Development of a Decision Support Tool for Intelligent Manufacturing using Classification and Correlation Analysis

Abstract

The development of the Production Unit Performance Management Tool (PUPMT), a monitoring and diagnostic tool to select basic signals that have a significant impact on output KPIs, is based on a hierarchical analysis of key performance indicators presented in the ISO 22400 standard. The growing number of quantified signals, such as ambient temperature or even the frequency of vibrations makes decision support tools more intelligent. Moreover, they also include predictive analysis in addition to online presentation of the current state of resources. PUPMT tool allows identifying key events that have a significant impact on current or future benefit in production output. It also allows what-if type analysis, running the simulation of the impact of the proposed changes, and the results of this simulation depend on the effects of similar changes that occurred in the past in a given smart manufacturing environment. Thanks to the automatic identification of potential dependencies, the proposed solution adapts to the specifics of a given environment or even a selected production unit. The paper describes the most important methods featured in the tool. Specifically, it considers a signal classification method, a signal correlation analysis and selection of key factors affecting the efficiency of the production process. In order to verify the methods various tests were carried out to calculate and control of OEE index on 20 production units. Finally, we conclude current results and future research directions.