KR-102962053-B1 - Real-time Fault Prediction System and Method for Marine Rotating Machinery Based on Vibration-Current Correlation Analysis

KR102962053B1KR 102962053 B1KR102962053 B1KR 102962053B1KR-102962053-B1

Abstract

The present invention relates to a real-time fault prediction system and method for a ship rotor based on vibration-current correlation, comprising: a vibration sensor attached to the rotor to collect multi-axis vibration data; a current sensor to measure the current applied to the rotor; a data acquisition unit that sorts and stores vibration and current data based on the same time; a machine learning learning unit that extracts characteristic values for each rotor and learns a prediction model; a correlation analysis unit that analyzes the correlation between current and vibration and calculates estimated vibration based on the model; a real-time prediction unit that receives real-time current data and calculates the possibility of signs of failure; and an integrated notification unit that aggregates prediction results and performs status determination and notification according to set criteria.

Inventors

김응곤
최양열
양성미

Assignees

주식회사 지노스

Dates

Publication Date: 20260507
Application Date: 20250708

Claims (14)

A vibration sensor (110) installed on the shaft of a rotating body to measure multi-axis vibration data in real time; A current sensor (120) that measures the current applied to the above-mentioned rotating body in real time; A data acquisition unit (130) that stores data collected from the above vibration sensor (110) and current sensor (120) in a synchronized form by aligning them based on the same time standard; A machine learning learning unit (140) that preprocesses the above-mentioned synchronized data according to the type of rotating body, extracts multidimensional vibration and current characteristic values for each rotating body, and learns a prediction model; A correlation analysis unit (150) that analyzes the correlation between the prediction result based on vibration data and the prediction result based on current data for each rotating body type based on the prediction model generated by the machine learning learning unit (140), and calculates the estimated vibration signal using the current data as input; A real-time prediction unit (160) that calculates the probability of a fault indication occurring based on real-time input current data by referring to the correlation analysis results for each rotating body generated by the above correlation analysis unit (150); and It includes an integrated notification unit (170) that aggregates fault indication information for each rotating body calculated by the real-time prediction unit (160), determines the state of the rotating body according to a pre-set boundary standard, and transmits it to an external system. The above vibration sensor (110) is installed on the shaft surface of the ship's rotating body and is composed of an inertial measurement module that integrates an accelerometer and a gyroscope to individually measure acceleration in the linear direction and angular velocity in the rotational direction, and separates and outputs vibration components according to the rotational state of the ship's rotating body from the measured multi-axis signal. The above vibration sensor (110) includes a piezoelectric or MEMS-based acceleration sensor and is configured to independently measure three-axis reference vibrations corresponding to the rotation axis direction, vertical direction, and horizontal direction of the ship's rotating body. The above vibration sensors (110) are distributed in multiple different locations, and each vibration sensor is defined with a unique identifier (ID), mounting location information, and coordinate-based direction information as preset metadata. The above vibration sensor (110) operates with a synchronization signal according to the setting of a reference sampling period for monitoring the operating status of the ship's rotating body, and when sampling of 1 kHz or more is set, records the amount of change in vibration magnitude at time intervals. Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
delete
In paragraph 1, The above current sensor (120) is, It is composed of a current sensing element placed around a power supply line to detect changes in current applied to a rotating body in a non-contact manner, and A method for continuously measuring changes in current waveforms over time by selecting and applying at least one of the electromagnetic induction method or the Hall effect method depending on the type of rotating body. Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
In paragraph 1, The above data acquisition unit (130) is, The method involves sorting data collected from vibration sensors (110) and current sensors (120) according to the collection time criteria of each sensor, and generating fixed section unit analysis input data that enables state analysis of each rotating body from the synchronized data. Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
In paragraph 1, The machine learning learning unit (140) above is, Based on vibration and current data collected for each rotating body, multiple vibration frequency components and current components representing the failure characteristics of the rotating body are extracted, and Constructing independent learning models according to each rotating body type to perform learning for fault determination per rotating body, Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
In paragraph 1, The above correlation analysis unit (150) is, Calculating a statistical correlation coefficient between vibration signals and current signals extracted for each rotating body, and generating a correlation model for each rotating body so that vibration signals can be estimated using current data as input only for failure types exceeding a preset threshold, Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
In paragraph 1, The above integrated notification unit (170) is, Collecting fault indication information for each rotating body calculated by the real-time prediction unit (160), and performing an alarm output that distinguishes the state of the rotating body according to a preset judgment criterion and transmits it to an external system. Vibration-Current Correlation-Based Real-Time Ship Rotor Failure Prediction System
A step of measuring multi-axis vibration data generated in the shaft of a rotating body through a vibration sensor; A step of measuring the current applied to the rotating body in real time through a current sensor; A step of storing the vibration data and current data in a synchronized form by aligning them based on the same time standard through a data acquisition unit; A step of preprocessing the synchronized data by rotating body type through a machine learning training unit, and extracting multidimensional vibration and current characteristic values for each rotating body to train a prediction model; A step of analyzing the correlation between a prediction result based on vibration data and a prediction result based on current data through a correlation analysis unit based on the prediction model, and calculating a vibration signal estimated using the current data as input; A step of calculating the probability of a fault indication occurring based on current data input in real time through a real-time prediction unit; and The method includes the step of aggregating the fault indication information through an integrated notification unit, determining the state of the rotating body according to a pre-set boundary criterion, and transmitting it to an external system. The step of measuring the multi-axis vibration data comprises: individually measuring linear acceleration and rotational angular velocity on the shaft surface of the rotating body through a vibration sensor, and separating and outputting vibration components according to the rotational state from the measured multi-axis signal. The above vibration sensor is installed on the shaft surface of the ship's rotating body and is composed of an inertial measurement module integrating an accelerometer and a gyroscope that individually measure acceleration in the linear direction and angular velocity in the rotational direction, and separates and outputs vibration components according to the rotational state of the ship's rotating body from the measured multi-axis signal. The above vibration sensor includes a piezoelectric or MEMS-based accelerometer and is configured to independently measure three-axis reference vibrations corresponding to the rotation axis direction, vertical direction, and horizontal direction of the ship's rotating body. The above vibration sensors are distributed in multiple different locations, and for each vibration sensor, a unique identifier (ID), mounting location information, and coordinate-based direction information are defined as pre-set metadata. The above vibration sensor operates with a synchronization signal according to the setting of a reference sampling period for monitoring the operating status of the ship's rotating body, and records the amount of change in vibration magnitude at time intervals when sampling of 1 kHz or higher is set, Real-time ship rotor failure prediction method based on vibration-current correlation.
delete
In paragraph 8, The step of measuring the above current in real time is, A step comprising: detecting a change in current applied to a rotating body in a non-contact manner through a current sensor, through a current sensing element placed around a power supply line, and continuously measuring a current waveform over time by applying one of the current sensor's sensing methods according to the type of rotating body. Real-time ship rotor failure prediction method based on vibration-current correlation.
In paragraph 8, The step of storing the above current data in a synchronized form by aligning it based on the same time standard is: A step comprising: sorting data collected from vibration sensors and current sensors through a data acquisition unit according to the collection time standard for each sensor, and generating analysis input data in fixed time interval units capable of state analysis for each rotating body based on the sorted data; Real-time ship rotor failure prediction method based on vibration-current correlation.
In paragraph 8, The step of training the above prediction model is, A step comprising: extracting multiple vibration frequency components and current components representing failure characteristics based on vibration and current data collected for each rotating body through a machine learning learning unit, configuring independent learning models according to each type of rotating body, and performing learning for failure determination for each rotating body; Real-time ship rotor failure prediction method based on vibration-current correlation.
In paragraph 8, The step of calculating an estimated vibration signal using the above current data as input is: The method comprises the step of calculating a statistical correlation coefficient between vibration signals and current signals extracted for each rotating body through a correlation analysis unit, and generating a correlation model for each rotating body so that vibration signals can be estimated using current data as input only for failure types that are above a pre-set standard. Real-time ship rotor failure prediction method based on vibration-current correlation.
In paragraph 8, The step of determining the state of the above-mentioned rotating body and transmitting it to an external system is, A step comprising: aggregating fault indication information for each rotating body calculated by the real-time prediction unit through the integrated notification unit, classifying the state according to preset judgment criteria, and transmitting an alarm message to an external system; Real-time ship rotor failure prediction method based on vibration-current correlation.

Description

Real-time Fault Prediction System and Method for Marine Rotating Machinery Based on Vibration-Current Correlation Analysis The present invention relates to a vibration-current correlation-based real-time fault prediction system and method for a ship rotor. More specifically, the invention relates to a vibration-current correlation-based real-time fault prediction system and method for a ship rotor configured to enable real-time diagnosis of the rotor's condition and linkage with external alarms by collecting multi-axis vibration data and current data generated from the shaft of a rotor within a ship in real-time based on sensors, synchronizing the collected heterogeneous data in time, and then estimating vibration signals from current data and determining signs of failure through a machine learning-based prediction model. Generally, rotating equipment installed on ships, such as generators, pumps, propellers, and fans, can fail due to complex factors including vibrations in the operating environment, load fluctuations, and fluid resistance; consequently, disruptions in ship operations have a significant impact on safety and operational efficiency. Predictive technologies have been developed in various forms to detect the condition of such rotating equipment in advance and determine whether any abnormalities exist. Conventionally, the primary method used involved collecting vibration data via vibration sensors attached to rotating bodies and determining abnormalities by analyzing changes in frequency components or amplitude. In some technologies, current sensors were used in parallel to identify sudden fluctuations in driving current or abnormal deformation of the current waveform as signs of failure. However, vibration sensor-based methods often made it difficult to reliably identify signs of failure due to issues such as installation location constraints, sensitivity to noise, and interference from the external environment; furthermore, relying solely on current data had limitations in precisely estimating signs resulting from changes in the structural state of the rotating body. In addition, the method of simply collecting vibration and current data in parallel and analyzing them individually had the problem that it was difficult to accurately predict the timing or type of failure of rotating bodies because it failed to fully utilize the correlation between the two datasets. In particular, given that various types and load conditions exist due to the characteristics of ship rotors, there is a need for advanced prediction technology that precisely constructs prediction models considering the specific characteristics of each rotor and analyzes heterogeneous sensor data by aligning and integrating it based on time. FIG. 1 is a diagram showing the overall configuration of a ship rotor real-time fault prediction system (100) based on vibration-current correlation according to one embodiment of the present invention. FIG. 2 is a flowchart showing the entire processing flow step by step through the vibration-current correlation-based real-time ship rotor failure prediction system (100) illustrated in FIG. 1. FIG. 3 is a configuration diagram showing the installation configuration of a multi-sensor for each ship rotor according to one embodiment of the present invention. FIG. 4 is a flowchart illustrating a machine learning learning model generation process for each rotating body according to an embodiment of the present invention. FIG. 5 is a diagram showing the results of vibration-current correlation analysis and modeling for each rotating body according to one embodiment of the present invention. FIG. 6 is a flowchart showing the real-time current-based rotating body failure prediction process according to one embodiment of the present invention in steps. FIG. 7 is a flowchart illustrating a method for predicting real-time failures of a ship rotor based on vibration-current correlation according to an embodiment of the present invention in a series of sequences. In describing the embodiments of this specification, if it is determined that a detailed description of known configurations or functions could obscure the essence of the embodiments of this specification, such detailed description is omitted. Additionally, parts of the drawings unrelated to the description of the embodiments of this specification have been omitted, and similar parts are denoted by similar reference numerals. In the embodiments of this specification, when a component is described as being "connected," "combined," or "joined" with another component, this may include not only a direct connection but also an indirect connection in which another component exists in between. Furthermore, when a component is described as "comprising" or "having" another component, this means that, unless specifically stated otherwise, it does not exclude the other component but may include additional components. In the embodiments of this specification, ter