Enhancing the Quality of Process Chains in a Non Stable Production Environment using Data Mining Methods

V. Bettin and H.D. Osuna (Germany)


Process Control, Process Modelling, Data Mining, Quality Control Loop, Prediction of Quality, Dynamic Inspection


The goal of efficient production processes is top quality output with low rate of scrap. To reach this goal model based quality control loops can be used. Because of the complexity of process chains tools for modelling are used for the improvement of adjustment, optimisation, conversion and control. Each production step has its special behaviour. Some of them – the stable processes – can be often described with standard quality management tools. The prediction of the output – the properties of the product – is possible if the right model ling is used and all the input values are known. The others – the non-stable processes – need different tools. In consequence of a non-stable process each pro duced part has to be checked. In some processes – i.e. light metal foundry – the produced parts can last hours in the cooling phase. The quality results can first be achieved at that time. With the combination of captured production data and known quality results predictions for part and process quality are possible and usable for process regulation in a quality control loop. Using Data Mining methods models can enhance processes to reach production’s goals. With the known predictions problem reaction times are lowered and stable parts of the process are defined.

Important Links:

Go Back