Development of a Standardized Fluid Management Dashboard with Predictive Analytics Using Machine Learning

Publicerad

Typ

Examensarbete för masterexamen
Master's Thesis

Modellbyggare

Tidskriftstitel

ISSN

Volymtitel

Utgivare

Sammanfattning

This master’s thesis explores the design and evaluation of a real-time monitoring dashboard for industrial oil filtration systems, with a focus on integrating predictive analytics to support data-driven maintenance strategies. The primary goal was to develop dashboards that provide operators with useful forecasts and visualizations to better understand system performance. To assess different implementation ap proaches, two cloud-based platforms, Microsoft Fabric and Databricks, were used to build proof-of-concept solutions. This allowed for a comparative analysis of their strengths in managing real-time data streams and supporting AI integration. The analytical part of the project centered on two key predictive tasks: estimating the remaining useful life of the filtration systems and forecasting future sensor values. While both platforms successfully supported real-time dashboard function ality, the performance of the predictive models was limited. This was largely due to data-related constraints, specifically insufficient historical data, limited laboratory sampling, and weak correlations between sensor readings and actual filter conditions. In particular, attempts to predict filter condition from the available sensor data proved unreliable, highlighting the need for more comprehensive and representative datasets. Despite these challenges, the project revealed what worked well and what fell short when adding predictive features. The findings emphasize the importance of improv ing both the quantity and quality of data to enhance predictive accuracy. Although current constraints affect the reliability of the predictive components, the developed dashboard solutions offer a solid foundation for improving operational visibility and enabling smart maintenance practices. If improved, these systems could support SKF’s predictive maintenance work and broader sustainability goals more effectively, while also strengthening the company’s competitive edge.

Beskrivning

Ämne/nyckelord

Real-time monitoring, Predictive analytics, Machine learning, Remaining useful life estimation, Condition-based maintenance, Dashboard systems, Industrial oil filtration, Microsoft Fabric, Databricks, Sustainability

Citation

Arkitekt (konstruktör)

Geografisk plats

Byggnad (typ)

Byggår

Modelltyp

Skala

Teknik / material

Index

item.page.endorsement

item.page.review

item.page.supplemented

item.page.referenced