State Observer for Nonlinear Systems: Application to Grinding Process Control

Due to the measurement problems encountered in mineral processes, observers are appropriate ingredients of advanced model based control algorithm. The measurement problem can be solved by designing nonlinear observer. This paper discusses the way in which a state observer may be designed to control a special class of nonlinear systems. Focus is put on the pertinent applicability of the scope of these techniques, to control the dynamics of mills in mineral processes. The approach uses a small number of parameters to control the mill power draw affected by sudden changes within the system. It provides with principles and ability of the system to adapt to changing circumstances due to intermittent disturbances (like for instance changes in hardness of the raw material). Performance and stability analysis was developed. Using a generalised similarity transformation for the error dynamics, it is shown that under boundedness condition the proposed observer guarantees the global exponential convergence of the estimation error. This way, the nominal performance of the process is improved but the robust stability is not guaranteed to fully avoid the mill plugging.