Factory operation data monitoring device and method

Abstract

The present invention provides a plant operation data monitoring device and method, which are suitable for operating data with high independence. The plant operation data monitoring device is composed of a computer, which has: an input unit for inputting the operation data of the plant; a database for storing the input operation data; and a calculation unit for executing a program, the calculation unit stores the input operation data in a first database in time series, divides the peak value of the stored operation data into positive and negative, calculates the slope of the operation data, and stores it in a second database for each positive and negative in time series, determines a threshold for abnormality judgment for each positive and negative slope, and divides each positive and negative slope into a normal value and an abnormal value and stores them in the second database for each positive and negative.

Description

Technical Field

[0001] The present invention relates to a plant operation data monitoring device and method, which monitors operation data from equipment constituting a plant and can detect the occurrence of abnormality at an early stage. Background Art

[0002] The following operations are widely performed in various factories: monitoring the operating data from the equipment that constitutes the factory and detecting the occurrence of abnormalities at an early stage

[0003] Most of these methods monitor operating data or use estimated data from operating data to determine an abnormality when the data deviates from a predetermined threshold or a predetermined threshold range, or utilize physical correlation between operating data to determine an abnormality when the correlation relationship does not hold.

[0004] As an example of a specific implementation method for this purpose, “device degradation state determination system and device degradation state determination method” of Patent Document 1 is known.

[0005] Patent Document 1: Japanese Patent No. 06474564

[0006] The method using the correlation with other operating data described in Patent Document 1 can be said to be an effective method when the physical correlation between the operating data is high.

[0007] On the other hand, for operating data with low physical correlation (low physical and engineering relevance) and high independence (low dependency), it is difficult to judge by association with other operating data. Therefore, a method is desired to judge abnormalities or signs of abnormalities based only on the operating data itself. Summary of the invention

[0008] In view of the above circumstances, an object of the present invention is to provide a plant operation data monitoring device and method that are particularly suitable for operation data with low physical correlation and high independence.

[0009] The present invention is a plant operation data monitoring device, which is composed of a computer. The computer has: an input unit for inputting plant operation data; a database for storing the input operation data; and a calculation unit for executing a program, the calculation unit stores the input operation data in a first database in a time series, divides the stored operation data into positive and negative values ​​according to the peak value, calculates the slope of the operation data, and stores it in a second database for each positive and negative value in a time series, determines a threshold value for abnormality judgment for each positive and negative slope, and divides each positive and negative slope into a normal value and an abnormal value and stores them in the second database for each positive and negative value.

[0010] In addition, the present invention is a factory operation data monitoring method, which inputs the factory operation data, uses the input operation data to monitor the factory abnormalities, stores the input operation data in time series, divides the stored operation data into positive and negative according to the peak value, calculates the slope of the operation data and stores it in time series, determines the threshold value for abnormality judgment for each positive and negative slope, distinguishes each positive and negative slope into normal value and abnormal value and stores them according to each positive and negative.

[0011] According to the present invention, it is possible to provide a plant operating data monitoring device and method that are particularly suitable for operating data with low physical correlation and high independence. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 An example of the equipment configuration of a thermal power plant is shown as a specific example of an application target of the present invention.

[0013] Figure 2 A configuration example of a general plant operation data monitoring device is shown.

[0014] Figure 3 An example of the time change of the input operating data is shown.

[0015] Figure 4 An example of processing and storage of operating data stored in the database DB is shown (front side).

[0016] Figure 5 An example of processing and storing the operating data stored in the database DB (negative side) is shown.

[0017] Figure 6 Indicates the processing contents of the processing program Pg.

[0018] Figure 7 Indicates the processing content of determining whether to store in the database DBB+ having a positive slope or in the database DBB- having a negative slope.

[0019] Figure 8 Further, the processing contents of the processing program Pg, especially the processing step S4, are described. DETAILED DESCRIPTION

[0020] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0021] [Example]

[0022] Figure 1 An example of the equipment configuration of a thermal power plant is shown as a specific example of an application target of the plant operation data monitoring device of the present embodiment.

[0023] Figure 1If the illustrated thermal power plant is roughly divided into its structural equipment, it can be said that it is composed of a generator G, a turbine T, and a boiler B. These are called the so-called main machines. In contrast, the equipment and equipment groups installed for the purpose of maintaining the functions of these main machines are called auxiliary machines. In the example of the figure, in addition to the auxiliary generator GA, the auxiliary turbine TA, and the auxiliary boiler BA, the condenser C and the generator excitation circuit G1 are sometimes also used as auxiliary machines.

[0024] A factory is usually composed of a combination of many such structural devices. In this embodiment, the operation data of such factory structural devices is focused on and appropriately input into a computer via sensors, communication units, input units, etc. The structure for this purpose may be a generally known structure.

[0025] Figure 2 The following is a configuration example of a general plant operation data monitoring device. The plant operation data monitoring device 1, which is composed of a computer, takes in the plant operation data D from a sensor (not shown) installed in the plant equipment 2 into the computer via a communication unit, an input unit, etc. The operation data D at this time sometimes targets operation data with low physical correlation (low physical and engineering correlation) between operation data, and can be set to be taken in at any period regardless of the period of taking in other operation data. In addition, as operation data with low physical correlation (low physical and engineering correlation) between operation data and high independence, the conductivity of the stator cooling water of the generator, the sealed hydrogen concentration of the generator, the hydrogen purity, the hydrogen pressure, etc. are exemplified.

[0026] The operating data D is stored in a time series in an internal database DB in the computer 1. In this case, the storage format in the initial stage is as exemplified by the database DBA, where at least the data collection time and the name and value of the operating data are associated with each other and stored in a time series.

[0027] Furthermore, a processing program Pg, which is a processing procedure for processing the input operating data D, is stored in the computer 1 , and the input operating data is processed sequentially according to the processing program Pg.

[0028] exist Figure 6 In the processing procedure Pg, which is the processing sequence for processing the operating data, the operating data stored in time series is processed into operating data evaluated by slope, and the storage format is changed and the data is stored again.

[0029] Figure 3 An example of temporal change of the input operating data D is shown. Figure 3The upper part of the vertical axis represents the size of the operating data D, and the horizontal axis represents time. It is an example of a trend that increases at even-numbered moments t0, t2, t4, etc. and decreases at odd-numbered moments t1, t3, t5, etc. Here, the peak values ​​at each moment are Figure 3 The first half (from time t0 to time t8) is represented by A1 to A4, and the second half (from time t9 to time t17) is represented by B1 to B4. In addition, in order to distinguish the peaks, each peak is given a general name. These are the upper peaks A1, A, A3, and the lower peaks A1, A, A3.

[0030] In this embodiment, the operation data D stored in the database DBA in time series is focused on whether the data is Figure 3 The operating data D shown in the upper part repeatedly increases and decreases in time series. Figure 3 The lower part illustrates the change rate (indicated as slope in the figure) of the increase or decrease of the operation data D in each time period. In addition, the period of the first half (from time t0 to time t8) is set here to represent the normal change period, and the increase change rate dn1 and the decrease change rate dn2 in the normal period represent a predetermined fixed value or a value within a predetermined range.

[0031] In contrast, the second half of the period (from time t9 to time t17) means the period of variation during the abnormal period, and the increase change rate da1 and the decrease change rate da2 during the abnormal period can generally be arbitrary values ​​that are not constrained to normal values. In the example shown in the figure, as shown in the example between t8 and t9, the increase change rate da1 during the abnormal period becomes a very large value compared to the increase change rate dn1 during the normal period, or as shown in the example between t10 and t11, the increase change rate da1 during the abnormal period becomes a very small value compared to the increase change rate dn1 during the normal period. In addition, in the example shown in the figure, as shown in the example between t11 and t12, the decrease change rate da2 during the abnormal period becomes a very large value compared to the decrease change rate dn2 during the normal period, or as shown in the example between t9 and t10, the decrease change rate da2 during the abnormal period becomes a very small value compared to the decrease change rate dn2 during the normal period. It can be considered that the changes in the change rate in these increasing and decreasing directions all show a tendency to predict and infer certain abnormalities in the plant structure equipment.

[0032] exist Figure 6 In the processing program Pg, the time series operation data D stored in the database DBA is evaluated and processed from the viewpoint of the slope, thereby changing the storage format of the database DBB. The processing program Pg is processed by the calculation unit inside the computer 1.

[0033] exist Figure 2The operation data D is first stored in the database DBA in the computer 1 together with the data collection time information. In contrast, the processed storage information in the processing step S1 includes the information of the slope of the operation data, and the slope is appropriately evaluated as needed, and then the operation data in normal state and the operation data in abnormal state are divided and the storage area is divided, and the operation data is appropriately processed and stored.

[0034] Figure 4 , Figure 5 An example of processing and storing the operating data D stored in the database DBB is shown. Figure 4 and Figure 5 It is shown that the processed operating data D is divided into a data group when the slope is positive (increasing change rate da1 and dn1) and a data group when the slope is negative (decreasing change rate da2 and dn2), and the area of ​​the database DB is divided and stored. In these figures, the stored data includes the peak value common name D11, the value of the operating data D12, the slope D13, and the judgment result (distinction between abnormal and normal) D14. In addition, these data can also appropriately include time information.

[0035] exist Figure 6 In the processing of , the time series of one operation data D of interest is used in the processing step S1 to detect the repetition of continuous maximum and minimum values. For the combination of the continuous maximum and minimum values, the time and value of the upper and lower peaks (maximum and minimum) and the slope of the increase and decrease direction during the period are calculated and stored in the subsequent processing.

[0036] In processing step S2, the continuous upper and lower peaks are retrieved in sequence as a group, and in processing step S3, the slope is calculated based on the time and the difference between the peaks during the period. In addition, in processing step S4, the data of many slopes calculated in the past and the threshold derived therefrom are used to determine whether the slope calculated this time is a normal slope or an abnormal slope, and the determination result is marked for each slope. And, if the slope data marked with the determination result is positive, it is saved in Figure 4 The slope of the database is positive in DBB+, if the slope is negative, it is stored in Figure 5 The slope is negative in the database DBB-.

[0037] These processes are performed sequentially for all combinations of consecutive maximum and minimum values ​​of a specific operation data. In addition, these processes are also performed sequentially for other operation data. Figure 4 , Figure 5 The processed slope data exemplified are sequentially expanded to form a database DBB.

[0038] Figure 7The specific configuration example of the processing part in the processing program Pg, in particular, for distinguishing whether to store in the database DBB+ with a positive slope or in the database DBB- with a negative slope, is illustrated. In this processing, in the processing step S10, it is distinguished whether the group of the continuous upper and lower peaks taken out is generated in the order of upper peak, lower peak, or conversely in the order of lower peak, upper peak. Based on the result of this distinction, in the processing step S11, the former is stored in the database DBB- with a negative slope, and the latter is stored in the database DBB+ with a positive slope.

[0039] Figure 8 The processing contents of the processing program Pg, especially the processing step S4, are described in more detail. Here, when a normal value and an abnormal value are identified, processing related to the threshold setting for identifying the normal value and the abnormal value is performed. In addition, several methods are considered in the method of setting the threshold, and one example is shown here.

[0040] First, in the processing stage of step S21, it is assumed that Figure 4 , Figure 5 The peak list of the slope D13 is formed by the peak name D11, the value D12 of the operation data, and the data of the slope D13, and does not store the judgment results (distinction between abnormal and normal) D14 of each peak. In the processing step S21, for example, Figure 4 In addition, after the series of treatments are completed, we focus on Figure 5 The slope of the negative peak.

[0041] In processing step S22, Figure 4 Among the multiple slopes included in the positive peak of the slope, for example, the slope indicating a value that is discrete from the average value is focused on. And, depending on the degree of dispersion, for example, the slope indicating a value that is about 20% greater or less, and the slope indicating a value that is about 20% less are judged as obvious abnormal values, and the deviation from the average value by plus or minus 20% is set as the threshold for abnormal determination. Various probability theory and statistical methods can be used in determining the threshold. For example, the standard deviation σ can be used as a reference to set the threshold, or 4 times the standard deviation σ can be taken to set plus or minus 4σ as the threshold for abnormal determination.

[0042] In the above description, the idea of ​​setting a threshold value for determining abnormality is described, but similarly, it is natural that a threshold value for identifying a normal value can be set based on a data set of normal data excluding abnormal data. In addition, a threshold value can also be derived based on obvious abnormal data. In addition, the occurrence of abnormalities in factory data sometimes changes greatly in an instant, but usually most of them show a tendency of foreshadowing that the degree of abnormality develops over time. If a situation showing a tendency of foreshadowing is considered, the threshold value is not set to 1 level, but is set to multiple levels, and is set so that the foreshadowing stage until the abnormality can be seen.

[0043] exist Figure 3 The slope of the operation data shown in the lower part of the diagram shows the values ​​dn1 and dn2 in normal state and the values ​​da1 and da2 in abnormal state. In addition, the thresholds for identifying the values ​​da1 and da2 in abnormal state are set as three levels of thresholds, with the upper and lower thresholds of the slope being shown in accordance with the positive and negative directions of each slope. Therefore, when a time trend showing that the second level threshold exceeds a larger threshold after exceeding the first level threshold is confirmed, it is a situation where the abnormality is developing continuously, and it can be grasped as a sign of abnormality.

[0044] In the processing step S23, the slope data for which the normal / abnormal determination flag D14 is not set is evaluated based on the set threshold. In the processing step S24, in particular, the abnormal or abnormal prediction stage of the slope data set as abnormal is constructed. Figure 4 , Figure 5 Then, Figure 5 The same process is performed for the negative peak of the slope.

[0045] According to the above series of processing, databases are formed for positive slopes and negative slopes, and different thresholds are determined for each data group according to the positive slope and negative slope, and the abnormal and abnormal sign levels are determined, and data classification is performed. In addition, the more the number of samples increases, the more accurately the abnormal and normal thresholds can be set, so it can be repeated at an appropriate period. Figure 8 processing.

[0046] According to the present embodiment, for example, in the case where the conductivity of the generator stator cooling water is the operating data, the case where the positive slope exceeds the lower limit and the case where the negative slope exceeds the upper limit can be monitored separately. In the former case, as the cause of the conductivity slope exceeding the lower limit, deterioration of the ion exchange resin, blockage of the line passing through the ion exchange resin, ionization of the return water part of the stator cooling water line, etc. can be considered, and in the latter case, as the cause of the conductivity slope exceeding the upper limit, excessive addition (over-injection) of the ion exchange resin, blockage of the line not passing through the ion exchange resin, etc. can be considered.

[0047] Furthermore, for example, regarding the purity of hydrogen sealed in the generator, when the positive slope exceeds the lower limit, the silica gel of the hydrogen dryer may be degraded. In addition, regarding the purity of hydrogen, when the slope is positive and when it is negative, there are two thresholds (upper and lower limits), and when the positive slope exceeds the lower limit, the silica gel of the hydrogen dryer may be degraded, and when the positive slope exceeds the upper limit, the heat of the heater in the hydrogen dryer or the wind of the blower may be in an abnormal state that is more than usual. In addition, the thresholds in these occurrences are usually different, so it can also help to determine the cause of the abnormal omen.

[0048] According to the embodiment described in detail above, the upper and lower limit peak values ​​of the operation data of the plant structure equipment are detected, the slope (positive / negative) thereof is calculated, and the slope is monitored in the process, thereby enabling the detection of equipment degradation and / or abnormal signs and / or abnormalities with high reliability. As a result, it is expected that the reliability of the plant structure equipment will be improved, which will contribute to the preventive maintenance of the plant structure equipment.

[0049] In addition, by distinguishing between the case where the slope is positive and the case where the slope is negative, an abnormality determination can be performed, and appropriate thresholds can be set for the case where the slope is positive and the case where the slope is negative. Specifically, an upper limit value and a lower limit value when the slope is positive, and an upper limit value and a lower limit value when the slope is negative can be set. Moreover, each upper limit value and a lower limit value can also be set to multiple levels. Moreover, in the case where the slope is positive and the case where it is negative, different abnormal signs and abnormalities can be detected by the upper limit value and the lower limit value. For example, a phenomenon different from an abnormal sign and abnormality that can be detected by the upper limit when the slope is positive can be detected by the lower limit.

[0050] Description of Reference Numerals

[0051] 1: Factory operation data monitoring device (computer)

[0052] 2: Factory Equipment

[0053] G: Generator

[0054] T: Turbine

[0055] B: Boiler

[0056] GA: Generator Auxiliary TA: Turbine Auxiliary

[0057] BA: Boiler Auxiliary Equipment

[0058] C: Condenser

[0059] G1: Generator excitation circuit.

Claims

1. A plant operation data monitoring device, comprising a computer, the computer comprising: an input unit for inputting plant operation data; a database for storing the input operation data; and a computing unit for executing a program, wherein: The operation unit stores the input operation data in a time series in a first database, retrieves consecutive upper and lower peak values ​​from the stored operation data in sequence as a group, calculates the slope of the operation data by dividing it into positive and negative according to the time within the group and the difference between the upper and lower peak values, stores each positive and negative slope obtained in sequence for the combination of consecutive upper and lower peak values ​​in a second database divided into positive and negative according to positive and negative, determines a threshold value for abnormal judgment of the positive and negative slopes, distinguishes each positive and negative slope into a normal value and an abnormal value, and stores them in the second database divided into positive and negative according to positive and negative.

2. The plant operation data monitoring device according to claim 1, characterized in that: The threshold value for performing the abnormality determination is set to different threshold values ​​when the slope is positive and when the slope is negative.

3. The plant operation data monitoring device according to claim 1 or 2, characterized in that: The threshold used for the abnormality determination is set to a multi-level threshold value when the slope is positive and when the slope is negative.

4. The plant operation data monitoring device according to claim 3, characterized in that: The calculation unit determines that the abnormality is progressing by confirming that the slope of the operating data shows a temporal tendency of exceeding a second level threshold value which is a larger threshold value after exceeding a first level threshold value.

5. A method for monitoring plant operation data, comprising inputting plant operation data and using the input operation data to monitor abnormalities of the plant, characterized in that: The input operation data is stored in time series, and continuous upper and lower peak values ​​are sequentially retrieved from the stored operation data as a group. The slope of the operation data is divided into positive and negative according to the time within the group and the difference between the upper and lower peak values. The positive and negative slopes sequentially obtained for the combination of continuous upper and lower peak values ​​are stored as positive and negative, and a threshold for abnormal judgment is determined for each positive and negative slope. Each positive and negative slope is distinguished as a normal value and an abnormal value and stored as positive and negative.

Images