Application of a statistical based cultivation and pattern recognition strategy for monitoring paper machine web breaks

Paper machine operation and control require continuous monitoring and adjustment of many process and product runnuability parameters. This research will focus on the development of a strategy using neural networks to evaluate the state of a selected paper machine and advise operators when adjustments are necessary. The state of a paper machine depends upon many variables. When these variables deviate far from normal conditions, there is a likelihood of a paper machine web break. The break index is a collective attempt to quantify the level of stress in the manufacturing process of paper. An experienced operator can usually predict this situation. Many different combinations of variables can lead to a break, which often can cause an operator confusion concerning what variables to adjust. By feeding these process variables into a neural network system, the process behavior can be learned approximately the same way the human brain learns it. Integration of artificial intelligence devices that know the behavior of the process and product variables will be key factors in maintaining process and product quality.