POWER PLANT MONITORING AID APPARATUS, POWER PLANT MONITORING AID METHOD, AND POWER PLANT MONITORING AID PROGRAM

Abstract

A plant monitoring assistance apparatus (10A) is provided with: an acquisition unit (15) that acquires process values ​​(P) of a monitoring target (20); a reproduction unit (16) that outputs a reproduced value (R) obtained by using a model (28) for performing data compression on the acquired process values ​​P in real time and data restoration on the compressed data; a first calculation unit (11) that calculates deviations (D) between the process values ​​(P) and the reproduced value; a detection unit (17) that detects an abnormality sign of the monitoring target (20) on the basis of a preset threshold (S) and the deviations (D); a second calculation unit (12) that calculates a change amount (V) of the deviations (D) on the basis of a time (T) at which the abnormality sign is detected;and a selection unit (18) which selects a process value (P) which belongs to another attribute and is related to the anomaly. Abstract figure: Fig.1;

Description

Title of the invention: POWER PLANT MONITORING ASSISTANCE APPARATUS, POWER PLANT MONITORING ASSISTANCE METHOD, AND POWER PLANT MONITORING ASSISTANCE PROGRAM FIELD OF THE INVENTION

[0001] Embodiments of the present invention relate to a power plant monitoring assistance technology that uses artificial intelligence (AI) to detect a sign of anomaly.

[0002] CONTEXT

[0003] In the case of the introduction of plant monitoring assistance technology in which 1TA-based abnormality sign detection has been put into practice, it is important not only to detect an abnormality sign early but also to be able to analyze the causes and act quickly. Furthermore, there is a need for such plant monitoring assistance technology to visualize process values ​​related to the detected abnormality.

[0004] Process values ​​are output by a large number of sensors provided for detecting the states of the constituent components of the power plant and are also output by a monitoring control device, and these process values ​​are centrally managed in a central monitoring room, for example. In a power plant, a wide variety of process values ​​are monitored in real time to check its operating state. When an abnormal sign is detected in a specific process value, other closely related process values ​​are also monitored at the same time, and then, the causes of the detected abnormal sign and its influence on the operating state of the power plant are investigated.

[0005] Since specifying such closely related process values ​​is difficult for human judgment due to an enormous number of targets, mechanical methods for refining them are considered. In order to achieve this, in a known technique, two data assumed to have a causal relationship are defined as input data and output data, and the refinement processing is performed on the basis of response characteristics of the output data with respect to the input data. In this case, the response characteristics of the output data with respect to a non-stationary change such as an increase, a decrease, and a fluctuation of the input data are refined for each type of change.

[0006] In another known technique, when a deviation between "a predicted value calculated from normal data" and the process value to be monitored exceeds a threshold, it is detected by 1TA as a sign of abnormality. In this case, it is also disclosed that the change trend of the deviation is visualized and simple hierarchy correlation coefficients between process values ​​are prioritized and displayed.

[0007] [Patent document 1] WO 2014 / 091955 [Patent Document 2] Japanese Patent No. 5669553 [Patent Document 3] Japanese Patent No. 7309548 [Patent Document 4] Japanese Patent No. 7391765

[0008] However, in the known techniques described above, in the case of extracting at least one process value which is related to an abnormal event or is to be influenced by the abnormal event, the extraction accuracy is so low that it causes four problems as follows.

[0009] First, even if the change trend of the deviation is visualized, it is difficult to extract at least one process value related to anomalies among many process values.

[0010] Second, when the simple hierarchy correlation coefficients are displayed as a hierarchy, process values ​​that remain changing or have a large deviation are displayed at the top of the hierarchy even if these process values ​​are not related to anomalies.

[0011] Third, the hierarchy changes due to a tiny change in process values ​​such as signal noise and pump pulsation.

[0012] Fourth, it is difficult to take into account that process values ​​related to an anomaly change with a time delay. Summary of the invention

[0013] In view of the circumstances described above, the embodiments of the present invention aim to provide a plant monitoring assistance technique which can efficiently and highly reliably refine process values ​​related to an abnormal event. The invention relates to a plant monitoring assistance apparatus comprising: an acquisition unit configured to acquire process values ​​from a monitoring target; a reproduction unit configured to output a reproduced value obtained using a model for performing data compression / restoration on the process values ​​acquired in real time; a first calculation unit configured to calculate deviations between the process values ​​and the reproduced value; a detection unit configured to detect an anomaly sign of the monitoring target based on a predefined threshold and deviations; a second calculation unit configured to calculate a change amount of the deviations according to a time at which the anomaly sign is detected; and a selection unit configured to use the change amount for the selection of a process value that belongs to another attribute and is related to the anomaly. Preferably, the plant monitoring assistance apparatus further comprises: a first processing unit configured to calculate the average of the deviations in the process values ​​acquired before the time at which the anomaly sign is detected; and a second processing unit configured to average the deviations in the acquired process values ​​after the time at which the abnormality sign is detected, wherein the amount of change is calculated using the averaged deviations. The plant monitoring assistance apparatus may further comprise a normalization unit configured to normalize calculated deviations to obtain normalized deviations, wherein the detection of the abnormality sign and the selection of the related abnormality are performed according to the normalized deviations. The plant monitoring assistance apparatus may further comprise a learning unit configured to cause a model to learn in such a manner that: normal values ​​of the process values ​​are compressed as input data; and restored output data agrees with the input data.The plant monitoring assistance apparatus may further comprise a display configured to display an attribute of a process value to be linked by an anomaly. It further relates to a plant monitoring assistance method comprising the steps of: . the acquisition of process values ​​from a monitoring target; outputting a reproduced value obtained using a model for performing data compression / restoration on process values ​​acquired in real time; the calculation of deviations between process values ​​and the reproduced value; detecting a sign of anomaly in the monitoring target based on a predefined threshold and deviations; calculating an amount of change in the deviations as a function of a time at which the anomaly sign is detected; and the use of the amount of change for the selection of a process value that belongs to another attribute and is related to the anomaly. It further concerns a computer-readable program for assisting with plant monitoring that enables a computer to perform: an acquisition process of acquiring process values ​​from a monitoring target; an output process of outputting a reproduced value obtained by using a model for performing data compression / restoration on process values ​​acquired in real time; a calculation process of calculating deviations between process values ​​and the reproduced value; a detection process consisting of detecting a sign of anomaly of the monitoring target based on a predefined threshold and deviations; another calculation process of calculating a change amount of the deviations based on a time at which the anomaly sign is detected; and a selection process of using the change amount for selection of a process value that belongs to another attribute and is related to the anomaly. Brief description of the drawings

[0014] In the attached drawings: [Fig.l] is a configuration diagram illustrating a plant monitoring assistance apparatus according to the first embodiment of the present invention; [Fig.2] is a graph illustrating a time series of deviations to be output by a first computing unit; [Fig.3] is a screen displaying attributes and quantities of process value change that are mutually related by an anomaly; [Fig.4] is a configuration diagram illustrating the plant monitoring assistance apparatus according to the second embodiment; [Fig.5] is a configuration diagram illustrating the plant monitoring assistance apparatus according to the third embodiment; and [Fig.6] is a flowchart illustrating the steps of a plant monitoring assistance method and an algorithm of a plant monitoring assistance program. DETAILED DESCRIPTION

[0015] Below, embodiments of the present invention will be described with reference to the accompanying drawings. [Fig.l] is a configuration diagram illustrating a plant monitoring assistance apparatus 10A (10) according to the first embodiment of the present invention.

[0016] The plant monitoring assistance apparatus 10A comprises: an acquisition unit 15 configured to acquire process values ​​P of a monitoring target 20; a reproduction unit 16 configured to output a reproduced value R obtained by using a model 28 for performing (a) data compression on the process values ​​P acquired in real time and (b) data restoration on the compressed data (hereinafter abbreviated as data compression / restoration); a first calculation unit 11 configured to calculate deviations D between the process values ​​P and the reproduced value R; a detection unit 17 configured to detect an abnormality sign of the monitoring target 20 on the basis of a predefined threshold S and the deviations D; a second calculation unit 12 configured to calculate a change amount V of the deviations D on the basis of a time T ([Fig.2]) at which the anomaly sign is detected; and a selection unit 18 configured to select a process value P which belongs to another attribute and related to the anomaly sign on the basis of the change amount V. .

[0017] A power plant serving as a monitoring target 20 is provided with a large number of sensors 25 for detecting respective states of its constituent components. Specifically, the process values ​​P to be output by the sensors 25 include temperature, pressure, water level (displacement), flow velocity (speed), and acceleration, for example. Furthermore, in a complex monitoring target 20 such as a nuclear power plant, the process values ​​P are also input and output to a large number of installed control devices 26. Such control devices 26 include an electric pump, an electric cylinder, and a valve, for example.

[0018] The operating status of the power plant is monitored by centrally managing these P process values ​​in a central monitoring room. In a power plant, P process values ​​having a wide variety of attributes are monitored in real time. When an abnormality sign is detected in a specific P process value, other closely related P process values ​​are also monitored at the same time to check the causes of the abnormality detection and / or the influence of the abnormality sign on the power plant.

[0019] The acquisition unit 15 acquires a plurality of process values ​​P of the monitoring target 20 in real time as time series data. Among the process values ​​P acquired by the sensors 25 and the control devices 26, analog signals are converted into digital data. A first accumulation unit 21 accumulates (i.e., stores) each of the acquired process values ​​P together with at least acquisition time information and identifier information of the sensor 25 or control device 26 from which this process value P is output.

[0020] A learning unit 27 constructs a model 28 that has been trained in the following manner. Among the stored process values ​​P, normal process values ​​P are compressed as input data, and output data is obtained by restoring the compressed data so that the output data matches the input data. Aspects of such a model 28 include an autoencoder. In some cases, a pre-built model 28 is used. Thus, the learning unit 27 is not a required component of the plant monitoring assistance apparatus 10.

[0021] Autoencoder is a type of a neural network. An autoencoder refers to an algorithm that compresses input data once to leave only important details and then restores the compressed data to its original dimensions as output data. This autoencoder receives input data at a node of an input layer and compresses the received input data into a hidden layer. At this stage, the input data is weighted according to its degree of importance, and data having low scores or a low degree of importance are filtered out (i.e., encoding). At the time of transition to an output layer, weighting is also applied, and the sum of the data that the node has received from a plurality of edges becomes the output data in the form of the final value (i.e., decoding).The learning unit 27 constructs the model 28 (i.e., autoencoder) in such a way that the output data matches the input data.

[0022] The reproduction unit 16 outputs a reproduced value R (i.e., output data), which is obtained by performing data compression / restoration on the process value P (i.e., input data) acquired in real time from the monitoring target 20 based on the model 28. In the first calculation unit 11, for each of the process values ​​P such as temperature and pressure, the difference between the process value P before input to the reproduction unit 16 and the reproduced value R after output from the reproduction unit 16 is calculated and output as the deviation D. The calculated and output deviations D are accumulated in the second accumulation unit 22 in time series so as to correspond to the respective process values ​​P.Preferably, each deviation D is also accumulated in the second accumulation unit 22 using common identifier information and time information so that it is linked to the corresponding process value P accumulated in the first accumulation unit 21.

[0023] The detection unit 17 detects an abnormality sign of the monitoring target 20 on the basis of the predefined threshold S and the deviations D. The threshold S is set for each of the plurality of process values ​​P, and these thresholds S are recorded in advance in a recording unit (not shown). When the deviation D for at least one process value P to be acquired in real time exceeds the corresponding threshold S, it is determined that a sign of anomaly is detected.

[0024] [Fig. 2] is a graph illustrating the deviation time series D of the process values. The second calculation unit 12 ([Fig. 1]) calculates the amount of change V (VA) of the deviation D (DA) at the time T at which the deviation D exceeds the threshold S (SA) and an abnormality sign is detected. The second calculation unit 12 divides the deviation time series data D based on the time T. The amount of change V is calculated by taking the difference between (i) an arbitrary value of the deviation time series data D after the division and (ii) an arbitrary value of the deviation time series data D before the division.

[0025] The calculated change quantity V is advantageously stored in a storage unit (not shown) using common identifier information and time information for each of the divided time series data of the process value P. The time series data of the deviations D can be divided by arbitrarily setting the time width forward or backward relative to the time T at which the anomaly sign is detected. This time width can be set differently or individually depending on the attribute of the process value P.

[0026] As shown in [Fig.2], the amount of change VB of the process value related to an anomaly is observed with a time delay relative to the amount of change VA of the process value for which the anomaly sign is detected.

[0027] Based on the amount of change V of each of the plurality of process values, the selection unit 18 selects the process values ​​P which are mutually related and have different attributes 37.

[0028] [Fig. 3] illustrates a screen of a display 29 that displays the amount of change V and attributes 37 of the process values ​​P that are mutually linked by an abnormality (including the process value P for which the abnormality sign has been detected). For example, the selection unit 18 can arbitrarily set the number of the process values ​​P that exceed the threshold S and are displayed on the display 29. Additionally or alternatively, they can be set as process values ​​P having a large amount of change regardless of the threshold S (the top 10 having the largest amount of change are shown in the case of [Fig. 3]). Furthermore, when the attributes 37 of mutually linked process values ​​are linked to each other in advance and an abnormality sign is detected in the attribute 37 of a certain process value, the attribute 37 of the linked process value can be displayed as a choice for linked detection and the amount of change V can be observed.

[0029] Although the display 29 displays the attribute 37 of the process values ​​mutually related to the anomaly so that their change amounts V are additionally displayed, the display 29 may also additionally display the process value P, the deviation D, the threshold S and the time T, for example. Although not illustrated in the drawings, a simple hierarchy correlation coefficient between the attributes 37 of the process values ​​may also be displayed.

[0030] In this manner, when an abnormality sign is detected in any of the process values ​​P, the display 29 displays not only that process value P for which the abnormality sign is detected but also other process values ​​P to be related by that process value P due to the abnormality. This configuration can provide a more intense warning to an observer. This configuration further eliminates the influence of the process values ​​which are the monitoring targets and have deviations D which remain large, and enables the observer to accurately recognize the process values ​​to be mutually related by an abnormality.

[0031] (Second embodiment) Next, the second embodiment of the present invention will be described with reference to [Fig. 4]. [Fig. 4] is a configuration diagram illustrating a plant monitoring assistance apparatus 10B (10) according to the second embodiment. The plant monitoring assistance apparatus 10B of the second embodiment has a configuration in which a first processing unit 31 and a second processing unit 32 are further added to the configuration of the first embodiment described above. In [Fig. 4], components having the same configuration or function as those in [Fig. 1] are designated by the same reference signs, and duplicate descriptions are omitted.

[0032] The first processing unit 31 calculates the average of the deviations D in the process values ​​P acquired before the time T at which the abnormality sign is detected. The second processing unit 32 calculates the average of the deviations D in the process values ​​P acquired after the time T at which the abnormality sign is detected. The duration of the averaging processing of the first processing unit 31 is arbitrarily determined for each of the attributes 37 of the process values, and the same is true for the duration of the averaging processing of the second processing unit 32.

[0033] In the second calculation unit 12, the amount of change V is calculated by taking the difference between the deviation D averaged over a predetermined period before time T and the deviation D averaged over a predetermined period after time T. Therefore, the configuration of the second embodiment can eliminate not only the influence of process values ​​having deviations that remain large but also the influence of their noise and time delay, and thus, can enable the observer to recognize highly accurately the process values ​​to be mutually related by an anomaly.

[0034] (Third embodiment) Next, the third embodiment of the present invention will be described with reference to [Fig. 5]. [Fig. 5] is a configuration diagram illustrating a plant monitoring assistance apparatus 10C (10) according to the third embodiment. The plant monitoring assistance apparatus 10C of the third embodiment has a configuration in which a normalization unit 35 is further added to the configuration of the first embodiment and the second embodiment described above. In [Fig. 5], components having the same configuration or function as those in [Fig. 1] or [Fig. 4] are designated by the same reference signs, and duplicate descriptions are omitted.

[0035] The normalization unit 35 normalizes each of the calculated deviations D so as to obtain a normalized deviation E. The normalized deviations E outputted from the normalization unit 35 are accumulated in time sequence in the second accumulation unit 22 for each of the plurality of process values ​​P. On the basis of the normalized deviations E, the detection unit 17 detects the process value P which has a sign of anomaly, and the selection unit 18 selects the process value(s) P to be mutually linked by an anomaly ([Fig.2]).

[0036] The normalization unit 35 calculates the normalized deviation E by dividing the deviation D by the variation in the process values ​​P. As the variation in the process values ​​P to be used for the denominator of the division, the standard deviation of the process value P and / or the standard deviation of the deviation D used in the learning unit 27 is adopted. However, the denominator of the division to be used for normalization is not limited to these variations in the process values ​​P.

[0037] Therefore, the configuration of the third embodiment can eliminate not only the influence of process values ​​having deviations that remain large but also the influence of an accidental change in the process values, and which can enable the observer to recognize highly accurately the process values ​​to be mutually related by an abnormality. The change amount V can be easily compared even between process values ​​having different attributes 37 (for example, when the physical quantities to be measured are different, such as pressure and flow rate, or when the physical quantities are the same but the targets are different).

[0038] Based on the flowchart of [Fig.6], the steps of the plant monitoring assistance method and the algorithm of the plant monitoring assistance program will be described.

[0039] First, in step S1 1, the trained model 28 is constructed in such a manner that: the normal values ​​among the process values ​​P for the monitoring targets 20 are compressed as input data; and the restored output data agrees with the input data.

[0040] Next, the process values ​​Pm(tn) of m = 1 to M in total are acquired from the monitoring target 20 in a time series of n = 1 to N. First, "m = 1 and n = 1" are set in step S12, and the process value Pm(tn) is acquired in step S13.

[0041] Based on the model 28, the process value Pm(tn) is subjected to data compression / restoration, and the reproduced value Rm(tn) is output in step S14.

[0042] Next, the deviation Dm(tn) between the reproduced value Rm(tn) and the process value Pm(tn) is calculated in step S15, then the preset threshold S is acquired in step S16, and then the two are compared in step S17.

[0043] As long as the deviation Dm(tn) does not exceed the threshold S (YES in step S17), the processing proceeds to step S18 and thus the loop of steps S13 to S17 is executed for another process value Pm+l(tn), and further, the loop of steps S13 to S18 is executed for the time series process value Pm(tn+1) (S 19; No, Yes, END).

[0044] If the deviation Dm(tn) exceeds the threshold S (No in step S17), this means that a sign of anomaly is detected, and the amount of change Vm(tn) of the deviation Dm(tn) at this detection time T is calculated in step S20.

[0045] Further, the loop of steps S13 to S17 is repeated in the same manner (S18: No, Yes), and another process value Pm+l(tn) to be mutually related by an anomaly is selected based on its change amount Vm+l(tn) in step S21.

[0046] Additionally or alternatively, instead of using the threshold S, for example, the change amount Vm(tn) of the deviation Dm(tn) for all the process values ​​is calculated in step S20, and the process values ​​to be mutually related by an abnormality are selected as TOP10. The selected process amount is updated or added to the display content on the display 29 in step S22, and then, the loop of steps S13 to S18 is executed for the time series process value Pm(tn+1) (S19; No, Yes, END).

[0047] According to the plant monitoring assistance apparatus of at least one embodiment described above, the process values ​​to be related by an anomaly can be refined efficiently and highly reliably by selecting the attributes of mutually related process values ​​from the amount of change of deviations, which are calculated based on the time of detection of the anomaly sign.

[0048] Although certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the inventions. Indeed, the novel methods and systems described herein may be embodied in a variety of other forms; further, various omissions, substitutions, and changes in the form of the methods and systems described herein may be made without departing from the spirit of the inventions. The appended claims and their equivalents are intended to cover such forms or modifications as would fall within the scope and spirit of the inventions.

[0049] The plant monitoring assistance apparatus described above comprises: a control device in which one or more processors, such as a dedicated chip, an FPGA (field programmable gate array), a GPU (graphics processing unit), and a CPU (central processing unit), are highly integrated; a memory such as a ROM (read-only memory) and a RAM (random access memory); an external storage device such as an HDD (hard disk drive) and an SSD (solid state disk drive); a display; an input device such as a mouse and a keyboard; and a communication interface. The plant monitoring assistance apparatus can be realized by a hardware configuration based on a general-purpose computer. Thus, the components of the plant monitoring assistance apparatus can be obtained by a processor of a computer and can be operated by a plant monitoring assistance program.

[0050] The plant monitoring assistance program may be provided in the form in which it is pre-embedded in a ROM or similar device. Additionally or alternatively, the plant monitoring assistance program may be provided in an installable format or an executable file stored in a computer-readable storage medium such as a CD-ROM, CD-R, memory card, DVD, and floppy disk (FD).

[0051] In addition, the plant monitoring assistance program according to the present embodiment can be stored on a computer connected to a network such as the Internet so as to be provided by being downloaded via the network. Furthermore, the plant monitoring assistance apparatus can also be configured by connecting separate modules together, which independently perform the respective functions of the components, via a network or dedicated lines and combining these modules so that these modules work in combination.

Claims

Claims

1. A plant monitoring assistance apparatus comprising: an acquisition unit configured to acquire process values ​​of a monitoring target; a reproduction unit configured to output a reproduced value obtained by using a model for performing data compression / restoration on the acquired process values ​​in real time; a first calculation unit configured to calculate deviations between the process values ​​and the reproduced value; a detection unit configured to detect an abnormality sign of the monitoring target according to a preset threshold and the deviations; a second calculation unit configured to calculate a change amount of the deviations according to a time at which the abnormality sign is detected;and a selection unit configured to use the amount of change for selection of a process value that belongs to another attribute and is related to the anomaly.;

2. The plant monitoring assistance apparatus according to claim 1, further comprising: a first processing unit configured to calculate the average of the deviations in the process values ​​acquired before the time at which the abnormality sign is detected; and a second processing unit configured to calculate the average of the deviations in the process values ​​acquired after the time at which the abnormality sign is detected, wherein the amount of change is calculated using the averaged deviations.

3. A plant monitoring assistance apparatus according to claim 1 or claim 2, further comprising a normalization unit configured to normalize calculated deviations to obtain normalized deviations, wherein the detection of the anomaly sign and the selection of the related anomaly are performed according to the normalized deviations.

4. A plant monitoring aid apparatus according to claim 1 or claim 2, further comprising a learning unit configured to cause a model to learn in such a way that: normal values ​​of the process values ​​are compressed as input data; and restored output data agree with the input data.

5. A plant monitoring aid apparatus according to claim 1 or claim 2, further comprising a display configured to display an attribute of a process value to be linked by an anomaly.

6. A method for assisting plant monitoring comprising the steps of: acquiring process values ​​of a monitoring target; outputting a reproduced value obtained by using a model for performing data compression / restoration on the acquired process values ​​in real time; calculating deviations between the process values ​​and the reproduced value; detecting an abnormality sign of the monitoring target according to a predefined threshold and the deviations; calculating a change amount of the deviations according to a time at which the abnormality sign is detected; and using the change amount for selecting a process value that belongs to another attribute and is related to the abnormality.

7. A computer-readable program for assisting plant monitoring that enables a computer to perform: an acquisition process of acquiring process values ​​of a monitoring target; an output process of outputting a reproduced value obtained by using a model for performing data compression / restoration on the acquired process values ​​in real time; a calculation process of calculating deviations between the process values ​​and the reproduced value; a detection process of detecting an abnormality sign of the monitoring target based on a predefined threshold and the deviations; another calculation process of calculating an amount of change in the deviations based on a time at which the anomaly sign is detected; and a selection process of using the amount of change to select a process value that belongs to another attribute and is related to the anomaly.

Images