Interconnected services for time-series data management in smart manufacturing scenarios

Abstract

The rise of Smart Manufacturing, together with the strategic initiatives carried out worldwide, have promoted its adoption among manufacturers who are increasingly interested in boosting data-driven applications for different purposes, such as product quality control, predictive maintenance of equipment, etc. However, the adoption of these approaches faces diverse technological challenges with regard to the data-related technologies supporting the manufacturing data life-cycle. The main contributions of this dissertation focus on two specific challenges related to the early stages of the manufacturing data life-cycle: an optimized storage of the massive amounts of data captured during the production processes and an efficient pre-processing of them. The first contribution consists in the design and development of a system that facilitates the pre-processing task of the captured time-series data through an automatized approach that helps in the selection of the most adequate pre-processing techniques to apply to each data type. The second contribution is the design and development of a three-level hierarchical architecture for time-series data storage on cloud environments that helps to manage and reduce the required data storage resources (and consequently its associated costs). Moreover, with regard to the later stages, a thirdcontribution is proposed, that leverages advanced data analytics to build an alarm prediction system that allows to conduct a predictive maintenance of equipment by anticipating the activation of different types of alarms that can be produced on a real Smart Manufacturing scenario.