Operation management method and operation management device

Abstract

This operation management method for managing the operation of a system to be managed comprises: acquiring, from a system to be managed, time-series data of each of N types of quantities relating to the operation and state of the system, where N is an integer at least equal to 2 (steps 101, 106); calculating the average value of each set of time-series data for each of a series of averaging periods (steps 102, 107); and plotting the average values for each averaging period in an N-dimensional space (discriminant space) with each of the N types of quantities as a coordinate axis, thereby generating scattered data (steps 103, 108).

Description

Operation management method and operation management device

[0001] The present invention relates to an operation management method and an operation management device for monitoring the status of various systems, such as water treatment systems for producing pure water or ultrapure water, and managing the operation of these systems.

[0002] Water treatment systems, such as pure water production systems that produce pure water or ultrapure water from raw water and wastewater treatment systems that treat wastewater, are composed of a combination of various devices, such as filtration systems, activated carbon devices, ion exchange devices, reverse osmosis membrane devices, ultraviolet oxidation devices, coagulation sedimentation devices, and electrodeionization (EDI) devices, as well as piping, heat exchangers, pumps, and other components. Among these devices, activated carbon, which is packed into activated carbon devices, is a functional material that needs to be replaced timely. Similarly, ion exchange resins used in ion exchange devices are also functional materials that need to be regenerated or replaced timely. Such functional materials are replaced based on, for example, operating time or treated water volume. However, replacing functional materials based on operating time or treated water volume can result in replacing functional materials that are still usable, necessitating the optimization of the replacement timing of functional materials. To optimize the replacement timing of functional materials, for example, it is possible to measure the water quality of treated water and analyze fluctuations in the water quality data to evaluate the performance of the functional materials and determine whether they have deteriorated. Furthermore, in water treatment systems, malfunctions can occur in various devices, including pumps, and there is a need to detect such abnormalities early and identify the cause of the malfunction.

[0003] In a water treatment system, evaluating the performance of functional materials based on actual measured data, determining when to replace functional materials, detecting abnormalities and identifying the causes of the failures, etc. are all part of managing and optimizing the operation of the water treatment system. In the following explanation, optimizing operation is also considered to be part of managing operation.

[0004] In the past, analyzing water quality data required skilled techniques to evaluate the performance of functional materials based on water quality data, determine when to replace functional materials, detect abnormalities, and identify the causes of failures. Therefore, various methods have been proposed to make it easier to manage the operation of water treatment systems.

[0005] Patent Document 1 discloses a method for evaluating the performance of anion exchange resins packed in ion exchange resin towers in a water treatment system, in which a mass transfer coefficient of the anion exchange resin relative to inorganic carbonate is calculated from measured values ​​of inorganic carbonate concentrations in the inlet water and the outlet water of the ion exchange resin tower, and the dynamic characteristics and degree of deterioration of the anion exchange resin are evaluated based on this mass transfer coefficient. Patent Document 2 discloses a method for evaluating the performance of an ion exchange resin by mixing and contacting a salt aqueous solution of a predetermined concentration with a predetermined amount of ion exchange resin, and measuring the change over time in the electrical conductivity of the salt aqueous solution after mixing and contact. Patent Document 3 discloses a water treatment system having a plurality of ion exchange resin towers of the same type and treatment capacity installed in parallel, in which the same water to be treated is passed through each ion exchange resin tower to obtain treated water. The system applies a higher load to some ion exchange resin towers than to other ion exchange resin towers, and when deterioration in the water quality of the treated water from the highly loaded ion exchange resin tower occurs, the system predicts the lifespan of the other ion exchange resin towers based on the cumulative load at that time, and subsequently controls the amount and time of water passing through each ion exchange resin tower based on the prediction result.

[0006] Patent Documents 4 to 8 disclose that when an ion exchange device is operated to repeatedly cycle between a water flow process and a regeneration process, the water quality of the raw water is measured, a prediction is made based on the measured values, and switching from the water flow process to the regeneration process is performed based on the prediction. Patent Document 9 discloses that in a moving-bed ion exchange device in which ion exchange resin moves between a water flow tower and a regeneration tower, the amount of water flowing through the water flow tower, the conductivity of the raw water, and the silica concentration of the ion-exchanged water are detected, the amount of regenerating agent to be injected into the regeneration tower is calculated from the water flow rate and conductivity, the calculated value is corrected by the silica concentration, and the amount of regenerating agent injected is adjusted based on the corrected calculated value.

[0007] Systems other than water treatment systems are also required to reliably and easily detect the occurrence of abnormalities in those systems. Patent Document 10, which addresses this problem, discloses a method for diagnosing abnormalities in equipment that operates in a strictly repeating pattern, such as a sizing press, a type of continuous rolling equipment, by sampling time-series data indicating the state of the equipment, which repeats a pattern corresponding to the specific pattern, and diagnosing the equipment abnormality based on the sampled time-series data. In the equipment abnormality diagnosis, a first principal component analysis is performed on one pattern in the sampled time-series data to determine whether or not there is a first abnormality. If the first determination indicates an abnormality, a second principal component analysis is performed on multiple temporally consecutive patterns, including the pattern subjected to the first principal component analysis, to determine whether or not there is a second abnormality. If the second determination indicates an abnormality, the equipment abnormality is determined to exist.

[0008] Japanese Patent Application Laid-Open No. 2002-48776 Japanese Patent Application Laid-Open No. 2015-13276 Japanese Patent Application Laid-Open No. 2018-111058 Japanese Patent Application Laid-Open No. 2023-34710 Japanese Patent Application Laid-Open No. 2014-188456 Japanese Patent Application Laid-Open No. 2021-3675 Japanese Patent Application Laid-Open No. 5-277382 Japanese Patent Application Laid-Open No. 6-55082 Japanese Patent Application Laid-Open No. 59-31377 International Publication No. 2021 / 145156

[0009] Although many proposals have been made regarding methods for predicting the timing of ion exchange resin regeneration or replacement in ion exchange units for managing the operation of water treatment systems, few proposals have been made regarding methods for easily detecting abnormalities and identifying the cause of the failure. The abnormality diagnosis method described in Patent Document 10 is intended for equipment that operates in a strictly fixed cycle, and is therefore difficult to apply to water treatment systems in which, for example, the operating conditions change depending on the quality of the raw water, and the timing of ion exchange resin regeneration varies accordingly. Some water treatment systems do not have a clear cycle in their operation, such as a cycle consisting of a water flow process and a regeneration process, and the technique described in Patent Document 10 is also difficult to apply to such water treatment systems. Furthermore, while water quality data obtained from a water treatment system can also be considered time-series data, when processing time-series data obtained from a managed system to determine the timing of functional material replacement or detect the occurrence of an abnormality in the system, the accuracy of the determination or detection may be reduced due to variations in values ​​in the time-series data due to factors such as hunting.

[0010] The object of the present invention is to provide an operation management method and an operation management device that can accurately and easily manage the operation of a system, regardless of the system being managed, based on time-series data obtained from the system regarding the operation and status of the system.

[0011] Time series data from a managed system regarding a quantity x related to the system's operation and status may be discrete with respect to time t, but can be treated as a function x = f(t) with time t as the independent variable and quantity x as the dependent variable. Traditionally, when attempting to detect the occurrence of an abnormality in a system using time series data, analysis has been performed by detecting changes in the value of the function f(t), which explicitly includes time t. However, time series data obtained from a managed system can contain effects such as hunting, which can reduce the accuracy of the analysis. Furthermore, analyzing the time series data itself often requires skilled techniques.

[0012] Therefore, in the present invention, multiple time-series data x1 = f1(t), x2 = f2(t), ..., each of which is time-series data for different types of quantities, are acquired from the managed system, and a dimension reduction technique is used to reduce the dimension of time t. The values ​​x1, x2, ... of each time-series data at the same time t are plotted as points P(x1, x2, ...) in a space whose coordinate axes are the quantities represented by each time-series data, generating scatter data. The space in which the scatter data is generated is called a discriminant space. In practice, it is difficult to measure each quantity at exactly the same time in the managed system, and it is also desirable to reduce the effects of hunting, etc. Therefore, an averaging period is set, and averages m1, m2, ... of the time-series data x1 = f1(t), x2 = f2(t), ... are calculated for each averaging period, and points P(m1, m2, ...) are placed in the discriminant space to generate scatter data. The averaging period can be appropriately set depending on the operating mode of the managed system, as long as it is common to multiple types of time-series data.

[0013] Therefore, one aspect of the operation management method of the present invention is an operation management method for managing the operation of a system to be managed, which acquires time series data for each of N types of quantities related to the operation and status of the system from the system, where N is an integer of 2 or greater, calculates the average value of each of the time series data for each of a series of averaging periods defined in the system, and generates scatter data by plotting the average values ​​for each averaging period in an N-dimensional space, i.e., a discriminant space, with each of the N types of quantities as its coordinate axis.

[0014] Once the scatter data is generated in this manner, the occurrence of an abnormality in the managed system can be detected from the arrangement of each plot point constituting the scatter data in the discriminant space. Because points are plotted in the discriminant space for each of a series of averaging periods, changes in the status of the managed system can be identified by tracking how the positions of the plotted points in the discriminant space change over time. In particular, when N is 2 or 3, the scatter data is represented in two-dimensional or three-dimensional space, allowing the arrangement of each plot point in the scatter data to be visually confirmed, making it easy to identify the occurrence of an abnormality in the managed system. Even when N is 4 or greater, the occurrence of an abnormality can be detected by determining whether a plot point exists in a specific region in the discriminant space. Furthermore, the accuracy of detecting an abnormality and identifying the location of a fault can be improved by using, for example, machine learning techniques.

[0015] Any data continuously obtained from the system to be managed can be used as time-series data. When the system to be managed is a water treatment system, data such as flow rate, temperature, pressure, pH, concentration of a specific component, conductivity (or resistivity), and turbidity can be used as time-series data. The time-series data can be obtained directly from a sensor installed in the system to be managed, or can be obtained by performing calculations on data obtained from the sensor. For example, when an absorbance sensor is installed, the concentration of a specific component calculated from absorbance can be used as time-series data. When two pressure sensors are installed, the differential pressure, which is the difference between the measurements of the two sensors, can be used as time-series data.

[0016] Any average value may be used as long as it can represent the data values ​​within the averaging period. For example, the average value may be the center of gravity of the data values ​​within the averaging period, i.e., the arithmetic mean value, the geometric mean value, or the harmonic mean value. It may also be an average value calculated by taking the logarithm of each data value and then calculating the logarithm of the logarithm.

[0017] An operation management device according to one embodiment of the present invention is an operation management device used to manage the operation of a system to be managed, and comprises: an average value calculation means for acquiring time series data for each of N types of quantities related to the operation and status of the system from the system, where N is an integer of 2 or greater, and calculating an average value for each of the time series data for each of a series of averaging periods defined in the system; and a scatter data generation means for generating scatter data by plotting the average values ​​for each averaging period in an N-dimensional space with each of the N types of quantities as its coordinate axis.

[0018] According to the present invention, whatever the system to be managed may be, the operation of that system can be accurately and easily managed based on time-series data obtained from that system regarding the operation and status of that system.

[0019] FIG. 1 is a flowchart illustrating an operation management method according to an embodiment of the present invention. FIG. 2 is a diagram illustrating the principle of the operation management method. FIG. 3 is a diagram illustrating an example of the configuration of an operation management device. FIG. 4 is a flow chart illustrating an example of the configuration of a water treatment system. FIG. 5 is a diagram illustrating an example of analysis based on time series data from the water treatment system. FIG. 6 is a diagram illustrating an example of analysis based on time series data from the water treatment system. FIG. 7 is a diagram illustrating an example of analysis based on time series data from the water treatment system.

[0020] Next, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a flowchart illustrating an operation management method according to an embodiment of the present invention. Fig. 2 explains the principle of the operation management method by taking as an example a case where the system to be managed is a water treatment system and two types of time-series data, namely, water quality values ​​(e.g., resistivity values) and differential pressure, are obtained from the water treatment system. In the following description, the system to be managed will be referred to as the target system.

[0021] The operation management method of this embodiment is a method for determining the replacement timing of functional materials in a target system, detecting the occurrence of abnormalities, and identifying fault locations, based on time series data obtained from the target system. As shown in Fig. 1, this operation management method first acquires time series data for each of multiple types of quantities related to the operation and status of the system from the target system in step 101. Let N be the number of types of time series data to be acquired, where N is an integer equal to or greater than 2. In the example shown in Fig. 2, two types of time series data, water quality values ​​and differential pressure, are acquired from the target system, so N=2.

[0022] Next, in step 102, the average value of each time series data within the averaging period is calculated. As mentioned above, the averaging period is also called a cycle. The averaging period must be a common period for multiple time series data, but the number of data points included in each time series data for each averaging period may differ for each time series data. The averaging period can be set as appropriate, but is preferably determined according to the operating mode of the target system. For example, if the target system includes an ion exchange device that alternately performs a water flow process and a regeneration process, one water flow process of the ion exchange device can be used as one averaging period. Of course, one water flow process can also be divided into multiple averaging periods. The lengths of the averaging periods may be different from one another. Any calculation method can be used to calculate the average value, such as the arithmetic mean, geometric mean, or harmonic mean. However, typically, the arithmetic mean value (i.e., the center of gravity value) of the data values ​​of the time series data within the averaging period is used as the average value. In FIG. 2, reference numeral 2A denotes the time series data for differential pressure and the time series data for water quality values ​​during the first averaging period as measurement data for the first cycle. In the graph of each time series data shown as the measurement data 2A of the first cycle, the center of gravity value G (i.e., the arithmetic mean value) calculated from each time series data is also shown by a double circle ("◎").

[0023] Next, in step 103, the average values ​​calculated in step 102 are plotted in a discriminant space, whose coordinate axes represent the quantities of the time-series data, to obtain scatter data. Because the number of types of time-series data is N (where N is ≥ 2), the discriminant space is an N-dimensional space. In the example shown in FIG. 2, N = 2, so the discriminant space is a two-dimensional space. Being two-dimensional, the scatter data is also called a scatter plot. In FIG. 2, the discriminant space obtained from the two time-series data sets, the measurement data 2A from the first cycle, is shown as a scatter plot 2B. In this case, the time-series data consists of two types of data: water quality values ​​and differential pressure. Therefore, one coordinate axis (the horizontal axis in the figure) of the discriminant space represents the differential pressure, and the other coordinate axis (the vertical axis in the figure) represents the water quality value. Then, as shown by [1] in the figure, points are plotted in the discriminant space at positions represented by the average differential pressure and the average water quality values, to obtain a scatter plot 2B.

[0024] Once the scatter data (or scatter diagram) is obtained in step 103, the occurrence of an abnormality in the system being managed is detected based on the scatter data in step 104. The detection of an abnormality here includes determining whether the functional materials used in the target system need to be replaced. For example, if the positions of the plotted points in the scatter data are more than a certain distance away from their expected normal positions, it can be determined that an abnormality has occurred in the target system. Furthermore, depending on the direction in which the positions of the plotted points in the discriminant space deviate from their expected normal positions, it is also possible to identify the location of a fault or the cause of the fault. If the only purpose is to obtain scatter data, step 104 may be skipped and step 105 may be skipped and step 104 may be skipped and step 105 may be executed directly. If the number of plotted points in the scatter data is small, step 104 may be skipped and step 105 may be executed directly.

[0025] In step 105, it is determined whether or not the same processing as that described in steps 101 to 104 will be performed during the next averaging period. If processing will not be performed during the next averaging period, the processing of the operation management method ends here. On the other hand, if processing will be performed during the next averaging period, in step 106, as in step 101, each time-series data item during the next averaging period is acquired, and in step 107, as in step 102, an average value during the averaging period is calculated for each time-series data item. In FIG. 2, each time-series data item during the second averaging period (i.e., the second cycle) and its corresponding average value (i.e., the center of gravity value G) are shown as measurement data 2C for the second cycle. Once the average value has been calculated, in step 108, as in step 103, the average value calculated in step 107 is plotted in the discriminant space. In this case, since there is scatter data in which the average values ​​calculated during the previous averaging periods have already been plotted in the discriminant space, plot points are added to the scatter data, and the scatter data is updated. Scatter diagram 2D ​​is obtained by adding plot points based on the average values ​​calculated from measurement data 2C in the second cycle to scatter diagram 2B. The added plot points are indicated by [2]. After step 108 is completed, the process from step 104 is repeated.

[0026] Repeating steps 104 through 108 increases the number of plot points included in the scatter data in the discriminant space. That is, the average value of each time-series data is calculated for each averaging period, and the average value for each averaging period is plotted in the discriminant space, generating scatter data containing a large number of plot points. In the case shown in FIG. 2 , the scatter data, i.e., the scatter diagram obtained by repeating steps 104 through 108 is shown as scatter diagram 2E. In scatter diagram 2E, the horizontal axis represents differential pressure and the vertical axis represents water quality values. When the plot points are positioned away from the normal position, it can be assumed that an abnormality has occurred in the target system. As will be described later, it is not necessary to determine in advance the position in the scatter data corresponding to a normal system. If the plot points are dispersed beyond a certain range, this in itself suggests the occurrence of an abnormality in the target system.

[0027] As the target system deteriorates or an abnormality occurs, the position of the plotted points on the scatter diagram may gradually shift from their normal positions. In the example shown in Figure 2, the position of the plotted points on scatter diagram 2E gradually shifts from the normal plotted points toward the lower right of the diagram. If we call this a trend of event A, event A, which involves a small increase in differential pressure but a deterioration in water quality, can be considered to correspond to, for example, a decrease in the reactivity of the ion exchange resin due to the adsorption of impurity ions onto the ion exchange resin, aging of the ion exchange resin due to use, detachment of ion exchange groups due to oxidizing substances, or poor regeneration. If the occurrence of event A is observed, it can be determined that the time for replacement or regeneration of the ion exchange resin is approaching. Furthermore, the position of the plotted points may shift toward the right of the diagram from the normal plotted points. If we call this a trend of Event B, then Event B, which involves an increase in the differential pressure across the water flow despite a small change in the water quality value, can be considered to correspond to, for example, clogging of the ion exchange resin layer due to fracture of the ion exchange resin, clogging due to swelling and shrinkage of the ion exchange resin, or clogging due to the inflow of turbidity from pretreatment equipment, etc. In other words, by detecting the deviation of the plotted points from their normal positions, we can detect the occurrence of an abnormality in the target system and identify its cause.

[0028] If the target system includes an ion exchange device that repeatedly performs a water flow process and a regeneration process, the entire water flow process between the regeneration processes can be used as the averaging period, as described above. However, the interval between regeneration processes in an ion exchange device can be as short as 24 hours or as long as 2-3 weeks or more. If a single water flow process is used as an averaging period, the averaging period would be approximately one day to several weeks. The operational management method of this embodiment detects the occurrence of an abnormality in the target system based on the average value calculated for each averaging period and its trend. The longer the averaging period, the longer the time required to determine whether an abnormality has occurred. Therefore, by dividing a single water flow process into multiple sections and using each divided section as an averaging period, the time required to determine whether an abnormality has occurred can be shortened. The criteria and methods for dividing the process into multiple sections can be time, water flow rate, other related parameter values, or any combination thereof. Other related parameters include, for example, time-series data such as pressure and conductivity. For example, if the interval between regeneration steps is 24 hours and this is divided into five so that one water flow step consists of five averaging periods, then when determining whether or not an abnormality has occurred based on the average value obtained in one averaging period, the determination can be made in just over four hours at the earliest, and even when determining based on the trend in the average values ​​over two or three averaging periods, the determination can be made in about 10 hours.Even if the interval between regeneration steps is longer than 24 hours, by setting the length of each averaging period to three hours and repeating the averaging period in one water flow step, it is possible to determine whether or not an abnormality has occurred in as little as three hours.

[0029] As described above, the operation management method of this embodiment makes it possible to automatically determine whether a target system is operating normally or if an abnormality has occurred, and to analyze the cause of the failure. In particular, by tracking changes in the positions of plot points in scatter data over a long period of time, it is possible to automatically detect the occurrence of an abnormality, without the need to prepare training data corresponding to normal conditions in advance. This facilitates predictive analysis using techniques such as machine learning. The method of this embodiment does not interpret individual time series data as quantitative or functional relationships, but summarizes the data by removing the time dimension from each time series data, allowing multiple time series data to be combined and understood graphically. This makes it possible to eliminate the effects of hunting and other phenomena, and to easily identify not only the presence or absence of an abnormality but also the cause of the failure from the positions and transitions of plot points in the discriminant space.

[0030] 3 shows an example of the configuration of an operation management device 10 that implements the above-described operation management method. As described above, if the system to be managed is referred to as a target system, then this operation management device 10 is used to manage the operation of the target system 20. The target system 20 is provided with multiple sensors 21. The sensors 21 acquire time series data on quantities related to the operation and status of the target system 20. Different sensors 21 acquire different types of time series data, and input this time series data into the operation management device 10.

[0031] The operation management device 10 includes an average value calculation unit 11 that acquires multiple types of time-series data from the target system 20 and calculates an average value for each of the time-series data for each averaging period, and a scatter data creation unit 13 that plots the average values ​​for each averaging period in a discriminant space to generate scatter data. The average value calculation unit 11 includes a calculation unit 12 provided for each piece of input time-series data, and the calculation unit 12 calculates an average value for the corresponding time-series data for each averaging period. The generated scatter data is output from the scatter data creation unit 13. An administrator operating the target system 20 can determine whether or not an abnormality has occurred in the target system 20, including the arrival of the time to replace functional materials, by observing and analyzing the scatter data, such as a scatter diagram, output from the scatter data creation unit 13.

[0032] The operation management device 10 may further include a discrimination unit 14 that discriminates whether an abnormality has occurred in the target system 20 based on the arrangement of plot points in the scatter data in the discriminant space. The discrimination unit 14 outputs a discrimination result indicating whether an abnormality has occurred. The discrimination unit 14 can determine whether a plot point exists in a specific region in the discriminant space to determine whether an abnormality has occurred in the target system 20 and identify the cause of the failure. The discrimination unit 14 can also detect signs of an abnormality in the target system 20 based on the trend of how the positions of the plot points change over a series of multiple averaging periods in the discriminant space. Furthermore, by incorporating a machine learning engine that performs machine learning into the discrimination unit 14, the accuracy of detecting an abnormality and identifying the location of a failure can be improved. When N is 4 or more, i.e., the number of dimensions of the discriminant space is 4 or more, it is difficult for a person to visually observe and analyze the scatter data. However, by using the discrimination unit 14, it is possible to easily detect an abnormality in the target system 20 and identify the cause of the failure, even when N is 4 or more.

[0033] Next, we will explain an example of applying the above-mentioned operation management method to an actual system. Here, we will explain the case where the target system is a water treatment system that produces pure water from raw water such as city water or well water. Figure 4 shows an example of the configuration of a water treatment system.

[0034] A storage tank 51 is provided for storing raw water, and the raw water in the storage tank 51 is pumped by a pump 52, and solids and the like are removed by a filter 53. The outlet water of the filter 53 is stored in a storage tank 54. Recovered water recovered from devices and facilities that use pure water is also supplied to the storage tank 54. The water in the storage tank 54 is pumped by a pump 55 to an activated carbon tower 56 for activated carbon treatment. The outlet water of the activated carbon tower 56 is supplied to an ion exchange device 60. The ion exchange device 60 includes a first cation exchange resin tower 61 filled with a cation exchange resin and supplied with the outlet water of the activated carbon tower 56, a decarbonation tower 62 supplied with the outlet water of the first cation exchange resin tower 61, an anion exchange resin tower 63 filled with anion exchange resin and supplied with the outlet water of the decarbonation tower 62, and a second cation exchange resin tower 64 filled with a cation exchange resin and supplied with the outlet water of the anion exchange resin tower 63. The outlet water of the second cation exchange resin tower 64 is the outlet water of the ion exchange device 60. As a result of the ion exchange treatment in the first cation exchange resin tower 61, the outlet water of the first cation exchange resin tower 61 contains hydrogen ions (H ＋ ) and becomes acidic, and the carbonate ions (CO3 2- ) and bicarbonate ions (HCO3 - ) is converted into free carbon dioxide. The decarbonation tower 62 is provided to remove this free carbon dioxide by blowing in air or by membrane deaeration. As a result of providing the decarbonation tower 62, water that does not contain carbon dioxide components is supplied to the anion exchange resin tower 63.

[0035] The ion exchange device 60 further includes a flow meter (FI) 65 provided at its inlet for measuring the flow rate of water supplied to the ion exchange device 60, i.e., the flow rate of treated water in the ion exchange device 60, a conductivity meter (CI) 66 for measuring the conductivity of the inlet water of the first cation exchange resin tower 61, a pressure meter (PI) 67 for measuring the pressure at the inlet of the anion exchange resin tower 63, a pressure meter (PI) 68 for measuring the pressure at the outlet of the anion exchange resin tower 63, and a resistivity meter (RI) 69 for measuring the resistivity of the outlet water of the second cation exchange resin tower 64. The outlet water of the ion exchange device 60 is fed by a pump 71 to a reverse osmosis membrane device 72. Pure water is discharged from the reverse osmosis membrane device 72.

[0036] The flowmeter 65, conductivity meter 66, pressure gauges 67 and 68, and resistivity meter 69 are all sensors that output time-series data related to the operation and status of the water treatment system. The difference between the pressure measurements of pressure gauges 67 and 68 indicates the differential pressure across the anion exchange resin tower 63, and this differential pressure also constitutes time-series data related to the operation and status of the water treatment system. Figure 5A shows time-series data obtained by measuring the quality of the outlet water from the second cation exchange resin tower 64 using the resistivity meter 69 during long-term continuous operation of the water treatment system. The time-series data shown here covers the period from March 3 to March 8. The periods labeled "Water Flow" and "Regeneration" indicate whether the ion exchange device 60 was undergoing the water flow process or the regeneration process, respectively. The water treatment system operated normally until around noon on March 7, but the water quality values ​​were unstable after that due to a malfunction.

[0037] In the past, assuming that time-series data such as that shown in Figure 5A were obtained, an abnormality in a water treatment system was detected by, for example, comparing the data values ​​with a threshold value. Figure 5B shows an example of applying a threshold value to the time-series data shown in Figure 5A. To avoid false detection of hunting in the measured values, it is necessary to set the threshold value to a value that is significantly different from the normal value. However, in this case, there is a risk that an abnormality that brings about a small change in the time-series data will be overlooked. Furthermore, it is difficult to identify the cause of the failure simply by comparing with the threshold value.

[0038] One conventional anomaly detection method is to use the gradient of change in time-series data. Figure 5C shows an example of applying anomaly detection based on the gradient of change to the time-series data shown in Figure 5A. In anomaly detection based on the gradient of change, a threshold value for the gradient is set, and an anomaly is determined to have occurred when the gradient of change in the time-series data exceeds the threshold value. However, this method often results in false positives, and it is difficult to identify the cause of the failure.

[0039] FIG. 6 shows an example in which the operation management method of this embodiment is applied to the time-series data shown in FIG. 5A. The operation management method of this embodiment requires at least two types of time-series data. In this example, the time-series data includes time-series data 6A of the water quality values ​​of the outlet water from the second cation exchange resin tower 64, as well as time-series data of the treated water flow rate measured by a flow meter (not shown). Scatter data 6B is obtained from these two types of time-series data. Because there are two types of time-series data, the scatter data 6B is two-dimensional and expressed as a scatter plot. In this scatter data 6B, the horizontal axis represents the treated water flow rate and the vertical axis represents the water quality values. Furthermore, the period of one water flow process is considered to be one averaging period, or cycle, and average values ​​are plotted for a series of averaging periods. In the time-series data 6A, #1 to #6 indicate the results of numbering each averaging period. In practice, many more averaging periods were used.

[0040] The average value obtained from averaging period #3, which indicates normal operation, falls within the range of the normal value plotted in the upper right portion of spray data 6B. In contrast, during averaging periods #5 and #6, which correspond to the time of the fault, the treated water flow rate significantly decreased, and the water quality also decreased. The decrease in water quality was particularly significant during averaging period #5. In this water treatment system, the raw water conductivity did not fluctuate significantly from the afternoon of March 5 to the afternoon of March 7. However, it was confirmed that a large amount of organic carbon components flowed into the ion exchange device 60 during this period. Therefore, the decrease in treated water flow rate during averaging periods #5 and #6 is likely due to fouling occurring within the ion exchange device 60, which slowed the ion exchange reaction rate and resulted in leakage of ions that would normally be adsorbed by ion exchange. The significant decrease in water quality during averaging period #5 is likely due to the inflow of ionic organic carbon components, which depleted the ion exchange capacity of the ion exchange resin. After the end of averaging period #5, the ion exchange resin is regenerated, and in the subsequent averaging period #6, the water quality values ​​tend to improve.

[0041] The above describes detecting an abnormality in the water treatment system shown in FIG. 4 using time-series data on the treated water flow rate and time-series data on the water quality of the outlet water of the second cation exchange resin tower 64. In this case, a conductivity meter or resistivity meter is installed at the outlet of the anion exchange resin tower 63 to obtain time-series data on the water quality of the outlet water of the anion exchange resin tower 63, and this data can be used instead of the time-series data on the water quality of the outlet water of the second cation exchange resin tower 64 obtained by the resistivity meter 69. Furthermore, in this embodiment, the multiple time-series data that can be used to detect an abnormality in the water treatment system shown in FIG. 4 are not limited to these. For example, scattering data may be generated using time-series data on the electrical conductivity of the inlet water of the ion exchange device 60 obtained by the conductivity meter 66, time-series data on the differential pressure of the anion exchange resin tower 63, which is obtained as the difference between the measurements of the pressure gauges 67 and 68, and time-series data on the resistivity of the outlet water of the second cation exchange resin tower 64 obtained by the resistivity meter 69, to detect an abnormality in the water treatment system. In this case, because time-series data of three types of quantities are used, the discriminant space is a three-dimensional interval, and the scatter data is also three-dimensional. Here, time-series data related to the water quality at the inlet of the ion exchange device 60 is used, so if an abnormality occurs in the water treatment system, it can be determined whether the quality of the water supplied to the ion exchange device 60 is affecting the cause of the failure. When the number of types of time-series data used is increased in this way, it becomes easier to associate multiple unit operations performed in the water treatment system and analyze the data.

[0042] REFERENCE SIGNS LIST 10 Operation management device 11 Average value calculation unit 12 Calculation unit 13 Spray data creation unit 14 Determination unit 20 Target system 21 Sensor 51, 54 Storage tank 52, 55, 71 Pump 53 Filter 55 Activated carbon tower 60 Ion exchange device 61, 64 Cation exchange resin tower 62 Decarbonation tower 63 Anion exchange resin tower 65 Flow meter 66 Conductivity meter 67, 68 Pressure gauge 69 Resistivity meter 72 Reverse osmosis membrane device

Claims

1. An operations management method for managing the operation of a managed system, comprising the steps of: acquiring from the system time series data for each of N types of quantities related to the operation and status of the system, where N is an integer of 2 or greater; calculating an average value of each of the time series data for each of a series of averaging periods defined in the system; and plotting the average value for each of the averaging periods in an N-dimensional space with each of the N types of quantities as a coordinate axis to generate scatter data.

2. The operational management method according to claim 1, further comprising detecting the occurrence of an abnormality in the system from the arrangement of plot points in the scatter data in the N-dimensional space.

3. An operational management method as described in claim 2, further comprising detecting the occurrence of an abnormality in the system based on changes in the positions of the plot points in the N-dimensional space for each averaging period over time.

4. An operational management method according to any one of claims 1 to 3, wherein the average value is expressed as a centroid value of the corresponding values ​​of the time series data within the averaging period.

5. An operation management method according to any one of claims 1 to 3, wherein the system is a water treatment system including at least an ion exchange device, and the time series data includes data regarding water quality within the water treatment system.

6. The operation management method according to claim 5, wherein each water passing step in the ion exchange apparatus is divided into a plurality of periods, and each of the divided periods is set as the averaging period.

7. An operation management device used to manage the operation of a system to be managed, comprising: average value calculation means for acquiring time series data for each of N types of quantities related to the operation and status of the system from the system, where N is an integer of 2 or greater, and calculating an average value of each of the time series data for each of a series of averaging periods defined in the system; and scatter data generation means for generating scatter data by plotting the average values ​​for each of the averaging periods in an N-dimensional space having each of the N types of quantities as its coordinate axis.

8. An operation management device as described in claim 7, further comprising a determination means for determining the presence or absence of an abnormality occurring in said system from the arrangement of plot points in said scatter data in said N-dimensional space.

9. An operation management device as described in claim 8, wherein the discrimination means discriminates whether or not an abnormality has occurred in the system based also on the change in position of the plot points in the N-dimensional space for each averaging period over time.

10. An operation management device according to any one of claims 7 to 9, wherein the average value calculation means calculates, as the average value, a center of gravity value of the corresponding values ​​of the time series data within the averaging period.

Images