Applying linear model predictive control to crushing circuit simulations
Paper in proceedings, 2019
In minerals processing control is vital to increasing performance and output of a plant. Model predictive control (MPC) is one of the most effective control strategies used in process industry. Sbárbaro (Sbárbaro and Del Villar, 2010), Asbjörnsson (Asbjörnsson, 2015), and Légaré (Légaré et al., 2016) have for example highlighted the possibility to simulate, develop and tune control systems using plant simulators within minerals processing. In more recent years the use of model predictive controllers have increased, and many process industries now utilize more advanced control methods to regulate their processes, raising the question regarding how to, build, test and verify a proposed control solution successfully. In this research, a linear model predictive controller is formulated and applied to a simulator of an existing tertiary crushing circuit. The circuit has in previous work by Johansson (Johansson, 2017) been modeled in MATLAB Simulink and calibrated relative plant data to an average error of less than 10% over 8 hours of production. The work demonstrates how to formulate the controller model structure to describe conveyor belts, storage components, screens, crushers, and control objectives for a minerals processing circuit. The work illustrates how to apply and successfully evaluate possible upsides in a circuit. The advantage of a crushing circuit simulator is that it can be configured, tuned and modified in any possible way, which is of great benefit to developers and engineers. Additionally, examples of how to use estimated measures to improve the performance of the plant are demonstrated. The control performance is finally evaluated for three different controller configurations.
Model Predictive control