MaXCept -- Decision Support in Exception Handling through Unstructured Data Integration in the Production Context: An Integral Part of the Smart Factory

Abstract

Today, data from different sources and different phases of the product life cycle are usually analyzed in isolation and with considerable time delay. Real-time integrated analytics is especially beneficial in a production context. We present an architecture for data-and analytics-driven exception escalation in manufacturing and show the advantages of integrating unstructured data.

Business Intelligence in Industry 4.0: State of the art and research opportunities

“…Technological limitations such as insufficient database infrastructure [5] or the complexity of integrating real time data [23] were cited by some authors as limitations in data integration and analysis in manufacturing processes. This is reflected in the number of articles focusing on the proposition of a standardized manufacturing BI architecture capable of real-time analysis.…”

“…Analysis of BI and related technological aspect reveals 33 out of 42 articles included real time or near real-time data processing. In many of these articles, authors emphasized the technical difficulty of processing machine data in real time, because of database limitations [5], the integration of unstructured data [23], limits to the acceptable visual complexity [46] or the number of variables required to develop a sufficiently precise model [35], [45].…”

The Social Factory: Connecting People, Machines and Data in Manufacturing for Context-Aware Exception Escalation

“…Building on the factory context model developed in [5], [6] defines a reference architecture and detailed data collection and analytics processes for automated exception escalation in the smart factory. This includes the integration of structured data, e.g., sensor data, with unstructured data, e.g., text data.…”

“…This includes a decision logic for ranking exceptions by urgency and resolving conflicting information, as well as data mining capabilities to automatically discover unusual situations from process and sensor data. The Social Factory concept is derived from the concept and architecture in [6]. It deliberately excludes some of its functionalities, focusing instead on the development of a concrete social-media-like communication infrastructure for shop floor management and problem resolution.…”

“…One important focus of the Social Factory is exception detection and handling as described in [6]. This is best achieved by creating and managing a knowledge base about historical exceptions or solutions drawn from manuals.…”