Abstracts Track 2022


Area 1 - Databases and Information Systems Integration

Nr: 1
Title:

Evaluation Methodology and Concept Implementation of Enhanced Social Media Data Analytics for Predictive Maintenance Models

Authors:

Jens Grambau

Abstract: A paper which describes how the generalized Reference Model is implemented to use machine learning methods for preprocessing and analyze Social Media data in the context of service to improve Predictive Maintenance services and analytics. Furthermore, it will be described how the evaluation methodology looks like and how the evaluation will be executed from a research point of view.

Nr: 6
Title:

Management of Unplanned Changes in Production Processes: AI Control Systems

Authors:

Ċ½ilvinas Svigaris

Abstract: The stability of each production depends almost directly on the management of unplanned changes. Although preparing the original plan, even if it is very complex, is relatively simple, unexpected reproductions, raw material defects, or unforeseen shortcomings, disruptions or changes to equipment, personnel, or other resources make the original plan unworkable and require not only urgent solutions, but also replanning of production, and sometimes reproductions. These situations severely affect costs, which are often disproportionately high because of the last-minute solutions. Examining 30 (50-100 employees) companies in furniture, printing, blending, and food industries, it is evident that they often face themselves working not in a planning mode but instead in a replanning one. In other words, it is not only the hypothetical, theoretical plan that matters most of all. Critical is anticipating and managing change and adjusting the plan to reality and alterations. It is extremely difficult to predict changes in the production plan due to the scale of the information involved in industrial processes. Still, it is possible to learn from historical situations and analyze the data that "remembers" previous cases that adversely have affected the production plan. The analysis of historical cases is possible by storing data on process parameters and recording statuses, states, events that can affect unforeseen production changes that harm the production plan. However, the volume of this information can be analyzed only by the artificial intelligence system that can compare different situations and states from the past with the current production situation by informing about possible disturbances and process stops or significant production defects before they occur or at very early stages when it is possible to avoid overproduction or significant losses and substantial changes in the overall production plan. In the presentation, we will explore an artificial intelligence system analyzing such information and giving forewarning of possible critical production statuses before they happen.