• Smart Structures and Systems
• /
• v.29 no.6
• /
• pp.757-766
• /
• 2022

To train deep learning algorithms, a sufficient number of data are required. However, in most engineering systems, the acquisition of fault data is difficult or sometimes not feasible, while normal data are secured. The dearth of data is one of the major challenges to developing deep learning models, and fault diagnosis in particular cannot be made in the absence of fault data. With this context, this paper proposes an anomaly detection methodology for rotating machines using only normal data with self-labeling. Since only normal data are used for anomaly detection, a self-labeling method is used to generate a new labeled dataset. The overall procedure includes the following three steps: (1) transformation of normal data to self-labeled data based on a pretext task, (2) training the convolutional neural networks (CNN), and (3) anomaly detection using defined anomaly score based on the softmax output of the trained CNN. The softmax value of the abnormal sample shows different behavior from the normal softmax values. To verify the proposed method, four case studies were conducted, on the Case Western Reserve University (CWRU) bearing dataset, IEEE PHM 2012 data challenge dataset, PHMAP 2021 data challenge dataset, and laboratory bearing testbed; and the results were compared to those of existing machine learning and deep learning methods. The results showed that the proposed algorithm could detect faults in the bearing testbed and compressor with over 99.7% accuracy. In particular, it was possible to detect not only bearing faults but also structural faults such as unbalance and belt looseness with very high accuracy. Compared with the existing GAN, the autoencoder-based anomaly detection algorithm, the proposed method showed high anomaly detection performance.

• Geomechanics and Engineering
• /
• v.31 no.5
• /
• pp.439-452
• /
• 2022

A forecast of slope behavior during catastrophic events, such as earthquakes is crucial to recognize the risk of slope failure. This paper endeavors to eliminate the significant supposition of predefined slip surfaces in the slope stability analysis, which questions the relevance of simple conventional methods under seismic conditions. To overcome such limitations, a methodology dependent on the slip line hypothesis, which permits an automatic generation of slip surfaces, is embraced to trace the extreme slope face under static and seismic conditions. The effect of earthquakes is considered using the pseudo-static approach. The current outcomes developed from a parametric study endorse a non-linear slope surface as the extreme profile, which is in accordance with the geomorphological aspect of slopes. The proposed methodology is compared with the finite element limit analysis to ensure credibility. Through the design charts obtained from the current investigation, the stability of slopes can be assessed under seismic conditions. It can be observed that the extreme slope profile demands a flat configuration to endure the condition of the limiting equilibrium at a higher level of seismicity. However, a concurrent enhancement in the shear strength of the slope medium suppresses this tendency by offering greater resistance to the seismic inertial forces induced in the medium. Unlike the traditional linear slopes, the extreme slope profiles mostly exhibit a steeper layout over a significant part of the slope height, thus ensuring a more optimized solution to the slope stability problem. Further, the susceptibility of the Longnan slope failure in the Huining-Wudu seismic belt is predicted using the current plasticity approach, which is found to be in close agreement with a case study reported in the literature. Finally, the concept of equivalent single or multi-tiered planar slopes is explored through an example problem, which exhibits the appropriateness of the proposed non-linear slope geometry under actual field conditions.

• Journal of Korean Society of Environmental Engineers
• /
• v.27 no.1
• /
• pp.93-100
• /
• 2005

Comparative thickening and dewatering characteristics of municipal wastewater sludge were investigated in terms of separated and combined treatment of primary and secondary BNR sludge. Also, various conditioning methods such as cation polymer addition, steam and ultrasonication treatment were examined to improve dewaterability of sludge. The dewaterability was measured by using specific resistant test, wedge zone simulator and centrifuge. The result of the sludge thickening test revealed that separated thickening was better in terms of solids recovery and supernatant quality. Particularly, the thickening of primary sludge with high solids (about 3.5% TS) showed very poor solid separation. The addition of cation polymer showed better conditioning characteristic for dewatering and the optimal polymer dosage was 0.26% for primary sludge, 0.43% for secondary sludge and 0.38% for combined sludge. Contrary to the result of the thickening, the dewatering test revealed that dewatering of the combined sludge is better than that of separated sludge, representing better solids separation and filtrate quality. The polymer addition was essential to improve dewaterability in filter (belt) press type dewatering but it was inefficient for the dewatering of secondary sludge only. The centrifuge type dewatering showed better performance and the dewaterability was slightly improved when the polymer was added. Based on the results of this research a sustainable sludge treatment process, particularly in terms of the recycle water quality and solids recovery, was proposed.