KR-20260063536-A - SYSTEM AND METHOD FOR DIAGNOSING ABNORMAL CONDITIONS IN ROTATING EQUIPMENT

Abstract

A rotating body abnormality diagnosis system and method are introduced, comprising: a data collection unit for collecting rotational time series data according to the rotation of a rotating body subject to abnormality diagnosis; a preprocessing unit for determining filtered time series data obtained by filtering the rotational time series data using a previously trained first model and determining a plurality of characteristic values based on the filtered time series data; and a judgment unit for determining whether the rotating body is abnormal based on the plurality of characteristic values using a previously trained second model and outputting an abnormality signal when the judgment unit determines an abnormality, wherein the preprocessing unit divides the rotational time series data into a plurality of datasets according to length, determines at least one time series characteristic that maintains a feature within each of the plurality of datasets using the first model, and determines the time series characteristic determined for each of the datasets as the plurality of characteristic values.

Inventors

• 방태형
• 조수봉
• 조재훈
• 노서연

Assignees

• 현대위아 주식회사

Dates

Publication Date

20260507

Application Date

20241030

Claims (20)

1. A data collection unit that collects rotational time series data according to the rotation of a rotating body subject to abnormal diagnosis; A preprocessing unit that determines filtered time series data obtained by filtering the rotational time series data using a previously trained first model, and determines a plurality of characteristic values based on the filtered time series data; and A judgment unit that uses a previously trained second model to determine whether the rotating body is abnormal based on the plurality of characteristic values, and outputs an abnormality signal when the judgment unit determines that it is abnormal, The above preprocessing unit is, A rotating body abnormality diagnosis system that divides the filtered time series data into multiple datasets based on time intervals, determines at least one time series characteristic that maintains a feature within each of the multiple datasets using the first model, and determines the time series characteristic determined for each of the datasets as the multiple characteristic values.
2. In Article 1, The above first model is, A rotating body anomaly diagnosis system that has been trained using a deep learning or machine learning-based non-linear time series analysis technique and a clustering technique, utilizing a plurality of training datasets and a plurality of training filtering time series datasets corresponding to each of the plurality of training datasets.
3. In Article 2, The above nonlinear time series analysis technique is, It includes non-linear time series alignment (DTW: Dynamic Time Warping), and The above clustering technique is, Rotating body anomaly diagnosis system including the K-means algorithm.
4. In Paragraph 3, The above time series characteristics are, A rotating body anomaly diagnosis system comprising at least one of the length, mean, standard deviation, maximum value, minimum value, maximum value location, skewness, kurtosis, Root Mean Square (RMS), Peak-to-Peak value, Crest Factor, and cluster information derived from the K-means algorithm of the filtered time series data.
5. In Article 1 The above characteristic value is, A rotating body abnormality diagnosis system further comprising an identification number of the rotating body subject to the above abnormality diagnosis.
6. In Article 1, The above second model is, A rotating body anomaly diagnosis system based on a deep learning technique, which is trained using multiple training feature values selected from filtered time series data filtered through a previous first model.
7. In Article 6, The above deep learning technique is, Rotating body anomaly diagnosis system including a One-Class SVM technique.
8. In Article 6, The above second model is, A rotating body abnormality diagnosis system that has been trained by resetting the hyperparameters of the second model until the abnormality determination result of the second model satisfies a preset optimization score, based on a plurality of verification feature values excluding the plurality of learning feature values from filtered time series data filtered through the previous first model.
9. In Article 1, A rotating body abnormality diagnosis system further comprising a learning unit that collects data in which the output abnormal signal is mis-detected, and retrains the first model and the second model when the mis-detected data exceeds a preset number or preset frequency.
10. In Article 1, The above data collection unit is, A rotating body abnormality diagnosis system that collects rotational time series data based on at least one of a numerical control device and a sensor.
11. A step in which a data collection unit collects rotational time series data according to the rotation of a rotating body subject to abnormal diagnosis; A preprocessing unit determines filtered time series data obtained by filtering the rotational time series data using a previously trained first model, and determines a plurality of characteristic values based on the filtered time series data; and A step in which a judgment unit determines whether there is an abnormality in the rotating body based on the plurality of characteristic values using a previously trained second model; and The above-mentioned judgment unit includes a step of outputting an abnormal signal when an abnormality is determined, A method for diagnosing an abnormality in a rotating body, wherein the preprocessing unit divides the filtered time series data into a plurality of datasets based on time intervals, determines at least one time series characteristic that maintains a feature within each of the plurality of datasets using the first model, and determines the time series characteristic determined for each of the datasets as the plurality of characteristic values.
12. In Article 11, The above first model is, A rotating body anomaly diagnosis method that has been previously trained using a deep learning or machine learning-based non-linear time series analysis technique and a clustering technique, utilizing a plurality of training datasets and a plurality of training filtering time series datasets corresponding to each of the plurality of training datasets.
13. In Article 12, The above nonlinear time series analysis technique is, It includes non-linear time series alignment (DTW: Dynamic Time Warping), and The above clustering technique is, A rotating body anomaly diagnosis method including a K-means algorithm.
14. In Article 13, The above time series characteristics are, A method for diagnosing anomalies in a rotating body, comprising at least one of the length, mean, standard deviation, maximum value, minimum value, maximum value location, skewness, kurtosis, Root Mean Square (RMS), Peak-to-Peak value, Crest Factor, and clustering information derived from the K-means algorithm of the filtered time series data.
15. In Article 11 The above characteristic value is, A method for diagnosing abnormalities in a rotating body, further comprising an identification number of the rotating body subject to the above-mentioned abnormality diagnosis.
16. In Article 11, The above second model is, A rotating body anomaly diagnosis method based on a deep learning technique, which is trained using multiple training feature values selected from filtered time series data filtered through a previous first model.
17. In Article 16, The above deep learning technique is, A rotating body anomaly diagnosis method including a One-Class SVM technique.
18. In Article 16, The above second model is, A method for diagnosing anomalies in a rotating body, which is previously trained by resetting the hyperparameters of the second model until the result of determining whether the second model is anomaly satisfies a preset optimization score, based on multiple verification feature values excluding multiple learning feature values from filtered time series data filtered through the previous first model.
19. In Article 11, A rotating body abnormality diagnosis method further comprising the step of a retraining unit collecting data in which the output abnormal signal is mis-detected, and retraining the first model and the second model when the mis-detected data exceeds a preset number or preset frequency.
20. In Article 11, The step of collecting the above data is, A method for diagnosing abnormalities in a rotating body, wherein the data collection unit collects rotational time series data based on at least one of a numerical control device and a sensor.

Description

System and Method for Diagnosing Abnormal Conditions in Rotating Equipment The present invention relates to a system and method for diagnosing abnormalities in a rotating body based on a previously learned model. A device that performs various mechanical tasks through rotational motion can be called a rotating body. Rotating bodies are widely used in various fields such as machine tools, generators, electric motors, and aircraft engines, and if a malfunction occurs in such a rotating body, it can have a significant impact on the stability and performance of the systems in those fields. There are various causes for abnormalities in rotating bodies, which primarily include mechanical defects resulting from wear, imbalance, misalignment, and bearing damage. In such cases, failure to diagnose these abnormalities early can lead to serious failures. Therefore, there is a need to propose a method for diagnosing abnormalities in rotating bodies. The matters described above as background technology are intended only to enhance understanding of the background of the present invention and should not be construed as an acknowledgment that they constitute prior art already known to those skilled in the art. FIG. 1 shows a rotating body abnormality diagnosis system according to an embodiment of the present invention. FIG. 2 shows an example of a rotating body abnormality diagnosis system according to one embodiment of the present invention performing a rotating body abnormality diagnosis. FIG. 3 is a flowchart of a rotary body abnormality diagnosis system according to one embodiment of the present invention. FIG. 4 is a flowchart illustrating the learning process of a first model according to an embodiment of the present invention. FIG. 5 is a flowchart illustrating the learning process of a second model according to one embodiment of the present invention. FIGS. 6 to 11 illustrate an example of clustered rotational time series data according to an embodiment of the present invention. FIG. 12 illustrates filtered time series data according to one embodiment of the present invention. Specific structural or functional descriptions of the embodiments of the present invention disclosed in this specification or application are merely illustrative for the purpose of explaining embodiments according to the present invention, and embodiments according to the present invention may be implemented in various forms and should not be interpreted as being limited to the embodiments described in this specification or application. Since embodiments according to the present invention may be subject to various modifications and may take various forms, specific embodiments are illustrated in the drawings and described in detail in this specification or application. However, this is not intended to limit embodiments according to the concept of the present invention to specific disclosed forms, and it should be understood that they include all modifications, equivalents, and substitutions that fall within the spirit and scope of the present invention. Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification. Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components regardless of drawing symbols are given the same reference number, and redundant descriptions thereof will be omitted. In the description of the following embodiments, the term "pre-set" means that the numerical value of a parameter is predetermined when the parameter is used in a process or algorithm. Depending on the embodiment, the numerical value of the parameter may be set when the process or algorithm starts or during the period in which the process or algorithm is executed. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification, and do not inherently possess distinct meanings or roles. In describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art may obscure the essence of the embodiments disclosed in this specification, such detailed description is omitted. Furthermore, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification, and the technical concept disclosed in this specification is not limited by the attached drawings; it should be understood that they include all modifications, equivalents, and substit