JR Centre for Knowledge-Assisted Visual Analytics for Industrial Manufacturing Data

This JR Centre strives to optimise the interaction between man and machine in order to gain relevant insights from industrial manufacturing data and production data more effectively and efficiently. On this basis, the associated manufacturing processes can then be improved, optimised and also partially automated.

Due to the digitalisation of the manufacturing industry, massive amounts of data records are collected in all phases of an industrial production process, most of which also have a temporal context. If these are placed in relation to each other, to the overall process and to end products, and the observations are interpreted in a relevant way, valuable information can be gained from them. This allows the entire production process to be optimised, for example with regard to dealing with environmental influences or the maintenance of production machines.

The difficulty of such an endeavour is that the sheer volume of data can be overwhelming for human experts, partly because it can appear contradictory. For computer systems, on the other hand, recognising trends and patterns in large amounts of data is much easier, but they simply lack the expert knowledge of humans to interpret the numerous patterns that occur in a meaningful way.

In order to improve the collaboration between man and machine in this context, methods from the field of visual analytics (VA) are to be used to visualise such data in order to derive insights. To this end, the JR Centre team is working on making a previously developed theoretical model from this field usable in practice in industrial production: The model was designed in the successfully completed project "KAVA-Time" (Knowledge-Assisted Visual Analytics Methods for Time-Oriented Data), which aimed to extract explicit expert knowledge, formalise it and store it in a VA system so that it can be used for automatic data analyses and provide information gains in the results.

Putting this principle into practice is the goal of the JR Centre, whereby great importance is also attached to an intuitive user interface that can be operated by the domain experts through visual interaction options such as drag and drop gestures. In this way, the experts can be made aware of their implicit knowledge and converted into explicit knowledge, which can then be fed into the KAVA system. The model is to be tailored to different user groups: Process technicians, quality managers, mechanical engineers and sales managers. Its practical use holds great potential for optimising the adjustment of production machines, sales and operational planning and the search for causes of errors in the manufacturing industry. Furthermore, such systems are also intended to counteract the loss of knowledge in host institutions due to migration or waves of retirements, or to train new employees using the stored knowledge.