Paper web quality profile analysis tool based on artificial neural networks

Abstract

In order to guarantee the uniform quality of paper across the whole cross-direction of the paper machine, online measurements and control equipment related to different paper quality profiles are necessary. In modern paper machines, only the basis weight, the moisture and the caliber profiles are controlled by an online control system. However, there are several important profiles that cannot be easily measured or controlled directly. As an example, the fiber-orientation profile is difficult to measure online and, in addition to this, the control mechanism is not clear. This is mainly due to insufficient knowledge about the complex relationship between the orientation profile and the other profiles. Artificial neural networks (ANNs) can be used to model such difficult complex systems where only input-output data is available. In this case study the ANN methods are applied to the fiber-orientation profile analysis and its control. The same method can be used to control other profiles as well.