooperation layer after network structure is carried out Intranet equivalence;
G2, cooperation layer, after each subsystem Intranet Equivalent Model of acquisition, set up the external network equivalent model of each subsystem in conjunction with interconnection information, and are sent to corresponding subsystem;
G3, subsystem accept External network model, form this subsystem electrical network and calculate topology;
After G4, calculating start, in each iterative process of each subsystem, border coordination information corresponding for this second advanced computation is sent to cooperation layer, cooperation layer asks for merging parameter according to border coordination information, and boundary node state amount is sent to corresponding subsystem, each subsystem completes iterative computation in conjunction with boundary information.
Wherein, the border coordination information that different computing application exchanges is different, and wherein dynamic power flow is calculated as border impedance matrix inverse matrix diagonal element, and state estimation is calculated as information matrix inverse matrix diagonal element.
Communication mode between described cooperation layer and subsystem and user is:
Establish a communications link between cooperation layer and each subsystem;
Cooperation layer and establishing a communications link between each user and shadow user;
Clog-free interference between the multiple communication connection of cooperation layer;
Communication connection between user with cooperation layer is that asynchronous communication is connected.
Described method also comprises user management step, and described user management step comprises:
User starts writing to each other of rear foundation and cooperation layer;
Record current user state;
After receiving User Status handoff request, await a response to cooperation layer Forward-reques;
Accept and recording user state switching result;
Calculating object is set up after accepting to start computations.
Below by way of more concrete sample calculation, technique effect of the present invention is described, those skilled in that art all know, the topological structure in example, data are only illustrative.
Sample calculation analysis is carried out with IEEE118 node system data instance.Programmed by C++, be aided with the mode that CORBA technology solves communication process, access distributed dynamic trend and distributions are estimated to verify distributed parallel system parallel computation effect.
Embodiment is: calculated by concrete dynamic power flow and be packaged into storehouse for calling with state estimation computing application program; And design the exchange process that corresponding communication interface completes border coordination information; Based on object logics relation in Fig. 5 and mating reaction mode, five class objects in distributed parallel system are carried out computing machine realization with C++ programming; Access dynamic power flow and state estimation are applied, and test with IEEE118 node data.
Fig. 8 is the deployment way figure of electric system Distributed Parallel Computing management system in the specific embodiment of the invention.In test, the computing system built under distributed parallel system administrative mechanism is deployed as 4 PCs, wherein a PC runs cooperation layer program, serves as control center's server of cooperation layer in simulation practical power systems; Three sub-System Control Centers simulated by other three PCs.Each subsystem PC runs a subsystem server process, as management subsystem object; The simultaneously several consumer process of each operation on each subsystem PC, is equivalent to now each subsystem has user to access, can sends computation requests to cooperation layer.
Fig. 9 is that the subsystem principal and subordinate mechanism of electric system Distributed Parallel Computing management method in the specific embodiment of the invention divides schematic diagram.Distributed Parallel Computing method is tested on IEEE118 nodal analysis method, will divide the data of IEEE118 nodal analysis method.For subsystem principal and subordinate mechanism, partition sub-system carries out the division of subsystem according to master subsystem interconnection extrapolation rule such as Fig. 9.Concrete interconnection situation is as shown in Figure 10 table.
For programming realization in the manner described above and the distributed parallel system disposed, carry out the test of parallel computation effect according to the scheme designed in Figure 11 table.In order to have good test for parallel computation, making the user 4 of subsystem 1 be in uninterrupted cycle calculations state in test, deliberately manufacturing possible conflicting when initiating to calculate with subsystem 2 user 5, test effect is as follows.
Multi-user's parallel computation effect is as shown in Figure 12 table: contrast user 4 is known with the result of calculation of user 5: when each user calculates, three subsystems have participated in once calculating, the computation process of each user be according to multiple subsystem between decomposition and cooperation carry out; User 4 is different from user 5 result of calculation iterations, and result of calculation comprises nodes difference, carried out the extrapolation of interconnection, and the result of calculation of master subsystem user is correct, demonstrates the realization in computing systems of slave system according to principal and subordinate's mechanism; Calculate respectively based under different model, result is all correct, and guarantees that calculating and sending is carried out can manufacture to calculate under conflict situations, demonstrates the validity of electric system Distributed Parallel Computing management method of the present invention.
Many application parallel computation test effect is as shown in Figure 13 table: user 4 starting state of subsystem 1 is estimated to calculate, the user 5 dynamic power flow calculating simultaneously of subsystem 2.State estimation calculating calculates with dynamic power flow and restrains respectively, occurs without waiting in line situation.Illustrate that distributed parallel system administrative mechanism can realize the parallel computation of many application, there is not impact each other, test result and multi-user test similar.
Therefore, electric system Distributed Parallel Computing management method of the present invention, by multi-control central decomposition coordination, computation schema is generalized to multi-user, many applications are put into practice, solve the parallel computation problem that following distributed energy management solutions system carries out Distributed Calculation, for the distributed many application of ripe electric system provide suitable access platform, solution is provided to three-dimensional coordinated energy management system of new generation.
It should be noted that; above-mentioned embodiment is only the present invention's preferably embodiment; the restriction to distance protection scope of the present invention can not be understood as, not depart under concept thereof of the present invention, with modification, distance protection scope of the present invention is all belonged to any subtle change that the present invention does.

Claims (10)

1. an electric system Distributed Parallel Computing management method, described method comprises:
When one user of A, a subsystem initiates senior computation, described user sends computation requests by its user management object to coordinated management object;
B, coordinated management object assign start-up simulation instruction to other subsystems;
The management subsystem object of C, other subsystems described opens up new calculation procedure in respective subsystem, and creates shadow user in process, coordinates initiation to calculate user and carries out Distributed Calculation;
User management object and the described coordinated management object of D, shadow user connect, and prepare composition decomposition and calculate;
The user management object of E, described user and shadow user sets up primary user's calculating object and from user's calculating object, coordinated management object is set up and coordinated calculating object;
F, coordinated management object are that all calculating objects distribute three-dimensional coordinate, and whole calculating object is mutually located by three-dimensional coordinate and communicated;
G, coordination calculating object complete the composition decomposition between each calculating object, and each calculating object completes iterative computation according to composition decomposition mechanism with subsystem principal and subordinate mechanism;
After H, calculating terminate, primary user's calculating object exports result of calculation;
In I, computing system, any one user's any instant initiates any one senior computation, completes a computation process according to above-mentioned steps A-step H.
2. according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, in described method, each calculated examples comprises calculating object by three-dimensional coordinate mark formation data hierarchy, belong to identical or the different user of different sub-systems can start identical or different senior computations simultaneously, mutually do not have an impact, parallel between each user's computation process completes, and described three-dimensional coordinate is time, space and applying three-dimensional spatial coordinate, wherein
Time dimension coordinate represents the different user participating in calculating,
Space dimension coordinate represents the different sub-systems that calculating object is subordinate to,
Application dimension coordinate represents different senior computations.
3. according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, decomposition and cooperation between multi-control center is expanded to multi-user and many applications by described method, when guaranteeing that the many Distributed Application of energy management system access, multi-user starts, the correctness of each subsystem Distributed Calculation, described senior computation to be connected to the distributed senior application of the various electric system of energy management system, comprises state estimation calculating, Load flow calculation and static security analysis.
4., according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, also comprise before described steps A:
A01, arrange the cooperation layer of Distributed Calculation at superior control centre, described cooperation layer possesses coordinated management function and coordinates computing function;
A02, utilize subordinate's All Control Center to serve as subsystems, each subsystem connects multiple user, and subsystem possesses management subsystem function, and user possesses subscriber management function and user's computing function.
5. according to the electric system Distributed Parallel Computing management method described in claim 4, it is characterized in that, management function and the computing function of described cooperation layer, subsystem and user are realized by object, cooperation layer has coordinated management object and coordinates calculating object, described subsystem has management subsystem object, described user has user management object and user's calculating object, wherein
Coordinated management object is for managing whole computing system, and comprise sub-system state and User Status manages, all computation process manages, be in core status, the duration is whole computing system;
Coordinate calculating object to be used for serving as each subsystem communication agency, and complete the border coordination process in concrete computation process, the duration is the duration once calculated;
Management subsystem object is for managing this subsystem state, the establishment of leading subscriber process, the foundation of calculating object, descending process, and the duration is the duration of subsystem;
User management object for managing this User Status, set up communication process between user and cooperation layer, the duration is the duration of user;
User's calculating object has been used for actual computation process, and communicated with coordination calculating object by interface, the duration is the duration once calculated.
6., according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, step G comprises further:
Before G1, calculating start, cooperation layer reads interconnection information between each subsystem, and according to subsystem principal and subordinate mechanism partition sub-system, each subsystem only has this subsystem network model, is sent to cooperation layer after network structure is carried out Intranet equivalence;
G2, cooperation layer, after each subsystem Intranet Equivalent Model of acquisition, set up the external network equivalent model of each subsystem in conjunction with interconnection information, and are sent to corresponding subsystem;
G3, subsystem accept External network model, form book system-computed topology;
After G4, calculating start, in each iterative process of each subsystem, border coordination information corresponding for this computation is sent to cooperation layer, cooperation layer asks for merging parameter according to border coordination information, and boundary node state amount is sent to corresponding subsystem, each subsystem completes iterative computation in conjunction with boundary information.
7. according to the electric system Distributed Parallel Computing management method described in claim 6, it is characterized in that, by current sub-system former interconnection outer boundary node or border plant stand, utilize dominant eigenvalues or boundary node voltage relative to the Decoupling And Linearization distribution factor of outer mesh element, determine buffer network, and count the detailed modeling of this subsystem Intranet.
8. according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, the border coordination information that different computing application exchanges is different, and wherein dynamic power flow is calculated as border impedance matrix inverse matrix diagonal element, and state estimation is calculated as information matrix inverse matrix diagonal element.
9. according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, the communication mode between described cooperation layer and subsystem and user is:
Establish a communications link between cooperation layer and each subsystem;
Cooperation layer and establishing a communications link between each user and shadow user;
Clog-free interference between the multiple communication connection of cooperation layer;
Communication connection between user with cooperation layer is that asynchronous communication is connected.
10. according to the electric system Distributed Parallel Computing management method described in claim 1, it is characterized in that, described method also comprises user management step, and described user management step comprises:
User starts writing to each other of rear foundation and cooperation layer;
Record current user state, User Status comprises registration and activates, and characterizes authority that active user has;
After receiving User Status handoff request, await a response to cooperation layer Forward-reques;
Accept and recording user state switching result,
Calculating object is set up after accepting to start computations.
CN201510239699.1A 2015-05-12 2015-05-12 Electric system Distributed Parallel Computing management method Expired - Fee Related CN104809593B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510239699.1A CN104809593B (en) 2015-05-12 2015-05-12 Electric system Distributed Parallel Computing management method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510239699.1A CN104809593B (en) 2015-05-12 2015-05-12 Electric system Distributed Parallel Computing management method

Publications (2)

Publication Number Publication Date
CN104809593A true CN104809593A (en) 2015-07-29
CN104809593B CN104809593B (en) 2019-01-15

Family

ID=53694401

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510239699.1A Expired - Fee Related CN104809593B (en) 2015-05-12 2015-05-12 Electric system Distributed Parallel Computing management method

Country Status (1)

Country Link
CN (1) CN104809593B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106022655A (en) * 2016-06-20 2016-10-12 华北电力大学 Distributed energy management system integrated platform construction method
CN109948185A (en) * 2019-02-01 2019-06-28 全球能源互联网研究院有限公司 A kind of decoupling emulation mode of electric system
CN111967844A (en) * 2020-08-07 2020-11-20 江苏海岸线互联网科技有限公司 Cross-organization project tracking and collaboration management system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101521596A (en) * 2009-04-03 2009-09-02 南京邮电大学 Management structure for distributed dynamic self-organizing network
CN102185311A (en) * 2011-04-29 2011-09-14 华北电力大学 Method for constructing distributed dynamic power flow computing system for energy management of electric power system
US20110270459A1 (en) * 2010-04-28 2011-11-03 Masahiko Murai Power consumption management system, and power consumption manager, power consumption management method, power supply manager, and power supply management method, which are used for the power consumption management system
CN103955801A (en) * 2014-05-15 2014-07-30 华北电力大学 Electric power system distributed parallel computing management method based on time-space dimension

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101521596A (en) * 2009-04-03 2009-09-02 南京邮电大学 Management structure for distributed dynamic self-organizing network
US20110270459A1 (en) * 2010-04-28 2011-11-03 Masahiko Murai Power consumption management system, and power consumption manager, power consumption management method, power supply manager, and power supply management method, which are used for the power consumption management system
CN102185311A (en) * 2011-04-29 2011-09-14 华北电力大学 Method for constructing distributed dynamic power flow computing system for energy management of electric power system
CN103955801A (en) * 2014-05-15 2014-07-30 华北电力大学 Electric power system distributed parallel computing management method based on time-space dimension

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
张海波等: "实用化分布式动态潮流计算系统的设计与实现", 《电力系统自动化》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106022655A (en) * 2016-06-20 2016-10-12 华北电力大学 Distributed energy management system integrated platform construction method
CN109948185A (en) * 2019-02-01 2019-06-28 全球能源互联网研究院有限公司 A kind of decoupling emulation mode of electric system
CN111967844A (en) * 2020-08-07 2020-11-20 江苏海岸线互联网科技有限公司 Cross-organization project tracking and collaboration management system

Also Published As

Publication number Publication date
CN104809593B (en) 2019-01-15

Similar Documents

Publication Publication Date Title
WO2023246568A1 (en) Digital twin architecture, model evaluation system, and model running method
Ciraci et al. FNCS: A framework for power system and communication networks co-simulation
Lin et al. Power system and communication network co-simulation for smart grid applications
CN102209074B (en) All-digital dynamic simulation system of power system
Min et al. Total transfer capability computation for multi-area power systems
Cao et al. Real‐time cyber− physical system co‐simulation testbed for microgrids control
CN102354981B (en) Distributed computation based voltage stability assessment method of sub-networks in interconnected power network
CN101527455B (en) Interconnected electric network distributed current calculating method on the basis of alternation and iteration of current module
CN106154865A (en) Vehicle-mounted bus simulation model self-generation method and system
CN109583002A (en) Aviation electronics integral design method
CN113055232B (en) Network configuration deployment method, device and equipment
Wang et al. Interfacing issues in multiagent simulation for smart grid applications
CN104809593A (en) Distributed type parallel computation management method of power system
CN103955801B (en) Electric power system distributed parallel computing management method based on time-space dimension
CN110299762A (en) Active distribution network Robust filter method based on PMU near-realtime data
CN105843991B (en) Stage HLA simulated members combined system and method
Zhang et al. A distributed multi-control-center dynamic power flow algorithm based on asynchronous iteration scheme
CN109474023A (en) Intelligent distribution network section real time updating method, system, storage medium and terminal
CN108873734A (en) A kind of power distribution network information physical union simulation platform and method
CN101996282B (en) Multi-mode data transformation service (DTS) interconnection method based on power flow matching and computation model splicing
Ilić et al. Scalable electric power system simulator
CN103763142B (en) Multimode isomery DTN simulation verification system and method
CN109344439A (en) A kind of modeling of building block formula and its simulative construction method based on BIM
CN102360393B (en) Method for sharing models between scheduling automatic systems
Sztipanovits et al. Model-based integration technology for next generation electric grid simulations

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
EXSB Decision made by sipo to initiate substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20190115