Water plant operation support system and method

Abstract

To automatically determine an appropriate alarm response priority order upon issuance of multiple alarms in the same time window.SOLUTION: On the basis of alarm history data indicating a log related to alarms that were generated at a facility provided in the water plant, operation history data indicating a log related to operations during operation of the facility, and a plant model that models a water plant including the facility, an alarm operation pattern in which an operation with a predetermined causal relation to each of the alarms is associated with a time series is created and stored in a memory. On the basis of the alarm history data, plant monitoring history data indicating a log related to measurement values at the facility, and the plant model, a chain alarm pattern in which an alarm with a predetermined causal relation is associated with the time series is created and stored in the memory. Using the alarm operation pattern and the chain alarm pattern, a priority level evaluation result list indicating candidates for alarms that can possibly occur in future in the chain alarm pattern is created and stored in the memory. On the basis of the created priority level evaluation result list, a priority order of alarms to be addressed is determined .SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to a water plant operation support system and method. [Background technology]

[0002] In industrial plants, the emergence of the Distributed Control System (DCS) in the 1970s made it possible to set alarms at low cost using software, and since then it has become common to determine the plant's condition and operate it based on alarms.

[0003] On the other hand, because alarms have become easier to set, many alarms have been set without careful consideration of their necessity, and the number of alarms set has increased dramatically. This has led to a situation known as "alarm flooding," where multiple alarms occur at the same time, or where low-urgency noise alarms are mixed in, causing operators to miss important alarms or make operational decisions incorrectly, which can lead to plant accidents.

[0004] To improve the situation described above, alarm management has become increasingly necessary. The most fundamental approach to alarm management to date has been to redesign the alarms and reconfigure the DCS alarms. While this is an ideal approach, it is extremely time-consuming and, in some cases, requires temporarily shutting down the DCS, disrupting plant operations. Therefore, technologies exist that rationalize alarm processing by adding new tools (systems) while leaving the alarms issued by the DCS unchanged. One common technique is to pre-classify alarms issued by the DCS according to importance and priority, and then filter or shelve the alerts displayed on the operation screen. This method allows for efficient operation because the response order is clear even in the event of an alarm flood. However, this method requires a human to pre-classify each alarm, requiring experts with in-depth knowledge of plant operations and is time-consuming.

[0005] There is a method that focuses on "chain alarms" to rationally prioritize alarm responses and enable even inexperienced operators to determine them. Chain alarms are a phenomenon in which, after one alarm occurs, multiple different alarms related to it are triggered in a chain reaction. By dealing with the upstream alarms in this chain alarm, it is possible to deal with the alarm that is the root cause, thereby enabling efficient alarm response. Therefore, when multiple alarms occur within the same time period, the alarms are displayed on the operation screen in chronological order of their occurrence, treating the alarms as a chain alarm and encouraging operators to deal with the alarms that occurred earlier (see Patent Document 1). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-93049 Summary of the Invention [Problem to be solved by the invention]

[0007] It is reasonable to use a technique that focuses on chain alarms and responds to alarms that are believed to be the root cause. However, the actual causal relationship of alarms cannot be identified based on chain information based on the alarm activation time alone. For example, if alarm A is activated at 10:00 and alarm B is activated at 10:10, it is unclear whether alarm A caused alarm B to be activated.

[0008] Furthermore, multiple alarms that are triggered within the same time period do not necessarily belong to the same chain alarm group. For example, there may be parallel chain alarms or unchained individual alarms. In such cases, it is unclear which alarm to respond to first. Therefore, with conventional methods, there is uncertainty in determining alarm response priority, making it difficult to determine an efficient alarm response order.

[0009] Therefore, an object of the present invention is to provide a technology that can automatically determine an appropriate alarm response priority when multiple alarms are issued in the same time period. [Means for solving the problem]

[0010] The water plant operation support system of the present invention, which solves the above-mentioned problems, is a water plant operation support system that uses a computer having a processor and memory to support responses to alarms that have occurred in the water plant, wherein the processor creates an alarm operation pattern that chronologically associates operations that have a predetermined causal relationship with the alarm based on alarm history data that indicates logs related to alarms issued from equipment of the water plant, operation history data that indicates logs related to operations in the operation of the equipment, and a plant model that models the water plant including the equipment, and stores the pattern in the memory; creates a chain alarm pattern that chronologically associates the alarms that have a predetermined causal relationship with the alarm based on the alarm history data, plant monitoring history data that indicates logs related to measurement values ​​of the equipment, and the plant model, and stores the pattern in the memory; uses the alarm operation pattern and the chain alarm pattern to create a priority evaluation result list that indicates candidate alarms that may occur in the future in the chain alarm pattern, and stores the list in the memory; and determines the priority of alarms to be responded to based on the created priority evaluation result list. [Effects of the Invention]

[0011] According to the present invention, it is possible to automatically determine an appropriate alarm response priority when multiple alarms are issued within the same time period. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a diagram illustrating a network configuration including a water plant operation support system according to an embodiment of the present invention. [Figure 2] 1 is a diagram illustrating an example of a hardware configuration of a water plant operation support system according to an embodiment of the present invention. [Figure 3] FIG. 3 is a diagram illustrating an example of the configuration of alarm history data according to the present embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of the configuration of operation history data according to the present embodiment. [Figure 5] FIG. 2 is a diagram showing an example of the configuration of plant monitoring history data in the present embodiment. [Figure 6] FIG. 2 is a diagram illustrating an example of a plant model in the present embodiment. [Figure 7] FIG. 2 is a diagram illustrating an example of the configuration of an absolute priority alarm table according to the present embodiment. [Figure 8] 1 is a diagram illustrating an example of a flow and functional configuration of a water plant operation support system according to an embodiment of the present invention. [Figure 9] FIG. 2 is a diagram showing an example of the configuration of an alarm-operation-recovery model pattern list in the present embodiment. [Figure 10] FIG. 1 is a diagram illustrating an example of a processing concept in this embodiment. [Figure 11] FIG. 1 is a diagram illustrating an example of a processing concept in this embodiment. [Figure 12] FIG. 10 is a diagram showing an example of the configuration of a list of chain alarm transition time patterns in the present embodiment. [Figure 13] FIG. 1 is a diagram illustrating an example of a processing concept in this embodiment. [Figure 14] FIG. 10 is a diagram showing an example of the configuration of a priority evaluation result list in the present embodiment. [Figure 15] FIG. 4 is a diagram illustrating an example of the configuration of an alarm evaluation list according to the present embodiment. [Figure 16] FIG. 10 is a diagram illustrating an example of output in this embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The following description and drawings are examples for explaining the present invention, and some omissions and simplifications have been made as appropriate for clarity of explanation. The present invention can be implemented in various other forms. Unless otherwise specified, each component may be singular or plural.

[0014] In order to facilitate understanding of the invention, the position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

[0015] In the following explanation, various types of information may be described using expressions such as "database," "table," and "list," but the various types of information may also be expressed in data structures other than these. To indicate that the information is not dependent on the data structure, "XX table," "XX list," etc. may be referred to as "XX information." When describing identification information, expressions such as "identification information," "identifier," "name," "ID," and "number" are used, and these are interchangeable.

[0016] When there are multiple components with the same or similar functions, they may be described using the same reference numeral with different subscripts. However, when there is no need to distinguish between these multiple components, the subscripts may be omitted.

[0017] Furthermore, in the following description, processing performed by executing a program may be described, but the program is executed by a processor (e.g., a CPU or a GPU (Graphics Processing Unit)) to perform the specified processing while appropriately using storage resources (e.g., memory) and / or interface devices (e.g., communication ports), and therefore the processor may be the subject of the processing. Similarly, the subject of the processing performed by executing a program may be a controller, device, system, computer, or node having a processor. The subject of the processing performed by executing a program may be any computing unit, and may include a dedicated circuit (e.g., an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit)) that performs specific processing.

[0018] A program may be installed on a device such as a computer from a program source. The program source may be, for example, a program distribution server or a computer-readable storage medium. If the program source is a program distribution server, the program distribution server may include a processor and storage resources for storing the program to be distributed, and the processor of the program distribution server may distribute the program to be distributed to other computers. Also, in the following description, two or more programs may be realized as one program, and one program may be realized as two or more programs.

[0019] <System configuration> FIG. 1 is a diagram showing an example of a system configuration including a water plant operation support system 100 according to this embodiment, and FIG. 2 is a diagram showing the functional configuration of the water plant operation support system 100 according to this embodiment.

[0020] 1 and 2 is a computer that can automatically determine appropriate alarm response priorities when multiple alarms are issued within the same time period. As shown in FIG. 1, water plant operation support system 100 of this embodiment is communicably connected to user terminal 200 via network 1. Therefore, these may be collectively referred to as water plant operation support system 100.

[0021] Water supply plant operation support system 100 of this embodiment is an information processing device that is connected via a network to an operation system already installed in a water supply plant, such as a water purification plant or a sewage treatment plant, and is used by an operator who operates the water supply plant. Water supply plant operation support system 100 of this embodiment is connected to user terminal 200, which is equipped as a user interface with a display for displaying screens and devices such as a keyboard and mouse for users to input information. Meanwhile, user terminal 200 is a terminal operated by the operator. This user terminal 200 is a device that displays information output from water supply plant operation support system 100. Specifically, a personal computer, a tablet terminal, a smartphone, etc. can be envisioned.

[0022] <Hardware configuration> As shown in the configuration examples of Figures 1 and 2, the water plant operation support system 100 of this embodiment includes an auxiliary memory device 101, a main memory device 103, a CPU 104 which serves as a computing device, a display interface 105, an input / output interface 106, and a network interface 107.

[0023] Of these, the auxiliary storage device 101 is composed of an appropriate non-volatile storage element such as an SSD (Solid State Drive) or a hard disk drive, while the main storage device 103 is composed of a volatile storage element such as a RAM (Random Access Memory), and the CPU (Central Processing Unit) 104 is a computing device that reads out a program 102 stored in the auxiliary storage device 101 into the main storage device 103 and executes it, thereby controlling the device itself and performing various judgments, calculations, and control processes.

[0024] The display interface 105 is an interface for displaying processing data between the user terminal 200. The input / output interface 106 is an interface for receiving key inputs and mouse operations from the user and returning processing results between the user terminal 200. The network interface 107 may be a network interface card or the like that is connected to an appropriate network 1 such as the Internet and handles communication processing with the user terminal 200 and other systems.

[0025] In addition to a program 102 for implementing the functions required for the water plant operation support system of this embodiment, the auxiliary storage device 101 also stores at least various databases and data such as alarm history data 120, operation history data 121, plant monitoring history data 122, a plant model 123, and an absolute priority alarm table 124. These will be described in detail later.

[0026] Furthermore, by executing the above-mentioned program 102, functionally, an alarm-operation correspondence extraction process 110, a chain alarm pattern extraction process 111, an operation-recovery model calculation process 112, a chain alarm time transition calculation process 113, a priority evaluation process 114, and a priority display process 115 are implemented.

[0027] <Data structure example> Various types of information used by the water plant operation support system 100 of this embodiment will be described.

[0028] 3 shows an example of the configuration of alarm history data 120 in this embodiment. The alarm history data in this embodiment stores logs related to alarm issuance obtained from a water plant. For example, the data is a table that shows the contents of the alarm, the time the alarm was issued (year, month, date, and time of the alarm), the time the alarm was restored (year, month, date, and time of the restoration), the name of the facility where the alarm was issued, and so on.

[0029] In Figure 3, the following information is stored in association with the alarm log: an ID for identifying the alarm; the date and time the alarm was issued; the date and time the alarm was restored after responding to the alarm; an alarm ID for identifying the alarm; the alarm name associated with the alarm; and the equipment for which the alarm was issued. For example, the alarm log identified by ID "10100" indicates that the alarm was issued at 8:03 AM on October 1, 2021, and that the alarm was restored at 9:34 AM on October 1, 2021, after responding to the alarm. The alarm is identified by alarm ID "2032," indicating that it was issued from "Sedimentation Tank 1." Alarms like these are sent from various equipment managed by this system and are accumulated as needed.

[0030] 4 shows an example of the configuration of operation history data 121 in this embodiment. The operation history data in this embodiment stores logs related to operations performed by operators in the operation of a water plant. For example, the data represents, as a table, the date and time of the operation, the operation content, the state of the plant due to the operation, the name of the equipment operated on, and so on.

[0031] In Figure 4, the ID for identifying the operation history log, the date and time of the operation, the operation content indicating the content of the operation, the equipment that was the target of the operation, and the state of the equipment at the time of the operation are stored in association with each other. For example, the operation identified by ID "20100" indicates that it was performed on equipment "Filtration Basin 1" at 8:06 on October 1, 2021. The content of the operation is "XX", indicating that the state was "XX abnormal" before the operation was performed. Operations like this are sent from various equipment managed by this system and are accumulated as needed.

[0032] 5 shows an example of the configuration of the plant monitoring history data 122 in this embodiment. The plant monitoring history data in this embodiment is a table showing various measurement values ​​acquired at regular intervals from plant monitoring measuring devices installed in a water plant.

[0033] In Figure 5, the date and time when the plant was monitored are stored in association with one or more pieces of equipment being monitored. For example, at 6:00:01 on October 1, 2021, the equipment "Intake Reservoir A," "Intake Reservoir B," "Intake Pump A Voltage," and "Intake Pump A Current" were measured as "8," "5," "50," and "10," respectively. These various measurement values ​​are accumulated every moment from the various pieces of equipment managed by this system.

[0034] An example of the plant model 123 in this embodiment is shown in Fig. 6. In the plant model of this embodiment, a water supply plant expert has previously modeled the entire water supply plant using a graph, by representing the plant equipment to be operated as nodes and the order of the equipment through which water flows as edges.

[0035] Figure 6 shows that the entire water plant, including multiple plant facilities (e.g., water intake tower 1, which is plant facility 601, and water intake reservoir 1, which is plant facility 602) and edges connecting each of the plant facilities (e.g., node 603), is stored as a plant model.

[0036] 7 shows an example of the configuration of the absolute priority alarm table 124 in this embodiment. The absolute priority alarm table in this embodiment is a list of alarms that should be addressed with the highest priority, prepared in advance by a water plant operation expert based on past operational know-how. Note that if there are no absolute priority alarms in the target water plant, the absolute priority alarm table may not be necessary.

[0037] In FIG. 7, an alarm ID for identifying an alarm and an alarm content indicating the specific content of the alarm are stored in association with each other.

[0038] <Flow example> The actual procedures of the water plant operation support system according to this embodiment will be described below with reference to the accompanying drawings. The various operations performed by the water plant operation support system described below are executed by programs that are read into a main memory or the like and executed by the water plant operation support system 100. These programs are composed of code for performing the various operations described below.

[0039] FIG. 8 is a diagram showing an example of the flow and functional configuration of the water plant operation support system according to this embodiment.

[0040] <Pre-operational processing> First, alarm and operation correspondence extraction processing 110 and chain alarm pattern extraction processing 111 of water plant operation support system 100 receive alarm history data 120, operation history data 121, and plant monitoring history data 122 as inputs, respectively.

[0041] In the alarm-operation correspondence extraction process 110, for each alarm that has been issued in the past, a statistical method is used to extract operations that have been performed with a statistically significant frequency, i.e., operations with a high probability of occurrence, from among the operations performed during the alarm issuance time of that alarm. First, the alarm ID is recorded in column 1250 and the operation name is recorded in column 1251 of the alarm-operation-recovery model pattern list 125 shown in Figure 9 (S801).

[0042] Fig. 9 is a diagram showing an example of the alarm-operation-recovery model pattern list 125. As shown in Fig. 9, the alarm-operation-recovery model pattern list 125 refers to the alarm history data 120 shown in Fig. 3 and the operation history data 121 shown in Fig. 4, and records, in association with an alarm, a group of operations with a certain or higher occurrence probability from among a plurality of groups of operations obtained by chronologically ordering the year, month, and date of the operation in the operation history data 121 included in the range from the alarm occurrence date and time to the recovery date and time in the alarm history data 120. For example, in Fig. 9, "operation 1," "operation 3," and "operation 5" are extracted and recorded in this order as a group of operations for alarm A.

[0043] Then, in the operation / recovery model calculation process 112, for each row of the alarm / operation / recovery model pattern list 125 shown in Fig. 9 recorded in S801, the plant model 123 shown in Fig. 6 is further referenced to calculate the distance between the equipment where the alarm occurred and the equipment operated by the operator. Furthermore, in the operation / recovery model calculation process 112, it is determined whether the calculated distance is greater than the distance in a certain set model, and if it is determined to be greater than the distance in the certain set model, it is rejected as an operation without a causal relationship, and the rejected operation is deleted from the operations in column 1251 (S802, see Fig. 10). S802 will be explained in detail below.

[0044] FIG. 10 is a conceptual diagram illustrating the distance between the equipment where an alarm occurred and the equipment operated by the operator on the model. As shown in FIG. 10, the operation / recovery model calculation process 112 refers to the plant model 123 and the operation history data 121 to identify the locations of the equipment where operations 1, 2, 3, and 5 were performed. Here, operations 1 and 3 are operations on equipment "chlorine tank 1," operation 5 is an operation on equipment "distributing tank 1" that is within a certain distance from equipment "chlorine tank 1," and operation 2 is an operation on equipment "distributing tank 3" that is a certain distance away from the equipment. Therefore, the operation / recovery model calculation process 112 determines (D1) that "operation 2," which is an operation on equipment "distributing tank 3," is an equipment that is a certain distance or more away from the equipment where the other operations were performed, and deletes "operation 2" from the operations recorded in the alarm / operation / recovery model pattern list 125.

[0045] Furthermore, Fig. 11 shows the behavior of the relevant plant monitoring history data 122 from each alarm to the time of subsequent operation and the alarm recovery after operation. In the operation / recovery model calculation process 112, the plant monitoring history data 122 corresponding to the recovery time T shown on the left side of Fig. 11 is mathematically formulated. Typical functions, such as linear functions, logarithmic functions, and exponential functions, are prepared in advance, and fitting is performed using the least squares method or the like to express the recovery time T, and the recovery time T is expressed by the function that best expresses it. As shown on the right side of Fig. 11, the expressed function is used as the recovery model.

[0046] For example, in FIG. 11, the operation and recovery model calculation process 112 plots function C, which is a time series plot of plant monitoring history data 122 for a certain piece of equipment where an alarm has occurred, and identifies an alarm region, which is a region above a certain threshold where an alarm is generated, in the monitoring data of the target plant. Then, a partial function C', which is a function in the identified alarm region, is extracted, and part C" of the extracted partial function C' at recovery time T, which is the time from the last operation X of a group of operations to alarm recovery, is fitted to one of the typical functions above, and the best-fitting function is determined as the recovery model. Once a recovery model for an alarm has been determined, it is recorded in column 1252 of FIG. 9. FIG. 9 shows that the best-fitting function for the time from the last operation "operation 5" in the group of operations "operation 1 → operation 3 → operation 5" for alarm A to alarm recovery is "function α." The operation and recovery model calculation process 112 performs this process for each alarm.

[0047] Returning to FIG. 8, in the chain alarm pattern extraction process 111, occurrence transition patterns of alarms that have been issued in the past are extracted using, for example, a Bayesian network, and those that appear with a certain degree of probabilistic significance are first recorded in column 1260 of the chain alarm transition time pattern list 126 shown in FIG. 12 (S803).

[0048] Fig. 12 is a diagram showing an example of the chain alarm transition time pattern list 126. As shown in Fig. 12, the chain alarm transition time pattern list 126 stores chain alarms that indicate a series of alarms that have transitioned over time, in association with statistical values ​​(Uave) of the transition times of the chain alarms. In Fig. 12, the chain alarm pattern extraction process 111 refers to the alarm history data 120 shown in Fig. 3, and, for example, identifies alarm B that appears next when alarm A appears, and records "A → B" that indicates the order in which these alarms appear as a chain alarm.

[0049] Furthermore, in the chain alarm time transition calculation process 113, for each row of the chain alarm pattern in the above column 1260, the distance between the equipment in which the first alarm (for example, alarm A) occurred and the equipment in which the later alarm (for example, alarm B) occurred is calculated using the plant model 123. Then, in the chain alarm time transition calculation process 113, if the calculated distance is greater than a distance on a certain set model, the chain alarm is rejected as having no causal relationship, and the rejected chain alarm is deleted from column 1260 in Fig. 12 (S804, see Fig. 13).

[0050] FIG. 13 is a conceptual diagram illustrating the distance between the equipment in which an alarm first occurred and the equipment in which an alarm subsequently occurred in the model. As shown in FIG. 13, the alarm chain time transition calculation process 113 identifies the locations of the equipment in which alarms A, B, C, and D occurred by referring to the plant model 123 and the alarm history data 120. Here, alarms A and B occurred in equipment "chlorine tank 1," alarm D occurred in equipment "distribution tank 1" within a certain distance from equipment "chlorine tank 1," and alarm C occurred in equipment "filtration tank 2" at a certain distance from the equipment in question. Therefore, the alarm chain time transition calculation process 113 makes a determination D2 that "alarm C," which occurred in equipment "filtration tank 2," is an equipment located at a certain distance or more from the equipment in which the other alarms occurred, and deletes the chain alarms recorded in the alarm chain transition time pattern list 126 that have "alarm C" downstream.

[0051] Returning to Fig. 8, in the chain alarm time transition calculation process 113, when all of the chain alarm patterns shown in column 1260 of the chain alarm transition time pattern list 126 shown in Fig. 12 are extracted from the alarm history data, an average value Uave of the difference between the occurrence time of the first alarm and the occurrence time of the last alarm, which is a statistical value between the occurrence time of the first alarm and the occurrence time of the last alarm, is calculated, and this is recorded as the alarm transition time in column 1261 of the alarm transition time pattern list 126 shown in Fig. 12 (S804). Note that, for example, if the alarm is related to water quality or water quantity, in addition to calculating the alarm transition time in the above process, the alarm transition time may also be calculated by performing a plant simulation.

[0052] <Processing during operation> 8, the alarm-operation-recovery model pattern list 125 and the chain alarm transition time pattern list 126 created in the pre-operation processing are used as inputs to perform the priority evaluation processing 114 (S805). A specific description will be given below.

[0053] In the priority evaluation process 114, for each alarm that is issued during operation, the currently issued alarm is first recorded in column 1270 of the priority evaluation result list 127 shown in FIG. 14. For example, alarms A to E are recorded in FIG. 14. Furthermore, in the priority evaluation process 114, for each alarm, the recovery model in the alarm-operation-recovery model pattern list 125, the current value of the monitoring data stored in the plant monitoring history data 122 (FIG. 5) corresponding to each currently issued alarm stored in the alarm history data 120 (FIG. 3), and the time of the last operation of a group of operations for the alarm recorded in the alarm-operation-recovery model pattern list 125 (FIG. 9) are used to calculate the alarm recovery time from the last operation (see the right side of FIG. 11), and record the recovery time Trec for each alarm in column 1273 of the priority evaluation result list 127 shown in FIG. 14. For example, it can be seen that the time until alarm A is recovered is 35 minutes.

[0054] Furthermore, the priority evaluation process 114 obtains the shortest Uave from the list of chain alarm transition time patterns 126 among the alarms that are predicted to chain immediately after each alarm currently being issued, and records this in column 1275. That is, in the priority evaluation process 114, if there are multiple combinations of a certain alarm (e.g., alarm A) and the next alarm (e.g., alarm B), the shortest Uave (e.g., "18") among these combinations is calculated and recorded in the above column 1275. This example shows that the shortest time from when alarm A occurs until alarm B occurs is 18 minutes.

[0055] Furthermore, the priority evaluation process 114 links all currently activated alarms with chain alarms from the chain alarm transition time pattern list 126. If there are multiple chain alarm groups, the chain alarm groups are numbered starting from 1, and the chain alarm group number to which each alarm belongs is entered in column 1271 of the priority evaluation result list 127 shown in FIG. 14. The hierarchy level, counting from the upstream of the chain alarm group to which each alarm belongs, is entered in column 1272 of the priority evaluation result list 127. Furthermore, for an activated alarm, the number of chain alarms predicted to occur thereafter is entered in column 1274 of the priority evaluation result list 127. Furthermore, if there is an activated alarm that corresponds to an alarm entered in the absolute priority alarm table 124 shown in FIG. 7, this is entered in column 1276 of the priority evaluation result list 127. In FIG. 14, alarm A is stored in the absolute priority alarm table 124 shown in FIG. 7, and therefore "T" indicating high priority is recorded in "Priority" in column 1276.

[0056] In FIG. 14 , alarms A, B, and C are a single group of alarms, so the same ID “001” is assigned to the “Chained Alarm Group” in column 1271. Since the chained alarm group is a group of alarms that are chronologically chained from “A to B” and “A to C,” the “Layer” in column 1272 records “1” for alarm A, indicating the highest level, and “2” for alarms B and C, indicating the next level. The recovery times Trec described above for alarms A, B, and C are recorded as “35,” “14,” and “22,” respectively. Furthermore, the “Number of Possible Chained Alarms” in column 1274 records “7,” “3,” and “2” alarms as future occurrences for alarms A, B, and C, respectively. This “Number of Possible Chained Alarms” records the number of alarms that may occur after the current alarm out of the total number of connected chained alarms recorded in the chained alarm transition time pattern list 126, as the number of alarms that may occur in the future. In the example of FIG. 14, when "alarm A" occurs, it is shown that of the alarms that are candidates to occur after the alarm, seven alarms including alarms B and C are likely to occur thereafter. Furthermore, it is shown that alarm B is most likely to occur 18 minutes after alarm A recorded in column 1275, which is the shortest time. As described above, the value in column 1275 is the smallest value (i.e., the shortest time from the previous alarm) among the values ​​in column 1261 of alarm transition time pattern list 126 shown in FIG. 12, and also shows that alarm A is a high-priority alarm. By performing the above-described processing, priority evaluation result list 127 shown in FIG. 14 is completed.

[0057] In the priority evaluation process 114, a comprehensive evaluation is then performed for each alarm, that is, for each row of the priority evaluation result list 127, using columns 1272 to 1276. As an example of a comprehensive evaluation method, the values ​​of each column are normalized so that they can be processed in the same column, and the weighted sum of the normalized values ​​is used as the comprehensive evaluation value for that alarm. Then, in the priority evaluation process 114, the comprehensive evaluation values ​​of each alarm are compared, and priorities are assigned in descending order of comprehensive evaluation value, i.e., the evaluation is highest, and an alarm evaluation list 128 shown in FIG. 15 is created. For example, in the priority evaluation process 114, the alarm with the largest weighted sum of the recovery times Trec of each alarm in column 1273, i.e., the alarm that takes the longest time to recover, and the alarm with the smallest weighted sum of the alarm transition time Uave in column 1275, i.e., the next alarm that occurs earliest, is assigned the priority as the alarm that should be addressed with the highest priority.

[0058] At this time, even if the above priority rankings have been assigned in the priority evaluation process 114, if there is an alarm stored in the absolute priority alarm table 124 shown in Fig. 7, the priority of that alarm is reset as an alarm with an even higher priority than the highest alarm among the alarms to which priorities have been assigned. This makes it possible to respond to alarms in descending order of priority, taking into account the need for a response to alarms as determined by an expert based on past experience.

[0059] Fig. 15 is a diagram showing an example of the alarm evaluation list 128. As shown in Fig. 15, the alarm evaluation list 128 associates alarms with response priorities, which are the priorities corresponding to the alarms, and the associated alarms are sorted and stored in descending order of the response priority. Fig. 15 shows that the alarm that should be responded to with the highest priority is "Alarm A." In this way, in S805, the priority at which each alarm that constitutes a chain alarm should be responded to is set.

[0060] Then, in the priority display process 115, the data obtained in the process underway is output on the screen (S806). An example of the final screen configuration is shown in FIG.

[0061] Fig. 16 is a diagram showing an example of a screen displayed on the display interface 105 in the priority display processing 115. As shown in Fig. 16, the priority display processing 115 displays an area 1601 showing alarms shown in the alarm evaluation list 128 created in S805 and their response priorities, and an area 1602 displaying, when an alarm shown in the area 1601 is selected, the alarm, the time of subsequent operation, and information showing the behavior of the corresponding plant monitoring history data 122 after the operation until the alarm is recovered (for example, the information shown on the left in Fig. 11). In addition, an area 1603 displaying the priority evaluation result list 127 shown in Fig. 14 is displayed on the display of the user terminal 200.

[0062] Furthermore, the priority display process 115 determines the chain relationships of alarms currently being issued from the chain alarm transition time pattern list 126, identifies alarms predicted to occur in a chain, and displays these in a graph in an area 1604 on the display of the user terminal 200. FIG. 16 shows that, when the priority display process 115 determines the chain relationships of alarms included in the chain alarm transition time pattern list 126, there are two alarm groups, alarm group 1 and alarm group 2. It can be seen that, in alarm group 1, alarm A will occur first, followed by alarms B and C, followed by alarms L, M, N, and so on. The time (e.g., 15 minutes from now) of alarms predicted to occur in the future can be set by referring to column 1275 of the priority evaluation result list 127 shown in FIG. 14. Similarly, it can be seen that, in alarm group 2, alarms D, E, and so on are predicted to occur.

[0063] Additionally, for each active alarm, a predicted alarm recovery time may be displayed on the display of the user terminal 200, with the horizontal axis representing time and the vertical axis representing values ​​in the plant monitoring history data 122. For example, when a predicted future alarm (e.g., alarm L) is selected in area 1604, information indicating the behavior of that alarm, including the recovery time, may be displayed, as in area 1602 above.

[0064] As described above, in the water plant operation support system of this embodiment, as described with reference to FIGS. 8 to 15 etc., in water plant operation support system 100, which supports responses to alarms generated in a water plant by a computer having a processor and a memory, the processor generates alarm operation patterns (for example, alarm operation patterns in alarm-operation-recovery model pattern list 125 shown in FIG. 9) in which operations having a predetermined causal relationship with the alarm are associated in time series (for example, a relationship that the positional relationship between the facilities is a predetermined condition, and the facilities are within a certain distance range) based on alarm history data (for example, alarm history data 120 shown in FIG. 3) indicating a log related to alarms generated from facilities of the water plant, operation history data (for example, operation history data 121 shown in FIG. 4) indicating a log related to operations in the operation of the facilities, and a plant model (for example, plant model 123 shown in FIG. 6) in which the operations having a predetermined causal relationship with the alarm (for example, an operation on facilities within a certain distance range as a predetermined condition) are associated in time series. column 1251) is created and stored in the memory, and a chain alarm pattern (for example, column 1260 of the chain alarm transition time pattern list 126 shown in FIG. 12) that chronologically associates the alarms that have a predetermined causal relationship (for example, a relationship in which alarms are generated from equipment that is within a certain distance as a predetermined condition of the positional relationship between the equipment) based on the alarm history data, plant monitoring history data (for example, plant monitoring history data 122 shown in FIG. 5) that indicates a log related to measurement values ​​of the equipment, and the plant model is created and stored in the memory, and a priority evaluation result list (for example, priority evaluation result list 127 shown in FIG. 14) that indicates candidate alarms that may occur in the future in the chain alarm pattern is created using the alarm operation pattern and the chain alarm pattern and stored in the memory, and the priority order of alarms to be responded to is determined based on the created priority evaluation result list (for example, alarm evaluation list 128 shown in FIG. 15 is created). As a result, even if alarms that are close in time to each other are generated from multiple pieces of equipment, it is easy to know the order in which alarms should be responded to, and responses can be performed in an appropriate order to prevent accidents, etc.

[0065] 8, 14, 15, etc., the processor determines the priority of the alarms to be handled based on statistical values ​​of the time related to the occurrence or recovery of each alarm included in the priority evaluation result list (for example, values ​​obtained by normalizing the recovery time Trec of each alarm in column 1273 and the transition time Uave of each alarm in column 1275). This makes it possible to determine the priority of the alarms to be handled, taking into account the recovery time of the alarms and the transition time from the previous alarm to the next alarm.

[0066] 9, 10, etc., in creating the alarm operation pattern, the processor extracts from the operation history data operations that have a certain probability of occurrence or higher, and determines that operations on equipment that is less than a certain distance from the equipment on which the extracted operation was performed are operations that have the predetermined causal relationship, and records these operations in the alarm operation pattern. This makes it possible to identify operations that should be targeted by this system according to the distance between specific equipment.

[0067] 11 and other figures, the processor mathematically expresses the behavior of the plant monitoring data recorded in the log related to the alarm from after the operation on the alarm to alarm recovery, based on the alarm history data, the operation history data, and the alarm operation pattern, and defines, based on the mathematically expressed behavior of the plant monitoring data and a predetermined mathematical formula (for example, a typical function such as a linear function, a logarithmic function, or an exponential function), a mathematical formula that satisfies a certain condition (for example, approximation is equal to or greater than a predetermined threshold) from after the operation to the alarm recovery for the plant monitoring data, as a recovery model for the operation, and creates the alarm operation pattern by calculating the recovery time from after the operation to alarm recovery (for example, the recovery time Tave shown in column 1252) using the defined recovery model. This makes it possible to express the recovery model as a typical function, and to estimate the time from alarm occurrence to recovery using the time calculated by the recovery model.

[0068] 12, 13, etc., in creating the chain alarm pattern, the processor extracts alarms with a certain probability of occurrence from the alarm history data, determines that alarms that have occurred in equipment that is less than a certain distance from the equipment in which the extracted alarm occurred are alarms that have the predetermined causal relationship, and records these in the chain alarm pattern. This makes it possible to identify alarms that should be targeted by this system depending on the distance between specific pieces of equipment.

[0069] 12, 13, etc., the processor calculates a temporal statistical value (for example, the alarm transition time Uave shown in column 1275) based on the occurrence time of the earlier alarm and the occurrence time of the later alarm for the alarms having the predetermined causal relationship, and records the calculated statistical value as the alarm transition time in the chain alarm pattern. This makes it possible to estimate the time until the next alarm occurs based on the statistical transition times of the previous and next alarms that make up the chain alarm.

[0070] 14, 15, etc., the memory stores in advance an alarm list (for example, the absolute priority alarm table 124 shown in FIG. 7) that is a list of alarms that should be dealt with on a priority basis, and the processor determines the priority of alarms that should be dealt with based on the alarm list and the priority evaluation result list. This makes it possible to deal with alarms in a priority order that takes into account the knowledge of experts.

[0071] 16 and other figures, the processor outputs at least the priority evaluation result list (e.g., area 1603), the determined priorities of the alarms to be addressed (e.g., area 1601), and a graph showing candidate alarms that may occur in the chain alarm pattern (e.g., area 1604) to a display device connected to the computer. This allows the user to grasp at a glance the priorities of the alarms to be addressed, and makes it easy to predict how much time remains until a future alarm occurs.

[0072] In other words, the water supply plant operation support system of this embodiment makes it possible to automatically determine appropriate alarm response priorities when multiple alarms are issued within the same time period. Furthermore, if the priorities for responding to alarms cannot be determined appropriately, for example, it takes time to appropriately set the water distribution volume and chlorine concentration to the water reservoir, requiring extra power and resources for water management. However, with this system, such problems do not occur, and water management can be performed with minimal power and resources without placing a burden on the environment.

[0073] Although the embodiments for carrying out the present invention have been specifically described, the present invention is not limited to these and can be modified in various ways without departing from the spirit of the present invention. [Explanation of symbols]

[0074] 100 Water Plant Operation Support System 200 user terminals 110 Alarm and operation response extraction processing (program) 111 Chain alarm pattern extraction process (program) 112 Operation and recovery model calculation processing (program) 113 Chain alarm time transition calculation process (program) 114 Priority evaluation process (program) 115 Priority display processing (program) 120 Alarm history data 121 Operation history data 122 Plant monitoring history data 123 Plant Model 124 Absolute Priority Alarm Table 125 Alarm, Operation, and Recovery Model Pattern List 126 List of Chain Alarm Transition Time Patterns 127 List of Priority Evaluation Results 128 Alarm Rating List

Claims

1. A water plant operation support system that supports responses to alarms generated in a water plant by a computer having a processor and a memory, The processor: based on alarm history data indicating a log related to alarms generated from equipment in the water plant, operation history data indicating a log related to operations in the operation of the equipment, and a plant model that models the water plant including the equipment, an alarm operation pattern is created in which operations having a predetermined causal relationship with the alarm are associated in chronological order, and the created alarm operation pattern is stored in the memory; creating a chain alarm pattern that associates the alarms having a predetermined causal relationship in time series based on the alarm history data, plant monitoring history data indicating a log related to measurement values ​​in the facility, and the plant model, and storing the pattern in the memory; a priority evaluation result list indicating candidates for alarms that may occur in the future in the chain alarm pattern is created using the alarm operation pattern and the chain alarm pattern, the list is stored in the memory, and the priority order of alarms to be addressed is determined based on the created priority evaluation result list; A water plant operation support system characterized by:

2. The processor: determining the priority of alarms to be addressed based on statistical values ​​of the time taken for each alarm to occur or recover from the alarms included in the priority evaluation result list; 2. The water plant operation support system according to claim 1.

3. The processor: In creating the alarm operation pattern, operations having a certain probability of occurrence or higher are extracted from the operation history data, and operations on equipment that are less than a certain distance from the equipment on which the extracted operation was performed are determined to be operations having the predetermined causal relationship, and are recorded in the alarm operation pattern.

2. The water plant operation support system according to claim 1.

4. The processor: Based on the alarm history data, the operation history data, and the alarm operation pattern, the behavior of the plant monitoring data recorded in the log related to the alarm from after the operation on the alarm until the alarm is recovered is mathematically formulated, and based on the mathematically formulated behavior of the plant monitoring data and a predetermined mathematical formula, a mathematical formula that satisfies certain conditions for the plant monitoring data from after the operation until the alarm is recovered is defined as a recovery model for the operation, and the defined recovery model is used to create the alarm operation pattern in which a recovery time from after the operation until the alarm is recovered is calculated.

2. The water plant operation support system according to claim 1.

5. The processor: In creating the chain alarm pattern, alarms with a certain probability of occurrence or higher are extracted from the alarm history data, and alarms that have occurred in equipment that is less than a certain distance from the equipment in which the extracted alarm occurred are determined to be alarms having the predetermined causal relationship, and are recorded in the chain alarm pattern.

2. The water plant operation support system according to claim 1.

6. The processor: For the alarms having the predetermined causal relationship, a temporal statistical value is calculated based on the occurrence time of the earlier alarm and the occurrence time of the later alarm, and the calculated statistical value is recorded in the chain alarm pattern as an alarm transition time.

6. The water plant operation support system according to claim 5.

7. The memory includes: An alarm list, which is a list of alarms that should be dealt with on a priority basis, is stored in advance. The processor: determining the priority of alarms to be addressed based on the alarm list and the priority evaluation result list; 2. The water plant operation support system according to claim 1.

8. The processor: outputting at least the priority evaluation result list, the determined priorities of the alarms to be dealt with, and a graph showing candidates for alarms that may occur in the future in the chain alarm pattern to a display device connected to the computer; 2. The water plant operation support system according to claim 1.

9. A water plant operation support method for supporting a response to an alarm that has occurred in a water plant, the method being carried out by a computer having a processor and a memory, comprising: based on alarm history data indicating a log related to alarms generated from equipment in the water plant, operation history data indicating a log related to operations in the operation of the equipment, and a plant model that models the water plant including the equipment, an alarm operation pattern is created in which operations having a predetermined causal relationship with the alarm are associated in chronological order, and the created alarm operation pattern is stored in the memory; creating a chain alarm pattern that associates the alarms having a predetermined causal relationship in time series based on the alarm history data, plant monitoring history data indicating a log related to measurement values ​​in the facility, and the plant model, and storing the pattern in the memory; a priority evaluation result list indicating candidates for alarms that may occur in the future in the chain alarm pattern is created using the alarm operation pattern and the chain alarm pattern, the list is stored in the memory, and the priority order of alarms to be addressed is determined based on the created priority evaluation result list; A water plant operation support method comprising:

Images