• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.44 no.4
• /
• pp.227-233
• /
• 2021

Predictive maintenance has been one of important applications of data science technology that creates a predictive model by collecting numerous data related to management targeted equipment. It does not predict equipment failure with just one or two signs, but quantifies and models numerous symptoms and historical data of actual failure. Statistical methods were used a lot in the past as this predictive maintenance method, but recently, many machine learning-based methods have been proposed. Such proposed machine learning-based methods are preferable in that they show more accurate prediction performance. However, with the exception of some learning models such as decision tree-based models, it is very difficult to explicitly know the structure of learning models (Black-Box Model) and to explain to what extent certain attributes (features or variables) of the learning model affected the prediction results. To overcome this problem, a recently proposed study is an explainable artificial intelligence (AI). It is a methodology that makes it easy for users to understand and trust the results of machine learning-based learning models. In this paper, we propose an explainable AI method to further enhance the explanatory power of the existing learning model by targeting the previously proposedpredictive model [5] that learned data from a core facility (Hyper Compressor) of a domestic chemical plant that produces polyethylene. The ensemble prediction model, which is a black box model, wasconverted to a white box model using the Explainable AI. The proposed methodology explains the direction of control for the major features in the failure prediction results through the Explainable AI. Through this methodology, it is possible to flexibly replace the timing of maintenance of the machine and supply and demand of parts, and to improve the efficiency of the facility operation through proper pre-control.

• Transactions of the Korean hydrogen and new energy society
• /
• v.31 no.1
• /
• pp.73-81
• /
• 2020

For the fault tree analysis (FTA) analysis of the packaged hydrogen filling station, the composition of the charging station was analyzed and the fault tree (FT) diagram was prepared. FT diagrams were created by dividing the causes of events into external factors and internal factors with the hydrogen event as the top event. The external factors include the effects of major disasters caused by natural disasters and external factors as OR gates. Internal factors are divided into tube tailer, compressor & storage tank, and dispenser, which are composed of mistakes in operation process and causes of accidents caused by parts leakage. In this study, the purpose was to improve the hydrogen station. The subjects of this study were domestic packaged hydrogen stations and FTA study was conducted based on the previous studies, failure mode & effect analysis (FMEA) and hazard & operability study (HAZOP). Top event as a hydrogen leaking event and constructed the flow of events based on the previous study. Refer to "Off shore and onshore reliability data 6th edition", "European Industry Reliability Data Bank", technique for human error rate prediction (THERP) for reliability data. We hope that this study will help to improve the safety and activation of the hydrogen station.

• Proceedings of the KSME Conference
• /
• 2004.11a
• /
• pp.962-966
• /
• 2004

Suction transfer system with air suctioning is widely used and continuously developed in production automation. Air suctioning has some drawbacks in use. To generate vacuum in the suction cup with air suctioning, complex of mechanical component like as air compressor, air tube, air value is need, and it needs continuous air supply. And if the failure of the suction in a cup in the multi-suction cup system which is generally used occurs then the suctions of all of the cup will be fail. To overcome these drawbacks, new suction mechanism which uses permanent magnet for the movement of the suction cup is proposed. The proposed mechanism activates each suction cup separately, so the air leakage of a cup is not critical. The proposed suction system wasdesigned and fabricated in real world. With some experiments, the usability and performance of the suction mechanism was proved. The strong points of the proposed suction mechanism are simple structure, high energy efficiency, and discrete energy supply.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.29 no.12 s.243
• /
• pp.1645-1652
• /
• 2005

Pick-and-place systems using suction cups have been being widely used and continuously developed in production automation. There are, however, some drawbacks in constructing such systems. One of them is that it generates high level noise due to air compressors. And the system must have complex constitutions of mechanical component such as air compressors, air tubes, air valves, etc. Moreover, it needs continuous air supply to maintain vacuum in suction cups. If there is a failure in any suction cup, the total suction system may fail owing to air leakage. To overcome these drawbacks, we propose PMS (Permanent Magnet Suction) mechanism which has permanent magnets for vacuuming suction cups with no air compressor. The basic concept of PMS mechanism is to rotate permanent magnets with fixed angle. Simple rotation of permanent magnets changes the direction of the magnetic force applied at the suction cups. Since each suction cup has no direct connection with any of the others, the air leakage at one suction cup is not critical. The proposed suction mechanism was designed and fabricated. With some experiments, the feasibility and performance of the PMS mechanism was shown. The strong points of the PMS mechanism are in its simple structure, generating low noise, high energy efficiency, and no need of continuous energy supply.

• Proceedings of the Korean Nuclear Society Conference
• /
• 1997.05a
• /
• pp.587-591
• /
• 1997

The DGs (Diesel Generators) in NPP (Nuclear Power Plant) has been used for the emergency electric power source to shut down the nuclear reactor safely in case of station blackout. The RCM (Reliability Centered Maintenance) has been applied to DGs for increasing the safety of NPP. The structured defects of DG were not remedied by the improvement of maintenance method. As the first stage of RCM, to find the structured defects, its failure modes were searched and analyzed through the ten year maintenance information. The structured defects such as the air compressor, the lubricating oil pressure, and the insufficient load were the root causes of main failures. The air reservoir reinstallation, the lubricating oil tube modification, the load bank installation, and the qualitative instrumentation were the solutions for the hardware oriented RCM of DGs. There remains the software oriented RCM such as the rejection of useless maintenance, the preventive maintenance, the database of maintenance information, and the predictive maintenance.

• Progress in Superconductivity and Cryogenics
• /
• v.15 no.4
• /
• pp.6-9
• /
• 2013

Korea Superconducting Tokamak Advanced Research (KSTAR) has a helium refrigeration system (HRS) with the cooling capacity of 9 kW at 4.5 K. Main cold components are composed of 300 tons of superconducting (SC) magnets, main cryostat thermal shields, and SC current feeder system. The HRS comprises six gas storage tanks, a liquid nitrogen tank, the room temperature compression sector, the cold box (C/B), the 1st stage helium distribution box (DB#1), the PLC base local control system interconnected to central control tower and so on. Between HRS and cold components, there's another distribution box (DB#2) nearby the KSTAR device. The entire KSTAR device was constructed in 2007 and has been operated since 2008. This paper will present the maintenance result of the KSTAR HRS during the campaign and discuss the operation record and maintenance history of the KSTAR HRS.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2017.05a
• /
• pp.877-880
• /
• 2017

Impeller blades in the centrifugal compressor are subjected to static loads due to the high-speed rotation and steady aerodynamic forces. At the same time, aerodynamic excitations by the interaction between the impeller and the diffuser vanes(DV) periodically excite the impeller blades in resonant conditions, which may lead to high cycle fatigue (HCF) and eventually result in failure of the blades. In order to predict the structural response accurately, the aerodynamic excitation and the major resonant conditions were predicted by performing the unsteady flow analysis and modal analysis using ANSYS. Next, a unidirectional forced vibration analysis was performed by using fluid-structure interaction (FSI) method, and the safety of HCF was evaluated based on the results.

• Journal of the Korean Society of Safety
• /
• v.12 no.4
• /
• pp.214-219
• /
• 1997

The mechanical seals, which are installed in rotating machines like pump and compressor, are generally used as sealing devices in the many fields of industries. The failure of mechanical seals such as leakage, crack, breakage, fast and severe wear, excessive torque, and squeaking results in big problems. To identify abnormal phenomena on mechanical seals and to propose the proper monitoring parameter for the failure of mechanical seals, sliding wear experiments were conducted. Acoustic emission, torque, and temperature were measured during experiments. Optical microstructure was observed for the wear processing after every 10 minute sliding at rotation speed of 1750 rpm and scanning electron microscopy was also observed. Except for the initial part of every experiment, the variation of acoustic emission was well coincided with torque variation during the experiments. This study concludes that acoustic emission and torque are proper monitoring parameters for the failure of mechanical seals. The intensity of acoustic emission signals is measured in root mean square voltage. Temperature of sealing face will be used as a parallel parameter for increasing the reliability of monitoring system.

• Journal of the Korean Institute of Gas
• /
• v.24 no.2
• /
• pp.9-14
• /
• 2020

This paper is a purpose to study the failure examples for a large bus vehicle fire. The first example, the researcher certified the fact that the absorbing material break away from the upper side of engine room because of weaken durability and the fire was produced in engine. The second example, it sought the fact that the fire breaks out by electric short because of over-load of compressor. The third example, it found the fact that the fire took place by heating of bellows upper part that was connected with muffler and exhaust manifold. The fourth example, it knew the fact that the fire occurred because of the electric short inside junction box of crash body part that was located to driver seat rightside. Therefore, the fire of a large bus occurring by decrepit of absorbing material and electric short have to thoroughgoingly manage the damage and dangerousness if it happens.

• The KIPS Transactions:PartC
• /
• v.16C no.1
• /
• pp.57-64
• /
• 2009

In order to efficiently provide IP based Wireless Broadband services in Mobile WiMAX, schemes for more efficiently utilizing a limited bandwidth in radio links are needed. Robust Header Compression (ROHC), a standard header compression scheme proposed by IETF, Bi-directional Optimistic (O) mode provides higher compression efficiency and robustness compared with Payload Header Suppression (PHS) which is an optional header compression scheme for mobile WiMAX [1, 2]. However, if consecutive packet losses occur because of a Shadow Region of Mobile WiMAX or bit errors, header decompression failure rate of the ROHC O mode increases due to inconsistency of the compression information between the Compressor and the Decompressor. Therefore, a complementary mechanism is needed. In this paper, we propose an approach for the dynamic adjustment of an optimistic parameter, which has an effect on both the compression efficiency and the robustness, for improving the performance of ROHC O mode. We also analyze the performance of the proposed approach using an OPNET simulator.