KR102336648B1 - Method for prediting the lifetime of high temperature parts based on cloud circumstance and cloud server therefor - Google Patents

Abstract

The present invention relates to a cloud environment-based high-temperature component life prediction method and a cloud server for the same, and the cloud environment-based high-temperature component lifetime prediction method according to an embodiment of the present invention includes (a) for evaluation of damage grade of high-temperature components of power generation facilities constructing a deep learning model for each high-temperature component material and storing and managing the high-temperature component damage rating evaluation result in a database; (b) deriving a damage grade probability distribution with respect to the operation time or the number of start-stops using the high-temperature component damage grade evaluation result; and (c) deriving a future damage rate estimation curve through the damage grade probability distribution.

Description

A method for predicting the lifespan of high-temperature parts based on a cloud environment and a cloud server for the same

The present invention relates to a cloud environment-based high-temperature component life prediction method and a cloud server for the same, and more particularly, by checking the damage grade evaluation result and high-temperature component life prediction for high-temperature component material images in the field, it is accurate for power generation facilities And to quickly evaluate the damage grade of high-temperature parts and predict the lifespan of high-temperature parts, to a cloud environment-based high-temperature parts life prediction method and a cloud server for the same.

Since most parts of power generation equipment are exposed to high temperature operation for a long time, it is necessary to accurately evaluate and manage the condition of high temperature parts for stable operation.

For example, boiler tubes and pipes are exposed to a high temperature environment for a long time because they are used for the purpose of generating and transporting steam. Thus, boiler tubes and piping are constantly subject to deterioration due to material damage, particularly creep damage.

Creep damage means permanent deformation at high temperature over time, and high stress is generated for a long time in an actual high-temperature environment, which reduces the strength of the part and ultimately shortens the life of the part.

That is, when creep damage occurs in the boiler tube and piping, pores grow in the microstructure to create cracks, and the size of the carbides increases, thereby increasing the gap between the carbides and decreasing the hardness.

In the past, the condition of high-temperature parts is evaluated by copying the damaged part of the material with a film at the power plant site and then observing it using a microscope in the laboratory or evaluating it with the naked eye.

In this evaluation method, not only the evaluation result is influenced by the subjective judgment and experience of an individual, but also it takes a long time from measurement in the field to analysis and judgment in the laboratory.

In addition, since it is difficult to clearly distinguish between the grades used in the evaluation criteria in this evaluation method, it is necessary to introduce a quantitative, systematic and objective evaluation method.

Therefore, in order to accurately and quickly evaluate damage to high-temperature components in actual power plants and industrial facilities, there is a need for a method capable of immediately executing accurate damage analysis after acquiring image data from the field.

Korean Patent Publication No. 10-2009-0071685

It is an object of the present invention to accurately and quickly evaluate the damage grade of high-temperature parts for power generation facilities by confirming the damage grade evaluation result and the high-temperature part lifetime prediction for the high-temperature part material image in the field, and to predict the high-temperature part lifespan, An object of the present invention is to provide a method for predicting the lifespan of high-temperature parts based on a cloud environment and a cloud server for the same.

The cloud environment-based high-temperature component lifetime prediction method according to an embodiment of the present invention comprises (a) a deep learning model for each high-temperature component material to evaluate the high-temperature component damage grade of a power generation facility, and stores the high-temperature component damage grade evaluation result in a database and managing; (b) deriving a damage grade probability distribution with respect to the number of start-stops using the high-temperature component damage grade evaluation result; and (c) deriving a future damage rate estimation curve through the damage grade probability distribution.

The step (a) may include a step of evaluating the damage grade of the evaluation target as the high-temperature component material image of the evaluation target is input.

According to an embodiment, after the step (c), when the number of times of starting and stopping of the evaluation target is determined and input, predicting the lifespan of the evaluation target according to the future damage rate estimation curve; may further include .

The step (a) may include performing machine learning on the deep learning model using the high-temperature component material images collected from each power generation facility in a remote location.

The deep learning model may be a Convolution Neural Network (CNN) model.

Variables of the CNN model may include variables representing the number of repetitions and reduction rates among the condition values for existing materials and new materials of high-temperature parts, and variables for classifying damage evaluation grades for each material.

The variable of the CNN model may be assigned a weight that is set and assigned by a classification criterion to determine the degree of damage to the high-temperature component.

The damage grade probability distribution may be expressed as a probability distribution with respect to the crack length according to the number of start-up stops.

In addition, as a cloud server according to an embodiment of the present invention, at least one or more processors; and a memory for storing computer-readable instructions, wherein the instructions, when executed by the at least one processor, cause the cloud server to dip each high-temperature component material in order to evaluate the high-temperature component damage rating of the power generation facility. The learning model is configured to store and manage the damage grade evaluation result of high-temperature parts in the database, and the damage grade probability distribution for the operation time or the number of start-stops is derived using the damage grade evaluation result of the high-temperature parts, and the damage grade It may be to derive a future damage rate estimation curve through probability distribution.

The instructions, when executed by the at least one processor, cause the cloud server to store and manage the high-temperature component damage grade evaluation result, damage to the evaluation target as the high-temperature component material image of the evaluation target is input It may be to evaluate the rating.

When the instructions are executed by the at least one processor, the cloud server causes the future damage rate estimation curve to be derived, and then, when the operation time or the number of start-stops of the evaluation target is determined and input, the future It may be to predict the lifespan of the evaluation target according to the damage rate estimation curve.

When the instructions are executed by the at least one processor, when the cloud server stores and manages the high-temperature component damage grade evaluation result, using the high-temperature component material image collected from each power generation facility in a remote location, the It may be to perform machine learning on a deep learning model.

In addition, as a computer-readable storage medium in which the program code according to an embodiment of the present invention is recorded, a deep learning model is constructed for each high-temperature part material to evaluate the damage grade of high-temperature parts of power generation facilities, and the high-temperature parts damage grade evaluation result is stored in the database. storage and management; deriving a damage grade probability distribution with respect to the operation time or the number of start-stops using the high-temperature component damage grade evaluation result; and deriving a future damage rate estimation curve through the damage grade probability distribution; it may be a computer-readable storage medium in which a program code for executing a cloud environment-based high temperature component life prediction method including a computer-readable storage medium is recorded.

The present invention can accurately and quickly evaluate the damage grade of high-temperature components for power generation facilities and predict the high-temperature component lifespan by confirming the damage grade evaluation result and the high-temperature component lifetime prediction for the high-temperature component material image in the field.

In addition, the present invention learns the damage grade determination result of the target material in a cloud environment using a deep learning technique, and uses the related model to automatically determine the damage grade of high-temperature parts only with images obtained through a portable microscope and smartphone camera. .

In addition, according to the present invention, after acquiring an image of a high-temperature component in the field, it is possible to evaluate the damage grade of the high-temperature component and to predict the lifespan within a short time.

In addition, the present invention can reduce the cost of equipment diagnosis required due to the metal texture evaluation test that is conducted regularly for high-temperature parts of power generation equipment.

In addition, the present invention can accurately evaluate the damage grade of high-temperature components compared to the subjective evaluation criteria by experts through mutual comparison with the visual discrimination criteria.

In addition, the present invention can be broadly applied to power facilities operated for a long time at high temperatures as well as power plant boiler tubes through image data learning of new materials.

1 is a diagram showing a system network according to an embodiment of the present invention;
2 and 3 are diagrams illustrating the configuration of computer readable instructions stored in a memory;
4 is a view showing a future damage rate estimation curve;
5 to 8 are views showing application or website screens provided to a user terminal in a cloud server;
9 is a diagram illustrating a cloud environment-based high-temperature component lifetime prediction method according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, detailed descriptions of well-known functions or configurations that may obscure the gist of the present invention in the following description and accompanying drawings will be omitted. Also, it should be noted that, throughout the drawings, the same components are denoted by the same reference numerals as much as possible.

The terms or words used in the present specification and claims described below should not be construed as being limited to conventional or dictionary meanings, and the inventor shall appropriately define his/her invention as terms for the best description of his/her invention. Based on the principle that it can be done, it should be interpreted as meaning and concept consistent with the technical idea of the present invention.

Accordingly, the embodiments described in this specification and the configurations shown in the drawings are only the most preferred embodiment of the present invention, and do not represent all of the technical spirit of the present invention. It should be understood that there may be equivalents and variations.

In the accompanying drawings, some components are exaggerated, omitted, or schematically illustrated, and the size of each component does not fully reflect the actual size. The present invention is not limited by the relative size or spacing drawn in the accompanying drawings.

In the entire specification, when a part "includes" a certain element, this means that other elements may be further included, rather than excluding other elements, unless otherwise stated. Also, when a part is said to be “connected” to another part, it includes not only a case in which it is “directly connected” but also a case in which it is “electrically connected” with another element interposed therebetween.

The singular expression includes the plural expression unless the context clearly dictates otherwise. Terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification is present, and includes one or more other features, number, or step. , it should be understood that it does not preclude in advance the possibility of the presence or addition of an operation, component, part, or combination thereof.

Also, as used herein, the term “unit” refers to a hardware component such as software, FPGA, or ASIC, and “unit” performs certain roles. However, "part" is not meant to be limited to software or hardware. A “unit” may be configured to reside on an addressable storage medium and may be configured to refresh one or more processors. Thus, by way of example, “part” includes components such as software components, object-oriented software components, class components and task components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. The functionality provided within components and “parts” may be combined into a smaller number of components and “parts” or further divided into additional components and “parts”.

Hereinafter, with reference to the accompanying drawings, the embodiments of the present invention will be described in detail so that those of ordinary skill in the art can easily implement them. However, the present invention may be embodied in several different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a diagram showing a system network according to an embodiment of the present invention, FIGS. 2 and 3 are diagrams for explaining the configuration of computer readable instructions stored in a memory, and FIG. 4 is a diagram showing a future damage rate estimation curve .

1, the system network according to an embodiment of the present invention stores data in a cloud server 110 connected through a wired/wireless communication network, and a user terminal 130 connects to the cloud server 110 through a wired/wireless communication network. It implements and configures a cloud environment in which data can be used anytime, anywhere simply by accessing it.

The cloud server 110 includes at least one processor 111 , a memory 112 for storing computer readable instructions, and a communication unit 113 enabling communication with the user terminal 130 through a wired/wireless communication network.

When the computer readable instructions stored in the memory 112 are executed by the at least one or more processors 111, the cloud server 110 may perform a cloud environment-based high-temperature component life expectancy method according to an embodiment of the present invention. .

Here, the processor 111 may also be referred to as a controller, a microcontroller, a microprocessor, a microcomputer, or the like.

In addition, the processor 111 may be implemented by hardware, firmware, software, or a combination thereof.

In addition, the memory 112 may be a single storage device, or may be a collective term for a plurality of storage elements. The computer readable instructions stored in memory 112 may be executable program code or parameters, data, and the like.

2 and 3, here, among the frameworks available in the form of a library, open source libraries TensorFlow and CUDA (Compute Unified Device Architecture, CUDA), web framework Django, and development tools (IDE) Use PyCharm, but use the programming language Python anaconda.

In addition, the memory 112 may include random access memory (RAM), or may include non-volatile memory (NVRAM) such as magnetic disk storage or flash memory.

Hereinafter, a process of performing the cloud environment-based high-temperature component lifetime prediction method according to an embodiment of the present invention in the cloud server 110 , that is, the processor 111 will be described in detail.

Specifically, the processor 111 may configure a deep learning model for each high-temperature component material for damage evaluation of high-temperature components of power generation facilities. For example, the processor 111 may use a convolutional neural network (hereinafter referred to as 'CNN') model. The CNN model consists of a convolution layer and a maxpooling layer.

Here, the variables of the CNN model include variables representing the number of repetitions and reduction rates among the condition values for the existing materials and new materials of high-temperature parts, and variables for classifying the damage evaluation grade for each material. These CNN model variables are given weights that are set and assigned by classification criteria to determine the degree of damage to high-temperature components.

At this time, the processor 111 executes machine learning on the CNN model by using the high-temperature component material image collected from each of the power generation facilities in the remote location. Accordingly, the reliability of the CNN model in evaluating the damage grade of high-temperature components for each high-temperature component material is improved. Here, the high-temperature component material image may be a high-temperature component image taken through a camera provided in the user terminal 130 or a mounted portable microscope.

Then, the processor 111 evaluates the high-temperature component damage grade for the evaluation target (high-temperature component material image) through the CNN model and stores it in the database 120 . At this time, the database 120 stores and manages the high-temperature component damage grade for each high-temperature component material.

In addition, the processor 111 uses the evaluation result of the damage grade of the high-temperature component stored and managed in the database 120 to derive an estimation curve of the damage grade of the high-temperature component according to the operating time or the number of start-stops to predict the lifespan of the high-temperature component. can

Specifically, the processor 111 may derive an estimation curve according to the damage grade probability distribution by using the evaluation result on the damage grade of the high-temperature component accumulated and managed in the database 120 .

First, the processor 111 may be expressed as a damage grade probability distribution (%) with respect to the crack length (l) according to the operation time (t) or the number of start-stops (N).

That is, the processor 111 may synthesize the respective damage grade probability distributions for the driving time or the number of start-stops for the i evaluation objects, and may represent the damage grade probability distribution with a correlation equation as in Equation 1 below.

Here, l is the crack length, ti is the operation time, and Ni is the number of start-stops.

Referring to FIG. 4 , the processor 111 derives a future damage rate estimation curve (regression) through the correlation expression representing the damage grade probability distribution with respect to the operation time or the number of start-stops of Equation 1 above.

Thereafter, when the operation time or the number of start-stops of an arbitrary evaluation target is determined, the processor 111 may derive a future damage rate estimation curve, and may predict the lifespan of the evaluation target accordingly.

As described above, the processor 111 is capable of deriving a damage grade probability distribution and a future damage rate estimation curve for an arbitrary driving time or number of start-stops without additional analysis.

On the other hand, the processor 111 may perform high-temperature component damage evaluation by classifying the CNN model for the new material into advanced users who can add or modify and general users who can perform simple evaluation.

Here, an advanced user can store and manage variables related to machine learning and variables for each neural network node that determines damage rating in constructing a CNN model. In addition, the general user can only use the function to receive the evaluation result for the high-temperature component image to be evaluated in the form of a report.

Next, the database 120 stores and manages the damage grade evaluation result for each material of the high-temperature component as described above.

In addition, the database 120 stores and manages information such as basic information, equipment information, operation information, evaluation date, and evaluator for evaluation by the user, the number of repetitions among the condition values for the existing material and the new material of the high-temperature part, It stores and manages the main variables of the CNN model such as the reduction rate, and stores and manages the variables of the CNN model that classifies the damage grade by material.

Next, the user terminal 130 evaluates the damage grade of the high-temperature component, and uploads the high-temperature component material image of the evaluation target for predicting the lifespan of the high-temperature component to the cloud server 110 .

In addition, the user terminal 130 informs the user of the damage grade and the probabilistic evaluation result according to the standards for each material and each damage grade.

To this end, the user terminal 130 is pre-loaded with an application for evaluating the damage grade of high-temperature components and predicting the lifespan of high-temperature components to access the cloud server 110 or to the cloud through a website provided by the cloud server 110 . It is possible to connect to the server 110 . The cloud server 110 provides an application or website. 5 to 8 are diagrams illustrating screens of applications or websites provided from a cloud server to a user terminal. The user terminal 130 may receive the damage grade evaluation and life expectancy prediction result of the high-temperature component from the cloud server 110 in the form of a report (refer to FIG. 8 ).

The user terminal 130 is a terminal accessible to the cloud server 110 through a wired/wireless communication network, and may be, for example, a desktop PC, a notebook computer, a smart phone, an IPTV, a tablet PC, a smart watch, smart glasses, a wearable device, etc. have.

9 is a diagram illustrating a cloud environment-based high-temperature component lifetime prediction method according to an embodiment of the present invention.

The cloud server 110 configures a deep learning model, that is, a CNN model for each high-temperature component material for damage evaluation of high-temperature components of power generation facilities (S201). At this time, the cloud server 110 performs machine learning on the CNN model by using the high-temperature component material image collected from each power generation facility in a remote location.

Thereafter, the cloud server 110 may evaluate the damage grade of the evaluation target when the high-temperature component material image of the evaluation target is input and check the result (S202, S203).

The cloud server 110 stores and manages the damage rating evaluation result in the database 120 (S204). Accordingly, as the cloud server 110 integrates and manages the damage grade evaluation results of high-temperature parts remotely by various users from individual places, the database 120 is continuously updated, so the accuracy of the damage grade evaluation results of high-temperature parts is improved. It is also possible to predict the tendency of the damage grade.

Thereafter, the cloud server 110 derives the damage grade probability distribution for the operating time or the number of start-stops by using the damage grade evaluation result of the high-temperature component stored in the database 120 (S205).

Then, the cloud server 110 derives a future damage rate estimation curve through a correlation expression representing the damage grade probability distribution with respect to the driving time or the number of times of starting and stopping ( S206 ).

At this time, when the operation time or the number of start-stops of the evaluation target is determined and input (S207), the cloud server 110 may predict the lifespan of the evaluation target according to the future damage rate estimation curve (S208).

The method according to some embodiments may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CDROMs and DVDs, and magneto-optical disks such as floppy disks. hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like.

Although the foregoing description has focused on novel features of the invention as applied to various embodiments, those skilled in the art will recognize the apparatus and method described above without departing from the scope of the invention. It will be understood that various deletions, substitutions, and changes are possible in the form and details of Accordingly, the scope of the invention is defined by the appended claims rather than by the description above. All modifications within the scope of equivalents of the claims are included in the scope of the present invention.

110 ; cloud server
111; processor
112 ; Memory
113 ; communication department
120 ; database
130 ; user terminal

Claims (17)

(a) constructing a deep learning model for each high-temperature component material to evaluate the high-temperature component damage grade of the power generation facility, and storing and managing the high-temperature component damage grade evaluation result in a database;
(b) deriving a damage grade probability distribution with respect to the number of start-stops using the high-temperature component damage grade evaluation result; and
(c) deriving a future damage rate estimation curve through the damage grade probability distribution; and
(d) predicting the lifespan of the evaluation target according to the future damage rate estimation curve when the number of times of starting and stopping of the evaluation target is determined and input;
The step (a) is,
performing machine learning on the deep learning model using the image of the high-temperature component material collected at each remote location by the power generation facility;
Including; evaluating the damage grade of the evaluation target as the image of the high-temperature component material to be evaluated is input;
The damage grade probability distribution is a cloud environment-based high temperature component life prediction method that is represented by a probability distribution for crack length according to the number of start-up stops.
delete delete delete The method of claim 1,
The deep learning model is
A method of predicting the lifespan of high-temperature components based on a cloud environment, which is a Convolution Neural Network (CNN) model.
6. The method of claim 5,
The parameters of the CNN model are,
A cloud environment-based high-temperature component life prediction method that includes a variable representing the number of repetitions and a reduction rate among the condition values for the existing material and the new material of the high-temperature component, and a variable for classifying the damage grade for each material.
7. The method of claim 6,
The parameters of the CNN model are,
A cloud environment-based high-temperature component life prediction method in which a weight assigned to set by a classification criterion to determine the degree of damage to the high-temperature component is given.
delete delete delete delete delete delete delete delete delete delete

Images