Zero-defect manufacturing by means of a learning supervision of process chains
Highly productive and low-stock or "just-in-time" manufacturing systems require safe manufacturing also in the process chain. Starting from a systematic analysis of defects and their causes in single-product and series production, a method has been developed enabling the supervision of quality in the process chain as well as its optimization by means of a learning management system. For this, quality data collected are evaluated and represented in mathematical process models. From these data, logic patterns are derived by means of cluster analyses or statistical analyses, which provide information on the causes and the assessment of nonconformities. Based on this and with the help of high-performance parallel computing, it has become possible to realize zero-defect manufacturing in process chains by means of a knowledge and neuronal-network-based learning system.