Predictive Functional Control Using a Blending Approach

IOP Conference Series: Materials Science and Engineering, 2020

In this paper, a novel combination of a multi-model predictive controller (MMPC) and an adaptive integral controller is usedto achieve offset-free control of a nonlinear process. The idea is to avoid the more complex tuning that comes with an offset-free control based on an observer. To create an easily tuned controller based on a piecewise linear (PWL) description of an MPC setup, which utilizes a Bayesian weighting approach. The PWL models are also used to design the separate the I-controller that is made adaptive by using the Bayesian weighting again. The MPC and the I-controller are then acting in parallel. The setup is implemented and tested using a simulation of a pH neutralization process.

Computers & chemical …, 2006

Journal of Process Control, 2004

The real-time implementation of a set of multi-linear model-based control design methodologies is studied using a bench-scale pH neutralization system that exhibits nonlinear dynamics. It is envisaged that advanced model-based control strategy based on the multi-linear models presents a promising paradigm to design controllers for complex nonlinear plants. The multi-linear modeling philosophy is based on the selection of a set of linear models, complemented with an adaptation mechanism to explain the nonlinear plant behavior in the whole operating range. Practical implications of each control strategy are evaluated and discussed.

2013

This paper deals with the development of a multivariable predictive control structure for improving the nitrogen removal of a biological wastewater treatment plant while reducing the operational costs. A simple dynamic matrix control algorithm is utilised as predictive controller and applied to a full-scale municipal wastewater treatment plant for controlling nitrogen concentrations at the end of the biological process. The complex calibrated model of the process is implemented in a commercial simulator that acts as a real-time testing platform for the proposed control structure, and allows the identification of the multivariable inputoutput model for the predictive control. Simulation results show the potentialities of the chosen predictive control, which allows the reduction of ammonia peaks in the effluent and at the same time permits a reduction of the energy consumption costs.

Computer Aided Chemical Engineering, 2015

This paper describes a procedure to find the best economically controlled variables for the activated sludge process in a wastewater treatment plant despite the load disturbances. A further controllability analysis of those variables including a nonlinear model predictive controller (NMPC) has been performed. The self-optimizing methodology has been applied, considering the most important measurements of the process. A first pre-screening of those measurements has been done based on the nonlinear model of the process and typical disturbances, in order to avoid non feasible operation. The NMPC performance has been compared with a distributed NMPC-PI structure.

Water Science and Technology

This paper presents a generalized predictive control (GPC) technique to regulate the activated sludge process found in a bioreactor used in wastewater treatment. The control strategy can track dissolved oxygen setpoint changes quickly, adapting to the system uncertainties and disturbances. Tests occur on an Activated Sludge Model No. 1 benchmark of an activated sludge process. A T filter added to the GPC framework results in an effective control strategy in the presence of coloured measurement noise. This work also suggests how a constraint on the measured variable can be added as a penalty term to the GPC framework which leads to improved control of the dissolved oxygen concentration in the presence of dynamic input disturbance.

This paper presents an advanced predictive method control of water treatment plants. The model for prediction if obtained by identifi-cation methods based on neural networks, tree partitioning and wavelet networks. The model describes the interaction between the pH factors of the raw water, the amount of lime and ferrous sulfate as inputs/input disturbances and the pH factor of the processed water at the end of the process. Since the derived models are highly nonlinear, we will use linearization in order to implement linear model predictive controller. We have identified several different working regimes of the water treatment plant, and we design different model predictive controller for each of these regimes. At the end we confirm the design with simulation results.

2016 American Control Conference (ACC), 2016

Journal of Process Control, 2000

A plant-wide control strategy based on integrating linear model predictive control (LMPC) and nonlinear model predictive control (NMPC) is proposed. The hybrid method is applicable to plants that can be decomposed into approximately linear subsystems and highly nonlinear subsystems that interact via mass and energy¯ows. LMPC is applied to the linear subsystems and NMPC is applied to the nonlinear subsystems. A simple controller coordination strategy that counteracts interaction eects is proposed for the case of one linear subsystem and one nonlinear subsystem. A reactor/separator process with recycle is used to compare the hybrid method to conventional LMPC and NMPC techniques. #

IFAC Proceedings Volumes, 2014

The need for processes to be operated under tighter performance specifications and satisfy constraints have motivated the increasing applications of nonlinear model predictive control (MPC) by the process industry. Nonlinear MPC conveniently meets the higher product quality, productivity and safety demands of complex processes by taking into account the nonlinearities and constraints in the processes. This paper examines the application of a nonlinear MPC to a multi-variable coagulation chemical dosing unit for water treatment plants. A nonlinear model of the dosing unit based on mechanistic modelling and identified by nonlinear autoregressive with external input (NLARX) estimator was developed. The simulation of the MPC based control system showed very good performance for set-point tracking and disturbance rejection. The closed loop performance of the nonlinear MPC (NMPC) compares favourably with the unconstrained and linearised nonlinear MPC (LTIMPC). The results of this study shows the suitability of nonlinear MPC for process control in the water treatment industry.

IFAC Proceedings Volumes, 2000

Process control is considered an important means to meet the increasingly tighter demands that are being placed on sewage treatment in most western countries, especially with respect to nutrients. This paper investigates stability robustness of a wastewater treatment plant controlled with Model Predictive Control (MPC). Aims of this model study are to study possibilities and limitations of advanced control application to wastewater treatment, to obtain insight in the process factors that affect control system robustness , and to find tuning rules to improve MPC robustness. A simple plant model was studied. A structured uncertainty description of parameter and state uncertainty was used to avoid conservative results. ;.t-Analysis was used to compute robustness bounds of the closed loop system. The results show that achievable robustness improvement by tuning is limited and indicate that nonlinearity has a stronger effect on stability than parameter errors.

Control Engineering Practice, 2007

IFAC Proceedings Volumes, 2002

This paper focuses on the design of a model-based predictive control (MPC or MBPC) technique to regulate the concentration levels of nitrate in both anoxic and aerobic zones of a pre-denitrifying activated sludge plant, aiming to improve the nitrogen (N)removal from wastewater. The synthesis of the MPC controller is based on a linear extended state-space model of the process, where an identification horizon is added to include a sequence of past inputs/outputs. This sequence can be used to estimate the model or the updated state of the process, thus eliminating the need for a state observer. The linear state-space model was obtained through subspace identification methods. The controller performance is tested by simulation and the results show the efficiency of the proposed strategy.

Industrial & Engineering Chemistry Research, 2004

Scheduling quasi-minmax model predictive control is an MPC algorithm developed by (Lu and Arkun, 2000) initially for linear parameter varying (LPV) system, then developed for nonlinear systems in (Lu and Arkun, 2002). In this paper, real-time application of the scheduling quasi-minmax MPC algorithm onto a benchscale pH neutralization reactor is presented. The control performance is compared with multi-linear model based :tvIPC modified from the algorithm in (Kwon and Pearson, 1978) and scheduling IMC-PID controller in which tuning parameters are from IMC design in (t-/lorari and Zafirioll, 1989

IFAC-PapersOnLine

In this paper, an optimal Proportional-Integral-Plus (PIP) controller based on State-Dependent Parameter (SDP) model is developed to control a highly nonlinear and time-varying pH neutralization process. Since the reaction invariants of the pH process are unavailable for direct measurement, a state observer is required for their estimation. Unfortunately, a closed-loop state observer cannot be designed for the pH process due to the decoupled nature of the reaction invariant dynamics. However, this problem is circumvented by engaging the power of State Variable Feedback (SVF) in the context of a Non-Minimum State Space (NMSS) framework. Furthermore, to compensate for process nonlinearities and unmeasured disturbances, an SDP model is identified using data collected from open-loop experiment, and a straightforwardly tuned optimal PIP controller is subsequently designed based on the SDP model to obtain the SDP-PIP controller. For benchmark purposes, a digital PI controller is designed and results are compared with those from the SDP-PIP controller. Simulation results show that SDP-PIP outperforms PI both in set point and disturbance changes.