Generalized Predictive Controller for Ball Mill Grinding Circuit in the Presence of Feed-grindability Variations

Abstract

Feed-grindability variations (clinker hardness) in cement grinding circuit affects the product quality and productivity of the cement plant. This investigation proposes a generalized predictive controller for cement grinding circuit that is more robust to feed-grindability variations. The main building block of the proposed controller is a new model for cement grinding circuit that directly relates product quality with elevator current and main drive load. The advantage of the model is that the effect of feed-grindability variations on product quality can be easily observed from the output. To develop such a model, this investigation adapts a data driven modelling approach. Experimental data obtained as measurements from cement grinding circuit in a cement mill located near Chennai, India is used to develop the transfer function model based on least squares approach. The model obtained from data driven modelling is used to design a generalized predictive controller whose objective is to optimize the product quality in the presence of feed-grindability variations without breaching physical and operational limits of the cement grinding circuit. The tuning parameters of the proposed generalized predictive controller is adjusted to meet performance metrics specific to cement industries. Our results show that the proposed controller provides better product quality in the presence of feed-grindability variations than other optimization based controllers such as the linear quadratic regulator.

Predictive Controller Design for a Cement Ball Mill Grinding Process under Larger Heterogeneities in Clinker Using State-Space Models

“…This is the measure of clinker hardness. The nominal grindability factor for the clinker is 33, which is a dimensionless quantity [29]. The grindability factor of the clinker is analysed in the laboratory.…”

“…However, the controllers developed based on these non-parametric models lead to computational complexity in the real-time environment. A generalised predictive controller was suggested for the cement grinding process [29] based on the transfer function models.…”

“…To test the GPC discussed in [29] and the state-space model predictive controller in the simulation; …”

Affordable Control Platform with MPC Application

“…The introduced control has three main characteristics: (1 or model prediction [5,6], for example, can be implemented easily using Python, as the language offers various facilities to manipulate generic data, like easy to use lists and dictionaries, generators, comprehensions, and offers good support to object-oriented and functional programming. Advanced control techniques like state-feedback [7] or even those with intensive computational requirements like fuzzy-control 8 and model predictive control [9,10] can then be easily implemented.…”

Constrained multiobjective PSO and T-S fuzzy models for predictive control

“…According to MPC strategy, the generating of the control actions is established by minimizing a cost function, which uses an internal process model to forecast the behavior of the system over the prediction horizon H P [2]. Although MPC was widely studied [1][2][3][4][5][6], this control technique is still an open field in the area of the constrained control problems typical with nonlinear multivariable systems.…”