nach oben

2015 | Buch

Kapitel lesen Erstes Kapitel lesen

Identifying Product and Process State Drivers in Manufacturing Systems Using Supervised Machine Learning

verfasst von: Thorsten Wuest

Verlag: Springer International Publishing

Buchreihe : Springer Theses

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

The book reports on a novel approach for holistically identifying the relevant state drivers of complex, multi-stage manufacturing systems. This approach is able to utilize complex, diverse and high-dimensional data sets, which often occur in manufacturing applications, and to integrate the important process intra- and interrelations. The approach has been evaluated using three scenarios from different manufacturing domains (aviation, chemical and semiconductor). The results, which are reported in detail in this book, confirmed that it is possible to incorporate implicit process intra- and interrelations on both a process and programme level by applying SVM-based feature ranking. In practice, this method can be used to identify the most important process parameters and state characteristics, the so-called state drivers, of a manufacturing system. Given the increasing availability of data and information, this selection support can be directly utilized in, e.g., quality monitoring and advanced process control. Importantly, the method is neither limited to specific products, manufacturing processes or systems, nor by specific quality concepts.

Inhaltsverzeichnis

Frontmatter

Chapter 1. Introduction

Abstract

One way to improve manufacturing processes is to look at the data and information involved and how this information is put to use (Hicks et al. 2006). As stated by Albinoet al. (2002), the successful coordination of a manufacturing process is mostly based on a successful handling of information to support process management and other tasks involved. With today’s advanced ICT it becomes possible to process, transfer and store large amounts of data and information for a reasonable price (Krcmar 2005). But too much information can be a threat for improved process quality as it can e.g., distract from the main issues/causalities or lead to delayed or wrong conclusions about appropriate actions (Lang 2007). Jansen-Vullers et al. (2003) emphasize the importance of the availability of the right information for quality during manufacturing processes. Hence the question is: What is the right and relevant information in the case of distributed manufacturing process chains and high tech industrial products?

Thorsten Wuest

Chapter 2. Developments of Manufacturing Systems with a Focus on Product and Process Quality

Abstract

In this section MS as well as recent developments in the area of holistic IM and related topics will be presented. Furthermore, certain basic aspects of manufacturing, MS and related areas are described in detail in order to allow readers to familiarize themselves with the fundamental terms and definitions used throughout this dissertation. In each subsection, concluding paragraphs summarize how the described topic is relevant to the research and putting it in perspective. Main principles and how they are utilized throughout this dissertation is summarized there.

Thorsten Wuest

Chapter 3. Current Approaches with a Focus on Holistic Information Management in Manufacturing

Abstract

In this section, the focus is laid on existing approaches and concepts that try to address some of the identified challenges of MS when it comes to transparent and product specific information and data management. The main focal methods and concepts are PDM, PLM and quality monitoring in manufacturing. The presented domain specific knowledge is discussed within this section as it has strong relations with the later concept development. In order to allow the reader to easily identify the relation of the individual method to the product state concept, a short conclusion after each section highlights the relevancy and connection to the topic. The final sub-section of this third section will furthermore briefly summarize the complete section and help the reader with the transition towards the next section where the product state concept is presented.

Thorsten Wuest

Chapter 4. Development of the Product State Concept

Abstract

In this section, the product state concept and its development will be illustrated from a theoretical perspective. The main intension is to provide a general understanding of the goals and basic pillars of the concept and its argumentation. Another major goal of this section is to discuss and present the challenges and limitations to the application of the presented theoretical approach in practice. This outcome is crucial for the selection of appropriate methods and the following approach to identify state drivers despite the knowledge gap concerning process intra- and inter-relations using ML which will bring the product state concept to life.

Thorsten Wuest

Chapter 5. Application of Machine Learning to Identify State Drivers

Abstract

In this section, the application of ML is investigated in further detail. First ML is briefly introduced in more detail with respect to the manufacturing domain. Based on this brief general elaboration, SVM algorithms are selected as a suitable ML technique to match the detailed requirements of the stated research problem. In the final subsection, the application of SVM is discussed towards its objective of identification of state drivers in manufacturing programmes. Within this last subsection, the application and evaluation approach of the SVM application are presented and the derived hypotheses are detailed based on the decision to use the SVM algorithm to conclude the section.

Thorsten Wuest

Chapter 6. Application of SVM to Identify Relevant State Drivers

Abstract

In this section, the previously derived hypotheses are evaluated by developing and analyzing three scenarios. The section is structured as follows: at first the scenarios are briefly introduced (for more detail refer to Annex Sect. A.2). The following two subsections focus on the application of the previously introduced research plan on the three scenarios. However, it has to be noted that the scenarios were not evaluated following the presented sequence during the analysis phase. The presented sequence (scenarios I–III) does not resemble the timely sequence of evaluation of the different scenarios. Therefore, it is possible that the background of and justification for some of the methods, tools and applications are explained in later sections even so they are applied beforehand. In such cases, reference is given to the more detailed explanation in later sections. The Chap. 7 presents and discusses the evaluation results and illustrates the limitations of the approach.

Thorsten Wuest

Chapter 7. Evaluation of the Developed Approach

Abstract

The evaluation results derived from the previous application section are presented in a condensed fashion and critically discussed within this section. The critical discussion is roughly structured along the previously presented research hypotheses. Following, the limitations identified during the evaluation and analysis including data pre-processing are highlighted. Within that section the implications of those limitations on the hypotheses and the research results are illustrated.

Thorsten Wuest

Chapter 8. Recapitulation

Abstract

Concluding, the presented product state concept allows to identify relevant state drivers of complex manufacturing systems. The concept is able to utilize complex, diverse and high-dimensional data sets which often occur in manufacturing applications. This fits nicely with current initiatives like ‘Industrie 4.0’, ‘Cyber Physical Systems’ in Europe and the ‘Industrial Internet’ and ‘Advanced Manufacturing Partnership’ in the US as well as the growing area of Big Data research. It can be safely said that in the near future, the amount of data derived from manufacturing operations will increase due to these developments. This offers both opportunities and challenges for manufacturing companies and manufacturing research. With the developed concept, the increasing data streams can be analyzed efficiently and applicable results can be derived. The analysis results present a direct benefit in form of the most important process parameters and state characteristics, the state drivers, of the manufacturing system. These can be directly utilized in, e.g., quality monitoring and advanced process control.

Thorsten Wuest

Backmatter

Titel: Identifying Product and Process State Drivers in Manufacturing Systems Using Supervised Machine Learning
verfasst von: Thorsten Wuest
Verlag: Springer International Publishing
Electronic ISBN: 978-3-319-17611-6
Print ISBN: 978-3-319-17610-9
DOI: https://doi.org/10.1007/978-3-319-17611-6

Premium Partner

Marktübersichten

Die im Laufe eines Jahres in der „adhäsion“ veröffentlichten Marktübersichten helfen Anwendern verschiedenster Branchen, sich einen gezielten Überblick über Lieferantenangebote zu verschaffen.

Zur Marktübersicht