Predictive Control of the mineral particle size with kernel-reduced volterra models in a balls mill grinding circuit

• Autor: Nieto Chaupis, Huber
• Formato: objeto de conferencia
• Fecha de Publicación: 2015
• Institución: Universidad de Ciencias y Humanidades
• Repositorio: UCH-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.uch.edu.pe:uch/376
• Enlace del recurso: http://repositorio.uch.edu.pe/handle/uch/376; http://dx.doi.org/10.1109/ISIE.2015.7281453; https://ieeexplore.ieee.org/document/7281453
• Nivel de acceso: acceso embargado
• Materia: Algorithms; Grinding (machining); Industrial electronics; Model predictive control; Monte Carlo methods; Particle size; Predictive control systems; Computational simulation; Mill-grinding; Mineral particles; Model based predictive control; Monte Carlo data; Output variables; Predictive control; Ball mills; Volterra model

We report the results of the application of the Model-based Predictive Control (MPC) algorithm for a 3×3 MIMO balls mill grinding system by using computational simulation and Monte Carlo data generation. For this purpose, the system has been identified through a reduced scheme of Volterra formalism by which the proposed methodology has required to employ up to 20 parameters. Subsequently, the model enters in a framework of MPC which targets to control the particle size, one of the most important output variables in this study. According to the simulation results the system identification error is of order of 3%, whereas the MPC scheme applied to control a desired set-point namely 75 %-200mesh is accompanied by a deviation of ±5%. Since the balls mill grinding circuit is a nonlinear system, it is expected that the system might collapse as consequence of the accumulated circulant load. The simulations have predicted that the MPC algorithm running with a Volterra-based model might surpass situations of stops and alarms system, even in those cases where the system is attacked by unexpected disturbs and random events.