Data Modeling and Knowledge Discovery in Process Industries

Data Modeling and Knowledge Discovery in Process Industries

Benjamin Klöpper (ABB Corporate Research Center, Germany), Marcel Dix (ABB Corporate Research Center, Germany), David Arnu (RapidMiner GmbH, Germany) and Dikshith Siddapura (ABB Corporate Research Center, Germany)
Copyright: © 2016 |Pages: 11
DOI: 10.4018/978-1-5225-0293-7.ch009
OnDemand PDF Download:
No Current Special Offers


Dispersed data sources, incompatible data formats and a lack of non-ambiguous and machine readable meta-data is a major obstacle in data analytics and data mining projects in process industries. Often, meta-information is only available in unstructured format optimized for human consumption. This contribution outlines a feasible methodology for organizing historical datasets extracted from process plants in a big data platform for the purpose of analytics and machine learning model building in an industrial big data analytics project.
Chapter Preview

A Typical Data Analytics Project In Process Industries

From a business model perspective, such a prediction feature for industrial control systems could be delivered to plant operators as a product (e.g. as a control system product extension), as a service (e.g. a data analytics consultation service for existing plants), or a combination of both (as so-called product-service-system or PSS, cf. Mont, 2001). In either way, the underlying development project to create prediction models in process industries will deal with similar tasks which are illustrated in Figure 1 and explained in the following. As we will see in this chapter, a key challenge in industrial data analytics is: understanding the customer problem and data, in order to prepare this data for the prediction model development.

Complete Chapter List

Search this Book: