Constrained Control and Estimation of Homogeneous Reaction Systems Using Extent-Based Linear Parameter Varying Models

Abstract

In this paper, a model predictive control (MPC) strategy for nonlinear homogeneous reaction systems is proposed. The nonlinear system is first transformed to a linear parameter varying (LPV) system by means of a linear transformation known as extents decomposition. The resulting LPV system is further converted to a linear time invariant (LTI) system by means of a parametric state feedback and feedforward control laws. The use of this description results in linear MPC with a quadratic performance index and nonlinear state parameter constraints. However, based on the polytopic nature of the parameter of the LPV systems, the constraints are transformed to a set of intersected polyhedrons. The final result is a linear MPC problem with linear constraints that can be easily converted and solved as a quadratic programming problem. Finally, the performance of the control strategy is illustrated in simulation and compared with a controller based on a constant-parameter LTI model.

Full-Scale Digesters: An Online Model Parameter Identification Strategy

“…However, despite the numerous advantages, such as the guarantee of identifiability of parameters, a rigorous identification procedure that covers a wide range of operational conditions, and the ability to validate the performance during transients, especially over unstable phases, this method exhibited several disadvantages. The most important being: the supposition of linearity using independent and dependent variables, the sensitivity to noise, and the presence of outliers and overfitting [14]. As the optimal parameter identification has shown better results [13,29], in the following section, a well-known parameter optimization method is presented [14,30]; however, further on, the alternative presented in the literature, a new parameter optimization method is proposed, aiming to increase considerably the performance in comparison with traditional methods.…”

“…Additionally, there is no evidence of inhibitions. Therefore, the influence of these parameters is discarded [10,14].…”

“…This controller was validated by full-scale experiments and showed promising results [13]. Nevertheless, the accuracy of NMPC depends on the capacity of the phenomenological-based model to represent reality, which derives from the accuracy of the model parameters [14].…”

Extension of the Set of Raw Sludges used in Industrial Anaerobic Digestion Reactors Implementable in Control Schemes Using an On-line Model Parameter Identification Strategy

“…The transient has been tested step by step once the inlet conditions change. However, despite the numerous advantages; the guarantee of identifiability of parameters, a rigorous identification procedure that covers a wide range of operational conditions, and the ability to validate the performance during transients, especially over unstable phases, the method exhibit disadvantages; the supposition of linearity related with independent and dependent variables, the sensitiveness to noise, the presence of outliers and overfitting [11].…”

Model Predictive Control with On-line Kinetic Parameter Identification Strategy on Anaerobic Digestion Reactors

“…Pross in-depth knowledge of the process kinetics is an extreme challenge in numerous engineering applications; therefore, there is a huge motivation to search for a category of observers which allows one to asymptotically reconstruct the missing states even when the kinetics are unknown. A cascade structure of an asymptotic observer to estimate the concentration of acidogens and methanogens is proposed followed by a kinetic parameters observer that estimates two reaction rates of the process [6][7][8].…”