Entwicklung eines ganzheitlichen Prognosemodells zur Kompensation von Varianzen in Prozessfolgen mittels Support Vektor Maschinen

The base for an effective planning and optimization in production is the understanding of working connections between process parameters, product qualities and the business organizational putting levers. The customer-oriented alignment of a production line with the target figures product cost, product quality and delivery time can be reached by means of classification methods. Thereby one can show and classify each relevant control variable of the production line, both on a technical and organizationally level. Through this a coupled analysis of the working connections, with regard to the result at the end of the process consequence and an assignment of measures is made possible. In this dissertation a method was developed, which at first permits an integral monitoring of all aim sizes looked at by means of a vectorial representation to be able to analyze and to classify the feature rooms with the aid of an approach from the artificial intelligence. This approach permits a generation of measures adapted on the situation and aligned with the complete aim sizes support vector method due to the very good forecast ability of these. In conclusion the method was transferred and the method verified on an exemplary process consequence