Method for automatically generating plant condition monitoring model of nuclear power plant using approximate nearest neighbor search

Abstract

This method for automatically generating a plant condition monitoring model of a nuclear power plant using approximate nearest neighbor search comprises the steps of: collecting a plurality of first monitoring signals received by a first equipment model of a plant condition monitoring model of a nuclear power plant and a plurality of second monitoring signals received by a second equipment model of the plant condition monitoring model; performing independent component analysis on each of the plurality of first monitoring signals and the plurality of second monitoring signals; reconstructing a first separation signal having a low correlation from among the plurality of first monitoring signals into a new signal monitoring model on the basis of the independent component analysis; and reconstructing a second separation signal having a low correlation from among the plurality of second monitoring signals into the new signal monitoring model on the basis of the independent component analysis.

Description

A method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search.

[0001] This paper relates to a method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search.

[0002] In general, the plant condition monitoring model is a model that monitors the condition of a nuclear power plant by using signals received from sensors installed in the facilities included in the nuclear power plant.

[0003] Plant condition monitoring models are generated using signals received from various sensors.

[0004] The conventional method of creating a plant condition monitoring model involves combining signals received from the sensors of the nuclear power plant based on the know-how and experience of engineers based on the physical classification system of the nuclear power plant and optimizing the model through repeated experiments, which increases the model creation time and cost.

[0005] In addition, the conventional method of creating a plant condition monitoring model has a problem in that if the monitoring signals received from the nuclear power plant to the plant condition monitoring model include monitoring signals with low correlation, the prediction accuracy of the entire plant condition monitoring model decreases, resulting in false alarms.

[0006] One embodiment provides a method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search, which automatically generates a plant condition monitoring model with improved prediction accuracy while reducing model generation time and generation cost, even when the plant condition monitoring model includes low-correlation monitoring signals among the monitoring signals received from the nuclear power plant.

[0007] One aspect provides a method for automatically generating a plant condition monitoring model, comprising the steps of collecting a plurality of first monitoring signals received in a first equipment model of a plant condition monitoring model of a nuclear power plant and a plurality of second monitoring signals received in a second equipment model of the plant condition monitoring model, performing independent component analysis on each of the plurality of first monitoring signals and the plurality of second monitoring signals, reconstructing a first separate signal having a low correlation among the plurality of first monitoring signals into a new signal monitoring model based on the independent component analysis, and reconstructing a second separate signal having a low correlation among the plurality of second monitoring signals into the new signal monitoring model based on the independent component analysis.

[0008] The above novel signal monitoring model can perform approximate nearest neighbor search prediction based on the physical classification system of the above nuclear power plant.

[0009] The step of performing the above independent component analysis can be performed by applying independent component analysis based on dimension reduction.

[0010] The step of collecting the plurality of first monitoring signals and the plurality of second monitoring signals may include the step of collecting the plurality of third monitoring signals received in the third equipment model of the plant status monitoring model.

[0011] The step of performing the independent component analysis may include the step of performing the independent component analysis of each of the plurality of third surveillance signals.

[0012] The method may further include a step of reconstructing a third separation signal with low correlation among the plurality of third surveillance signals based on the independent component analysis into the new signal surveillance model.

[0013] According to one embodiment, a method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search is provided, which automatically generates a plant condition monitoring model with improved prediction accuracy while reducing model generation time and generation cost, even if the plant condition monitoring model includes low-correlation monitoring signals among the monitoring signals received from the nuclear power plant.

[0014] Figure 1 is a flowchart illustrating a method for automatically generating a plant status monitoring model according to one embodiment.

[0015] FIG. 2 is a diagram for explaining a method for automatically generating a plant status monitoring model according to one embodiment.

[0016] FIG. 3 is a graph showing an example of independent component analysis used in a method for automatically generating a plant condition monitoring model according to one embodiment.

[0017] FIG. 4 is a diagram showing an example of approximate nearest neighbor search used in a method for automatically generating a plant condition monitoring model according to one embodiment.

[0018] FIG. 5 is a flowchart illustrating an example of a method for automatically generating a plant status monitoring model according to one embodiment.

[0019] Hereinafter, with reference to the attached drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily implement the present invention. The present invention may be implemented in various different forms and is not limited to the embodiments described herein.

[0020] Additionally, throughout the specification, whenever a part is said to "include" a component, this does not mean that it excludes other components, but rather that it may include other components, unless otherwise specifically stated.

[0021] Hereinafter, a method for automatically generating a plant status monitoring model according to one embodiment will be described with reference to FIGS. 1 to 5.

[0022] A method for automatically generating a plant condition monitoring model according to one embodiment may include a method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search.

[0023] A method for automatically generating a plant condition monitoring model according to one embodiment can automatically generate a plant condition monitoring model for monitoring the condition of a nuclear power plant, but is not limited thereto.

[0024] A method for automatically generating a plant condition monitoring model according to one embodiment may be performed by artificial intelligence driven by a computing device, but is not limited thereto.

[0025] Fig. 1 is a flowchart illustrating a method for automatically generating a plant condition monitoring model according to one embodiment. Fig. 2 is a diagram for explaining a method for automatically generating a plant condition monitoring model according to one embodiment.

[0026] Referring to FIGS. 1 and 2, first, a plurality of first monitoring signals (MS1) received by a first equipment model (100) of a nuclear power plant status monitoring model (1000), a plurality of second monitoring signals (MS2) received by a second equipment model (200) of the plant status monitoring model (1000), and a plurality of third monitoring signals (MS3) received by a third equipment model (300) of the plant status monitoring model (1000) are collected (S100).

[0027] For example, a plurality of first monitoring signals (MS1) are received by a first equipment model (100) of a plant condition monitoring model (1000) of a nuclear power plant. The plurality of first monitoring signals (MS1) may include existing monitoring signals received by the first equipment model (100). A plurality of second monitoring signals (MS2) are received by a second equipment model (200) of a plant condition monitoring model (1000) of a nuclear power plant. The plurality of second monitoring signals (MS2) may include existing monitoring signals received by the second equipment model (200). A plurality of third monitoring signals (MS3) are received by a third equipment model (300) of a plant condition monitoring model (1000) of a nuclear power plant. The plurality of third monitoring signals (MS3) may include existing monitoring signals received by the third equipment model (300). A plurality of first monitoring signals (MS1), a plurality of second monitoring signals (MS2), and a plurality of third monitoring signals (MS3) are collected.

[0028] As another example, a plurality of first monitoring signals (MS1) received by a first equipment model (100) of a nuclear power plant status monitoring model (1000) and a plurality of second monitoring signals (MS2) received by a second equipment model (200) of the plant status monitoring model (1000) can be collected.

[0029] Next, independent component analysis (ICA) is performed on each of the plurality of first surveillance signals (MS1), the plurality of second surveillance signals (MS2), and the plurality of third surveillance signals (MS3) (S200).

[0030] For example, independent component analysis (ICA) is performed on each of the plurality of first surveillance signals (MS1), the plurality of second surveillance signals (MS2), and the plurality of third surveillance signals (MS3). The ICA may include various known ICAs. For example, various known ICAs may be performed by artificial intelligence driven by a computing device.

[0031] For example, the step of performing independent component analysis of each of a plurality of first surveillance signals (MS1), a plurality of second surveillance signals (MS2), and a plurality of third surveillance signals (MS3) can be performed by applying independent component analysis based on dimension reduction.

[0032] FIG. 3 is a graph showing an example of independent component analysis used in a method for automatically generating a plant condition monitoring model according to one embodiment.

[0033] Referring to Figure 3, for example, AI can automatically classify signal components by applying dimensionality reduction-based independent component analysis (ICA). AI can perform sophisticated predictions by grouping only strongly correlated signals and restoring the signal. ICA based on dimensionality reduction can produce superior prediction results compared to existing statistical techniques. AI can also automatically separate weakly correlated signals and perform separate predictions. ICA based on dimensionality reduction is more efficient and objective than the conventional manual separation method by engineers.

[0034] As another example, independent component analysis (ICA) can be performed on each of the plurality of first surveillance signals (MS1) and the plurality of second surveillance signals (MS2).

[0035] Next, based on independent component analysis, the first separation signal (SS1) with low correlation among the plurality of first surveillance signals (MS1) is reconstructed into a new signal surveillance model (400) (S300).

[0036] For example, based on independent component analysis, at least one first separation signal (SS1) among multiple first monitoring signals (MS1) with low correlation is reconstructed into a new signal monitoring model (400). The new signal monitoring model (400) can perform an Approximate Nearest Neighbor Search (ANNS) prediction based on the physical classification system of the nuclear power plant.

[0037] The new signal monitoring model (400) can automatically reorganize adjacent signals based on a physical model of a nuclear power plant. For example, the new signal monitoring model (400) can automatically reorganize adjacent signals, such as vibration and pressure signals of the same pump or temperature and pressure signals of the same system line, based on a physical model of a nuclear power plant. The new signal monitoring model (400) can perform predictions using an approximate nearest neighbor search, which quickly searches for the most similar values ​​from recent past information, rather than a conventional correlation-based statistical technique.

[0038] FIG. 4 is a graph showing an example of approximate nearest neighbor search used in a method for automatically generating a plant condition monitoring model according to one embodiment.

[0039] Referring to Fig. 4, for example, artificial intelligence can perform prediction using approximate nearest neighbor search. The artificial intelligence can perform approximate nearest neighbor search by vectorizing past signal information, dividing the hyperplane to index the data distribution area into a tree structure, vectorizing signal measurement information, performing an index search, and deriving adjacent values. Since nuclear power plants repeat operation and planned preventive maintenance (OH) at regular intervals, the artificial intelligence can easily utilize the latest data from the previous cycle for the new signal monitoring model (400). The new signal monitoring model (400) can increase the correlation from the physical model perspective and can have excellent prediction accuracy and computation speed even at low correlations. The Approximate Nearest Neighbor Search (ANNS) prediction of the new signal monitoring model (400) can have higher prediction accuracy than the conventional Auto Associative Kernel Regression (AAKR) statistical prediction.

[0040] Approximate nearest neighbor search (ANN) can include various known ANN methods. For example, various ANN predictions can be performed by AI powered by a computing device. AI can perform sophisticated predictions for highly correlated signals through signal reconstruction using independent component analysis (ICA) of multivariate signals, compared to statistical techniques, and automatically separate out signals with low correlation. The separated signals can then be combined into signal clusters from adjacent equipment signals based on the plant's physical system structure. Using ANN methods, AI can then find the most similar values ​​from past data to validate the prediction quality.

[0041] Next, based on independent component analysis, a second separation signal (SS2) with low correlation among multiple second surveillance signals (MS2) is reconstructed into a new signal surveillance model (400) (S400).

[0042] For example, based on independent component analysis, at least one second separation signal (SS2) with low correlation among multiple second surveillance signals (MS2) is reconstructed using the new signal surveillance model (400) described above. The new signal surveillance model (400) can perform approximate nearest neighbor search prediction based on the physical classification system of the nuclear power plant.

[0043] The new signal monitoring model (400) can automatically reorganize adjacent signals based on a physical model of a nuclear power plant. For example, the new signal monitoring model (400) can automatically reorganize adjacent signals, such as vibration and pressure signals of the same pump or temperature and pressure signals of the same system line, based on a physical model of a nuclear power plant. The new signal monitoring model (400) can perform predictions using an approximate nearest neighbor search, which quickly searches for the most similar values ​​from recent past information, rather than a conventional correlation-based statistical technique.

[0044] Referring to Fig. 4, for example, artificial intelligence can perform prediction using approximate nearest neighbor search. The artificial intelligence can perform approximate nearest neighbor search by vectorizing past signal information, dividing the hyperplane to index the data distribution area into a tree structure, vectorizing signal measurement information, performing an index search, and deriving adjacent values. Since nuclear power plants repeat operation and OH at regular intervals, the artificial intelligence can easily utilize the latest data from the previous cycle for the new signal monitoring model (400). The new signal monitoring model (400) can increase the correlation from the physical model perspective and have excellent prediction accuracy and computation speed even at low correlations. The Approximate Nearest Neighbor Search (ANNS) prediction of the new signal monitoring model (400) can have higher prediction accuracy than the conventional Auto Associative Kernel Regression (AAKR) statistical prediction.

[0045] Approximate nearest neighbor search (ANN) can include various known ANN methods. For example, various ANN predictions can be performed by AI powered by a computing device. AI can perform sophisticated predictions for highly correlated signals through signal reconstruction using independent component analysis (ICA) of multivariate signals, compared to statistical techniques, and automatically separate out signals with low correlation. The separated signals can then be combined into signal clusters from adjacent equipment signals based on the plant's physical system structure. Using ANN methods, AI can then find the most similar values ​​from past data to validate the prediction quality.

[0046] Next, based on independent component analysis, the third separation signal (SS3) with low correlation among multiple third surveillance signals (MS3) is reconstructed into a new signal surveillance model (400).

[0047] For example, based on independent component analysis, at least one third separation signal (SS3) with low correlation among multiple third surveillance signals (MS3) is reconstructed using the new signal surveillance model (400) described above. The new signal surveillance model (400) can perform approximate nearest neighbor search prediction of the first separation signal (SS1), the second separation signal (SS2), and the third separation signal (SS3) based on the physical classification system of the nuclear power plant.

[0048] The new signal monitoring model (400) can automatically reorganize adjacent signals based on a physical model of a nuclear power plant. For example, the new signal monitoring model (400) can automatically reorganize adjacent signals, such as vibration and pressure signals of the same pump or temperature and pressure signals of the same system line, based on a physical model of a nuclear power plant. The new signal monitoring model (400) can perform predictions using an approximate nearest neighbor search, which quickly searches for the most similar values ​​from recent past information, rather than a conventional correlation-based statistical technique.

[0049] Referring to Fig. 4, for example, artificial intelligence can perform prediction using approximate nearest neighbor search. The artificial intelligence can perform approximate nearest neighbor search by vectorizing past signal information, dividing the hyperplane to index the data distribution area into a tree structure, vectorizing signal measurement information, performing an index search, and deriving adjacent values. Since nuclear power plants repeat operation and OH at regular intervals, the artificial intelligence can easily utilize the latest data from the previous cycle for the new signal monitoring model (400). The new signal monitoring model (400) can increase the correlation from the physical model perspective and have excellent prediction accuracy and computation speed even at low correlations. The Approximate Nearest Neighbor Search (ANNS) prediction of the new signal monitoring model (400) can have higher prediction accuracy than the conventional Auto Associative Kernel Regression (AAKR) statistical prediction.

[0050] Approximate nearest neighbor search (ANN) can include various known ANN methods. For example, various ANN predictions can be performed by AI powered by a computing device. AI can perform sophisticated predictions for highly correlated signals through signal reconstruction using independent component analysis (ICA) of multivariate signals, compared to statistical techniques, and automatically separate out signals with low correlation. The separated signals can then be combined into signal clusters from adjacent equipment signals based on the plant's physical system structure. Using ANN methods, AI can then find the most similar values ​​from past data to validate the prediction quality.

[0051] FIG. 5 is a flowchart illustrating an example of a method for automatically generating a plant status monitoring model according to one embodiment.

[0052] Referring to FIG. 5, as an example of a method for automatically generating a plant condition monitoring model according to one embodiment, an artificial intelligence driven by a computing device automatically generates a signal model by using an independent component analysis method of a dimensionality reduction method when constructing a condition monitoring signal model, and performs restoration prediction by automatically constructing a signal model by combining principal components with high correlation with adjacent equipment, and performs prediction by using an approximate nearest neighbor search. The artificial intelligence performs signal restoration prediction by grouping principal components with high correlation together, thereby improving prediction accuracy, and at the same time, for signals with low correlation, the artificial intelligence groups adjacent signals together into a signal model based on the physical classification system of the nuclear power plant, and performs prediction by finding the most similar value in past data through an approximate nearest neighbor search method.

[0053] In the case of conventional statistical analysis prediction, there is a problem that the prediction accuracy of the entire model is reduced when the correlation between signals is weakened, but the method for automatically generating a plant condition monitoring model according to one embodiment can solve the conventional problem and secure objectivity because it separates based on the correlation of pure data rather than the experience of an engineer.

[0054] In addition, the method for automatically generating a plant condition monitoring model according to one embodiment can operate the system more efficiently through automation.

[0055] For example, a method for automatically generating a nuclear power plant condition monitoring model using approximate nearest neighbor search is provided, which automatically generates a plant condition monitoring model with improved prediction accuracy while reducing model generation time and generation cost, even if the monitoring signals received from the plant condition monitoring model include monitoring signals with low correlation.

[0056] Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements made by those skilled in the art using the basic concept of the present invention defined in the following claims also fall within the scope of the present invention.

[0057] [Explanation of symbols]

[0058] Plant condition monitoring model (1000), first facility model (100), second facility model (200), third facility model (300), new signal monitoring model (400)

Claims

1. A step of collecting a plurality of first monitoring signals received in a first equipment model of a plant status monitoring model of a nuclear power plant and a plurality of second monitoring signals received in a second equipment model of the plant status monitoring model; A step of performing independent component analysis of each of the plurality of first surveillance signals and the plurality of second surveillance signals; A step of reconstructing a first separation signal with low correlation among the plurality of first surveillance signals based on the independent component analysis into a new signal surveillance model; and A step of reconstructing a second separation signal with low correlation among the plurality of second surveillance signals based on the independent component analysis into the new signal surveillance model. A method for automatically generating a plant condition monitoring model including:

2. In paragraph 1, The above new signal monitoring model is a method for automatically generating a plant condition monitoring model that performs approximate nearest neighbor search prediction based on the physical classification system of the nuclear power plant.

3. In paragraph 1, A method for automatically generating a plant condition monitoring model, wherein the step of performing the above independent component analysis is performed by applying independent component analysis based on dimension reduction.

4. In paragraph 1, A method for automatically generating a plant status monitoring model, wherein the step of collecting the plurality of first monitoring signals and the plurality of second monitoring signals includes the step of collecting the plurality of third monitoring signals received in a third equipment model of the plant status monitoring model.

5. In paragraph 4, A method for automatically generating a plant status monitoring model, wherein the step of performing the independent component analysis includes the step of performing the independent component analysis of each of the plurality of third monitoring signals.

6. In paragraph 5, A method for automatically generating a plant condition monitoring model, further comprising a step of reconstructing a third separation signal with low correlation among the plurality of third monitoring signals based on the independent component analysis into the new signal monitoring model.

Images