Method and system for pre-checking abnormal signs of nuclear power plant equipment, including a processing procedure for determining equipment importance and alarm validity

Abstract

The present invention provides a method and system for early warning of abnormal signs of equipment, wherein the early warning includes determination of equipment importance and alarm validity. The method for early warning of abnormal signs of equipment includes: a first step: determining whether the value of the equipment monitoring signal exceeds the normal operating range by an early warning processing device using a weighted value based on importance data for each monitoring variable previously analyzed by an operator; a second step: generating an alarm by the early warning processing device when the value of the equipment monitoring signal exceeds the normal operating range; and a third step: determining whether the generated alarm is a valid alarm by an alarm determination device, and the valid alarm is to be subjected to alarm analysis and tracking.

Description

[0001] This application is a divisional application of a patent application with application number 201780053355.9, application date August 29, 2017, and invention name "Method and system for pre-checking abnormal signs of nuclear power plant equipment including a processing procedure for determining equipment importance and alarm validity". Technical Field

[0002] The present invention relates to warning of equipment abnormality signs, and more particularly to generating an alarm by determining equipment abnormality signs in a power plant and pre-detecting equipment abnormality signs in a nuclear power plant. Background Art

[0003] Conventionally, in order to generate an alarm by determining an abnormality in a power plant, alarm data such as temperature, pressure, etc. are acquired from measuring instruments, and the current value of the data is compared with the predicted value, thereby determining the abnormality of the equipment accordingly and generating an early alarm. The early alarm system currently used is implemented using the following principle: the difference between the current value of the equipment monitoring variable and the predicted value calculated by the early alarm program is expressed as a residual, and an alarm is generated when the residual exceeds the normal operating range of the equipment monitoring variable.

[0004] Figure 1 FIG. 1 is a block diagram showing a conventional early warning system. Figure 1 The conventional early alarm system includes: a measuring instrument 100 for measuring the status of power plant equipment; a data processing device 110 for receiving and storing monitoring signals (such as pressure, temperature, flow rate, etc.) obtained from the measuring instrument 100 and converting them into digital data; an early alarm processing device 120 for generating an early alarm by analyzing the converted data; and an operator computer 130 for the operator to check the equipment status, early alarm status, etc.

[0005] Figure 2 FIG. 1 is a flow chart showing a conventional early warning method. Figure 2 As shown, the data processing device 110 receives a measuring instrument signal from a measuring instrument (step S200 ).

[0006] The received measuring instrument signal is an analog signal, and thus the analog data is converted into digital data by undergoing a data processing step (step S210 ).

[0007] Then, the data processing device 110 calculates the current value of the measuring instrument signal in real time (step S220).

[0008] The early alarm processing device 120 receives historical measuring instrument signals within a predetermined period of time that have been subjected to a data processing step from the data processing device 110 and generates a pattern learning model in advance (step S230).

[0009] The calculated current value is transmitted to the early alarm processing device 120, and the early alarm processing device 120 calculates a predicted value by using a pattern learning model generated in advance (step S240).

[0010] The residual (ie, |current value−prediction value|) is calculated by using the current value and the prediction value (step S250).

[0011] The residual is compared with the pre-configured normal operating range of each monitoring variable (e.g., temperature, pressure, flow rate, etc.) (step S260). In the case where the residual exceeds the normal operating range, an alarm is generated (step S270), and in the case where the residual does not exceed the normal operating range, no alarm is generated. At this time, the normal operating range for generating an alarm can be adjusted by the operator.

[0012] When an alarm is generated, the early alarm processing device 120 transmits alarm current status information to the operator computer 130 so that the operator can check the alarm current status (step S280).

[0013] However, this type of early warning system has the following problems.

[0014] In a power plant, the number of monitoring target devices reaches several hundred, and the number of monitoring variables reaches 1000 to 2000 per device. Therefore, if an alarm is generated in the above manner, the alarm is generated very frequently, and it takes a lot of time for an operator to check and analyze the alarm.

[0015] In addition, since the importance of equipment is not taken into consideration, there is a problem that when an alarm is generated, an operator needs to determine the importance of the alarms one by one.

[0016] Furthermore, after an alarm is generated, it is impossible to distinguish whether the alarm is generated by a sign of abnormality in actual equipment or by a pre-scheduled equipment replacement operation or an alarm generated in an experiment, maintenance, etc. Therefore, there is a problem that every time an alarm is generated, an operator needs to analyze the alarm and determine the validity of the alarm.

[0017] Therefore, it should be made in advance that an alarm is generated only in an actual abnormal state of the equipment, and accordingly an operator needs to effectively operate the early alarm system.

[0018] In conventional abnormality sign detection systems for equipment in nuclear power plants, such as Figure 5a As shown, in the event that a variable used to monitor a device exceeds a normal operating range and the exceeded value reaches a preconfigured set point, an alarm is generated.

[0019] Figure 5b The following is a flowchart of the traditional alarm method. Figure 5b, the conventional alarm method includes: receiving measuring instrument data through a data processing device (step S100), and performing a data processing process of converting the received measuring instrument signal into a data signal (step S110). Then, the method includes: calculating a current value at a predetermined time interval (step S120), and determining whether the current value calculated in the early alarm processing device exceeds a configured value for each monitoring variable (step S130). As a result of the determination, in the case where the current value exceeds the configured value for each monitoring variable, an alarm is generated (step S140).

[0020] In the case of the conventional alarm method as described above, it is difficult for the nuclear power plant operator to know Figure 5a The abnormal state between the normal operation period 100 and the configuration value arrival period 110. In addition, the equipment may suddenly fail or an accident may occur during the transition period 120, causing the power plant to shut down. Therefore, effective pre-monitoring means are required between the normal operation period and the configuration value arrival period. Summary of the invention

[0021] Technical issues

[0022] The object of the present invention is to provide a method and system for early warning of abnormal signs of equipment, which includes the determination of equipment importance and alarm effectiveness, by adjusting the alarm generation frequency according to the importance of the equipment and distinguishing between alarms that need to be tracked and analyzed and alarms that do not need to be tracked and analyzed and providing them to operators.

[0023] The object of the present invention is to provide a method and system for pre-detecting signs of equipment abnormality, which is used to pre-detect subtle abnormal signals of equipment to prevent failure of nuclear power plant main equipment and minimize equipment failure and sudden shutdown of nuclear power plants by generating early alarms.

[0024] Technical Solution

[0025] According to one aspect of the present invention, there is provided a method for early warning of abnormal signs of equipment, wherein the early warning includes determining the importance of the equipment and the validity of the alarm. The method for early warning of abnormal signs of equipment comprises: a first step: determining whether the value of the equipment monitoring signal exceeds the normal operating range by an early alarm processing device using a weighted value based on the importance data for each monitoring variable previously analyzed by an operator; a second step: generating an alarm by the early alarm processing device when the value of the equipment monitoring signal exceeds the normal operating range; and a third step: determining whether the generated alarm is a valid alarm by an alarm determination device, and the valid alarm is to be subjected to alarm analysis and tracking.

[0026] The first step may include: performing analog-to-digital conversion on the monitoring signal received from the measuring instrument by a data processing device; and calculating a current value of the monitoring signal and sending the current value to the early warning processing device and the importance classification device at predetermined time intervals.

[0027] The first step may further include searching, by the importance classification device, an importance database for an importance level and a weighted value of the monitoring variable corresponding to the sent current value, and sending the importance level and the weighted value to the early warning processing device.

[0028] The first step may further include: calculating a residual by the early warning processing device using the transmitted current value, the weighted value, and a preconfigured pattern learning model.

[0029] The first step may further include determining, by the early warning processing device, whether the residual exceeds the normal operating range preconfigured by an operator.

[0030] The third step may include: when the alarm is generated, receiving, by the alarm determination device, current state information of the alarm, and classifying the current state information into an alarm that has been predicted and an alarm to be analyzed.

[0031] The predicted alarm may correspond to at least one of an alarm generated by a power plant control room, an alarm generated by an equipment replacement operation, and an alarm generated by a pre-scheduled test or maintenance.

[0032] If the alarm is an alarm to be analyzed, when it is determined as a result of the operator performing the alarm analysis and tracking that power plant maintenance is required, the alarm determination device can notify at least one of the power plant operator computer and the power plant operator personal wireless communication device of the generation of the alarm.

[0033] According to another aspect of the present invention, the present invention provides a system for early warning of abnormal signs of equipment, wherein the early warning includes determination of equipment importance and alarm validity. The system for early warning of abnormal signs of equipment includes: an early warning processing device, the early warning processing device is used to determine whether the equipment monitoring signal value exceeds the normal operating range by using a weighted value based on importance data for each monitoring variable previously analyzed by an operator, and generate an alarm when the equipment monitoring signal value exceeds the normal operating range; and an alarm determination device, the alarm determination device is used to determine whether the generated alarm is a valid alarm, and the valid alarm is to be subjected to alarm analysis and tracking.

[0034] The system for providing early warning of abnormal signs of equipment may also include a data processing device for performing analog-to-digital conversion on the monitoring signal received from the measuring instrument, and for calculating the current value of the monitoring signal and sending the current value to the early warning processing device and the importance classification device at predetermined time intervals.

[0035] The system for providing early warning of abnormal signs of equipment may also include an importance classification device, which is used to search the importance database for the importance level and weighted value of the monitoring variable corresponding to the current value sent, and send the importance level and weighted value to the early warning processing device.

[0036] The early warning processing means may calculate a residual by using the transmitted current value, the weighted value, and a preconfigured pattern learning model.

[0037] The early warning processing means may determine whether the residual exceeds the normal operating range preconfigured by an operator.

[0038] When the alarm is generated, the alarm determination device may receive current state information of the alarm and classify the current state information into an alarm that has been predicted and an alarm to be analyzed.

[0039] The predicted alarm may correspond to at least one of an alarm generated by a power plant control room, an alarm generated by an equipment replacement operation, and an alarm generated by a pre-scheduled test or maintenance.

[0040] If the alarm is an alarm to be analyzed, when it is determined as a result of the operator performing the alarm analysis and tracking that the power plant maintenance is required, the alarm determination device can notify at least one of the power plant operator computer and the power plant operator personal wireless communication device of the generation of the alarm.

[0041] According to another aspect of the present invention, the present invention provides a method for pre-detecting abnormal signs of nuclear power plant equipment, the pre-detection including determining the importance of the equipment and the effectiveness of the alarm. The method for pre-detecting abnormal signs of equipment includes: performing pattern learning using historical equipment monitoring data of a predetermined period of time by a pattern learning device and grouping together monitoring variables that show similar patterns for the same equipment; and generating a predicted value for current monitoring data received in real time using the pattern of the grouped monitoring variables by a predicted value calculation device.

[0042] The step of grouping the monitoring variables may include: a first step of grouping the monitoring variables by the same device; and a second step of grouping the monitoring variables grouped by the same device into groups respectively showing similar patterns.

[0043] The step of generating the predicted value may calculate the predicted value by comparing the current monitoring data with the pattern learning values ​​of a similar monitoring variable group belonging to the same device, and assigning a higher weighted value if the values ​​are more similar.

[0044] The historical equipment monitoring data and the current monitoring data may be digital data that undergoes analog-to-digital conversion.

[0045] The historical equipment monitoring data and the current monitoring data may be data related to monitoring variables that are early warning targets predetermined by an operator.

[0046] The method for pre-detecting abnormal signs of equipment may further include obtaining, by an early warning processing device, a residual by using an absolute value of a difference between the predicted value and the current data, and generating an early warning when the residual is greater than a normal operating range residual.

[0047] The normal operating range residual is pre-specified by an operator for each monitored variable and stored in a database.

[0048] According to another aspect of the present invention, the present invention provides a system for pre-detecting abnormal signs of nuclear power plant equipment, the pre-detection including determining the importance of the equipment and the effectiveness of the alarm. The system for pre-detecting abnormal signs of equipment includes: a pattern learning device for performing pattern learning by using historical equipment monitoring data of a predetermined period and grouping together monitoring variables showing similar patterns for the same equipment; and a predicted value calculation device for generating a predicted value for current monitoring data received in real time by using the pattern of the grouped monitoring variables.

[0049] Grouping the monitoring variables may include: a first step of grouping the monitoring variables by the same device; and a second step of grouping the monitoring variables grouped by the same device into groups that respectively show similar patterns.

[0050] Generating the prediction value may be performed by comparing the current monitoring data with the pattern learning values ​​of the similar monitoring variable group belonging to the same device, and assigning a higher weight value if the values ​​are more similar.

[0051] The historical equipment monitoring data and the current monitoring data may be digital data that undergoes analog-to-digital conversion.

[0052] The historical equipment monitoring data and the current monitoring data may be data related to monitoring variables that are early warning targets predetermined by an operator.

[0053] The system for pre-detecting abnormal signs of equipment may also include an early alarm processing device, which is used to obtain a residual by using the absolute value of the difference between the predicted value and the current data, and generate an early alarm when the residual is larger than a normal operating range residual.

[0054] The normal operating range residual is pre-specified by an operator for each monitored variable and stored in a database.

[0055] Effects of the Invention

[0056] According to an exemplary embodiment of the present invention, a method and system for early warning of signs of equipment abnormality (including determination of equipment importance and alarm effectiveness) can minimize the manpower requirements for early alarm analysis by reducing the alarm generation frequency for equipment with low importance, and even for equipment with high importance, can detect minor changes and generate alarms, thereby effectively preventing equipment and power plant shutdowns.

[0057] In addition, the generated alarms are divided into those that require tracking and analysis and those that do not require tracking and analysis, so that the early warning system of multiple devices can be effectively operated with only a small amount of manpower.

[0058] According to the conventional alarm method based on configuration values, even if a minor abnormality occurs in a device, an alarm is generated only when the abnormality reaches the configuration value of the preconfigured monitoring variable, and the operator is aware of it at this time. Since most of the monitoring variables have a significant difference between the normal operating range and the configuration value range, an alarm is not generated when a minor abnormality occurs in the device, and it is difficult for the operator to notice the minor abnormality. In addition, when such minor abnormalities of the device continue to accumulate, this can lead to serious malfunction problems.

[0059] According to the method and system for pre-detecting signs of equipment abnormality according to an embodiment of the present invention, residuals are calculated using predicted values ​​calculated in real time by a pattern learning model, and an alarm is immediately generated when the size of the residual exceeds the normal operating deviation range, so that subtle signs of equipment can be detected in advance.

[0060] In addition, the failure signs of nuclear power plant equipment are detected quickly, and the failure signs can be prevented from causing the failure of the equipment. Therefore, the sudden shutdown of the power plant can be minimized, and scheduled maintenance can be performed in advance by predicting the shutdown of the equipment, which helps to save maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 FIG. 2 is a block diagram showing a conventional early warning system.

[0062] Figure 2 FIG. 1 is a flow chart showing a conventional early warning method.

[0063] Figure 3 FIG. 4 is a block diagram showing an early warning system according to an embodiment of the present invention.

[0064] Figure 4 FIG. 4 is a flow chart showing an early warning method according to an embodiment of the present invention.

[0065] Figure 5a FIG. 1 is a diagram illustrating the scope of early warning generation of a conventional method and an exemplary embodiment of the present invention.

[0066] Figure 5b Flowchart of traditional alarm method.

[0067] Figure 6 2 is a diagram showing an early warning system according to an embodiment of the present invention.

[0068] Figure 7 A diagram illustrating a conventional monitoring variable grouping method and a monitoring variable grouping method according to an embodiment of the present invention.

[0069] Figure 8 FIG. 1 is a flow chart illustrating generating an early warning according to an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The present invention can have multiple variations and various embodiments, and will be shown in the drawings and described in detail in the detailed description. However, this does not limit the present invention to the specific embodiments, and it should be understood that the present invention covers all modifications, equivalents and replacements within the concept and technical scope of the present invention.

[0071] The terms including first, second, etc. are used to describe various components, but these components are not limited by these terms. These terms are only used to distinguish one component from another component. For example, the first component can be referred to as the second component, and similarly, the second component can be referred to as the first component without departing from the scope of the present invention. The term 'and / or' includes any one of a combination of multiple associated disclosures or multiple associated disclosures.

[0072] When describing that a component is "connected to" or "accesses" another component, the component may be directly connected to or accesses the other component or a third component may exist between the two. In contrast, it should be understood that when describing that a component is "directly connected to" or "directly accesses" another component, it should be understood that there are no other components between the component and the other component.

[0073] The terms used in this application are only used to describe specific embodiments, and are not intended to limit the present invention. If there is no clear opposite meaning in the context, the singular form may include the plural form. In this application, it should be understood that the term "including" or "having" indicates that the features, numbers, steps, operations, components, parts or combinations thereof described in this specification exist, but does not preclude the possibility of the existence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0074] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meanings as those generally understood by those of ordinary skill in the art. Terms defined in commonly used dictionaries should be understood to have the same meanings as in the context of the relevant technology, and are not understood as ideal or overly formal meanings, unless clearly defined in this application.

[0075] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, and when describing the preferred embodiments with reference to the accompanying drawings, the same figure numerals will refer to the same or corresponding components regardless of the figure numerals, and repeated descriptions will be omitted.

[0076] Figure 3 FIG. 2 is a block diagram showing an early warning system according to an embodiment of the present invention. Figure 3The early warning system of the exemplary embodiment of the present invention includes: a measuring instrument 100, the measuring instrument 100 is used to measure the state of the power plant machine; a data processing device 110, the data processing device 110 is used to receive and store the monitoring signal (such as pressure, temperature, flow rate, etc.) obtained from the measuring instrument 100 and convert it into digital data; an early warning processing device 120, the early warning processing device 120 is used to generate an early warning by analyzing the converted data and using the importance for each monitoring signal; an importance classification device 140, the importance classification device 140 is used to store The device comprises a computer system 150 for storing the importance of each monitoring signal data analyzed in advance and classifying the collected monitoring signals according to the importance; an alarm determination device 150, which is used to determine the validity of the alarm when the alarm is generated; an integration center operator computer 130, which is used by the operator of the integration center to receive and check the equipment status and early alarm status; a power plant operator computer 160, which is used to receive the current status of the alarm only in the case of a valid alarm; and a power plant operator personal wireless communication device 170.

[0077] The measuring instrument 100 is attached to or around a power plant device and measures the state of the device (e.g., pressure, temperature, flow rate, etc.) and transmits the monitoring data in a wired manner to the data processing device 110. The operator can transmit data at a predetermined transmission interval (e.g., 1 second interval).

[0078] The data processing device 110 includes an analog-to-digital converter and converts analog monitoring data received from the measuring instrument 100 into digital monitoring data. The digital monitoring data is sent to the early alarm processing device 120 and the importance classification device 140 through a wired communication link.

[0079] The early alarm processing means 120 generates a pattern learning model by using the transmitted historical monitoring data of a predetermined period (e.g., 18 months). In addition, the early alarm processing means 120 determines whether the operation is normal and generates an alarm by comparing the transmitted current monitoring data value with the predicted value of the previously generated pattern learning model. When comparing the current monitoring data value with the predicted value, the normal operation is determined by using the importance for each monitoring variable stored in the importance classification means.

[0080] The importance classification device 140 includes an importance database.

[0081] The importance database stores all monitoring variables of equipment operated in the nuclear power plant, and stores information that classifies the importance for each monitoring variable according to an importance classification standard specified in advance by an operator.

[0082] For example, the information may be stored wherein each monitored variable is classified into three types, including variables that affect power plant shutdown (level A), variables that affect single equipment shutdown but not power plant shutdown (level B), and variables that do not affect equipment shutdown (level C).

[0083] For example, the water supply pump bearing temperature is a variable that may cause the water supply pump to shut down when it suddenly increases and may be classified as level B, and the worm gear bearing vibration is a variable that may cause the power plant to shut down when the vibration increases and may be classified as level A.

[0084] In addition, in the importance database, each weight value for each level specified by the operator for calculating the predicted value is stored. For example, each weight value for each level can be specified as 1.2 to 1.5 for level A, 0.9 to 1.1 for level B, and 0.6 to 0.8 for level C.

[0085] When the weighting value is thus specified, there is an effect that for a monitoring variable with high importance, the residual (|current value-predicted value|) can become larger, and an alarm is generated even for a small change. Therefore, for a monitoring variable with high importance, even a slight change can be detected.

[0086] This has the effect that for monitored variables with low importance, the residuals become smaller and alarms are generated only in the case of large changes, so the alarm frequency can be significantly reduced.

[0087] The importance levels and the respective weighting values ​​for the levels may be arbitrarily determined by the operator based on the contents described in all types of operation procedure documents and the operator's operation experience.

[0088] The importance classification device 140 determines the importance of the monitoring data transmitted from the data processing device 110 based on the importance information stored in the importance database.

[0089] When it is determined that the generated alarm is a valid alarm that needs to be notified to the power plant and measures to be taken, the alarm determination device 150 automatically sends the current state of the alarm to the power plant operator computer 160 and the power plant operator personal wireless communication device 170. In the determination of the valid alarm, the following items are determined and only alarms that do not correspond to such cases are determined as valid alarms: whether the same alarm is generated even in the power plant Main Control Room (MCR) (i.e., the case where the alarm data value reaches a predetermined configuration value and an alarm is generated), whether the alarm is generated by an equipment replacement operation, and whether the alarm is generated by a pre-planned test or maintenance.

[0090] Figure 4 FIG. 1 is a flow chart showing an early warning method according to an embodiment of the present invention. Figure 4 , the data processing device 110 receives a measuring instrument signal from the measuring instrument (step S400).

[0091] Since the received measuring instrument signal is an analog signal, the measuring instrument signal undergoes an analog-to-digital conversion step (step S410 ) in which it is converted into digital data by the data processing device 110 . The converted digital data is sent to the early warning processing device 120 .

[0092] Then, the data processing device 110 calculates the current value of the measuring instrument signal at predetermined time intervals (eg, 1 second intervals) (step S520 ), and sends the current value to the early warning processing device 120 and the importance classification device 140 .

[0093] The importance classifying device 140 determines the importance of the current value transmitted from the data processing device 110 (step S460). The importance determination is performed by searching for a value stored in the importance database of the importance classifying device 140 for the importance level and weighted value of the monitoring variable corresponding to the current value. For example, in the case where the transmitted current value corresponds to the water supply pump bearing temperature data, the importance level and weighted value of the water supply pump bearing temperature stored in the importance database are searched.

[0094] The early alarm processing device 120 generates a pattern learning model (step S430) by using the predetermined period (eg, 18 months) of historical monitoring data transmitted from the data processing device 110 and subjected to the analog-to-digital conversion step (step S410).

[0095] The early warning processing device 120 calculates the predicted value by using the transmitted current value and the pre-generated pattern learning model (step S440). That is, the early warning processing device 120 compares the current value with the learning value in the pattern learning model at a predetermined time interval (e.g., 1 second interval) and finds a similar pattern, and then calculates the predicted value.

[0096] Next, the residual a (ie, |current value-predicted value|) is calculated by using the current value and the predicted value (step S450).

[0097] The residual b (weighted value×residual a) is calculated by using the calculated residual a and the weighted value corresponding to the importance of the monitoring variable determined in the importance classification device 140 (step S470). The early warning processing device 120 includes a function of calculating the residual a in real time and finding the weighted value of the same monitoring variable from the importance classification device 140, so the residual b can be calculated in real time.

[0098] The calculated residual b is compared with the pre-configured normal operating range of each monitoring variable (step S480).

[0099] An alarm is generated when the residual error b exceeds the normal operating range (step S490), and no alarm is generated when the residual error b does not exceed the normal operating range.

[0100] At this time, the normal operating range for generating an alarm can be adjusted by the operator. For example, the normal operating range of a water supply pump is 45 (minimum) to 55 (maximum), and the deviation between the minimum and maximum values ​​is 10. In the case where the program is to generate an alarm when the residual exceeds 30% (±3) of the [maximum-minimum] deviation of the normal operating range of 10, when the input current value is 60 and the calculated predicted value is 52, the residual is 8, which exceeds 30% of the deviation, and an alarm is generated. On the other hand, when the current value is 43 and the predicted value is 45, the residual is 2 and no alarm is generated.

[0101] When an alarm is generated, the early alarm processing device 120 sends the alarm current state information to the integrated center operator computer 130 monitoring the entire power plant so that the operator can check (step S500). At the same time, the early alarm processing device 120 sends the alarm current state information to the alarm determination device 150.

[0102] The alarm determining device 150 receives the alarm current state information and performs a validity determination process of the alarm.

[0103] First, the alarm determination device 150 classifies the type of the received alarm and determines the analysis target (step S510 ).

[0104] For example, the alarm determination device 150 determines whether the generated alarm is generated even in the main control room (MCR) of the power plant, whether the alarm is generated by an equipment replacement operation, whether the alarm is generated by a pre-planned test or maintenance, and in the case where the alarm corresponds to any of these items, the alarm is stored in the list of predicted alarms, and in the case where the alarm does not correspond to any of these items, the alarm is stored in the list of alarms to be analyzed.

[0105] The integrated center operator alarm tracking / analyzes the information stored in the list of alarms to be analyzed (step S520). Thus, if it is determined that power plant maintenance is required, the fact that the alarm was generated is notified to the power plant operator computer 160 and the personal wireless communication device 170 of the pre-designated power plant operator (or person in charge of the equipment) (step S530). If it is determined that power plant maintenance is not required, trends are continuously monitored and additional analysis is performed if necessary.

[0106] Various embodiments of the present invention have been described so far. However, it is to be understood that this description is intended to illustrate and not to limit the scope of the present invention, which is defined by the scope of the appended claims. Therefore, other embodiments are also within the scope of the appended claims. For example, various embodiments may be performed without departing from the scope of the present invention. In addition, some of the steps described above are not order dependent and may be performed in an order different from that described above.

[0107] Figure 6 FIG. 2 is a diagram showing an early warning system according to an embodiment of the present invention. Figure 6 , an early warning system includes: a measuring instrument 200 of a power plant equipment; a data processing device 210, the data processing device 210 is used to process a monitoring signal obtained from the measuring instrument; a monitoring variable selection / storage device 220, the monitoring variable selection / storage device 220 is used to select and store monitoring variables as early warning application objects; a monitoring variable pattern learning device 230, the monitoring variable pattern learning device 230 is used to learn the normal operation history of the selected monitoring variable within a predetermined period (for example, 18 months); a pattern database 240, the pattern database 240 is used to store monitoring variables, wherein monitoring variables of similar patterns are grouped together; a predicted value calculation device 250, the predicted value calculation device 250 is used to calculate a predicted value using a current value sent from the data processing device 210 in real time; an early warning processing device 260, the early warning processing device 260 is used to determine whether to generate an alarm by comparing the current value with the predicted value; and a normal operating range database 270, the normal operating range database 270 is used to store the normal operating range for each monitoring variable.

[0108] The data processing device 210 collects and stores monitoring signals (such as temperature, pressure, flow rate, etc.) acquired from the measuring instruments 200 of the power plant equipment, and converts these signals into digital signals.

[0109] The monitoring variable selection / storage device 220 selects and stores the monitoring variables as the early warning application object in a database. The number of monitoring variables collected in the data processing device 210 in the power plant is about 7000 to 10000, and the monitoring variable selection / storage device 220 selects the monitoring variables that can be effectively used for early warning among these variables and stores these monitoring variables. The selected objects are variables that can be used for early warning through pattern recognition, mainly monitoring variables (such as temperature, pressure, flow rate, vibration, etc.) of active equipment (such as pumps, valves, worm gears, generators, etc.), and important variables that are not active equipment but can evaluate the operating status of the power plant (such as power generation of the power plant, radioactivity concentration, etc.). However, the monitoring variables of the switch signals generated instantly or the equipment that slightly affects the safety and operation of the power plant are excluded. This is because when too much data is processed, it becomes difficult to process and transmit data in real time (units of 1 second).

[0110] In the monitoring variable pattern learning device 230, the normal operation history of the selected variable for the past year or more (e.g., 18 months) is learned. For the selected monitoring variables (about 1,300 for each power plant), the normal operation state values ​​(i.e., historical operation values) for a predetermined period (e.g., 18 months, which is one cycle operation period of the nuclear power plant) are collected, and pattern learning is performed. That is, when the normal operation history is learned by collecting the values ​​under the normal operation state within the predetermined period for each monitoring variable, the normal operation pattern of the corresponding variable can be completely learned. Each variable can have various operation modes during the predetermined period, and all such modes are learned.

[0111] For the learned patterns, monitoring variables showing similar patterns can be grouped together, and monitoring variables belonging to the same group can be regarded as having similar operating patterns. The grouped monitoring variables are stored in the pattern database 240 .

[0112] Figure 7 FIG. 2 is a diagram showing a conventional monitoring variable grouping method and a monitoring variable grouping method according to an embodiment of the present invention. Figure 7 , the conventional pattern learning group generation method shown on the left involves a method of forming a pattern learning group by pattern learning the total monitored variable (approximately 1000) data and classifying the data according to each pattern, in which method, as long as the patterns are similar, the variables are classified into the same pattern even if they do not belong to the same device.

[0113] In this case, there is the following problem: the reliability and accuracy of pattern learning are deteriorated in some parts. For example, problems may occur when the temperature, pressure and flow rate of the water supply pump and the water level of the oil supply tank of the power plant show similar patterns within one year and are grouped into similar pattern groups. The temperature, pressure and flow rate of the water supply pump show similar patterns in terms of physics and operating experience, which is reasonable, but the above three variables and the water level of the oil supply tank of the power plant do not have similarity in terms of physics and operating experience. However, when these variables are simply grouped into the same pattern group due to the similarity of the patterns of the variables, when the predicted values ​​are generated later, the monitoring variables of the similar pattern groups will affect each other, so the accuracy of the predicted values ​​deteriorates. In addition, in the case of the traditional method, when the current value is input, it is necessary to search for all target patterns grouped in the similar pattern group to see if there is a similar pattern, so it takes a lot of time to perform the analysis process. That is, it may be difficult to derive the required results in the second or many computers.

[0114] According to an embodiment of the present invention, in order to improve the accuracy of prediction value calculation by improving the reliability of the pattern learning model, two grouping processes are performed, such as Figure 3 That is, in the primary grouping, all variables that are the objects of pattern learning are first grouped by each device, and in the secondary grouping, variables that show similar patterns in the same device are grouped together.

[0115] When grouping is performed as described above, for example, when a current value is input, the variable is first grouped into the device to which the current value belongs, and a similar pattern is searched within the corresponding device. Therefore, the time for searching for similar patterns is reduced, and variables with high actual correlation are grouped together and used to calculate the predicted value, so the accuracy becomes high.

[0116] Refer again Figure 6 The data processing device 210 sends the currently measured variable value to the prediction value calculation device 250 and the early warning processing device 260 in real time (for example, at intervals of 1 second).

[0117] The predicted value calculation device 250 calculates the predicted value using the current value of the monitoring variable transmitted in real time from the processing device 210 .

[0118] As an algorithm for calculating the predicted value, for example, a technique such as AutoAssociative Kernel Regression (AAKR) can be used. When the current value of the measured monitoring variable is input, the device to which the variable belongs is identified, and these values ​​are compared with the pattern learning values ​​of the most similar pattern group among the grouping patterns of the corresponding devices. The predicted value is calculated by assigning weighted values ​​to all relevant pattern models (such as assigning a high weighted value to the most similar pattern model and a low weighted value to the least similar pattern model). For example, the predicted value can be calculated and totaled by assigning weighted values ​​so that the sum of the weighted values ​​for each relevant pattern model is 1.

[0119] The early warning processing means 260 calculates a residual by comparing the calculated predicted value with the current value.

[0120] The residuals are calculated by using the calculated predicted values ​​with Eq.

[0121] [Equation 1]

[0122] Residual = |predicted value - current value|

[0123] The calculated residual is then compared to the normal operating range residual for each monitored variable stored in the normal operating range database 270 , and an early alarm is generated when the calculated residual is greater than the normal operating range residual.

[0124] It can be configured to generate an early warning when the calculated residuals exceed the deviation range (%) configured by the operator relative to the normal operating maximum and minimum ranges of each variable (i.e., normal operating range residuals). The normal operating range residuals preconfigured by the operator are stored in the normal operating range database 270.

[0125] For example, the normal operating range of the water supply pump is 45°C (minimum value) to 55°C (maximum value), and the deviation between the minimum and maximum values ​​is 10°C. In the case where it is configured to generate an alarm when the residual exceeds 30% of the |maximum value-minimum value| deviation (i.e., 10°C) of the normal operating range, when the input current value is 60°C and the calculated predicted value is 52°C, the residual becomes 8°C, and then an alarm is generated. On the other hand, in the case where the current value is 43°C and the predicted value is 45°C, the residual becomes 2°C, and then no alarm is generated.

[0126] Figure 8 FIG. 1 is a flow chart showing the generation of an early warning according to an embodiment of the present invention. Figure 8 , after being converted into digital data in the data processing device 210, the predicted value calculation device 350 receives the current value of the monitoring signal (step S400).

[0127] Next, the device to which the monitoring variable of the current value belongs is identified (step S410). That is, when the current value is the bearing temperature value of the water supply pump, the device to which the monitoring variable of the current value belongs is identified as the water supply pump.

[0128] Among the pattern models stored in the pattern database, a similar pattern model is identified within the device to which the variable belongs (step S420).

[0129] When a similar pattern model is identified, the similarity between the learned value and the current value of each variable belonging thereto is analyzed (step S430).

[0130] By assigning a high weight value to a pattern model with high similarity and a low weight value to a pattern model with low similarity, as expressed in Equation 2, the sum of the weight values ​​of each similar pattern model multiplied by the learning value is obtained as a prediction value (step S440).

[0131] [Equation 2]

[0132] ∑(weighted value × learning value) = predicted value

[0133] (In this article, the weighted values ​​sum to 1.)

[0134] The early warning processing means 260 receives the predicted value calculated in the predicted value calculation means 250 and the current value from the data processing means 210 and calculates a residual by using Equation 1 described above.

[0135] Next, the normal operating range residual for the corresponding monitoring variable stored in the normal operating range database 270 is compared with the residual (step S460). When the calculated residual is greater than the normal operating range residual, an alarm is generated (step S470).

[0136] Various embodiments of the present invention have been described so far. However, it is to be understood that this description is intended to illustrate and not to limit the scope of the present invention, which is defined by the scope of the appended claims. Therefore, other embodiments are also within the scope of the appended claims. For example, various embodiments may be performed without departing from the scope of the present invention. In addition, some of the steps described above are not order dependent and may be performed in an order different from that described above.

[0137] [Description of Reference Signs]

[0138] 100: Measuring instrument 110: Data processing device

[0139] 120: Early warning processing device 130: Integration center operator computer

[0140] 140: Importance classification device 150: Alarm determination device

[0141] 160: Power plant operator computer 170: Power plant operator personal wireless communication device

[0142] 100-1: Equipment normal operation period 110-1: MCR alarm (based on configuration value)

[0143] 120-1: Transition State (Early Alert Period)

[0144] 200: Important equipment in nuclear power plants (including measuring instruments)

[0145] 210: Data processing device (computer in power station)

[0146] 220: Monitoring variable selection / storage device 230: Monitoring variable pattern learning device

[0147] 240: Pattern database 250: Prediction value calculation device

[0148] 260: Early warning processing device 270: Normal operating range database.

Claims

1. A method for pre-detecting abnormal signs of equipment, wherein the equipment is a nuclear power plant equipment, characterized in that: The method comprises: performing pattern learning using historical equipment monitoring data of a predetermined period by a pattern learning device and grouping together monitoring variables showing similar patterns for the same equipment; and Generate a predicted value for the current monitoring data received in real time by using the grouped patterns of the monitoring variables through a predicted value calculation device, The step of grouping the monitoring variables comprises: The first step: grouping the monitoring variables according to the same device; and The second step is to group the monitoring variables grouped by the same device into groups respectively showing similar patterns.

2. The method for pre-detecting abnormal signs of equipment according to claim 1, wherein: The step of generating the predicted value is to calculate the predicted value by comparing the current monitoring data with the pattern learning value of the similar monitoring variable group belonging to the same device, and assigning a higher weight value if the current monitoring data is more similar to the pattern learning value.

3. The method for pre-detecting abnormal signs of equipment according to claim 1, wherein: The historical equipment monitoring data and the current monitoring data are digital data that undergo analog-to-digital conversion.

4. The method for pre-detecting abnormal signs of equipment according to claim 1, wherein: The historical equipment monitoring data and the current monitoring data are data related to monitoring variables that are early warning targets predetermined by an operator.

5. The method for pre-detecting abnormal signs of equipment according to claim 1, further comprising: A residual is obtained by the early warning processing means by using an absolute value of a difference between the predicted value and the current monitoring data, and an early warning is generated when the residual is greater than a normal operating range residual.

6. The method for pre-detecting abnormal signs of equipment according to claim 5, wherein: The normal operating range residual is pre-specified by an operator for each monitoring variable and stored in a database.

7. A system for pre-detecting abnormal signs of equipment, wherein the equipment is a nuclear power plant equipment, characterized in that: The system comprises: Pattern learning means for performing pattern learning by using historical equipment monitoring data of a predetermined period and grouping together monitoring variables showing similar patterns for the same equipment; and a prediction value calculation device for generating a prediction value for current monitoring data received in real time by using the pattern of the grouped monitoring variables, Wherein, grouping the monitoring variables includes: grouping the monitored variables by the same device; and The monitored variables grouped by the same device are grouped into groups respectively showing similar patterns.

8. The system for pre-detecting abnormal signs of equipment according to claim 7, wherein: When generating the predicted value, the predicted value is calculated by comparing the current monitoring data with the pattern learning value of a similar monitoring variable group belonging to the same device, and assigning a higher weight value if the current monitoring data is more similar to the pattern learning value.

9. The system for pre-detecting abnormal signs of equipment according to claim 7, wherein: The historical equipment monitoring data and the current monitoring data are digital data that undergo analog-to-digital conversion.

10. The system for pre-detecting abnormal signs of equipment according to claim 7, wherein: The historical equipment monitoring data and the current monitoring data are data related to monitoring variables that are early warning targets predetermined by an operator.

11. The system for pre-detecting abnormal signs of equipment as described in claim 7 also includes an early alarm processing device, which is used to obtain a residual by using the absolute value of the difference between the predicted value and the current monitoring data, and generate an early alarm when the residual is larger than the normal operating range residual.

12. The system for pre-detecting abnormal signs of equipment according to claim 11, wherein: The normal operating range residual is pre-specified by an operator for each monitoring variable and stored in a database.

Images