Abstract

On-line particle size soft-sensors play an important role in the efficient operation of control systems in many industrial grinding plants. To cope with disturbances and changing operating points it will be necessary to adapt the soft-sensor parameters to the new conditions. This work proposes the use of constrained parameter estimation algorithms, in order to take into account some process prior knowledge and enhance the performance of the soft-sensor. Several experiments, using data taken from an industrial grinding circuit, illustrate the benefits of the proposed approach. © 2007.