SMU Data Science Review
Abstract
Work Order (WO) data from System Applications and Products in Data Processing (SAP) software contains valuable information about what WOs intend to accomplish. Using SAP work order data, with time-series machinery sensor data combined into the same dataset, provides an opportunity to optimize prediction models to increase performance. Ideally, WO data can be utilized to help predict machinery's anticipated performance and can help prioritize a WO among others based on the anticipated machinery performance. It is possible to identify anomalies in pump sensor data using the Isolation Forest algorithm as the method for anomaly detection. The relationship between the sensor data and the WO data is not straightforward due to scheduled maintenance programs, causing anomalies in the data and periods where a pump has experienced higher than normal performance. Autoregressive integrated moving average (ARIMA) provides additional insight from a time series perspective but may not necessarily provide different results. However, some anomalies did show that some advance notice or other factors that could be used for elevating the priority of a work order. Further analysis in what is considered to be a “good” and a “bad” anomaly may need additional research to enable a more efficient approach to detection with respect to WO prioritization of “bad” anomaly data.
Recommended Citation
Brionez, Jacob; Burnett, Amber; Kim, Cho; Whitney, Scott M.; Anderson, Thomas N.; and Treehan, Sumeet
(2020)
"Well Oiled Machine: Classifying Machinery Performance Reductions Using Work Order Data,"
SMU Data Science Review: Vol. 3:
No.
3, Article 2.
Available at:
https://scholar.smu.edu/datasciencereview/vol3/iss3/2
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
