Big data mining and analytics tools I


In this deliverable, we present the current development state of the task 5.1. We start by presenting a general background in this chapter. In chapter 2, we present the COMPOSITION scenarios where Data Analytics will be used, and the place of thesetechniques in the general COMPOSITION architecture. In chapter 3,we present the solution for Big Data Analysis. In chapter 4, we present the analysis of the possible representation of the Big Data Analysis. Finally, we end with a short conclusion in chapter 5

The manufacturing industry is being disrupted in what is already known as the 4th industrial revolution or Industry 4.0. This revolution is driven by the need of reduction of the time-to-market[1], increment of complexity, (mass customization)[2][3], and added value services[4]around the products --all together in a competitive globalized world [4].To solve these challenges, this revolution is introducing a set of new advanced networking technology, hardware, and more important, intelligent software. While in the 3rd industrial revolution, the manpower was replaced by simple "hardwired" automatization [5],which could not adapt to market trends fast enough -e.g. in mass customization where almost unlimited variations of a product can be produced [1], [6],[7], [8],[9]-the current revolution is creating so-called cyber-physical systems,where machines collect data, communicatewith each otherand jointly take decisions[10]. To succeed in this new industry, the technologies must be highly adaptable, manageable, and in many cases even self-managed and self-configured[11], [12]. To achieve this degree of intelligence, advanced algorithms have beenincorporatedinto the production process to achieve embeddedartificial intelligence (AI) within the process. This embedded AI has been constructed fromthe experiences obtained by the machines and usually designed by data scientists[13]. These techniques can be used in several manufacturing challenges such as predictive maintenance or product defect detection. However, while many efforts has been invested in tackling these challenges, few workshas been done in developing an integrated manageable platform to solve these problems[14], [15], [16], [17],[18]. Most of the solutions propose a heterogeneous set of technologies to achieve the goals, and mostly none provide tooling for a deploymentor to manage the solution in a deployed running system. Most existing solutions do not provide any tooling for collecting data and management for AI technologies, such as online machine learning (ML). Additionally, the solutions provide very little integrated deployment tools for the reproduction of ML methodsor modelsin other deployment environments.We believe in the need ofan integrated extensible solution that providesruntime management tools and is able to manage and configure itself. A platform that provides a set of mechanisms for real-time data collection, processing, and analysis. In this manner, it is possible to create common methodologies to reproduce and redeploy ML and other AI technologies reducing their cost and increasing their usage. In this deliverable, we present our solution first shownin[19], withina manufacturing environment. More specifically, we deploy the solution in Surface-Mount Technology production in BLSplant and in other process in the KLEproduction plant.

Ecosystem for Collaborative Manufacturing Processes _ Intra- and Interfactory Integration and Automation
Not specified (see website if available)
  • Not specified