Asset Reliability and Predictive Breakdown of Machinery..... using Machine Learning Algorithms (Supe

Using Data from SAP Plant Maintenance Application and Integrating with Machine Learning Algorithms to predict equipment (Pumps) failure.

A framework model to predict equipment failure has been keenly sought by Asset intensive organizations.

We performed an experiment by applying data extraction algorithm on equipment maintenance records residing in SAP application that consisted of condition monitoring measurements, spare part usage, elapsed time period since the last completed maintenance and the next closest preventive maintenance order scheduled in a future date.

We applied unsupervised learning technique of clustering and performed classes to cluster evaluation with 80 percent accuracy. As part of supervised learning, data from the finalized data model was fed into various Machine Learning (ML) algorithms where the classifier model was trained to predict preventive, corrective breakdown scenarios, and subsequently tested on mutually exclusive data sets.

The Support vector machine (SVM) and Decision Tree (DT) algorithms were able to classify and therefore, predict equipment breakdown with high accuracy and true positive rate (TPR) of more than 95 percent. SAP and Machine Learning integration has huge potential to be realized.