회전기계 이상 진단을 위한 로터 다이나믹스 시뮬레이션 시스템 연구

최현준,이승주,김세원,김영석 한국산학기술학회 2023 한국산학기술학회논문지 Vol.24 No.12

With recent advancements in industrialization, the use of machinery has significantly increased. Unforeseen faults of machinery can lead to process interruptions and production halts, making accurate and rapid condition diagnosis crucial. Approximately 90% of all machinery consists of induction motors, which are rotating machines. Therefore, fault diagnosis for rotating machinery is particularly important. Failures in rotating machinery are primarily attributed to mechanical defects such as misalignment, unbalance, and cracks, and they manifest in the form of vibrations and noise. In this paper, a 1D numerical model was developed for operating rotating machinery. Rotor dynamics simulations were conducted to assess conditions involving parallel misalignment, angular misalignment, and unbalance faults. Through frequency spectrum analysis, vibration characteristics were analyzed according to the root causes of faults, with an observed increase in amplitude at specific frequencies depending on the fault type. The proposed rotor dynamics simulation system is expected to be valuable in the initial stages of constructing rotating machinery, especially when abnormal vibration data is not available. Also, it enables the establishment of vibration characteristics associated with fault conditions, facilitating the diagnosis of machinery faults and the evaluation of their root causes.

Wavelet Transform-based Identification of Vibration Fault Signals in Rotating Machinery

Yaping Zhao 대한전자공학회 2023 IEIE Transactions on Smart Processing & Computing Vol.12 No.4

The study of fault identification of vibration signals from rotating machinery is essential for enhancing industrial production safety. A method combining a capsule network and frequency-slicing wavelet transform is proposed to improve the fault identification accuracy, considering the problem that the original vibration signal of rotating machinery carries multiple noises. The capsule network learning model was also optimized using a dynamic weighting method based on the channel attention mechanism, considering the variable operating conditions of rotating machinery. The dynamic weighting algorithm based on the channel attention mechanism used in the study achieved the highest fault recognition rates, with 99.65%, 99.25%, and 99.90% on sensor 1, sensor 2, and feature fusion data, respectively. Hence, the proposed model for fault identification in rotating machinery vibration signals is superior to other models.

보편적 특징의 추출 및 선택에 기반한 회전체 기계의 고장 검출 및 진단

김민기 사단법인 한국융합기술연구학회 2023 아시아태평양융합연구교류논문지 Vol.9 No.12

Since most mechanical equipment includes rotating machines, rapid failure detection and diagnosis of rotating machinery is essential to manage mechanical equipment to operate normally. Traditional signal processing methods have the inconvenience of having to analyze the data characteristics of the domain and extract valid features manually every time the domain changes in order to diagnose a specific machine failure. In contrast, the methods using deep learning automatically extract valid features regardless of the domain, but they face the challenge of securing a large amount of data to train a deep neural network. In this study, we extract universal features widely used in traditional signal processing methods and apply the Relief-F algorithm to automatically select valid features. Finally input them into a shallow multi-layer perceptron (MLP) classifier, which can be trained with relatively little training data, to detect and diagnose machine failures regardless of the domain. As a result of applying the proposed method to the MaFaulDa dataset, it showed an accuracy of 99.95% for both fault detection and diagnosis when using a 256-dimensional feature vector. Even when the feature vector was reduced to 64 dimensions, the fault detection and diagnosis accuracy were 99.75% and 99.65%, respectively. These results show that the proposed method is effective in detecting and diagnosing failures in rotating machinery. 기계 설비는 대부분 회전하는 기계를 포함하고 있으므로, 기계 설비가 정상적으로 동작하도록 관리하기 위해서는 회전체 기계에 대한 신속한 고장 검출 및 진단이 필수적이다. 전통적인 신호 처리 방식은 특정한 기계의 고장을 진단하기 위하여 도메인이 바뀔 때마다 해당 도메인의 데이터 특성을 분석하고 연구자가 수작업으로 유효한 특징을 추출해야 하는 번거로움이 있다. 이에 반하여 딥 러닝을 이용한 방식은 도메인에 무관하게 자동으로 유효한 특징을 추출하는데 비하여 심층신경망을 학습시키기 위하여 다량의 데이터를 확보해야 하는 과제를 안고 있다. 본 연구에서는 전통적인 신호 처리 방식에서 널리 사용되는 보편적인 특징들을 추출한 후 Relief-F 알고리즘을 적용하여 유효한 특징을 자동으로 선별하였다. 이렇게 추출된 특징을 상대적으로 적은 학습데이터로 학습시킬 수 있는 깊이가 얕은 다층퍼셉트론(MLP) 분류기에 입력하여 도메인에 무관하게 기계의 고장을 검출 및 진단할 수 있는 방법을 제안한다. 제안한 방법을 MaFaulDa 데이터세트에 적용하여 실험한 결과 256차원의 특징 벡터를 사용하였을 때 고장 검출과 진단 모두 99.95%의 정확도를 보였다. 특징 벡터를 64차원으로 줄인 경우에도 고장 검출과 진단 정확도는 각각 99.75%, 99.65%를 보였다. 이러한 결과는 제안한 방식이 회전체 기계의 고장 검출 및 진단에 효과적임을 보여준다.

A rule-based classifier ensemble for fault diagnosis of rotating machinery

Dongyang Dou,Jian Jiang,Yuling Wang,Yong Zhang 대한기계학회 2018 JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY Vol.32 No.6

To predict potential problems and avoid an unexpected breakdown of rotating machinery, a rule-based classifier ensemble approach is presented. Feature reduction was first implemented on a fault decision table using discernibility matrices and the genetic algorithm. The generated rules of the reducts were used to build the candidate base classifiers. Then, several base classifiers were selected according to their diversity and scale. The weights of the selected base classifiers were also calculated based on the support rate measurements. A classifier ensemble was constructed through an integration of the base classifiers using an improved weighted voting technique. Finally, the proposed classifier ensemble was verified based on the vibration data of bearing types SKF6203 and NU205. The accuracy for the SKF6203 bearing type reached 88.75 %, which is at least 5 % higher than that of the three base classifiers for this type of bearing. In addition, the recognition rate for the latter bearing type was 90 %. The reasoning process was much easier to comprehend owing to the semantic descriptions of the rules. The results show that this is a promising and transparent approach for diagnosing typical faults of rotating machinery.

Minimum Entropy Deconvolution(MED) 필터의 유사도 기반 시간동기평균화를 통한 회전설비 결함진단

하종문 한국비파괴검사학회 2023 한국비파괴검사학회지 Vol.43 No.4

The minimum entropy deconvolution (MED) filter is widely employed in the extraction of weak fault-related features in vibration signals measured from a rotating machinery. However, the MED filter is vulnerable to external noise because the filter is designed to extract all impulsive fault-like features in the signal. Time synchronous averaging (TSA) can be used to solve this problem because it can extract repetitive fault-induced features while reducing the random noise via the ensemble averaging of the MED-filtered signal for each machinery cycle. However, it is difficult to extract the fault-related features via the TSA process if the MED-filtered signal does not have similar patterns for each unit cycle of the system. To solve this challenge, a similarity-based TSA process for the MED-filtered signal is proposed in this study. The proposed method involves the division of the MED-filtered signal into multiple segments based on the rotating frequency of the system, and the selection of the fault-related segments based on the inter-similarity score. Subsequently, the TSA of the selected signals is used to effectively extract the faulty features while reducing the effect of the noise. The performance of the proposed method is then validated using a planetary gearbox testbed. Minimum Entropy Deconvolution(MED) 필터는 복잡한 진동 신호에서 충격성(Impulsive) 형태를 띄는회전 설비의 미세 결함특성을 추출하기 위해 널리 사용되는 신호분석 기법이다. 하지만 MED 필터는 신호내에 포함된 모든 충격성 결함 유사 특성을 강조하기 때문에 외부 노이즈에 민감하다는 단점을 갖는다. 이를해결하기 위해 회전설비의 회전주기별로 MED 필터링 결과를 분할한 후 앙상블(Ensemble) 평균을 취하는 시간동기평균화 기법을 도입하여 반복 결함신호의 추출 및 노이즈 제거가 가능하다. 하지만 회전설비의 특성상매 회전마다 같은 결함특성이 발생하지 않는다면 시간동기평균화를 통한 결함특성의 효과적인 추출이 불가하다. 이 논문에서는 이러한 문제를 해결하기 위해 MED 필터링 신호의 유사도 기반 시간동기 평균화기법을제안한다. 제안 기법은 MED 필터링 신호를 회전 주기별로 나누고, 각 신호에 대한 유사성에 기반하여 결함특성을 공통적으로 가지는 신호를 선별한다. 이후 선별된 신호에 대한 시간동기평균화를 통해 결함과 관련된특성만을 추출하고, 결함과 무관한 노이즈성 신호를 제거하는 것을 특징으로 한다. 유성기어박스 테스트베드를 활용한 검증 결과 제안된 기법은 기존의 MED 필터 및 시간동기평균화기법 대비 고장진단 성능이 뛰어남을 확인하였다.

자기조직화특징지도와 학습벡터양자화를 이용한 회전기계의 이상진동진단 알고리듬

양보석,서상윤,임동수,이수종 한국소음진동공학회 2000 소음 진동 Vol.10 No.2

The necessity of diagnosis of the rotating machinery which is widely used in the industry is increasing. Many research has been conducted to manipulate field vibration signal data for diagnosing the fault of designated machinery. As the pattern recognition tool of that signal, neural network which use usually back-propagation algorithm was used in the diagnosis of rotating machinery. In this paper, self-organizing feature map(SOFM) which is unsupervised learning algorithm is used in the abnormal defect diagnosis of rotating machinery and then learning vector quantization(LVQ) which is supervised learning algorithm is used to improve the quality of the classifier decision regions.

Feature-Based Analysis for Fault Diagnosis of Gas Turbine using Machine Learning and Genetic Algorithms

안병현(Byung Hyun Ahn),유현탁(Hyeon Tak Yu),최병근(Byeong Keun Choi) Korean Society for Precision Engineering 2018 한국정밀공학회지 Vol.35 No.2

Fault diagnosis and condition monitoring of rotating machines are important for the maintenance of the gas turbine system. In this paper, the Lab-scale rotor test device is simulated by a gas turbine, and faults are simulated such as Rubbing, Misalignment and Unbalance, which occurred from a gas turbine critical fault mode. In addition, blade rubbing is one of the gas turbine main faults, as well as a hard to detect fault early using FFT analysis and orbit plot. However, through a feature based analysis, the fault classification is evaluated according to several critical faults. Therefore, the possibility of a feature analysis of the vibration signal is confirmed for rotating machinery. The fault simulator for an acquired vibration signal is a rotor-kit based test rig with a simulated blade rubbing fault mode test device. Feature selection based on GA (Genetic Algorithms) one of the feature selection algorithm is selected. Then, through the Support Vector Machine, one of machine learning, feature classification is evaluated. The results of the performance of the GA compared with the PCA (Principle Component Analysis) for reducing dimension are presented. Therefore, through data learning, several main faults of the gas turbine are evaluated by fault classification using the SVM (Support Vector Machine).

Fault diagnosis of bent shaft in rotor bearing system

S. P. Mogal,D. I. Lalwani 대한기계학회 2017 JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY Vol.31 No.1

Bent shaft is the most common fault in rotating machinery. Bent shaft generates excessive vibration in a machine, depending on amount and location of the bend. In this paper, order analysis technique of vibration analysis used for bent shaft diagnosis is proposed. In order analysis, both phase and amplitude are obtained. From phase and amplitude, the fault type and location are usually identified. Experimental results show order analysis is an effective technique for bent shaft.

A new method of spectral analysis -- synchronous cycle-ratio spectral analysis

Junqing Fu,Yimin Shao 제어로봇시스템학회 2009 제어로봇시스템학회 국제학술대회 논문집 Vol.2009 No.8

A synchronous cycle-ratio analysis is presented in the paper. The new method considers 2π as a basic cycle, which embodies the characteristics of 2π cycle in rotating machinery. The new method overcame the shortcomings of order analysis. An experiment verified that the new method is very effective, which can excellently filter random noise by statistically synchronous averaging of samples, and the cycle-ratio leakage can be eliminated perfectly. The peak values of synchronous cycle-ratio spectrum is not only clear without noise and leakage, but can be used to identify therelated physical parameters, which will be very useful for trouble shooting of rotating machinery.

Machine Fault Diagnosis and Prognosis: The State of The Art

Tung, Tran Van,Yang, Bo-Suk Korean Society for Fluid machinery 2009 International journal of fluid machinery and syste Vol.2 No.1

Machine fault diagnostic and prognostic techniques have been the considerable subjects of condition-based maintenance system in the recent time due to the potential advantages that could be gained from reducing downtime, decreasing maintenance costs, and increasing machine availability. For the past few years, research on machine fault diagnosis and prognosis has been developing rapidly. These publications covered in the wide range of statistical approaches to model-based approaches. With the aim of synthesizing and providing the information of these researches for researcher's community, this paper attempts to summarize and classify the recent published techniques in diagnosis and prognosis of rotating machinery. Furthermore, it also discusses the opportunities as well as the challenges for conducting advance research in the field of machine prognosis.