Perancangan Model Predictive Maintenance Berbasis Data Mining Terhadap Kerusakan Finish Water Pump (FWP) pada Instalasi Pengolahan Air

Abstract

Ensuring a smooth production process is essential for every company to maintain production stability. Machine performance is one of the key factors influencing production efficiency. This research aims to develop a predictive maintenance system for the finish water pump at IPAM Tirtanadi Martubung. Data mining applications are currently very effective in revealing hidden information from machine damage histories in a database using specific algorithms. This study examines company case data, where machine status history is analyzed to determine if the machine requires preventive maintenance using the naive Bayes classifier method. By implementing data mining with Anaconda Navigator and using an 80/20 data split, an accuracy rate of 98.60% was achieved in predicting the need for preventive maintenance. There are various conditions where the prediction results are accurate and others where they are not, influenced by the limited availability of data, which does not encompass all possible scenarios. Overall, this model provides valuable insights to users regarding potential machine failures and preventive maintenance scheduling