Predictive Analytics for Prioritizing Preventive Maintenance in Manufacturing Systems

Abstract

Maintenance planning is all about making sure that industrial equipment works properly by using money, people, and spare parts in the best way possible. Even though there have been improvements in how maintenance is managed over time, there are still some challenges. This is because the tasks involved in maintenance are quite complex, and also because maintenance itself can be expensive. In a manufacturing environment, resources are limited, so engineers need to decide which maintenance jobs should be done first. One tool that is commonly used for this is Failure Mode and Effect Analysis (FMEA). However, it has a big limitation because it relies on opinions rather than real data. To fix this, we developed a data-based method for planning preventive maintenance using something called clustering techniques. This method groups together past failure records that have similar impacts on how long machines are available and how reliable they are, as well as the costs of the maintenance actions that were taken. Then, these groups are ranked using a method called TOPSIS, which is a tool used for making decisions with multiple factors. Using these rankings, engineers can decide which maintenance tasks to do first and create a preventive maintenance plan, starting with the highest-ranked group. We tested this method using real data from a company in India that had kept records of failures over five years. The main benefit of this approach is that it helps in making a good balance between keeping machines available and reliable while also controlling maintenance costs. Also, different types of failures within the same group can be handled with the same maintenance action, which helps reduce the overall cost of preventive maintenance.