Invited Talk - Prof. Diego Galar

Big Data Issues for Mining Knowledge in Maintenance Information Systems Prof. Diego Galar Division of Operation and Maintenance Engineering Lulea University of Technology, Lulea, Sweden Date & Time: 12:00 - 01:00pm; July 22 (Tuesday), 2014 LOCATION: Ballroom 1

Abstract

Maintenance is a strategic process and/or service all over the word. The effective monitoring of the assets is a key task in order to guarantee efficient and safe exploitation. Current assets, with plenty of sensors' already installed and pervasive computing on them generate a huge of data along their day-today working as well as their maintenance. In this scenario, asset managers host a large number of diverse systems where data, regarding different aspects of their activity, are stored. In most cases, these data are captured, stored and processed by different, often incompatible systems , and further managed by independent departments and not shared at all.

In addition, it is fairly frequent that while large amounts of data are gathered only a small fraction of it is used for a specific purpose; the remainder is simply saved, or even worse, discarded. Important information and knowledge are buried within those extends of data. They could be discovered if the data were properly organized and processed. A number of unused techniques and paradigms in Information Technology allow for this knowledge discovery. On the one hand, Cloud Computing brings a new service delivery paradigm allows for pay-as-you-go services, which adapt to the customer needs, without requiring expensive data centers' infrastructures. From the customer point of view, Cloud Computing brings the possibility of requesting computing resources, storage, and network bandwidth - as needed. This paradigm can be used both from public services or even within-house private services. On the other hand, Data Mining techniques, developed along fifteen years have allowed the discovery of non-trivial knowledge from large databases. As the computing resources increase in their power and decrease in their prices, the capture and storage of data are becoming increasingly affordable leading to huge Big Data techniques to improve asset monitoring and management stores of data. This is increasing in several dimensions, not just size but also, variety - structure, semi-structured, non-structured -, speed of gathering, ... to face their management and processing new computing techniques are required. They are all included under a new term: Big Data. It refers to systems, algorithms, and procedures suitable to process data sets, which largely overcome the capacity of current single computers. Big Data is one-term drawing attention of many companies and institutions all over the world. Most organizations are speeding up their data processing strategies towards Big Data. This means a clear recognition by industry, agencies and public institutions. The goal of this talk is to address the current challenges of big data in maintenance; i.e. the analysis, design and implementation of systems that allow for the effective exploitation of data for asset managers.