JP2006246692A - Method and program for preparing system recovery procedure - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and program for preparing system recovery procedure that solves the problem, wherein it takes time until recovery, because the differences of switch states from the current state are reflected in the procedure, when preparing a system recovery procedure after power failure, based on information on equipment and components that constitute an electric power system. <P>SOLUTION: A system recovery procedure is prepared by handling operation objectives (objective operations) applied to electric power equipment and components comprising an electric power system. Also, the procedure includes extracting operation sets that satisfy local conditions, by inputting the system immediately after the occurrence of power failure, searching a feasible objective operations set, and creating operation candidates for the next system configuration, when a feasible objective operations set exists in any system section. Evaluation uses a perspective evaluation function, when the recovery operation is finished, when the feasible objective operations set is completed to obtain higher evaluations. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method for creating a system restoration operation procedure for restoring a system from a power system accident and a program therefor.

When an accident occurs in the power system and the circuit breaker trips due to relay operation, when the cause of the accident is removed and power is supplied again to the outage area, a recovery procedure is automatically created, and this procedure is automatically performed. Recovery has been done. As such an automatic recovery method, for example, the one disclosed in Patent Document 1 is known. Patent Document 1 discloses that both ends of an extracted switch are extracted for the purpose of improving the complexity of an operation procedure associated with extracting a switch unnecessary for an operation for restoring a power system at the time of creating the procedure. It is determined whether the equipment can be restored on the determined system, and means for proceeding to the next procedure only when restoration is possible is provided.
Japanese Patent Publication No. 7-20348

  A conventional method for generating a system restoration procedure including Patent Document 1 is a method of performing optimization calculation in a cross section after restoration of a power outage load and reflecting a difference in a switch state from the current state in the procedure. For this reason, since the procedure is created in two steps of creating a procedure after the final system is determined, there is a problem in the optimization of the operation process and a problem that takes time.

  An object of the present invention is to provide a method for creating a recovery procedure that maintains the overall optimum and a program therefor.

The first of the present invention is to create a system restoration procedure after a power failure based on the equipment information constituting the power system,
The system restoration procedure is created by handling the target operation performed on the power equipment constituting the power system.
A second aspect of the present invention is characterized in that the system configuration state of the power system handled in the target operation is a pre-operation state, an in-operation state, and an after-operation state.

  According to the third aspect of the present invention, an executable target operation set is extracted by inputting a system immediately after the occurrence of a power failure, and can be further executed for each system section obtained by executing each element of the executable target operation set. When the target operation set exists, an executable objective operation set for the system cross section is generated, and the next system cross section is formed from the system cross section by the executable objective operation, until there is no executable target operation set for any system cross section. By repeating this, it is determined that the restoration operation has ended when there is no set of executable target operations on the system section, and the system section that has determined the end of recovery is evaluated using the global evaluation function to reach the system section with the best evaluation. This is characterized in that the order of target operations is a recovery procedure to be obtained.

  A fourth aspect of the present invention is characterized in that the executable objective operation set satisfies both a system constraint condition and a local condition.

  According to a fifth aspect of the present invention, after the executable objective operation set is extracted, the elements of the executable objective operation set are narrowed down by the second global evaluation function based on the system operation, and the next system section is shifted. It is.

  A sixth aspect of the present invention is characterized in that evaluation points based on the global evaluation function are assigned to all system cross sections that appear in the search process.

  In the seventh aspect of the present invention, a system section formed by an executable objective operation selected by a second global evaluation function based on a system operation is extracted by inputting a system immediately after the occurrence of a power failure and extracting an executable objective operation set. When a search is performed with priority selection and the end of the recovery operation is determined in the generated system cross section, the process leading to this is regarded as the first optimal executable recovery operation, and optimal execution is performed for each non-selected executable target operation. Perform a search for only those that have obtained a better evaluation score by comparing the feasible restoration operation with the global evaluation function, and when the feasible restoration operation is found, the next optimum feasible restoration operation is performed and finally remains In other words, the recovery procedure is determined by seeking the optimum feasible recovery operation.

  According to an eighth aspect of the present invention, when searching for the feasible recovery operation, a finite search specified in advance is performed from an initial cross section of a given system, and each feasible recovery operation system is evaluated with a global function. It is characterized by extracting multiple possible recovery operation system candidates from superior to specified order, and branching by excluding unselected executable recovery operations from the operation candidates in the operation process connected to the extracted system Is.

  According to a ninth aspect of the present invention, the outage evaluation value obtained by accumulating the value obtained by multiplying the evaluation value for the pruning by the individual importance factor by the individual importance factor and integrating the elapsed time of the recovery operation. It is characterized by having an evaluation value including.

  According to a tenth aspect of the present invention, the branching by the evaluation value is performed when the evaluation value in the time section exceeding the determination start time exceeds the evaluation value of the feasible solution in the same time section. It is what.

  The eleventh aspect of the present invention is characterized in that the branching based on the evaluation value is a branching when the evaluation value of the peer system state searched in the past is exceeded.

  According to a twelfth aspect of the present invention, the branching by the peer is recognized as a completely peer system state when the system configuration / power generation / load state in the power system and the operation execution situation of the shift staff are equal, which is worse than the past evaluation value. It is characterized by the fact that it is sometimes cut into branches.

  According to the thirteenth aspect of the present invention, the branching by the peer is recognized as an incomplete peer state when the system configuration, power generation, and load state in the power system are the same and only the operating state of the shift operator is different. It is characterized in that a branch is cut when the state continues for a specified number of times.

  According to the fourteenth aspect of the present invention, when one operation performed on a system section is selected, a designated finite stage is searched by an executable target operation, and each system section is evaluated by a global evaluation function. The first target operation connected to a good system section is determined as the selected target operation, and the recovery procedure is determined by repeating this when selecting the next operation to be performed.

  According to the fifteenth aspect of the present invention, the system recognition for the purpose operation includes recognition of a partial system divided by a transformer constituting an electric power system and an open switch, recognition of a power supply terminal, recognition of an overload system, or It is characterized by any one of recognition of an operation failure system.

According to the sixteenth aspect of the present invention, in order to create a system restoration procedure after a power failure based on the equipment information constituting the power system,
A system that inputs a system immediately after the occurrence of a power failure and extracts an executable objective operation set that is a set of objective operations that satisfy local conditions and system constraint conditions, and extracts the next system section by the executable objective operation and performs the next execution Causing a computer to execute means for extracting a set of possible objective operations, means for repeating the same, means for calculating an evaluation value using a global evaluation function, and means for selecting a high evaluation value by judging the calculation result Is.

  The seventeenth aspect of the present invention is a means for narrowing down the elements of the executable objective operation set by the second global evaluation function based on the system operation after the extraction by the means for extracting the executable objective operation set, and the next after the narrowing by this means The system section is extracted or the evaluation value calculation means is executed by a computer.

  According to an eighteenth aspect of the present invention, after the executable objective operation set is extracted, means for preferentially selecting an executable objective operation selected by a second global evaluation function based on system operation, and extracting the next system section Alternatively, the computer is caused to execute evaluation value calculation means, means for selecting a branch having a good evaluation value, and means for generating a system cross section subsequent to only the selected branch.

  According to a nineteenth aspect of the present invention, means for generating a finite-stage restoration tree designated by each means, means for selecting a branch having a good evaluation value by the evaluation value calculating means, and this is repeated at the time of the next operation selection. The means is executed by a computer.

  As described above, according to the present invention, it is possible to obtain an optimal procedure that can easily generate a tree with few branches until recovery in the operation process by using the target operation. Furthermore, this shortens the search time and responds flexibly to changes in the basic system, and the operation is selected by a logical structure that is easy for the user to recognize, thereby facilitating understanding of the operation content.

Prior to specific description of the present invention, the theory will be described.
In the present invention, the system restoration operation is handled in “target operation units”, and “a recovery tree is constructed in the process of combining the target operations, and an optimum restoration procedure is obtained by an evaluation function”.
(1) About the target operation In the conventional method for generating the recovery procedure, the operation is a change in the open / close state of a specific switch. On the other hand, the “target operation” in the present invention deals only with the purpose of the operation performed on the power equipment constituting the power system. For example, it is assumed that there is a bus during power outage as shown in FIG. In the figure, G is a generator, A and B are busbars, CB1 and CB2 are circuit breakers, CB1 is open, and bus A and B are out of power. As shown in FIG. 11B, the power receiving operation of the bus is performed by inserting the circuit breaker CB1 to supply power to the bus A and B, and by opening the circuit breaker CB2 as shown in FIG. 11C. At least two methods are conceivable: a method of switching and inputting. Each method differs in the content of the operation specifically performed. However, when this is considered as the target operation, both are combined into one target operation of “use busbar”. In other words, the target operation is a clear operation only for the “purpose” for performing the operation, and the specific operation content for realizing the purpose is not specified. When the target operation is actually executed, what operation is executed depends on the individual equipment.
(2) Types of target operation If the system operation performed in the power failure recovery process is divided into the minimum units for each purpose, the equipment is divided into the following eight types.
Use operation: Operation in which a facility that is stopped is charged by executing this operation.
Combined operation: An operation in which both ends of the open circuit breaker are of the same system and are turned on.
Linking operation: An operation in which both ends of the open circuit breaker are of different systems and are turned on. Stop operation: Operation to stop the equipment. Of course, this includes the operation of stopping the facility being charged in a power failure state, but whether or not the facility is stopped is irrelevant to the prior power failure state. For example, when a wide range of power fails due to an accident on the power source side, the power switch may be out of power while the facility switch is in use. If the equipment is charged while it is in use, there may be a problem in operation. Therefore, even if there is a power outage, the operation may be stopped.
Release operation: An operation to release the circuit breaker in a facility where a power failure occurs while the circuit breaker is turned on.
Operation to unlock the combined use: Operation to release the circuit breaker in the same system. There will be no power outage after the release.
Loop switching operation: An operation to switch without stopping the current that flows through the circuit breaker, such as when switching the connected bus from A to B, or when switching the power receiving point at the two-line substation with constant standby power reception. .
Power failure switching operation: An operation to stop and switch the current flowing through the circuit breaker when switching the connected bus from, for example, A to B, or when switching the power receiving point at the two-line substation where standby power is always received.
That is, there are eight types of target operations necessary for generating a recovery operation, and the system recovery is possible by combining these eight operations. In addition, in order to improve operation efficiency, a target operation may be performed on a certain number of facilities such as a substation.
(3) Regarding local conditions and self-ordering of target operations All target operations have execution conditions for executing the operations (hereinafter referred to as local conditions). It is different from the constraints on the system (overload, voltage abnormality, etc.). For example, when it is going to perform a use cycle to the power transmission line which has stopped, use operation cannot be performed if there is no voltage in the bus line used as a power source. Whether or not there is a power source is a problem before considering whether an overload occurs in the system after power transmission. Thus, the local condition is fundamentally different from the problem of system constraints. In addition, since the operation has local conditions, in the above example, the operation flow of using the busbar → using the power transmission line is inevitably determined. This means that if the change in the system configuration is handled by the target operation, the number of system configurations that must be considered can be greatly reduced. Hereinafter, the property of the system operation that the order is arbitrarily determined is referred to as self-ordering. The target operation that can be executed is a target operation that satisfies both the local condition and the system constraint condition. Because system constraints are added. The number of system configurations that must be considered in the recovery process is further reduced.
(4) The states that the target operation can take are two states if only the switch is used, but the target operation has three states before operation, during operation, and after operation.
Pre-operation state: This is a pre-operation state. The system state satisfies the local condition of the target operation to be executed.
In-operation state: Usually, it takes time to change the switch state of the power system. That is, the in-operation state is in the middle of changing the state of the switch included in the operation. In the operating state, the system configuration has system configuration cross sections that differ by the number of switch status changes included in the operation.
Post-operation state: A state in which the target operation has been completed, and when there is a next target operation, the system configuration is a pre-operation state of the next operation.
When trying to process operations continuously, the previous operation needs to be in a post-operation state (self-ordering). Also, when processing operations in parallel, whether or not the state of the parallel operation is in operation can be one of important local conditions.
(5) About the general condition of the target operation The general condition of the target operation is a condition that embodies the idea of what the operator uses as the evaluation axis, and determines the recovery strategy for power failure. is there. For example, when there are multiple target operations that can be operated in a self-order while satisfying the two conditions of local conditions and system constraints in a section, they are generated at the stage of selection or selection. In addition, for the question of which one to select from among a plurality of recovery operations, conditions such as “minimum power outage” or “priority for current recovery of power outage load” are shown. A numerical expression of this condition as an evaluation function is called a global evaluation function. That is, the target operation itself has a meaning but has no will, so a part for determining the will is required.

FIG. 2 is a conceptual diagram of a power system for explaining the present invention.
In a certain power system, Ψ is a set of system configurations s composed of 2 ^x elements, where ψ is defined as a state space in which a system configuration that can be taken according to the state of a switch installed in the system is defined. Here, x is the number of included switches. Further, in this power system, P ₀ is an entire set of system operations that are the basis of combinations prepared in advance.
In the i-th examination section for searching for an i-th operation, Si⊆Ψ is a set of possible system configurations, and its element is sij.
Si = {si1, si2, ... sij ...}
Further, a set of system operations that satisfy the local condition in the system section sij is Pij⊆P ₀ ,
Pij = {pij1, pij2,... Pijk.
And Since the system operation pijk is a function for making a transition from the system configuration sij to si + 1, j, the following is described as follows.
pijk: sij → si + 1, j is si + 1, j = pijk · sij
pij · sij = {pij1 · sij, pij2 · sij, pij3 · sij ...}
Further, it is assumed that the constraint condition is satisfied when the system constraint conditional expression C is C (s)> 0.

FIG. 1 is an algorithm showing a first embodiment of the present invention, and shows a system restoration method by a combination of procedures. Reference numeral 1 denotes a system cross-section input means, which performs input processing in a computer as a system configuration set of a given system as si. Here, S1 = {sft}, where sft is the initial system state including the power failure load. Reference numeral 2 denotes system analysis means, which executes processing for extracting an operation set Pij that satisfies a local condition in sijεSi. Reference numeral 3 denotes an executable objective operation extracting means, which obtains an executable objective operation set Qij. here,
Qij = {pijk | pijkεPij, C (pijk · Sij)> 0}
Reference numeral 4 denotes operation candidate determination means. When the executable target operation set Qij in any system cross section Sij is Qij ≠ φ, that is, when there is an operation candidate, the next system configuration generation means 5 A system configuration set Si + 1 is obtained. Therefore, Si + 1 = Qi1si1∪Qi2si2∪... Is executed by the next system configuration generation means 5, and a recovery tree is formed by repeating the processing by these means 1-5. Reference numeral 6 denotes evaluation value calculation means. When the executable target operation set Qij of the system section sij is Qij = φ, it indicates that the recovery operation at the branch of the recovery tree has been completed. In the generation of the recovery tree, the search is performed a specified number of times from the search step where the branch for which the recovery operation has been completed first is found, and the recovery tree is determined. Of the generated branches, the evaluation value is calculated by the global evaluation function H for the series for which the restoration operation has been completed. The arguments of the global evaluation function H are the transition history from the initial cross section to Sij, the system status of the final cross section, and the like.
H = H (sij) = H (pijk · pi1jk · pi2jk ... P1jk ... s1j)
Reference numeral 7 denotes an optimum operation sequence selection means, which is a recovery procedure that finally obtains the transition history of each branch having the highest overall evaluation function H evaluation. It should be noted that the global evaluation function is arbitrarily determined by evaluation factors such as fewer system changes, better power transmission, and fewer operation steps until recovery.

  FIG. 3 shows an example of searching for an optimal solution by the state enumeration method. As a result of narrowing down from the system constraint condition C while a plurality of operation candidates are generated in a system cross section sij, the operation pijk is selected. A plurality of operation candidates are also generated in the next system cross section si + 1j, but the higher evaluation is selected by the global evaluation function H.

FIG. 4 is an algorithm showing the second embodiment of the present invention, in which two global evaluation functions are introduced to speed up the search. In addition, the same code | symbol is attached | subjected to the part same as FIG. 1, or a corresponding part. That is, in this embodiment, the operation content narrowing means 8 is provided between the executable target operation extracting means 3 and the operation candidate determining means 4. The operation content narrowing means 8 introduces the second global evaluation function H2 = H2 (Qij) in the executable objective operation set Qij, and sets the priority for the execution of PijkεQij. In FIG. 3, the operation candidates generated in a certain system cross section sij are further narrowed down by H 2 at this stage. This is for directly evaluating the meaning of the operation to determine the order and selecting an operation with a high evaluation value. For example, in a system section where an overload has occurred, the system changes a part of the load as an executable operation to eliminate the overload. Use (use together) unused equipment. If there are two operations, this corresponds to the case of selecting the use operation of the unused equipment unconditionally under the policy of “not affecting the surrounding system as much as possible”. As described above, a representative example of the second global evaluation function is an evaluation function including conditions for system operation.
Since the number of branches to be searched is greatly limited by the second global evaluation function H2, it greatly contributes to speeding up. In addition, while an optimal solution can be obtained within a limited range, the evaluation item of H2 is essentially different from that of the global evaluation function H. Therefore, a solution that optimizes H may be excluded by H2. is there. Therefore, the optimum solution of H cannot always be obtained in the second embodiment.

FIG. 5 is an algorithm showing the third embodiment, and shows a search method for an optimum procedure by branching and limiting in combination of the first and second embodiments.
The result obtained in FIG. 4 is used as the lower bound of the restoration tree, and branching is limited. This is because the problem of power failure recovery is that the longer it takes to recover, the more steps it takes, and the worse the evaluation, the more the results of FIG. It is intended to obtain a fast and optimal procedure.

An initial system cross section is input by the system cross section input means 1, and a lower bound z in FIG. 4 is obtained by generating a recovery procedure by the priority selection means for the target operation using the second global evaluation function (lower bound determination means 9). After obtaining the lower bound, the evaluation progression means 10 searches for other branches, and the examination system section generation means 11 generates the examination system section. The evaluation value calculation means 6 obtains an executable target operation set Qij from the generated system section sij. When the executable target operation set Qij is Qij = φ, the restoration operation at that branch ends. 12 is an evaluation lower bound judgment means. If PijkεQij and H (pijk · sij) <z, the search of this branch is terminated. If H (pijk · sij) ≧ z, pijk · sij is set to the next examination section Si. Add.
The system analysis means 2 executes the extraction processing of the operation set Pij that satisfies the local condition in sijεSi, and the executable objective operation extraction means 3 obtains the executable objective operation set Qij. Further, in the operation candidate determination unit 4, when the executable target operation set Qij in any system section Sij is Qij ≠ φ, the next system configuration generation unit 5 obtains the system configuration set Si + 1 of the next section. Then, after repeating these 11, 6, 12, 2, 3, 4 and 5, the lower bound is updated by the lower bound updating means 14, and finally through the means 15, 7 from the branches that have reached the restoration, The history of the highest evaluation is the optimal solution.

  FIG. 6 shows a processing image according to FIG. 5. Among a plurality of operation candidates generated in a system cross section sij, an operation candidate whose evaluation value is worse than the lower bound z at the stage of narrowing down with the system constraint condition C. The generator sij + ij = pijk · sij is discarded without being searched. A system configuration set is made with the remaining ones, and the ones with better evaluation than the lower bound z are evaluated.

In this embodiment, an evaluation value called a power outage evaluation value is introduced in the generation of the recovery tree, and the concept of branching is introduced in the search of the recovery tree.
The power failure evaluation value LPH (Lost Price Hour) is defined as follows.
LPH = Σ∫ (a · Pi) dt
Here, “a” is an importance coefficient, and this coefficient “a” is set for each individual demand based on the fact that the damage caused by the consumers at the time of a power failure is not constant for all the consumers. Pi is the magnitude of the load of each individual demand. In other words, the evaluation value LPH is obtained by multiplying the magnitude of each individual demand by the individual importance coefficient and integrating the values integrated by the power failure time. This evaluation value does not necessarily have to simulate damage.

In order to minimize the LPH, various optimization conditions can be reproduced by using the importance coefficient a. For example,
A). Let all a be a = 1.
The evaluation value LPH in this case is the total power outage, and the recovery procedure is a procedure aimed at minimizing the total power outage.

I). It is assumed that a = 1 / Pi.
The evaluation value LPH in this case is the total power outage time, and the restoration procedure is a procedure aimed at minimizing the total power outage time.

C). Set a individually.
When a is set individually, the evaluation value LPH takes the political consideration into consideration. For example, by setting large importance factors such as various lifelines such as police, fire fighting, large hospitals, and media that are considered to have a large social impact during a long blackout, these restorations are given priority. It is a procedure that reflects the desire to do so.

D). By defining the application a of other a as a function of power failure time as a = a (t), a case where the above events are intertwined, for example, as an initial response at the time of power failure, It is possible to obtain an optimal restoration procedure for the system such as giving priority to power supply and then minimizing the amount of power outage.
The evaluation value LPH for the recovery procedure search uses LPH that takes into account the amount of power outage and the overload evaluation.

  FIG. 12 is a conceptual diagram of the generation of a recovery tree. The system shift from the occurrence of a power failure accident to the recovery of the accident is represented as a recovery tree, and an optimal solution is obtained by searching for this. The recovery tree is a branch of the target operation that changes the system configuration, and the recovered system is the end. First, the target (system) operation Pijk that can be executed in the given system is generated, and the generated target operation is generated. The following system cross section is generated for each. When the restoration is completed, the target operation is not generated and becomes the end of the tree, and the process of change from the initial cross section to the end becomes the operation procedure.

  Of all the ends in the recovery tree, the process from the initial cross section to the highest system cross section is the optimum operation. As described above, each element of the executable operation set Qij in the recovery tree generation process includes the second operation. A global evaluation function H2 = H2 (Qij) is introduced, and a priority is set for execution of PijkεQij. In this method, the meaning of the operation is directly evaluated to determine the order, and the operation having a high evaluation value is selected. Based on the rank obtained by the second global evaluation function, the best priority selection is performed, and the selected executable operation is preferentially searched.

  FIG. 13 shows the state of this search. Each system section has an evaluation point, and the target operation also has an evaluation point different from the system section. If an objective operation with a high evaluation is continued to be executed, an initial solution can be obtained.

  FIG. 14 shows a state diagram in which the solution obtained by this best priority selection is used as a lower bound and the recovery tree is limited. The lower bound is the lower bound of X in the ordered set A when x ≧ a with respect to all elements x of the subset X (from Iwanami Mathematical Dictionary 3rd edition). In finding a combinatorial optimization problem, the evaluation value of the optimal solution is always greater than or equal to the lower bound. Therefore, if the lower bound is known, the search range of the optimal solution can be narrowed (branch and bound method). In the search process, when a solution better than the lower bound is found, the lower bound is sequentially updated, and finally the remaining lower bound becomes the optimal solution.

  In the search of a simple recovery tree, there is an infinite loop, so searching through the brute force is not a solution. Therefore, the initial executable solution generated by evaluating the target operation is set as a lower bound, and the size of the restoration tree is limited. Hereinafter, this is called branch cutting. In FIG. 14, the portion surrounded by the line does not exceed the evaluation value of the lower bound, and the other portions are branched, and there are the following four types of branching.

  Branching by target operation type: This is based on the evaluation function H. For example, each of the generated operation candidates is an operation candidate such as an operation of “uncombining use” and other operations during system interconnection. For this purpose, the candidates are sequentially judged based on whether the candidate is useful or unnecessary.

  Branching by evaluation value: Power failure evaluation value LPH (Lost Price Hour) = Σ∫ (a · Pi) dt, the objective function is LPH minimization, so it is used as an initial value, and the lower bound used in combination with other branching It will be judged to exceed.

Branch cutting by isotopic section determination: When a certain system state is an isotope (the same system configuration has appeared twice or more) that has already appeared in the past search assumption, this is called an isotopic section. There are two types of isotope sections: complete isotope section and incomplete isotope section. The complete isotope section is the same as the system configuration, and the operation status of the operating operator is the same. If it is bad, it will be debranched.
Pruning by a peer is considered to be a completely peered system state when the evaluation value of the peer system state searched in the past is exceeded and when the system configuration / power generation / load state in the power system and the operation execution status of the shift personnel are equal. It is done when it is recognized and worse than the past evaluation value.
When the system configuration, power generation, and load conditions in the power system are the same and only the operating state of the duty personnel is different, it is recognized as an incomplete peer state, and it is determined that this incomplete peer state is determined to be continued for the specified number of times. It is a feature.
Incomplete isotope cross sections are the same system configuration, but the operating conditions of the operating operators are different. In this case, the branching is the same in the system configuration, power generation and load status in the power system, and only the operating status of the shift personnel. If they are different, it is recognized as an incomplete isotopic state, and this incomplete isotopic state is performed for a specified number of times. That is, when the incomplete isotope cross section appears continuously for the designated number of times, branching is performed for the set number of times.

  Branch cutting by cross-border determination: Branching is performed when the LPH in the time section exceeds the LPH of the feasible solution in the same time section regardless of the system configuration and the operation status. FIG. 15A shows the LPH when the cross-border judgment start time until the power failure is eliminated. The solid line is the LPH of the feasible solution and the dotted line is the optimal solution. This is the feasible solution LPH indicated by the solid line. Because there is a part beyond In FIG. 15B, after the cross-border determination start time t, the optimal solution LPH indicated by the dotted line does not exceed the feasible solution LPH, and thus becomes the survival optimal solution. The cross border determination start time t is set by the user.

  FIG. 16 is a summary diagram for searching for an optimal solution. Target operations satisfying local conditions are extracted from a target operation set obtained in a system cross section Sij, and are narrowed down by the system constraint conditions shown in FIG. Then, the target operation having the highest evaluation is selected by ordering and branching by the second global evaluation function, and the search for the next system cross section Si + 1j is advanced with priority given to the operation, and this is repeated until a restored branch is found. . When a restored branch is found, this is set as the lower bound of the executable solution, and another branch is searched.

  In FIG. 16, the system cross section Si + 3j corresponds to the initial executable solution, and when the initial executable solution is found, evaluation of other branches by the global function is executed while going back as indicated by the dotted line. If a feasible solution with good evaluation by the global function is found from the lower bound, the lower bound is updated sequentially, and the remaining lower bound finally becomes the optimum solution to be obtained.

When evaluating the supply reliability of the system, the following six points are evaluated: overload evaluation, power outage evaluation, recovery (power outage) time, number of operations, difficulty, and optimization rate.
The overload evaluation is an LPH generated by an overload, and the power failure evaluation is an LPH generated by a power failure. The evaluation value obtained by the above-described method is used. The power outage time is the time elapsed until the power outage occurs when the power outage occurs. The other three requirements are parameters for the user to determine whether or not the operation tool is operating correctly. The number of operations is the number of recovery operations executed, and the greater the number of operations, the more time is required for recovery. The difficulty is the number of branches of the retrieved recovery tree, and the more difficult the search is. The optimization rate is a parameter of the ratio of the final solution LPH to the initial feasible solution LPH, and the larger the optimization rate, the more rewarding the examination efforts.

FIG. 7 is an explanatory diagram of the fifth embodiment. In the first to third embodiments, the optimum procedure is obtained. This embodiment is used as an inference engine. In the process of recovering from a power failure, system conditions such as equipment conditions are not always constant. For example, it is confirmed that a bus that was not usable immediately after the accident was confirmed to have no problem in use as a result of patrol. In other words, when these various conditions are updated, there is a possibility that the recovery procedure that was optimal until then will not be optimal. As a countermeasure, it is easy to come up with a method of establishing a recovery procedure by repeating the optimal calculation every time the condition is updated, but it takes time to repeat the optimal procedure search many times. Therefore, a search is made to a certain depth, and even if recovery is in progress, an evaluation in the process up to that point is performed and an operation is selected.

In FIG. 7, the system analysis in the system cross section at time t0 is executed, operation candidates A and B are extracted, and when operation A is selected in the optimal calculation, operation A is executed until time t1. As a result, when the operation candidates for C and D are extracted and the operation candidate for D is selected by the optimal calculation, the operation for D is executed from time t1. Similarly, E and F are extracted from the operation candidates extracted by the operation of D from time t2, and an operation is selected by optimal calculation.
In other words, evaluation based on an evaluation function such as fewer system changes, better power transmission, better operation steps until restoration, etc. One hand that seems to be optimal is selected. In the example of FIG. 7, three prefetching with diagonal lines is executed, and the best procedure is selected from the result. According to this, since the optimum calculation is performed every time, it can flexibly cope with changes in equipment conditions and the like, and since it is a finite stage, the calculation time is relatively fast.

In order to achieve the target operation of the present invention, it is necessary to prepare conditions for the operation. Since the operation is meaningful, the system information required for the operation is not limited to the switching state of the switch, voltage / current information, and the like. Therefore, the following four methods are conceivable as system recognition methods.
The first is a recognition method for partial systems. This is a recognition method in which a power system is divided by a transformer and an open circuit breaker, and each divided partial system is made into one group. In a switch with voltage at both ends, when the switch is operated, it is divided into groups such that when both ends are the same groove, it is a combined operation, and when they are different, it is a linked operation. Identify the operation.
The second is a recognition method using a power supply terminal. In this method, as shown in FIG. 8, in any bus, among the terminals connected to the bus, a terminal that can reach the reference power source when the terminal is traced back is recognized as a power source terminal.
The third is a recognition method using an overload system. FIG. 9 is an explanatory diagram of the overload, which is a system downstream in the tidal direction in the overload branch. The system is determined to be further downstream from the bus connected downstream (specifically, it is a power supply terminal). Power transmission system with no terminal) is recognized as an overload system. System operation for overload processing is performed between an overload system and a system adjacent thereto.
The fourth is an operation failure system recognition method. FIG. 10 is an explanatory diagram thereof. For example, when trying to transmit power to a system that is stopped, if the system cannot be executed due to system constraints or the like, the stopped system that is the target of power transmission is recognized as an operation failure system.

An algorithm showing an embodiment of the present invention (embodiment 1). The conceptual diagram for demonstrating this invention. The optimal solution search conceptual diagram. The algorithm (Example 2) which shows the Example of this invention. An algorithm (Example 3) showing an example of the present invention. Process conceptual diagram of Example 3. FIG. Process explanatory drawing by optimal calculation. System recognition explanatory drawing in target operation. System recognition explanatory drawing at the time of overload. System recognition explanatory drawing at the time of operation failure. Explanatory drawing of target operation. Recovery tree generation diagram. Search diagram of optimal solution. The restoration tree pruning state diagram by the lower bound. The branch cutting state diagram by cross-border judgment. Search diagram of optimal solution.

Explanation of symbols

1 ... System cross section input means
2 ... System analysis means
3 ... Extraction means of executable target operation
4 ... Operation candidate determination means
5 ... System configuration generation means
6 ... Evaluation value calculation means
7: Optimal operation sequence selection means
8: Narrowing means 9 ... Lower bound determining means 10 ... Evaluation progress means
11 ... Means for generating cross section of examination system
12 ... Evaluation limit judging means 14 ... Lower bound updating means

Claims (19)

In the method of creating the system restoration procedure after a power failure based on the information on the devices that make up the power system,
A system restoration procedure creation method, characterized in that a system restoration procedure is created based on a target operation performed on a power device constituting the power system. The method for creating a system restoration procedure according to claim 1, wherein the system configuration state of the power system handled by the target operation is a pre-operation state, an in-operation state, and a post-operation state. When the system immediately after the power failure occurs is input to extract the executable objective operation set, and each individual cross section obtained by executing each element of the executable objective operation set, when there is a further executable objective operation set, Generate a feasible objective operation set for the system cross section, form the next system cross section from the system cross section by the executable objective operation, and repeat this until there is no feasible objective operation set for any system cross section. It is judged that the restoration operation has ended when there is no set of executable objective operations, and the system cross section for which the restoration end has been judged is evaluated using the global evaluation function, and the order of the target operation that reaches the system section with the best evaluation is obtained. The method for creating a system restoration procedure according to claim 1 or 2, wherein the restoration procedure is used. 4. The method for creating a system restoration procedure according to claim 3, wherein the executable objective operation set satisfies both a system constraint condition and a local condition. 5. The system according to claim 3 or 4, wherein after the executable objective operation set is extracted, elements of the executable objective operation set are narrowed down by a second global evaluation function based on system operation, and the next system section is shifted to. How to create a recovery procedure. 6. The method for creating a system restoration procedure according to claim 3, wherein evaluation points based on the global evaluation function are assigned to all system cross sections that appear in the search process. A system immediately after the occurrence of a power failure is input to extract a set of executable objective operations. From this, the system section formed by the executable objective operations selected by the second global evaluation function based on system operation is selected and searched. If the end of the recovery operation is determined in the generated system cross section, the process leading up to this will be the first optimal executable recovery operation, and a global evaluation of the optimal executable recovery operation for each non-selected executable target operation Perform a search for only those that have obtained a better evaluation score by performing comparison with the function, and when the executable recovery operation is found, the next optimal executable recovery operation will be used. The method for creating a system restoration procedure according to claim 3, wherein the restoration procedure is obtained. At the time of searching for the feasible recovery operation, a finite search specified in advance is performed from the initial cross section of a given system, and each feasible recovery operation system is evaluated with a global function. The system recovery according to claim 7, wherein a plurality of executable recovery operation candidates are extracted, and unselected executable recovery operations are excluded from operation candidates in an operation process connected to the extracted system. How to create a procedure. The evaluation value for the pruning is an evaluation value including the power outage evaluation value obtained by multiplying the value of each individual demand by the individual importance coefficient and integrating the elapsed time of the restoration operation. The method for creating a system restoration procedure according to claim 8. The branching by the evaluation value is performed when the evaluation value in the time section exceeding the determination start time exceeds the evaluation value of the feasible solution in the same time section. How to create the described system restoration procedure. 11. The method for creating a system restoration procedure according to claim 8, wherein the branching by the evaluation value is performed when the evaluation value of the peer system state searched in the past is exceeded. The branching by the peer is recognized as a completely peered system state when the system configuration / power generation / load state in the power system and the operation execution situation of the shift staff are equal, and is determined to be branching when it is worse than the past evaluation value. The method for creating a system restoration procedure according to claim 11. The branching by the peer is recognized as an incomplete peer state when the system configuration, power generation, and load state in the power system are the same and only the operating state of the shift personnel is different, and this incomplete peer state continues for the specified number of times. 13. The method for creating a system restoration procedure according to claim 11 or 12, wherein the system is cut into branches. When selecting one operation to be performed on a system cross section, the specified finite stage is searched by the executable target operation, each system cross section is evaluated by the global evaluation function, and the first connected to the system cross section with the best evaluation value is obtained. 14. The system recovery procedure according to claim 1, wherein the target operation is determined as a target operation to be selected, and the recovery procedure is determined by repeating this operation when an operation to be performed next is selected. How to make. System recognition for the purpose operation may be any of recognition of a partial system divided by a transformer constituting an electric power system and an open switch, recognition of a power supply terminal, recognition of an overload system, or recognition of an operation failure system. 15. The method for creating a system restoration procedure according to claim 1, wherein: Based on the equipment information that composes the power system, the system restoration procedure after a power failure is created.
A system that inputs a system immediately after the occurrence of a power failure and extracts an executable objective operation set that is a set of objective operations that satisfy local conditions and system constraint conditions, and extracts the next system section by the executable objective operation and performs the next execution Causing a computer to execute means for extracting a set of possible objective operations, means for repeating the same, means for calculating an evaluation value using a global evaluation function, and means for selecting a high evaluation value by judging the calculation result Program for. Means for narrowing down the elements of the executable objective operation set by a second global evaluation function based on system operation after extraction by the means for extracting the executable objective operation set, and extracting or evaluating the next system cross section after narrowing down by this means The program according to claim 16, which causes a computer to execute the calculation means. Means for preferentially selecting an executable objective operation selected by a second global evaluation function based on system operation after extraction of the executable objective operation set; and extracting or evaluating value calculation means for the next system section; 17. The program according to claim 16, which causes a computer to execute means for selecting a branch having a good evaluation value and means for generating a system cross section that follows only the selected branch. A means for causing a computer to execute means for generating a restoration tree of a finite stage designated by each means, means for selecting a branch having a good evaluation value by the evaluation value calculating means, and means for repeating this when selecting the next operation. Item 16. The program according to items 16 to 18.

Images