KR-20260065147-A - ARTIFICIAL INTELLIGENCE LEARNING FEATURE EXTRACTION METHOD AND SYSTEM FOR MONITORING THE STATUS OF ROTATING EQUIPMENT AND DIAGNOSING FAULTS

KR20260065147AKR 20260065147 AKR20260065147 AKR 20260065147AKR-20260065147-A

Abstract

The present invention relates to an artificial intelligence-based method for monitoring the condition of rotating equipment and diagnosing faults, comprising: a data acquisition step of acquiring vibration data from a plurality of vibration sensors mounted on rotating equipment by at least one processor of a computer system; a CDP image generation step of generating a Correlated Dot Pattern (CDP) image by converting the acquired vibration data from the time domain to a polar coordinate system; a step of learning an artificial intelligence model related to monitoring the condition of rotating equipment and diagnosing faults based on the generated CDP image; and a step of verifying a fault diagnosis result based on the learning. Furthermore, the present invention relates to an artificial intelligence-based system for monitoring the condition of rotating equipment and diagnosing faults, comprising: a data extraction unit for acquiring vibration data from a plurality of vibration sensors mounted on rotating equipment; a CDP image generation unit for generating a CDP image by converting the acquired vibration data from the time domain to a polar coordinate system; an artificial intelligence model learning unit for learning an artificial intelligence model related to monitoring the condition of rotating equipment and diagnosing faults based on the generated CDP image; and a fault diagnosis unit for verifying a fault diagnosis result based on the learning. Accordingly, the performance of artificial intelligence-based monitoring the condition of rotating equipment and diagnosing faults can be significantly improved.

Inventors

장대식
이정한
고태영
전지현

Assignees

한국원자력연구원

Dates

Publication Date: 20260508
Application Date: 20241101

Claims (11)

In an artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method, by at least one processor of a computer system, A data acquisition step for acquiring vibration data from a plurality of vibration sensors mounted on a rotating equipment; A CDP image generation step that converts the acquired vibration data from the time domain to a polar coordinate system to generate a CDP (Correlated Dot Pattern) image; A step of training an artificial intelligence model related to monitoring the condition of rotating equipment and diagnosing faults based on the CDP image generated above; and Artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method comprising the step of verifying the fault diagnosis result based on the above learning.
An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method according to claim 1, characterized in that a plurality of vibration sensors mounted on the rotating equipment are composed of various combinations of one or more acceleration sensors and one or more displacement sensors.
In Article 1, In the above CDP image generation step, the following mathematical formulas (1), (2) and (3): (1) (2) (3) (Here, x represents vibration data from an arbitrary sensor, y represents vibration data from another arbitrary sensor other than x, i represents a data sequence, τ represents an arbitrary integer representing the increment value of the data sequence, r represents the polar coordinate radius, θ represents the polar coordinate angle, φ represents the polar coordinate angle interval setting parameter, and ζ represents the angle gain value) An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method characterized by converting to a polar coordinate system using
In Article 1, An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method characterized by further including the step of applying an autocorrelation function before converting the acquired vibration data into a polar coordinate system, and then converting it into a polar coordinate system to generate an autocorrelation CDP image.
In Article 4, In the step of generating the above autocorrelation CDP image, The following mathematical formula (4): (4) (Here, x represents vibration data from an arbitrary sensor, i represents a data sequence, and τ represents an arbitrary integer representing the increment value of the data sequence) An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis method characterized by applying an autocorrelation function using
In an artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system, A data extraction unit that acquires vibration data from multiple vibration sensors mounted on a rotating equipment; A CDP image generation unit that converts the acquired vibration data from the time domain to a polar coordinate system to generate a CDP (Correlated Dot Pattern) image; An artificial intelligence model learning unit that learns an artificial intelligence model related to monitoring the status of rotating equipment and diagnosing faults based on the CDP image generated above; and An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system comprising a fault diagnosis unit that verifies the fault diagnosis results based on the above learning.
An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system according to claim 6, characterized in that the plurality of vibration sensors mounted on the rotating equipment are composed of various combinations of one or more acceleration sensors and one or more displacement sensors.
In Article 6, In the above CDP image generation unit, the following mathematical formulas (1), (2) and (3): (1) (2) (3) (Here, x represents vibration data from an arbitrary sensor, y represents vibration data from another arbitrary sensor other than x, i represents a data sequence, τ represents an arbitrary integer representing the increment value of the data sequence, r represents the polar coordinate radius, θ represents the polar coordinate angle, φ represents the polar coordinate angle interval setting parameter, and ζ represents the angle gain value) An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system characterized by converting to a polar coordinate system using
In Article 6, An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system characterized by further including an autocorrelation CDP image generation unit that generates an autocorrelation CDP image by applying an autocorrelation function before converting vibration data acquired from the above data extraction unit into a polar coordinate system, and then converting it into a polar coordinate system.
In Article 9, In the above autocorrelation CDP image generation unit, The following mathematical formula (4): (4) (Here, x represents vibration data from an arbitrary sensor, i represents a data sequence, and τ represents an arbitrary integer representing the increment value of the data sequence) An artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system characterized by applying an autocorrelation function using
A computer-readable recording medium storing a computer program for executing a method according to any one of claims 1 to 5 on a computer system.

Description

Artificial Intelligence Learning Feature Extraction Method and System for Monitoring the Status of Rotating Equipment and Diagnosing Faulsions The present invention relates to a method and system for extracting features for artificial intelligence learning for monitoring the condition of rotating equipment and diagnosing defects. Various types of rotating equipment are installed and operated in industrial plant or process facilities. If unexpected shutdowns occur due to failures in rotating equipment, which plays a critical role in the operation of these plants and process facilities, it can result in significant economic losses and safety accidents. Accordingly, most facilities install Vibration Monitoring Systems (VMS) on key rotating equipment to monitor its status. When an abnormal condition is detected via VMS, experts perform precise diagnoses of defects and failures through vibration signal analysis, and decisions regarding the maintenance of rotating equipment are made based on the results. This diagnostic technique relies heavily on experts and can lead to errors where diagnostic results vary depending on the expert's experience and competence. Above all, there is a limitation in that it takes a significant amount of time from the moment a vibration signal corresponding to an abnormal condition is transmitted to an expert until a maintenance decision is made through signal analysis and precise diagnosis. To overcome these limitations and automate diagnostic systems, active research and development on AI-based rotating equipment diagnostic technology has recently been underway. Developing AI-based rotating equipment diagnostic technology requires vibration signal preprocessing and feature extraction, and diagnostic accuracy is significantly affected by the quality of the feature data used for training the AI model. Feature data used for training diagnostic models of rotating equipment include statistical moments, orbits, and time-frequency domain images; however, recently, many studies have been attempting to utilize Symmetrized Dot Pattern (SDP) images as features due to their advantage of being robust against noise. SDP is a technique that visualizes a time-domain vibration signal by converting it into a polar coordinate domain using its maximum, minimum, and angle gain values. However, while techniques utilizing SDP images as described above allow for signal normalization and are robust against noise, they have limitations in that they are less sensitive to defects; therefore, a new technology is needed to overcome these limitations. FIG. 1 is a conceptual diagram exemplarily showing the positions of sensors according to one embodiment of the present invention in a rotating equipment including a rotating part and a bearing. FIG. 2 is a flowchart showing each step of an artificial intelligence-based rotating equipment condition monitoring and defect diagnosis method according to one embodiment of the present invention. FIG. 3 is a configuration diagram showing each component of an artificial intelligence-based rotating equipment condition monitoring and fault diagnosis system according to one embodiment of the present invention. Figure 4 is a comparison diagram showing the difference in data sensitivity between SDP and CDP according to normal and abnormal states according to one embodiment of the present invention. Figure 5 is a comparison diagram showing the difference in data sensitivity between SDP and CDP according to the rotational speed of the shaft according to one embodiment of the present invention. Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. However, the present invention is not limited by the following embodiments. In addition, the same reference numerals are used for identical components in the drawings, and redundant descriptions thereof are omitted. FIG. 1 is a conceptual diagram exemplarily showing the positions of sensors according to an embodiment of the present invention in a rotating equipment including a rotating part and bearings. Referring to FIG. 1, a rotating equipment according to an embodiment of the present invention includes a rotating part composed of a blade (16) and a rotating shaft (15) and bearings (11a, 11b), and a plurality of vibration sensors, such as acceleration sensors (12a, 12b) and displacement sensors (13a, 13b, 14a, 14b), are mounted on the rotating equipment and can be used to acquire data for monitoring the condition of the rotating equipment and diagnosing defects. In this regard, according to one embodiment of the present invention, a plurality of vibration sensors mounted on a rotating equipment may be composed of a first acceleration sensor (12a) on one side bearing (11a) and a second acceleration sensor (12b) on the other side bearing (11b), a first x-axis displacement sensor (13a) and a first y-axis displacement sensor (14a) on a rotation axis (15) located between one side