PENERAPAN MACHINE LEARNING DALAM PREDICTIVE MAINTENANCE PADA UNIT L.O COOLER MESIN INDUK NKK SEMPT – PIELSTICK 12PC4-2V-570 PADA KAPAL PASSENGER

Abstract

Predictive maintenance (PdM) is a type of maintenance that is used to predict when a part
of equipment or system fails and assist in planning maintenance in advance of potential future failures.
Maintenance can be applied based on the equipment condition using Condition Based Monitoring
(CBM). Condition Based Monitoring (CBM) is maintenance performed by monitoring equipment
conditions using sensors, software, or the Internet of Things (IoT). The monitoring results are then
analyzed to determine whether a condition occured indicating damage or not, using predictive
maintenance. In this research, predictive maintenance is applied to the L.O cooler unit on the main
engine to predict failures on the unit based on engine log sheet data derived from the engine operational
condition. The application of predictive maintenance is carried out using the results of modelling from
artificial intelligence, namely machine learning with supervised learning type and the classifications
used are Support Vector machine (SVM) and logistic regression. Based on the results of the ratio and
method evaluation, it is found that the logistic regression method with a ratio of 90%:10% is the most
accurate method as evidenced by the MAE, RMSE, RAE, and Accuracy values of 0.0012, 0.035, 0.631%
dan 99.877%. The results of predictive maintenance modeling obtained that the LO cooler is predicted
to fail when 3292 hours of operation.