Long Range Predictive Control of Nonlinear Processes Based on Recurrent Neuro-Fuzzy Network Models

A type of recurrent neuro-fuzzy network is proposed in this paper to build long-term prediction models for nonlinear processes. The process operation is partitioned into several fuzzy operating regions. Within each region, a local linear model is used to model the process. The global model output is obtained through the centre of gravity defuzzification which is essentially the interpolation of local model outputs. This modeling strategy utilizes both process knowledge and process input output data. Process knowledge is used to initially divide the process operation into several fuzzy operating regions and to set up the initial fuzzification layer weights. Process input output data are used to train the network. Network weights are such trained so that the long-term prediction errors are minimized. Through training, membership functions of fuzzy operating regions are refined and local models are learn. Based on the recurrent neuro-fuzzy network model, a novel type of nonlinear model-based long range predictive controller can be developed and it consists of several local linear model-based predictive controllers. Local controllers are constructed based on the corresponding local linear models and their outputs are combined to form a global control action by using their membership functions. This control strategy has the advantage that control actions can be calculated analytically avoiding the time consuming nonlinear programming procedures required in conventional nonlinear model-based predictive control. The techniques have been successfully applied to the modeling and control of a neutralization process.

This work establishes the feasibility of using a multilayer perceptron for the development of a multimodel that combines structurally simple local models developed in different operating regions. The local models either are obtained by linearizing a first principles model or are identified from input−output data using a linear combination of Laguerre filters. In particular, it is shown that the proposed multimodels can capture dynamic and steady-state characteristics of a continuous fermenter, which exhibit input multiplicities and change in the sign of steady-state gains, fairly accurately over a wide operating range. The proposed multimodel is further used to develop a nonlinear model predictive control (NMPC) scheme. The effectiveness of the NMPC scheme is demonstrated by simulating a servo problem that requires the fermenter to be controlled at its optimum operating point, which happens to be singular points where the invertibility is lost. The proposed NMPC scheme is found to achieve a smooth transition for large-magnitude setpoint changes and control the systems at the singular operating point even in the presence of measurement noise. The NMPC scheme is also found to be robust to moderate variations in system parameters.

The article studies the trajectory tracking control problem of underactuated marine vehicles via the nonlinear model predictive control (NMPC) strategy, where practical control and state constraints present. It is a well-known challenging issue that the conventional NMPC is not applicable for underactuated marine vehicles, due to the fact that there does not exist a local static continuous state-feedback controller to stabilize the underactuated dynamics. To resolve this issue, this article proposes to construct an auxiliary time-varying tracking controller to aid terminal constraint design in the NMPC framework, where the time-varying tracking controller borrows the ideas from Lyapunov's direct method and backstepping approach. Based on this, a novel NMPC algorithm is designed to ensure trajectory tracking control of underactuated marine vehicles. Furthermore, a systematic parameter design approach is developed. Under the designed parameters, we show that the tracking error system is input-to-state stable (ISS). Finally, the effectiveness of the designed algorithm is verified by thorough simulation and hardware experiments.

Improving performance and increasing process profitability represent a priority for the long‐term operation of the crude oil distillation process. To increase process profitability and maintain control performance, in this paper, an improved economic‐based nonlinear model predictive control (NMPC) strategy is used to optimize the crude oil distillation (CDU) process. Considering the strong coupling and nonlinear dynamics of the CDU process, we make some modifications for the existing NMPC strategy. To reduce the complexity of the nonlinear prediction model, a support vector regression (SVR) based model is employed to describe the dynamical behaviour of the CDU process. Considering the coupling of the CDU process, we propose the concept of the coupling weighting factor to modify the objective function of the NMPC optimization problem. The simulation results demonstrate the effectiveness of the proposed improved NMPC algorithm for the CDU process.

Nonlinear MPC based on a Volterra series model for greenhouse temperature control using natural ventilation

A magnetic levitation system (Maglev) is a sensitive, multi-parameter, nonlinear, and unstable system that is utilized to levitate a ferromagnetic object in free space. Due to its vast applications, various research studies in the field of control strategy have become extremely important and challenging. This work proposes the design of a nonlinear model predictive (NMPC) control scheme for the object position control against the nonlinearities and uncertainties of a Maglev system. A novel bio-inspired Artificial Protozoa Optimization (APO) algorithm is used to fine-tune the NMPC parameters, which include best weighting matrices ( ), shorter prediction horizons ( ), and shorter time steps ( ) to minimize the objective cost function. The effective performance of the NMPC is verified using simulation-based results in MATLAB. The CasADi toolbox is utilized to solve nonlinear optimization problems and handle the nonlinearity of the Maglev system model. Simulations are implemented for three trajectories tracking (step, sine, and square) with 20% and without Maglev parameters perturbations. To prove the superiority of the proposed controller, comparisons are made with the conventional Linear Quadratic Regulator (LQR) and proportional-integral-derivative (PID) controllers. Two performance indices are introduced, Integral of Squared Error (ISE) and Integral of Absolute Error (IAE), to examine the tracking performances of the NMPC, LQR, and PID controller. The NMPC controller has shown more efficient performance and accurate results than other controllers. The contributions of this work include a new optimization technique of APO, a new engineering application of the APO integrated with NMPC to control a Maglev system, consideration of inherent nonlinearities and system constraints, and robustness improvement under perturbation.

SummaryThis article addresses the problem of designing a sensor fault‐tolerant controller for an observation process where a primary, controlled system observes, through a set of measurements, an exogenous system to estimate the state of this system. We consider sensor faults captured by a change in a set of sensor parameters affecting the measurements. Using this parametrization, we present a nonlinear model predictive control (NMPC) scheme to control the observation process and actively detect and estimate possible sensor faults, with adaptive controller reconfiguration to optimize the use of the remaining sensing capabilities. A key feature of the proposed scheme is the design of observability indices for the NMPC stage cost to improve the observability of both the state of the exogenous system and the sensor fault parameters. The effectiveness of the proposed scheme is illustrated through numerical simulations.

Integrated nonlinear model predictive fault tolerant control and multiple model based fault detection and diagnosis

In this paper, we propose a real-time nonlinear model predictive control (NMPC) strategy for stabilization of a vehicle near the limit of lateral acceleration using the rear axle electric torque vectoring configuration of an electric vehicle. A nonlinear four-wheel vehicle model that neglects the wheel dynamics is coupled with a nonlinear tire model to design three MPC strategies of different levels of complexity that are implementable online: one that uses a linearized version of the vehicle model and then solves the resulting quadratic program problem to compute the necessary longitudinal slips on the rear wheels, a second one that employs the real-time iteration scheme on the NMPC problem, and a third one that applies the primal dual interior point method on the NMPC problem instead until convergence. Then, a sliding mode slip controller is used to compute the necessary torques on the rear wheels based on the requested longitudinal slips. After analyzing the relative tradeoffs in performance and computational cost between the three MPC strategies by comparing them against the optimal solution in a series of simulation studies, we test the most promising solution in a high-fidelity environment.

The development of schemes for effectively examining unmeasured disturbances and plant-model mismatch in nonlinear model predictive control (NMPC) is an important area, which has attracted considerable attention over the past decade. In this work, we propose NMPC formulations for improving regulatory performance using nonlinear state space models identified from input-output data. The NARX-type MISO and MIMO nonlinear models identified in Srinivasarao et al.(17) are used to develop the state estimators and the predictors in the proposed NMPC formulations. We show that these models provide information about the covariances of the fast changing disturbances that are necessary for developing an extended Kalman filter (EKF) used in an NMPC scheme (EKF-NMPC). Rather than formulating an EKF to reject these disturbances, which we show has certain drawbacks, we proceed to propose an alternate prediction scheme based on the direct use of the identified noise model for future trajectory predictions. We show that the resulting NMPC formulation (DC-NMPC) is relatively easy to tune to achieve disturbance rejection as well as offset removal. The efficacy of the proposed control schemes is demonstrated using simulation studies on two benchmark control problems: (a) regulatory control of a continuously stirred tank reactor (CSTR) exhibiting input multiplicity at a singular point, and (b) servo control of a CSTR exhibiting output multiplicities. The analysis of simulation results reveals that the proposed EKF-NMPC and DC-NMPC schemes consistently generate significantly better regulatory performances, when compared to the performance of a conventional NDMC-type formulation, even in the presence of significant plant-model mismatch. In particular, the DC-NMPC performs significantly better than the EKF-NMPC formulation and can be tuned to achieve excellent regulatory performance without causing excessive input variability using parameters of a single disturbance filter. The conclusions reached through simulation studies are validated by conducting servo and regulatory control experiments on a laboratory-scale heater-mixer setup.

Nonlinear Model Predictive Control of Electrical Submersible Pumps based on Echo State Networks

In this article, we present a quasi-infinite horizon nonlinear model predictive control (MPC) scheme for tracking of generic reference trajectories. This scheme is applicable to nonlinear systems, which are locally incrementally stabilizable. For such systems, we provide a reference generic offline procedure to compute an incrementally stabilizing feedback with a continuously parameterized quadratic quasi-infinite horizon terminal cost. As a result, we get a nonlinear reference tracking MPC scheme with a valid terminal cost for general reachable reference trajectories without increasing the online computational complexity. The practicality of this approach is demonstrated with a benchmark example.

A nonlinear model predictive control (NMPC) strategy using particle swarm optimization (PSO) is presented in this paper. The PSO is improved by integrating chaotic local search (CLS) and roulette wheel mechanism (RWM). The new algorithm named RCPSO is adopted for solving the nonlinear optimization problem in NMPC. The NMPC based on RCPSO is implemented in the simulation of a neutralization reaction process for pH set-point tracking.

Optimal operation of a large-scale packed bed chemical-looping combustion process using nonlinear model predictive control

Nonlinear model predictive control, which is conceptually similar to its linear counterpart except that nonlinear dynamic model is used for process prediction and optimization, is a developing control field. The proposed nonlinear model predictive control scheme, which is called nonlinear sliding mode model predictive control, integrates model predictive control and sliding mode variable structure control. A feasible dual-mode control scheme is presented in the paper, sliding mode predictive controller is implemented while the system state is outside the terminal region, besides sliding mode variable structure control designed off-line is used. The resulting control scheme has strong points of the two control methods. By predicting the pre-designed switching function, the predictive control sequence can be found by solving constrained open-loop optimal control problem, and the current control is implemented. At next sampling time the optimizing procedure is repeated. By constraining terminal sliding mode with an inequality, the state is steered into the predesigned sliding mode region, then the closed-loop stability is proved, and furthermore the system performance is analyzed. With some conditions, the method can be extended to a quasi-infinite horizon formulation. Simulation results show that less calculation is needed.