Návrh a implementace BI řešení pro podporu rozhodování v údržbě lisovacích linek ve Škoda Auto

This bachelor's thesis addresses a specific operational problem in the maintenance of press lines 31 and 33 at Škoda Auto: vibrodiagnostic and production data are stored in two mutually isolated systems and maintenance staff have no direct access to vibrodiagnostic trends, which severely limits the maintenance department's ability to evaluate the technical condition of the lines. The aim of the thesis is to design and implement a Business Intelligence solution that enables maintenance staff to evaluate the technical condition of the lines independently by integrating vibrodiagnostic and production data. Based on an analysis of the existing environment and the requirements of the maintenance department, a solution was designed and implemented on the Microsoft Power BI platform, covering ETL transformations of heterogeneous data sources and user-initiated, controlled refresh of the dataset. A dimensional data model links the previously isolated domains and the report provides an overview of the status of individual measurement points, trend analysis with reference thresholds, and traffic-light logic that replaces isolated threshold alarms with a cumulative evaluation over a parameterised time window. The functionality of the solution was validated through a retrospective case study of a historical bearing failure, which the solution would have detected several months in advance, and through feedback from members of the maintenance department. The resulting tool is in production use and, within the existing vibrodiagnostic environment of the maintenance department, is the first to integrate vibrodiagnostic and production domains into a single self-service interface. Maintenance staff thus gain direct access to data for independent decision-making on maintenance actions, and the deployed solution also serves as a concrete argument for the full integration of vibrodiagnostic data into the company's central data platform.