Recently, due to the expansion of the logistics industry, demand for logistics automation equipment is increasing. The modern logistics industry is a high-tech industry that combines various technologies. In general, as various technologies are grafted, the complexity of the system increases, and the occurrence rate of defects and failures also increases. As such, it is time for a predictive maintenance model specialized for logistics automation equipment. In this paper, in order to secure the operational safety and reliability of the parcel loading system, a predictive maintenance platform was implemented based on the Naive Bayes-LSTM(Long Short Term Memory) model. The predictive maintenance platform presented in this paper works by collecting data and receiving data based on a RabbitMQ, loading data in an InMemory method using a Redis, and managing snapshot DB in real time. Also, in this paper, as a verification of the Naive Bayes-LSTM predictive maintenance platform, the function of measuring the time for data collection/storage/processing and determining outliers/normal values was confirmed. The predictive maintenance platform can contribute to securing reliability and safety by identifying potential failures and defects that may occur in the operation of the parcel loading system in the future.

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.

다양한 산업에서 강조되고 있는 정비의 중요성은 각 분야에 다양한 정비전략을 적용하도록 만들었다. 해양산업 역시 그에 따른 정비전략의 변화가 있었으나 타 산업 대비 그 속도가 느려 실제 적용이 되지 않은 채 과거 시행되고 있던 방식을 유지하는 경우가 많다. 특히 선박은 기존에 행해왔던 방식의 정비전략을 사용하고 있는 편이며 해상의 조건에서 선박은 새로운 정비전략의 개발을 필요로 하고있다. 이에 선박예지정비모델은 기기의 정비가 필요한 시점을 예지하여 조치할 수 있는 정비전략으로서 선박이 항해 중에 처할 수 있는 정비 관련 위험요소들을 줄여 주는 모델이다. 본 연구는 선박예지정비모델의 개발을 위한 연구 중의 하나로서, LNG선박 입거사양서의 텍스트 데이터 분석을 통한 결과를 원문의 내용을 바탕으로 해석해보았다. 공통된 정비항목 조합을 도출하여 선박 내 다른 기기들 사이에 작용하고 있는 상호연관성을 발견하고 이를 앞으로 개발될 선박예지정비모델에 적용하고자 한다.

해양 운송 산업은 특성상 항공 및 철도 등의 다른 운송 산업보다 비교적 늦게 신기술이 적용되는 산업이다. 현재 대부분의 선박은 기계장치 및 시스템에 문제가 발생하거나 운용 시간 기반으로 정비를 하는 사후 정비(Corrective Maintenance, CM)와 예방 정비 (Preventive Maintenance, PM)에 속하는 시간 기반 정비(TBM, Time Based Maintenance)가 적용되고 있다. 그러나 높은 유지보수 비용이 요구되고, 육상의 즉각적인 지원이 어려우며, 선박이 멈추면 즉시 위험에 노출되는 해양 환경에서 운영되는 선박에서 과도한 단순 정비로 인한 인력과 비용 낭비, 예측되지 못한 고장 및 결함으로 유발되는 사고 등으로 인해 운용 효율화 측면에서 기존 정비법에 대한 한계점이 문제시 되고 있다. 예지 정비(Predictive Maintenance, PdM)는 진보된 기술로 기계의 상태 및 성능을 모니터링하여 고장시기를 예측하여 정비하는 방법으로 핵심 기계장치가 항상 최상의 작동 상태를 효율적으로 유지할 수 있도록 한다. 본 논문은 해양 환경에서 PdM의 적용성에 중점을 둔 해양 예지 정비(MPdM, Maritime Predictive Maintenance)에 대해 고안하였으며, 제시된 MPdM은 지리적 고립과 극한 해양 상황 등 해양 운송 산업의 특수한 환경을 고려하여 설계되었다. 본 논문은 선진 미래 해양 운송을 가능하게 하는 MPdM이라는 개념과 그 필요성을 제안한다.

In the process of cutting large aircraft parts, the tool may be abnormally worn or damaged due to various factors such as mechanical vibration, disturbances such as chips, and physical properties of the workpiece, which may result in deterioration of the surface quality of the workpiece. Because workpieces used for large aircrafts parts are expensive and require strict processing quality, a maintenance plan is required to minimize the deterioration of the workpiece quality that can be caused by unexpected abnormalities of the tool and take maintenance measures at an earlier stage that does not adversely affect the machining. In this paper, we propose a method to indirectly monitor the tool condition that can affect the machining quality of large aircraft parts through real-time monitoring of the current signal applied to the spindle motor during machining by comparing whether the monitored current shows an abnormal pattern during actual machining by using this as a reference pattern. First, 30 types of tools are used for machining large aircraft parts, and three tools with relatively frequent breakages among these tools were selected as monitoring targets by reflecting the opinions of processing experts in the field. Second, when creating the CNC machining program, the M code, which is a CNC auxiliary function, is inserted at the starting and ending positions of the tool to be monitored using the editing tool, so that monitoring start and end times can be notified. Third, the monitoring program was run with the M code signal notified from the CNC controller by using the DAQ (Data Acquisition) device, and the machine learning algorithms for detecting abnormality of the current signal received in real time could be used to determine whether there was an abnormality. Fourth, through the implementation of the prototype system, the feasibility of the method proposed in this paper was shown and verified through an actual example.

In this research, we are developing a predictive maintenance model of the injection molding machines based on the prediction of trend of injection molding parameters. At first, we developed an interface method to directly monitor the real-time injection molding parameter data from injection molding machine controller. Second, we identified the principal injection parameters which mainly affect the quality of injection molding products and need to be monitored for maintenance. Third, based on the time series analysis, we developed the prediction models of the principal injection molding parameters, which are identified by previous statistical model to forecast its future patterns/trends and schedule its maintenance point in time. We adopted Nelson’s rules to identify abnormal patterns in predicted data. Finally, we used FTA (fault tree analysis) to relate the injection molding parameters to the parts of the injection molding machine, find out the equipment or parts to be corrected.