• Proceedings of the Korean Society of Computer Information Conference
• /
• 2017.07a
• /
• pp.9-10
• /
• 2017

본 논문에서 소음, 진동을 이용한 딥러닝 기반 기계 고장진단 임베디드 시스템을 제안하였다. 제안된 시스템은 기계로부터 취득된 소리와 진동을 바탕으로 학습한 DNN모델을 통해 실시간으로 기계 고장을 진단한다. 딥러닝 기술을 사용하여 학습에 따라 적용대상이 변경될 수 있도록 함으로써 특정 기계에 종속적이지 않고 가변적으로 다양한 기계에 대해 고장 예지 및 건전성 관리를 제공하도록 설계하였으며, 이를 증명하기 위해 액추에이터를 환풍기로 설정하여 정상상태와 4가지 비정상상태의 5가지상태를 학습하여 실험한 결과 93%의 정확도를 얻었다.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2023.05a
• /
• pp.275-276
• /
• 2023

산업 현장의 기계 시설물 고장 문제는 큰 인명피해와 경제적 손실을 초래할 수 있기 때문에, 기계 시설물의 상태를 기반하여 고장을 진단하는 것은 대단히 중요하다. 따라서, 본 연구에서는 전류 센서 데이터를 활용하여, 시설물의 고장 여부를 진단할 수 있는 알고리즘을 제안한다. 본 연구에 활용된 전류 센서 데이터는 x, y, z축을 가진 3상 전류 데이터로 구성되어 있으며, 2kHz로 1초간 샘플링 되어 있다. 본 연구에서는 2차원적 특성을 가지는 전류 센서 데이터를 분석하기 위해 CNN(Convolution Neural Network)을 활용한다. 시설물의 고장진단에 가장 적합한 모델을 선정하기 위해 CNN의 대표적인 백본 네트워크를 활용하여, 결과를 비교하였다. 실험 결과, 본 연구에서 구성한 후보 백본 네트워크 중 ResNet의 분류 정확도가 98.5%로 가장 높게 나타났다.

• Proceedings of the Safety Management and Science Conference
• /
• 1999.11a
• /
• pp.133-146
• /
• 1999

오늘날의 산업용 로봇, CNC 공작기계 및 여러 산업설비들은 시스템간에 관계가 복잡하게 연결되어 높은 신뢰성(reliability)을 달성하여 왔다. 그러나 가동시 발생하는 결과의 고장 가능성은 적은 반면에, 고장 발생의 파급 효과는 매우 높은 것으로 나타났다. 따라서 복잡한 구조의 산업설비들에 대한 안전진단 결과들을 적절하게 분석하고 관리할 필요성이 크게 대두되고 있다. 이러한 안전진단 작업은 여러 가지 정량적ㆍ정성적인 방법들을 포함하는 전형적인 분석방법이 필요하다. 최근에는 고장탐색, 진단처리 작업 및 신뢰성 분석 작업에 지식기반(knowledge-based)을 기초로한 퍼지 전문가 시스템을 적용하고자 하는 시도가 많이 이루어지고 있다. 안전진단 분석에 관한 일반화된 지식은 이들 후속 단계들에서 상당히 효율적일 수 있다. 그러나 이러한 연구를 수행하기에는 지금까지 상대적으로 열악한 계산 도구들을 이용하였기 때문에 안전진단 분석을 행하기에는 한계가 있었다. 그러나 오늘날 컴퓨터를 이용하여 위의 여러 단계들의 수행과정에 안전진단 분석을 행할 수 있는 적절한 방법으로써, 지식-기반(knowledge-base) 전문가 시스템들을 이용하는 방법을 연구하고 있다. 이에 본 연구는 시스템의 설계단계 뿐만 아니라, 시스템의 가동ㆍ유지ㆍ보수ㆍ수리시에도 비전문가가 고장안전진단을 수행할 수 있도록 하는데 목표를 두었다.

• Journal of the Institute of Convergence Signal Processing
• /
• v.20 no.2
• /
• pp.78-83
• /
• 2019

Sounds based machine fault diagnosis recovers all the studies that aim to detect automatically abnormal sound on machines using the acoustic emission by these machines. Conventional methods that use mathematical models have been found inaccurate because of the complexity of the industry machinery systems and the obvious existence of nonlinear factors such as noises. Therefore, any fault diagnosis issue can be treated as a pattern recognition problem. We propose here an automatic fault diagnosis method of hand drills using discrete wavelet transform(DWT) and pattern recognition techniques such as artificial neural networks(ANN). We first conduct a filtering analysis based on DWT. The power spectral density(PSD) is performed on the wavelet subband except for the highest and lowest low frequency subband. The PSD of the wavelet coefficients are extracted as our features for classifier based on ANN the pattern recognition part. The results show that the proposed method can be effectively used not only to detect defects but also to various automatic diagnosis system based on sound.

• Journal of the Institute of Convergence Signal Processing
• /
• v.21 no.3
• /
• pp.121-126
• /
• 2020

Sound-based machine fault diagnosis is the automatic detection of abnormal sound in the acoustic emission signals of the machines. Conventional methods of using mathematical models were difficult to diagnose machine failure due to the complexity of the industry machinery system and the existence of nonlinear factors such as noises. Therefore, we want to solve the problem of machine fault diagnosis as a deep learning-based image classification problem. In the paper, we propose a CNN-based automatic machine fault diagnosis method using Spectrogram images. The proposed method uses STFT to effectively extract feature vectors from frequencies generated by machine defects, and the feature vectors detected by STFT were converted into spectrogram images and classified by CNN by machine status. The results show that the proposed method can be effectively used not only to detect defects but also to various automatic diagnosis system based on sound.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.35 no.12
• /
• pp.1351-1358
• /
• 2011

To approximate the threshold of the fault detection and diagnosis (FDD) system, validation of the measurements is mandatory. Naturally, the system shows uncertainties due to measuring sensors - mostly thermocouples or RTDs - and due to repeatability. The uncertainty of a thermocouple comes from natural variation or a drift of the thermocouple measurement. Considering the natural variation behaves like zero-mean white noise, its natural variation can be characterized closely by the steady-state standard deviation. However, residuals between measurements and no-fault references in FDD systems show a statistical distribution with various uncertainties. In this paper, steady-state variations of measurement residuals were investigated by utilizing built-in temperature sensors in a heat pump for the model development and the final application.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2011.11a
• /
• pp.599-602
• /
• 2011

A propulsion controller for one-time flight vehicles should be designed robustly so that it can complete its missions even in case sensor failures. These vehicles improve their fault tolerance by back-up sensors prepared for the failure of major sensors, which raises the total cost. This paper presents the NARX model which substitutes vehicles' velocity sensors, and detects failure of sensor signals by using model based fault detection. The designed NARX model and fault detection algorithm were optimized and installed in TI's TMS320F2812 so that they were linked to HILS instruments in real-time. The designed propulsion controller made the vehicle to have better fault tolerance with fewer sensors and to complete its missions under a lot of complicated failure situations. The controller's applicability was finally confirmed by tests under the HILS environment.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.5 no.4
• /
• pp.443-449
• /
• 2010

Current testing is an effective method which offers higher fault detection and diagnosis capabilities than voltage testing. Since current testing requires much longer testing time than voltage testing, it is important to note that a fault is untestable if the two nodes have same values at all times. In this paper, we present an object oriented fault detection scheme for various fault models using current testing. Experimental results for ISCAS benchmark circuits show the effectiveness of the proposed method in reducing the number of faults and its usefulness in various fault models.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.20 no.2
• /
• pp.56-66
• /
• 2016

Robustness is essential for model based FDI (Fault Detection and Isolation) and it is inevitable to have modeling errors and sensor signal noises during the process of FDI. This study suggests an improved method by applying NARX (Nonlinear Auto Regressive eXogenous) model and Kalman estimator in order to cope with problems caused by linear model errors and sensor signal noises in the process of fault diagnoses. Fault decision is made by the probability of the trend of gradually accumulated errors applying Fuzzy logic, which are robust to instantaneous sensor signal noises. Reliability of fault diagnosis is verified under various fault simulations.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.40 no.9
• /
• pp.835-842
• /
• 2016

In Korea, urban railway cars are typically maintained using the strategy of predictive maintenance. In an effort to overcome the limitations of the existing strategy, there is increased interest in adopting the condition-based maintenance strategy. In this study, a novel method is proposed to detect faults in the solenoid valves of the braking system in urban railway vehicles. We determined the key component (i.e., solenoid valve) that leads to braking system faults through the analysis of failure modes, effects, and criticality. Then, an equivalent circuit model was developed with the compensation of the temperature effect on solenoid coils. Finally, we presented how to detect faults with the equivalent circuit model and current signal measurements. To demonstrate the performance of the proposed method, we conducted a case study using real solenoid valves taken from urban railway vehicles. In summary, it was shown that the proposed method can be effective to detect faults in solenoid valves. We anticipate the outcome from this study can help secure the safety and reliability of urban railway vehicles.