Simulation of a real-time process adaptation in the manufacture of high-density fibreboards using multivariate regression analysis and feedforward control

The industrial manufacturing of wood-based panels has become a highly technological process, where all parameters have to be perfectly adjusted to manufacture products of high quality. However, variations caused by differing wood characteristics as well as variations of single process parameters can cause out-of-control events. These undesirable events can be diminished by monitoring and controlling the entire manufacturing process using multivariate statistical techniques. Hence, a real-time process adaptation of an industrial scale fibreboard manufacturing process was simulated. Regression results revealed a mean normalised root mean squared error of prediction of 4.6 %, when predicting the internal bond strength of fibreboards. The regression model is regularly validated and, if necessary, recalibrated using the offline determined board properties (feedback control). Consequently, the process can immediately be adapted as soon as the board is produced (feedforward control). The investigations resulted in reliable models and revealed high potential for permanent industrial implementation.