Mining Big Data in Manufacturing: Requirement Analysis, Tools and Techniques

Abstract

Data Mining has tremendous potential and usefulness in improving the effectiveness of decision-making in manufacturing. Tools and techniques of data mining can be intelligently applied from product design analysis to the product repair and maintenance. Vast amount of data in the form of documents (text), graphical formats (CAD-file), audio/video, numbers, figures and/or hypertext are available in any typical manufacturing system. Our ultimate goal is to develop data-driven methodologies to solve manufacturing problems using data mining techniques. As a precursor, based on a literature study, this paper investigates selective manufacturing areas to identify the requirements for applying data mining techniques in solving potential manufacturing problems. The reviewed manufacturing areas are: (i) the “Design Intent” retrieval process for the product design and manufacturing, (ii) selection of materials, (iii) performance evaluations of manufacturing process design and operation management, and (iv) product inspection, and after-sales services (repair and maintenance). Industrial efforts towards addressing “Big Data” issues have also been briefly narrated in this paper. Lastly, the paper discusses two important data–related issues that may affect any applications of the data mining tools and techniques — (i) uncertainty involved in data collection, and (ii) interoperability of data collected at different levels of an enterprise.

Text data-driven new product development: a systematic mapping review

“…Data mining is a combination approach of statistics, pattern recognition, machine learning, visualization and other methods (Choudhary et al , 2009; Roy, Zhu et al , 2014). When data used for data mining is textual or non-structured, the approach used by data mining is changed to text mining and is used to explore knowledge from the context of these data (Aggarwal and Zhai, 2012; Marques et al , 2015).…”

A PLM-based data analytics approach for improving product development lead time in an engineer-to-order manufacturing firm

“…The first challenge comes from the data variety when considering the relevant data related to a product's full lifecycle. The product-, production-, and service-related data are available in various manufacturing information systems (PLM, MES, and ERP) [6], even the data may reside in external supply chain partners' systems. The second challenge is the inability of big data processing and fusion due to limitations of IT resources in a manufacturing firm [7].…”

A Knowledge-Enriched Computational Model to Support Lifecycle Activities of Computational Models in Smart Manufacturing

“…7 shows an example of the tree traversal strategy. Next section will present the detail of how to calculate the semantic similarity within a layer.…”