Evaluation system, method, and program

Abstract

To make it possible to accelerate failure calculation while maintaining sufficient pertinence to a variety of failure cases.SOLUTION: An evaluation system for evaluating fault tolerance of a network constituted of multiple devices comprises: a failure tree generation unit that generates a failure tree that shows relationships between a failure of a single device and failures of other devices that derive from that failure and cause multiple-device failures; and a failure tree management unit that executes single-precision failure calculation in accordance with the failure tree and executes double-precision failure calculation for sites where calculation results that make it possible to establish countermeasures against failure cannot be obtained through the single-precision failure calculation.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to a technique for evaluating a network that is a social infrastructure. [Background technology]

[0002] Networks that form the social infrastructure, such as power systems and transportation networks, require fault tolerance, and need to be adequacy and resilience against many expected failures. Hereinafter, networks that form the social infrastructure may be referred to as social infrastructure networks or infrastructure networks. It is important to evaluate the fault tolerance of social infrastructure networks and use the evaluation results in the operation of social infrastructure networks.

[0003] Fault tolerance evaluation of social infrastructure networks involves calculating indicators that show the state of the network when a failure occurs for each scenario that assumes various failures in the network. Hereinafter, calculating indicators that show the state of the network for scenarios that assume failures may be referred to as failure calculations. Also, scenarios that assume failures may be referred to as failure cases.

[0004] Calculating failures in social infrastructure networks requires computational costs. The computational costs increase if it is assumed that multiple devices will fail simultaneously. Simultaneous failures of multiple devices are hereinafter referred to as multiple failures. In large-scale social infrastructure networks, the number of devices that should be assumed to fail is enormous, increasing the possibility of multiple failures. For this reason, there has been an increasing need in recent years to evaluate multiple failures in social infrastructure networks, and making this possible is of great value to operators and owners of social infrastructure networks.

[0005] In this context, various techniques have been proposed to handle fault calculations for multiple faults in social infrastructure networks. For example, Patent Documents 1 and 2 disclose methods for speeding up fault calculations.

[0006] Patent Document 1 discloses a method for reducing the scale of calculations for fault calculations in a power system by generating an equivalent circuit of the power system and using the equivalent circuit.

[0007] Patent Document 2 discloses a method for using a semi-analytic solution in power system fault calculations, which enables fault evaluation without performing transient calculations that increase calculation time.

[0008] In addition, Non-Patent Documents 1 and 2 propose to speed up fault calculations by utilizing the parallel calculation performance of GPUs (Graphical Processing Units), which are a technology that has made rapid advances.

[0009] Non-Patent Document 1 proposes that in order to find a solution to the simultaneous equations required for fault calculation, a matrix is ​​prepared by incomplete LU decomposition, and then a convergence calculation called the BICGSTAB method is used as an additional step.

[0010] Non-Patent Document 2 proposes parallelizing the LU decomposition of a plurality of fault cases when performing circuit calculations for a power system. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] US2015 / 0378384 specification [Patent Document 2] US2017 / 0153682 specification [Non-patent literature]

[0012] [Non-Patent Document 1] K. Tang, S. Dong, B. Zhu, Q. Ni and Y. Song, “GPU-Based Real-time N-1 AC Power Flow Algorithm With Preconditioned Iterative Method,” 2018 IEEE Power & Energy Society General Meeting (PESGM), 2018, pp. 1-5, doi: 10.1109 / PESGM.2018.8586399. [Non-Patent Document 2] G. Zhou et al., “GPU-Based Batch LU-Factorization Solver for Concurrent Analysis of Massive Power Flows,” in IEEE Transactions on Power Systems, vol. 32, no. 6, pp. 4975-4977, Nov. 2017, doi: 10.1109 / TPWRS.2017.2662322. DISCLOSURE OF THEINVENTION [Problem to be solved by the invention]

[0013] The method of Patent Document 1 degenerates the network that is the target of the fault calculation, so that it is difficult to maintain sufficient validity of the calculation results in fault calculation for serious fault cases.

[0014] The method of Patent Document 2 reduces the calculation time by omitting transient calculations, and therefore the calculation time is not reduced in fault cases where the original fault calculation does not include transient calculations.

[0015] In the method of Non-Patent Document 1, the number of layers of convergence calculations is increased, which may reduce the convergence of the calculations.

[0016] In the method of Non-Patent Document 2, the failure cases that can be handled are limited, and therefore there is a risk that it may not be possible to adequately handle various failure cases.

[0017] An object of the present disclosure is to provide a technique that enables faster fault calculations while maintaining sufficient validity for a variety of fault cases. [Means for solving the problem]

[0018] The evaluation system disclosed herein is an evaluation system for evaluating the fault tolerance of a network composed of multiple devices, and includes a fault tree creation unit that creates a fault tree showing the relationship between a failure of a single device and the failure of other devices that derive from the failure and cause multiple device failures, and a fault tree management unit that performs single precision fault calculations according to the fault tree, and performs double precision fault calculations for points where the single precision fault calculations do not produce calculation results that enable the setting of countermeasures against failures. Effect of the Invention

[0019] According to the present disclosure, it is possible to provide a technique that enables faster fault calculations while maintaining sufficient validity for various fault cases. [Brief description of the drawings]

[0020] [Figure 1] FIG. 2 is a schematic diagram for explaining the relationship between an infrastructure network and an infrastructure network evaluation system according to the first embodiment and related parties thereof. [Diagram 2] FIG. 1 is a schematic diagram showing an example in which the infrastructure network is a power system. [Diagram 3] 1 is a block diagram showing an example of a functional configuration of an infrastructure network evaluation system. FIG. [Figure 4] 1 is a block diagram showing an example of a hardware configuration of an infrastructure network evaluation system. [Diagram 5] 4 is a flowchart showing an example of processing performed by the infrastructure network evaluation system according to the first embodiment. [Figure 6] FIG. 2 illustrates an example of a fault tree. [Figure 7]6 is a flowchart showing detailed processing of step S4 in FIG. 5. [Figure 8] Here is the fault tree when the first single precision calculation is performed: [Figure 9] Shows the fault tree when a double-precision calculation is performed after a single-precision calculation. [Figure 10] 4 shows the fault tree after all calculations have been performed. [Figure 11] FIG. 11 is a schematic diagram showing a first example of an output of a calculation result. [Figure 12] FIG. 11 is a schematic diagram showing a second example of an output of a calculation result. [Figure 13] FIG. 11 is a block diagram showing a configuration example of an infrastructure network evaluation system according to a third embodiment. [Figure 14] 13 is a flowchart showing an example of a process performed by a fault tree management unit according to the third embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] Hereinafter, some embodiments will be described with reference to the drawings.

[0022] (Embodiment 1) FIG. 1 is a schematic diagram for explaining the relationship between the infrastructure network, the infrastructure network evaluation system, and the parties involved.

[0023] In this embodiment, a case will be described where the infrastructure network SOL1 is a power system, however, the infrastructure network SOL1 is not limited to a power system.

[0024] The infrastructure network evaluation system SOL2 performs fault tolerance evaluation and enhancement proposals for failure cases occurring in the infrastructure network SOL1 based on network data acquired from the infrastructure network SOL1. A failure case is a scenario that assumes a failure, and includes information indicating, for example, the failure location, failure type, failure timing, and failure severity. For example, the failure cases may include a single device failure (N-1), multiple device failure (NX), and cascading failure (N-1-1) for devices that make up the infrastructure network SOL1.

[0025] An infrastructure network operator SOL3, for example an electric power system operator, checks the resistance of the infrastructure network SOL1 to failures with reference to a fault tolerance evaluation provided by the infrastructure network evaluation system SOL2, and operates the infrastructure network SOL1.

[0026] An infrastructure network owner SOL4, an example of which is a power line owner, considers augmenting the infrastructure network SOL1 with reference to an augmentation proposal provided by the infrastructure network evaluation system SOL2. At this time, the infrastructure network owner SOL4 may use the augmentation proposal to make an augmentation request to an infrastructure network augmentation approver SOL5, an example of a stakeholder of the power line owner. Furthermore, the infrastructure network augmentation approver SOL5 may decide whether or not to augment the infrastructure network SOL1 based on the augmentation proposal. If approval for the augmentation is obtained from the infrastructure network augmentation approver SOL5, the infrastructure network owner SOL4 may augment the infrastructure network SOL1.

[0027] Since the reinforcement of the infrastructure network SOL1 requires a great deal of time and effort, sufficient justification is required. The infrastructure network evaluation system SOL2 performs fault calculations for many more fault cases within a limited time and provides fault tolerance evaluations and reinforcement proposals, so that each of the players mentioned above, SOL3, SOL4, and SOL5, involved in the solution, can take more appropriate actions based on sufficient justification.

[0028] FIG. 2 is a schematic diagram showing an example in which the infrastructure network is a power system.

[0029] The power system G1 is an infrastructure network composed of buses, transmission lines, loads, generators, etc., and aims to provide a stable power supply from the power generation area to the consumption area (load).

[0030] A stable power supply requires that the transmission limit set for the power transmission route not be exceeded, that the bus voltage be kept within a certain range, and that the frequency of the AC voltage be kept within a certain range. If these conditions are not met, protective devices will be activated to prevent damage to the equipment that makes up the power system.

[0031] In this embodiment, a case where a fault tolerance evaluation of the power system G1 is performed for a single device failure (N-1), multiple device failures (N-2), and multiple device failures (N-3) in the power system G1, with L1-3, L1-2, L2-4, and L3-4 being devices that may fail, will be described. Note that hereinafter, multiple device failures may include both multiple failures in which multiple devices fail simultaneously and chained failures of multiple devices.

[0032] FIG. 3 is a block diagram of the infrastructure network evaluation system SOL2.

[0033] The infrastructure network evaluation system SOL2 has, as input data, a network data DB1, a fault case information DB2, and a countermeasure data DB3.

[0034] The infrastructure network evaluation system SOL2 has, as output data, a fault tree DB4, a fault tree archive DB5, and an enhancement proposal DB6.

[0035] The infrastructure network evaluation system SOL2 has, as processing units, a fault calculation number calculation unit 1, a fault tree creation unit 2, a fault tree management unit 3, a fault tree archive unit 4, a reinforcement location calculation unit 5, and an output unit 6.

[0036] Next, the input data will be described.

[0037] The network data DB1 includes configuration information indicating connections between devices in the infrastructure network SOL1, measurement information indicating items measured by each of the devices, and parameters related to the infrastructure network SOL1.

[0038] The failure case information DB2 is information related to failure cases. A failure case is indicated by information including at least one of information on the location where the failure occurs, the type of failure, the timing at which the failure occurs, and the severity of the failure.

[0039] The countermeasure data DB3 is data related to countermeasures. A countermeasure is defined by the state of the infrastructure network SOL1 in which the countermeasure should be executed and the procedure for executing the countermeasure. The countermeasure data DB3 may be registered in advance.

[0040] Next, the output data will be described.

[0041] The fault tree DB4 indicates the relationship between the failure of a single device constituting the infrastructure network SOL1 and the failure of other devices that derive from the failure and cause multiple device failures. The fault tree DB4 may also include the results of fault calculations related to fault tolerance evaluations obtained with each precision (e.g., single precision or double precision). In other words, the fault tree DB4 manages the relationship between expected faults. For example, when the infrastructure network SOL1 is a power system, the fault tree DB4 manages an N-1 failure and an N-2 failure or an N-1-1 failure that derives from the N-1 failure. Here, an N-2 failure indicates that two devices fail simultaneously, and an N-1-1 failure indicates that two devices fail with a time lag.

[0042] The fault tree archive DB5 includes the past fault trees DB4. For example, the fault tree archive DB5 associates the type of the result of fault calculation in single precision and / or double precision with each device in the fault tree DB4. The fault tree archive DB5 is used, for example, to generate the enhancement proposal DB6.

[0043] The augmentation proposal DB 6 includes information on augmentation proposals for the infrastructure network SOL 1. For example, the augmentation proposal DB 6 includes information indicating measures to improve the fault tolerance of an augmentation location in the infrastructure network SOL 1. In addition, the augmentation proposal DB 6 may include information indicating the basis for augmenting that location.

[0044] Next, each processing function will be described.

[0045] The fault tree creation unit 2 creates a fault tree DB4. The fault tree management unit 3 executes fault calculation with single precision according to the fault tree DB4, and executes fault calculation with double precision for parts where the fault calculation with single precision does not give a calculation result that enables setting of countermeasures against the fault. For example, the fault tree management unit 3 executes fault calculation based on fault case information. As a result, the fault calculation with single precision is executed first according to the fault tree DB4, and then the fault calculation with double precision is executed for parts where the calculation result that enables setting of countermeasures is not given, so that it is possible to speed up the fault calculation while maintaining sufficient validity for various fault cases.

[0046] Furthermore, the fault tree management unit 3 may perform fault calculations in double precision from the beginning for faults of multiple devices including faults calculated in double precision, rather than performing fault calculations in single precision. This allows efficient calculations to be performed by performing the fault calculations in double precision from the beginning, since it is known that a valid calculation result that enables setting of countermeasures cannot be obtained for faults of multiple devices including faults calculated in double precision using single precision fault calculations.

[0047] In addition, in the processing executed by the fault tree management unit 3, the processing for updating the fault tree DB4 may be referred to as “update tree update”, the processing for performing fault calculation in single precision may be referred to as “single precision fault calculation”, and the processing for performing fault calculation in double precision may be referred to as “double precision fault calculation”.

[0048] The fault calculation number calculation unit 1 calculates the number of fault calculations that can be performed simultaneously in parallel for each of single precision and double precision based on network data including configuration information indicating connections between devices in the infrastructure network SOL1 and measurement information indicating items measured by each device. Also, the fault tree management unit 3 may simultaneously perform as many fault calculations as can be performed simultaneously in parallel for each of single precision and double precision. As a result, the number of fault calculations that can be performed simultaneously in parallel for each precision is calculated based on the network data, and fault calculations of each precision are simultaneously performed as many as can be performed simultaneously in parallel, so that the characteristics of the dedicated calculation unit 93 (e.g., GPU) (see FIG. 4) can be utilized to efficiently perform fault calculations of each precision and increase the speed.

[0049] Furthermore, when the infrastructure network SOL1 is a power system, the fault calculation number calculation unit 1 may specify the amount of memory required for executing the power flow calculation and the optimal power flow calculation of the power system for each of single precision and double precision, and calculate the number of fault calculations that can be executed simultaneously in parallel based on the amount of memory. In this way, in the evaluation of the fault tolerance of the power system, the number of fault calculations that can be executed simultaneously in parallel is calculated according to the amount of memory required for the power flow calculation and the optimal power flow calculation for each of single precision and double precision, so that an appropriate number of fault calculations can be executed in parallel, and the evaluation calculation can be executed efficiently.

[0050] The fault tree archive unit 4 records a fault tree archive in which the type of fault calculation result using single precision and / or double precision is associated with each device in the fault tree DB4. The reinforcement location calculation unit 5 identifies, based on the fault tree archive DB5, locations where calculation results that enable setting countermeasures against faults cannot be obtained as reinforcement locations, and generates reinforcement proposal DB6 for measures to improve the fault tolerance of the reinforcement locations in the infrastructure network SOL1. As a result, the types of fault calculation results in past fault calculations are associated with the fault tree DB4, and reinforcement locations and reinforcement proposals for improving their fault tolerance are generated based on this information, so that accurate reinforcement proposals can be obtained by utilizing the history of past fault calculations.

[0051] Furthermore, the reinforcement location calculation unit 5 may generate information on the improvement of fault tolerance of the infrastructure network SOL1 by adopting the countermeasures at the reinforcement locations as basis information, together with the reinforcement locations and countermeasures, in the reinforcement proposal DB 6. This makes it possible to obtain the basis for the countermeasures in the reinforcement proposal.

[0052] The output unit 6 outputs the data stored in the fault tree DB 4, the fault tree archive DB 5, or the enhancement proposal DB 6.

[0053] FIG. 4 is a block diagram showing an example of a hardware configuration of the infrastructure network evaluation system SOL2.

[0054] The infrastructure network evaluation system SOL2 includes at least one computer and a communication network.

[0055] The infrastructure network evaluation system SOL2 includes a communication unit 91 and a measuring instrument G11. The measuring instrument G1 is installed in a power system G1, which is an example of an infrastructure network, and measures measurement information. The measuring instrument G11 is connected to the communication unit 91 through a predetermined communication network.

[0056] Moreover, in addition to the above-mentioned communication unit 91, the infrastructure network evaluation system SOL2 includes a network data DB1, a fault case information DB2, a countermeasure data DB3, a fault tree DB4, a fault tree archive DB5, an enhancement proposal DB6, a general-purpose calculation unit 92, a calculation unit 93, a memory 94, a various program database DB7, and an input / output unit 95. These components are connected to a bus B1 and can transmit and receive data to and from each other through the bus B1. Moreover, another system 96 and a GUI device 97 may be connected to the input / output unit 95.

[0057] The databases DB1 to DB6 store the various information described above as databases, and so a description thereof will be omitted here. The various program database DB7 stores computer programs and data for implementing the functions of the infrastructure network evaluation system SOL2 described above.

[0058] The general-purpose calculation unit 92 executes processes (eg, executes a computer program) for implementing the functions of the infrastructure network evaluation system SOL2. The general-purpose calculation unit 92 may be interpreted as a Central Processing Unit (CPU).

[0059] The dedicated arithmetic unit 93 may be configured to include at least one of a Graphical Processing Unit (GPU), a Field Programmable Gate Array (FPGA), and an Application Specific Integrated Circuit (ASIC) in addition to a CPU. The dedicated arithmetic unit 93 may be configured to have a different processing speed when performing single precision calculation and a different processing speed when performing double precision calculation. For example, the dedicated arithmetic unit 93 may be configured to operate at a first processing speed when a mode for performing single precision calculation is set, and to operate at a second processing speed when a mode for performing double precision calculation is set.

[0060] The memory 94 is a storage medium that temporarily stores data. Examples of the memory 94 include a Random Access Memory (RAM) or a Virtual Memory.

[0061] The GUI 97 receives input from a user. The GUI 97 also displays output results from the infrastructure network evaluation system SOL2. The GUI 97 may be a monitor, a PC, a smartphone, a tablet, or the like.

[0062] The other system 96 is a system different from the infrastructure network evaluation system SOL2, and may be, for example, another server.

[0063] The input / output unit 95 inputs information input to the GUI 97 to the infrastructure network evaluation system SOL2. The input / output unit 95 outputs information generated by the infrastructure network evaluation system SOL2 to the GUI 97. The input / output unit 95 may also input and output data to and from another system 96.

[0064] The processing performed by the infrastructure network evaluation system SOL2 may be processing performed by the general-purpose calculation unit 92 or the dedicated calculation unit 93 in cooperation with other components. Therefore, in this embodiment, the processing mainly performed by the infrastructure network evaluation system SOL2 or each block may be interpreted as processing mainly performed by the general-purpose calculation unit 92 or the dedicated calculation unit 93.

[0065] FIG. 5 is a flowchart showing an example of processing performed by the infrastructure network evaluation system SOL2 according to the first embodiment.

[0066] In step S1, the fault calculation number calculation unit 1 reads the network data DB1, the fault case information DB2, the countermeasure data DB3, and the like.

[0067] In step S2, the fault calculation number calculation unit 1 calculates the number of fault calculations that can be executed simultaneously in parallel for each of single precision and double precision, based on the network data DB1 including configuration information and measurement information regarding the infrastructure network SOL1.

[0068] In step S3, the fault tree creation unit 2 creates a fault tree DB4.

[0069] In step S4, the fault tree management unit 3 executes single precision fault calculation, double precision fault calculation, and fault tree update to update the fault tree DB4.

[0070] In step S5, the fault tree archive unit 4 stores the updated fault tree DB4 in the fault tree archive DB5.

[0071] In step S6, the reinforcement portion calculation unit 5 calculates reinforcement portions based on the fault tree archive DB5.

[0072] In step S7, the output unit 6 outputs the calculation result.

[0073] Each of the above steps will now be described in detail.

[0074] First, step S1 will be described in detail.

[0075] The network data DB1 includes, for example, connection state information indicating the connection state of each device in the power system, and measurement information including measurement results of the active power, reactive power, voltage, etc. of the power system. These pieces of information may be stored by item in the network data DB1, or may be held in a processed form. The processed form may be, for example, a state estimation result in which the system configuration and state variables of the system are estimated. Alternatively, the processed form may be a power flow snapshot in which the power flow of the power system is estimated based on the state estimation result.

[0076] The fault case information DB2 includes, for example, a power system fault target device, a fault type such as a busbar ground fault, timing between faults, and duration of the fault.

[0077] The countermeasure data DB3 includes, as types of faults, for example, that the complex power of the line exceeds a predetermined range, or that the voltage of the bus bar deviates from an upper or lower limit. The countermeasure data DB3 also includes, as countermeasures against faults, for example, adjusting the output of a generator, switching devices in the power system, cutting off the power transmission route with a relay, and so on. In this embodiment, a power flow calculation is performed to calculate the power flow of the power system in order to determine whether the apparent power of the power transmission route deviates from a predetermined range due to a fault. Also, an optimal power flow calculation, which is a modified version of the power flow calculation, is performed as a countermeasure against faults.

[0078] Next, step S2 will be described in detail.

[0079] In the above-described power system, in order to determine whether or not a fault has caused a deviation from the power flow limit, both a power flow calculation and an optimal power flow calculation are performed for each fault case.

[0080] In step S2, the dedicated calculation unit 93 specifies the memory amount when the power flow calculation and the optimal power flow calculation are performed in single precision and the memory amount when the power flow calculation and the optimal power flow calculation are performed in double precision.

[0081] The dedicated calculation unit 93, which is typified by a GPU, performs calculations by accessing a memory area different from that of the general-purpose calculation unit 92. Therefore, the maximum number of fault cases that can be held can be calculated by dividing the size of the device memory available to the dedicated calculation unit 93 by the amount of memory required for fault calculation for each of single precision and double precision. For example, assume that the size of the device memory is 2 GB, the amount of memory required for fault calculation in single precision is 250 MB per fault case, and the amount of memory required for fault calculation in double precision is 500 MB per fault case. In this case, the maximum number of fault cases in single precision is calculated to be 8 (=2 GB / 250 MB), and the maximum number of fault cases in double precision is calculated to be 4 (=2 GB / 500 MB).

[0082] In this way, by calculating the maximum number of fault cases in advance, it is possible to reduce the amount of data movement to the dedicated calculation unit 93 in the calculation for one cycle. In other words, it is possible to perform the processing efficiently.

[0083] Next, step S3 will be described in detail with reference to Fig. 6. Fig. 6 is a diagram showing an example of a fault tree.

[0084] The fault tree shown in Figure 6 starts with an N-1 failure, and then shows more severe failures such as an N-2 failure and an N-3 failure. The N-1 failure shows the individual failures of equipment L1-3, equipment L1-2, equipment L2-4, and equipment L3-4. The N-2 failure shows a further failure derived from the N-1 failure. The N-3 failure shows a more severe failure derived from the N-2 failure.

[0085] For example, in Figure 6, a circle in an N-1 failure indicates the failure of a single device included in the N-1 failure. A circle in an N-2 or N-3 failure indicates the failure of multiple devices, including the devices included in the N-2 or N-3 failure and the device on the root side of the tree. For example, the L1-2 circle at the top of an N-2 failure indicates multiple failures of two devices, L1-3 and L1-2. The L2-4 circle at the top of an N-3 failure indicates multiple failures of three devices, L1-3, L1-2, and L2-4.

[0086] In this embodiment, a fault is assumed to be a state in which a power transmission line is released for some reason. The extent to which NX (X is a positive integer) is to be analyzed is defined by the fault case information DB2. In this way, data is defined by a tree structure, making management easier.

[0087] Next, step S4 will be described in detail with reference to Fig. 7. Fig. 7 is a flow chart showing the detailed process of step S4 in Fig. 5.

[0088] In step S401, the fault tree management unit 3 reads various data.

[0089] In step S402, the fault tree management unit 3 determines whether or not there is an uncalculated fault case in the fault tree DB 4. If there is an uncalculated fault case (S402: Yes), the fault tree management unit 3 proceeds to step S403, and if there is no uncalculated fault case (S402: No), the fault tree management unit 3 ends this process.

[0090] In step S403, the fault tree management unit 3 executes single-precision fault calculation based on the number of fault calculations.

[0091] In step S404, the fault tree management unit 3 determines whether there is a fault case for which countermeasures cannot be taken using single-precision fault calculation. Whether a countermeasure is possible or not can be easily determined from the calculation result or calculation process. For example, whether a countermeasure is possible or not can be determined based on whether a feasible operational solution can be obtained in the power flow calculation. If there is no fault case for which countermeasures cannot be taken using single-precision fault calculation (S404: No), the fault tree management unit 3 returns the process to step S402, and if there is a fault case for which countermeasures cannot be taken using single-precision fault calculation (S404: Yes), the fault tree management unit 3 proceeds to step S405.

[0092] In step S405, the fault tree management unit 3 executes double-precision fault calculation for fault cases that cannot be solved by single-precision fault calculation.

[0093] In step S406, the fault tree management unit 3 judges whether or not there is a fault case that cannot be dealt with even by double-precision fault calculation. If there is no fault case that cannot be dealt with by double-precision fault calculation (S406: No), the fault tree management unit 3 returns the process to step S402, and if there is a fault case that cannot be dealt with by double-precision fault calculation (S406: Yes), the fault tree management unit 3 proceeds to step S407.

[0094] In step S407, the fault tree management unit 3 determines that the fault case and fault cases derived therefrom cannot be dealt with, and then returns the process to step S402.

[0095] That is, when the fault tree management unit 3 obtains a calculation result that enables setting a countermeasure against a fault, it may specify the state of the infrastructure network SOL1 from the calculation result, and select a countermeasure to be executed for the specified state from among the countermeasures registered in the countermeasure data DB3. In this way, countermeasures are stored in a database as the countermeasure data DB3, and a countermeasure to be executed according to the state is selected from the database, so that countermeasures can be easily determined.

[0096] Next, the series of procedures in step S4 will be described in detail using a fault calculation for a power system as an example.

[0097] In the power system shown in Figure 2, the state of the power system when a faulty device is present can be found by power flow calculation of the power system. In power flow calculation, state variables of the power system (e.g. bus voltage and phase) are found from, for example, the configuration of the power system, bus voltage, power consumption of the bus, generated power injected into the bus, etc., and the power flow is calculated based on the power flow equation. A commonly known method uses the Newton-Raphson method.

[0098] In the Newton-Raphson method, the state variable x is updated using the Jacobian J until the maximum error of the power flow error f becomes equal to or less than the threshold ε or the maximum number of steps kmax is reached. For example, it is calculated by the following algorithm.

[0099] ______________________________ Algorithm 1 =============================== While(|f(x)|>ε&&k<=Kmax) x=xJ(x)^-1*f(x) k++ End Powerflow = f2(x) =============================== Here, x, J, and f are as follows:

[0100]

number

number

number

number

[0101] In addition, Pi(x) and Qi(x) are determined by the conductance matrix G and the susceptance matrix B, although the explanation is omitted here.

[0102]

number

number

[0103] When performing fault calculations for a power system, various fault conditions can be grasped by changing the conductance matrix and susceptance matrix in the above-mentioned Newton-Raphson method.

[0104] In addition, by applying the above-mentioned power flow calculation, it is also possible to calculate measures to deal with abnormal conditions in the power grid, for example through the transfer of power generation.

[0105] When executing the single-precision fault calculation in step S403, the fault tree management unit 3 can simultaneously perform a series of power flow calculations in single precision for a plurality of fault cases.

[0106] The fault tree management unit 3 can perform this single precision power flow calculation at higher speed by using the dedicated calculation unit 93.

[0107] On the other hand, as described above, power flow calculations include a procedure that requires the inverse matrix of the Jacobian J. Therefore, when performing calculations in single precision, there is a possibility that the matrix will have a low rank. In such cases, it may be difficult to calculate fault cases and / or countermeasures.

[0108] Next, a procedure for associating the fault calculation results with the fault tree will be described with reference to Figures 8, 9, and 10. Figure 8 shows the fault tree when the first single precision calculation is performed. Figure 9 shows the fault tree when double precision calculations are performed after single precision calculations. Figure 10 shows the fault tree after all calculations have been performed. In Figures 8, 9, and 10, a white circle symbol indicates that for the fault case of the equipment with that symbol, a countermeasure can be taken using single precision calculation. A white circle with a diagonal line symbol indicates that for the fault case of the equipment with that symbol, a countermeasure cannot be taken using single precision calculation. A hatched circle symbol indicates that for the fault case of the equipment with that symbol, a countermeasure can be taken using double precision calculation. A hatched circle with a diagonal line symbol indicates that for the fault case of the equipment with that symbol, a countermeasure cannot be taken even using double precision calculation, or it has been determined that a countermeasure is impossible.

[0109] The single-precision calculation can be performed simultaneously for multiple fault cases. Therefore, when the fault tree management unit 3 performs the first single-precision calculation in step S403, it can simultaneously associate the results of eight single-precision calculations with the fault tree DB 4, as shown in FIG.

[0110] The fault tree DB4 shown in Figure 8 shows that in the N-1 fault case, countermeasures can be taken using single precision calculations for all of L1-3, L1-2, L2-4, and L3-4, but in the N-2 fault case, countermeasures cannot be taken using single precision calculations for the fault transitioning from L1-3 to L3-4.

[0111] In step S405, the fault tree management unit 3 performs double precision calculations on the fault cases of L3-4 of N-2 for which it was determined that no countermeasures could be taken by the single precision calculations as shown in Fig. 8. Although the double precision calculations increase the calculation time, they improve the convergence of the calculations, making it possible to determine whether or not a countermeasure can be taken.

[0112] As a result of the double precision calculation, it is assumed that the fault case of L3-4 of N-2 cannot be dealt with even by double precision calculation, as shown in Fig. 9. In this case, the fault tree management unit 3 may determine that L1-2 and L2-4 derived from L3-4, which cannot be dealt with even by double precision calculation, cannot be dealt with even by double precision calculation, without actually performing single precision or double precision calculation. Alternatively, when there is a fault case determined to be unable to be dealt with even by double precision calculation, the fault tree management unit 3 may prune (delete) more serious fault cases derived from that fault case in the fault tree DB4. This makes it possible to execute more fault calculations in a shorter time.

[0113] The fault tree management unit 3 executes the above-mentioned series of procedures, and finally generates a fault tree DB4 as shown in Fig. 10. By referring to the fault tree DB4 thus generated, it is possible to easily determine which faults can be addressed.

[0114] That is, according to the procedure of the present embodiment described above, more fault calculations can be performed in a shorter time, so that the fault tolerance of the target infrastructure network SOL1 can be grasped in a shorter time.

[0115] Next, step S5 will be described in detail.

[0116] In step S5, the fault tree archive unit 4 stores various fault trees DB4 generated by the above-mentioned method as archives in the fault tree archive DB5. This is because even if the configuration of the infrastructure network SOL1 is the same, the behavior at the time of a fault may differ depending on the usage situation of the infrastructure network SOL1.

[0117] Next, step S6 will be described in detail.

[0118] In step S6, the reinforcement location calculation unit 5 calculates reinforcement locations using the fault tree DB4 stored in the fault tree archive DB5. For example, the reinforcement location calculation unit 5 calculates locations in the infrastructure network SOL1 that would be impossible to address if a fault were to occur, using the fault tree DB4 stored in the fault tree archive DB5. Then, the reinforcement location calculation unit 5 proposes reinforcing a route that bypasses the locations where the countermeasures would be impossible. For example, when the reinforcement location calculation unit 5 calculates from the fault tree DB4 that L1-2 is prone to overload, it proposes reinforcing L1-2. At this time, the infrastructure network evaluation system SOL2 may repeat the procedure from step S1 to step S5 for a case where the location is reinforced, and calculate the result of the reinforcement.

[0119] That is, the reinforcement location calculation unit 5 may perform the following process. The reinforcement location calculation unit 5 determines, as an reinforcement location, a location where a calculation result that enables setting of a countermeasure cannot be obtained even with double-precision failure calculation, that is, a location where a countermeasure cannot be applied to the failure. Then, the reinforcement location calculation unit 5 formulates an reinforcement proposal to improve the fault tolerance of the infrastructure network SOL1 in the event of a failure at that location by adding or reinforcing a connection on the infrastructure network SOL1 that bypasses that location.

[0120] Next, step S7 will be described in detail with reference to Fig. 11 and Fig. 12. Fig. 11 is a schematic diagram showing a first example of output of the calculation result. Fig. 12 is a schematic diagram showing a second example of output of the calculation result.

[0121] In step S7, the output unit 6 outputs the calculation results to the monitor M1. The output unit 6 can select or adjust, for each user, which items of the calculation results are to be displayed.

[0122] For example, as shown in Fig. 11, the output unit 6 displays the current state of the power system or the calculation results of fault cases for the system in the near future (i.e., the real-time state of the power system) for the power system infrastructure network operator SOL3. At this time, the output unit 6 does not need to display reinforcement locations and the effects of reinforcement of those locations, which are important to the infrastructure network owner SOL4 but not so important to the infrastructure network operator SOL3. This allows the infrastructure network operator SOL3 to easily grasp the state of the infrastructure network SOL1.

[0123] For example, as shown in Fig. 12, the output unit 6 displays the reinforcement points and the effect of reinforcing those points (i.e., information obtained from the past performance of the power system) to the power system infrastructure network owner SOL4. At this time, the output unit 6 does not need to display the real-time status of the power system, which is important to the infrastructure network operator SOL3 but not so important to the infrastructure network owner SOL4. This allows the infrastructure network owner SOL4 to show the infrastructure network reinforcement approver SOL5 the reinforcement request and the basis for the reinforcement.

[0124] This embodiment can also be applied to infrastructure networks in fields other than power systems, such as waterworks and railways. By providing appropriate input data according to the infrastructure network of a target field, the same effect as that of this embodiment can be achieved in that field.

[0125] (Embodiment 2) In the second embodiment, a modified example of the infrastructure network evaluation system SOL2 will be described.

[0126] In the first embodiment, the fault tree management unit 3 performs calculations in single precision for all fault calculations once. However, the fault tree management unit 3 may perform calculations in double precision from the beginning for parts where it is known in advance that high precision is required, instead of performing calculations in single precision for all fault calculations once. This omits single precision calculations for parts where it is known in advance that high precision is required, making it possible to perform more fault calculations in a shorter time.

[0127] (Embodiment 3) In the third embodiment, a fail-safe for single-precision calculation in the infrastructure network evaluation system SOL2 will be described.

[0128] The procedure described in the first embodiment is based on the premise that there is no problem with the calculation accuracy when convergence calculation is possible with single-precision calculation. In the third embodiment, a method for preventing single-precision calculation from producing an optimistic solution and improving the reliability of the infrastructure network evaluation system SOL2 will be described.

[0129] Fig. 13 is a block diagram showing a configuration example of an infrastructure network evaluation system SOL2 according to embodiment 3. Next, with reference to Fig. 13, differences from the infrastructure network evaluation system SOL2 according to embodiment 1 shown in Fig. 3 will be described.

[0130] In the infrastructure network evaluation system SOL2 shown in Fig. 13, a single-precision calculation evaluation parameter DB8 is input to the fault tree management unit 3. The single-precision calculation evaluation parameter DB8 is a parameter for determining whether the single-precision calculation is valid or invalid. For example, the single-precision calculation evaluation parameter DB8 includes the route usage rate, the usage status of the network adjustment device, the state of the network after the fault evaluation, etc.

[0131] Fig. 14 is a flowchart showing an example of a process performed by the fault tree management unit 3 according to the third embodiment. Next, a modified example of the process shown in Fig. 7 (i.e., a modified example of the details of step S4 shown in Fig. 5) will be described with reference to Fig. 14. Note that here, differences from Fig. 7 will be described, and a description of commonalities with Fig. 7 will be omitted.

[0132] In the flowchart shown in FIG. 14, step S403b is added between step S403 and step S404 in the flowchart shown in FIG.

[0133] In step S403b, the fault tree management unit 3 evaluates the single precision calculation result according to the single precision calculation evaluation parameter DB8. For example, even if it is determined that the overload of the power transmission route can be eliminated in the case after the failure by the single precision calculation, if the usage of the equipment is high and there is insufficient margin for eliminating the overload, the fault tree management unit 3 invalidates the single precision calculation. For the part where the single precision calculation is invalidated in this way, the fault tree management unit 3 performs double precision calculation in step S405.

[0134] In this way, by providing a configuration that prevents optimistic solutions from being output in places where there is a high possibility that an optimistic solution would be obtained by single-precision calculation, the reliability of the infrastructure network evaluation system SOL2 can be improved. In other words, the infrastructure network evaluation system SOL2 can provide more reliable information to users.

[0135] The infrastructure network evaluation system SOL2 may use a plurality of different algorithms to perform the single-precision fault calculation and the double-precision fault calculation. For example, when the infrastructure network SOL1 is a power system, the infrastructure network evaluation system SOL2 may use the Gauss-Ciedal method in addition to the Newton-Raphson method described above. This allows the infrastructure network evaluation system SOL2 to provide more reliable information to the user. [Explanation of symbols]

[0136] SOL1...Infrastructure network, SOL2...Infrastructure network evaluation system, SOL3...Infrastructure network operator, SOL4...Infrastructure network owner, SOL5...Infrastructure network augmentation approver, DB1...Infrastructure network data, DB2...Fault case, DB3...Fault countermeasure procedure, DB4...Fault tree, DB5...Fault tree archive, DB6...Augmentation proposal, DB7...Various program databases, DB8...Single-precision calculation evaluation parameters, 1...Calculation unit for number of fault calculation cases by precision, 2...Fault tree creation unit, 3...Fault tree management unit, 4...Fault tree archive unit, 5...Augmentation location calculation unit, 6...Output unit, 91...Communication unit, 92...General-purpose calculation unit, 93...Dedicated calculation unit, 94...Memory, 95...Input / output unit, 96...Other systems, 97...GUI, M1...Monitor, G1...Power system (infrastructure network)

Claims

1. An evaluation system for evaluating fault tolerance of a network composed of a plurality of devices, comprising: a fault tree creation unit that creates a fault tree showing the relationship between a failure of a single device and failures of other devices that are derived from the failure and cause multiple device failures; a fault tree management unit that executes fault calculations in single precision according to the fault tree and executes fault calculations in double precision for points where the single precision fault calculations do not produce calculation results that enable setting of countermeasures against faults; An evaluation system having:

2. 2. The evaluation system according to claim 1, a fault calculation number calculation unit that calculates the number of fault calculations that can be performed simultaneously in parallel for each of single precision and double precision based on network data including configuration information indicating connections between the devices in the network and measurement information indicating items measured in each of the devices; the fault tree management unit executes fault calculations in parallel as many times as fault calculations can be executed in parallel in single precision and double precision, respectively; Rating system.

3. 3. The evaluation system according to claim 2, a fault tree archive unit that records a fault tree archive in which a type of a result of fault calculation using single precision and / or double precision is associated with each device in the fault tree; and an enhancement point calculation unit that identifies, based on the fault tree archive, a point where a calculation result that enables setting a countermeasure against a fault cannot be obtained as an enhancement point, and generates an enhancement proposal for a countermeasure that improves the fault tolerance of the enhancement point in the network. Rating system.

4. 2. The evaluation system according to claim 1, The fault is indicated by fault case information including at least one of information on a type of the fault, a timing at which the fault occurs, and a severity of the fault, in addition to information on a location at which the fault occurs; the fault tree creation unit indicates the fault in the fault tree by fault case information, The fault tree management unit performs fault calculation based on the fault case information. Rating system.

5. 2. The evaluation system according to claim 1, the countermeasure is registered in advance as countermeasure data defined by a state of the network in which the countermeasure should be executed and a procedure for executing the countermeasure; when a calculation result enabling a countermeasure against a fault to be set is obtained, the fault tree management unit specifies a state of the network from the calculation result, and selects a countermeasure to be executed against the specified state from among the countermeasures registered in the countermeasure data. Rating system.

6. 4. The evaluation system according to claim 3, the reinforcement portion calculation unit generates, in the reinforcement proposal, information on an improvement in fault tolerance of the network by adopting the countermeasure to the reinforcement portion, together with the reinforcement portion and the countermeasure, as basis information; Rating system.

7. 3. The evaluation system according to claim 2, the network is a power grid; the fault calculation number calculation unit specifies the amount of memory required when the power flow calculation and the optimal power flow calculation of the power system are executed for each of single precision and double precision, and calculates the number of fault calculations that can be executed simultaneously in parallel based on the amount of memory; Rating system.

8. 2. The evaluation system according to claim 1, the fault tree management unit does not execute fault calculation in single precision for faults of multiple devices including a fault calculated in double precision, but executes fault calculation in double precision from the beginning; Rating system.

9. 1. A method for evaluating fault tolerance of a network made up of a plurality of devices, comprising: Creating a fault tree showing the relationship between a single device failure and other device failures that result from the single device failure and cause multiple device failures; performing fault calculations using single precision according to the fault tree, and performing fault calculations using double precision for points where the fault calculations using single precision do not provide calculation results that enable setting of countermeasures against faults; A computer-implemented evaluation method.

10. An evaluation program for evaluating fault tolerance of a network composed of a plurality of devices, Creating a fault tree showing the relationship between a single device failure and other device failures that result from the single device failure and cause multiple device failures; performing fault calculations using single precision according to the fault tree, and performing fault calculations using double precision for points where the fault calculations using single precision do not provide calculation results that enable setting of countermeasures against faults; An evaluation program that causes a computer to perform the following:

Images