Dual particle filters for state and parameter estimation with application to a run-of-mine ore mill

Abstract

Measurements are not readily available for grinding mills owing to the nature of the milling operation. State and parameter estimation for a grinding mill which forms part of a run-of-mine ore milling circuit has been implemented.These estimates may then be used in an advanced control algorithm. The estimation was done with dual particle filters as well as with a simultaneous estimation scheme, on simulated data, to compare the performances. The sensitivity analyses for the different schemes show the class of systems in which dual estimation may produce superior results.

Throughput and product quality control for a grinding mill circuit using non-linear MPC

“…Hodouin (2011) describes how the peripheral tools of a control loop are as important as the controller itself to successfully control and optimise a mineral processing plant. The peripheral tools include disturbance observers for external disturbance rejection (Olivier et al, 2012a), model-plant mismatch detection (Olivier and Craig, 2013), and state and parameter estimation (Olivier et al, 2012b).…”

“…A particle filter was successfully applied to a grinding mill circuit in simulation in Olivier et al (2012b) to estimate the mill states. A comparison between EKF and particle filtering for grinding mill circuits can be seen in Cuevas and Cipriano (2008).…”

“…The particle filter method of Olivier et al (2012b) is used for the estimation of all five mill states. However, to use this method it is necessary to inferentially determine the mill discharge flow-rates as industrial mills have no direct measurements of these variables.…”

“…The particle filter of Olivier et al (2012b) is used to provide mill state estimates for water (X mw ), solids (X ms ), fines (X mf ), rocks (X mr ) and balls (X mb ). Only a brief description of the particle filter implementation is given here and further details can be found in Olivier et al (2012b).…”

“…Only a brief description of the particle filter implementation is given here and further details can be found in Olivier et al (2012b).…”