Optimization of the Fiber Cement Composite Process

This study reflects the success of combining focused beam reflectance measurement (FBRM) techniques and artificial neural networks (ANN) to make predictions of fiber cement properties, to optimize the industrial process. Three neural networks have been developed. The inputs of these networks are the FBRM sensor measurements and the densities taken from formed sheets. The outputs are final product properties, related to product resistance. With this work, a good prediction of final properties has been achieved. The conclusions reached with the analysis of the results of neural networks can be used in establishing optimal process conditions. The obtained results demonstrate that the FBRM probe can be considered to be a good soft sensor for predicting the on-line fiber cement resistance.