Background

Industry 4.0. is a collective term for automation and data exchange in the manufacturing sector. It covers the concepts cyber physical systems (CPS), the internet of things (IoT), the internet of services, all of which aimed at enabling “smart factories” to become a reality. The CPS in such a “smart factory” monitor, start and end processes and create a virtual representation of the physical surroundings and conditions. In the long run the different CPS and the internet of things will be used to create a intercommunicating system including data from machines, logistics, production and workers. The so called fourth industrial revolution not only targets single production steps but also the value chain as a whole.

Project Content

In the preceding project WiKant – a project dealing with the production of metal profiles – a system has been developed that is capable of predicting whether a specific part can be produced or not. More precisely, metal profiles are ordered in a web shop, then the system checks semi-automatically whether the parts can be produced, after that the data is transmitted to the production line and finally the parts are delivered. In the present project the digitization of this production chain is driven forward. Currently, the metal profiles have to be examined for production faults and failures by qualified workers in order to find out whether the parts meet the quality standards. This is time-consuming and slows down the production process, in particular when a part does not pass quality control and has to be produced again. However, a (semi-) automated quality check on the basis of modern sensor technologies, networked systems and image processing tools can lower the likelihood of such problems occurring.

Goals

It is the main aim of the project to create a device that automatically checks whether metal profiles conform with quality specifications. This ensures that every single metal profiles is scanned and that deviations from the production standards or any anomalies are reported back to the system. Therefore, it is possible to automatically inspect and approve production orders in real time and to quickly identify defects in the production chain. The collected data on anomalies and defects can also be used to determine the intervals for maintenance work. In summary, the combination of image processing tools and 3D scanning technologies gives an apparatus that thoroughly documents the quality of each part that has been produced.

Methods

A standardized measuring device (i.e., measuring table) is constructed. This device mainly consists of a 3D scanning apparatus whose data are processed by image recognition software tools (i.e., computer vision).  It is tested then whether the prototype version of this set-up is suitable for everyday operation. After the testing phase and making necessary adaptions the prototype is integrated with the production plant of the corporate partner. The whole set-up inspects and measures the produced parts in real time, compares the measurements with the 3D model of the products ordered, recognizes and records anomalies automatically and as a result ensures high quality standards. To meet the needs of the staff members with regard to design and usability, the device is developed according to the principles of user-centred design.

Results

Modern information and communication technologies provide the means to make industrial production more efficient, autonomous and flexible. The current project makes a contribution to that by fusing sensor technologies with computer vision to check the quality of metal profiles. Not only does this save time because quality checks on a random basis are replaced by an automated procedure, but also optimizes the use of raw material because problems are identified on the spot and the production of faulty parts can be stopped quickly. In the long run the whole value chain will be mapped and recorded digitally ranging from the purchase requisition, the production of the parts ordered, to the assembly on the construction site. The automation of the quality check and making the collected data accessible for further analyses is one step on this path.