Elevating Model Predictive Control Using Feedforward Artificial Neural Networks: A Review

Dynamic matrix control (DMC) has been widely applied in many industrial processes due to the simple design for multivariable process control. However, lack of the adaptability restricts its application in the highly nonlinear and complex processes. The nonlinear model predictive controller for processes is needed, but, from a computational perspective, it has quite comprehensive demands. In this paper, modified quadratic dynamic matrix control (MQDMC), which integrates DMC with the neural network model, is proposed. The predictive model control strategy linearizes the model by applying instantaneous linearization to the nonlinear neural network model at each sampling time. MQDMC has two advantages. First, less computation of linear DMC is used. Second, the nonlinear characteristics of neural networks can be incorporated into predictive control design. In the simulation studies, the performance of MQDMC matches that of nonlinear neural network model predictive control.

Model predictive control has been widely used in the industry. This can control the multivariable system with constraints on input and output variables but it needs online computation solver, and creates the non-convex solution in nonlinear plant due to the parameter uncertainties. The online computational problem and non-convex solution of the model predictive control are achieved via neural network model predictive control. The paper explores the speed control of permanent magnet synchronous motor (PMSM) by using neural network model predictive control (NNMPC) technique. The multi-layer artificial neural network is used to identify the dynamics of PMSM. The set point speed tracking control of PMSM is identified by using neural network model predictive control strategy. By using the set of input and output data obtained from the system, the multi input-output feed-forward neural network model is created. Levenberg-Marquardt algorithm is used to train the process models of the PMSM. That provides future plant output for control optimization of the predictive control. The overall system is developed and tested in the MATLAB/Simulink. To evaluate the efficiency of the controller proposed, it is compared with a constrained model predictive controller through the studies of simulation. The overshoot and settling time of the speed response of the PMSM are measured and analyzed for NNMPC and constrained MPC.

An investigation into neural network model predictive control is described in this paper. The control strategy developed is applied to a laboratory process to control temperature, pH and dissolved oxygen. The main difficulties in control of this process are nonlinearity, coupling effects among variables and long time-delay in the heat exchanger. Parallel neural models are developed from real process data for the use with online model predictive control and off-line simulations. The online control results are demonstrated.

A novel method of neural network model predictive control integrated process monitoring and applications to hot rolling process

Model Predictive Control is a well-suited control strategy; however, it needs to solve two main problems in order to be applied, first, it is necessary to have a suitable model of the plant to be controlled, in order to allow an adequate prediction, second, it is necessary to solve an optimization problem. These two problems not always can be solved particularly for nonlinear complex systems. Therefore, in this chapter we propose two variations for Model Predictive Control, in a first stage a recurrent high order neural network is proposed to obtain a fitting model for the plant to be controlled, and, at the same time this neural model identifies the system on-line through available measurements. Then, in a second stage, the optimization problem is solved using a particle swarm optimization algorithm. Using these two modifications, it is proposed a Neural Evolutionary Predictive Control for discrete-time nonlinear systems under disturbances under disturbances, and its effectiveness is shown in the experimental results by using data obtained from a linear induction motor prototype.

Model predictive control (MPC) is a method to formulate the optimal scheduling problem for grid flexibilities in a mathematical manner. The resulting time-constrained optimization problem can be re-solved in each optimization time step using classical optimization methods such as Second Order Cone Programming (SOCP) or Interior Point Methods (IPOPT). When applying MPC in a rolling horizon scheme, the impact of uncertainty in forecasts on the optimal schedule is reduced. While MPC methods promise accurate results for time-constrained grid optimization they are inherently limited by the calculation time needed for large and complex power system models. Learning the optimal control behaviour using function approximation offers the possibility to determine near-optimal control actions with short calculation time. A Neural Predictive Control (NPC) scheme is proposed to learn optimal control policies for linear and non-linear power systems through imitation. It is demonstrated that this procedure can find near-optimal solutions, while reducing the calculation time by an order of magnitude. The learned controllers are validated using a benchmark smart grid.

This study introduces an innovative approach to enhance the energy efficiency and position control performance of electro-hydraulic systems, employing a comprehensive comparative analysis. It presents and evaluates three control techniques: Proportional-Integral-Derivative (PID) control, Model Predictive Control (MPC), and Neural Network Model Predictive Control (NN-MPC). These methods are systematically assessed across varying load conditions. Notably, our research unequivocally establishes the exceptional performance of the NN-MPC approach, even when confronted with load variations. Furthermore, the study conducts an exhaustive examination of energy consumption by comparing a conventional system, where a flow control valve is not utilized as a hydraulic cylinder bypass, with a proposed system that employs a fully open Flow Control Valve (FCV). The results underscore the remarkable energy savings achieved, reaching up to 9% at high load levels.

In this paper, A neural network estimation-based nonlinear predictive model control(NN-NMPC) method is proposed to improve the station-keeping performance for a multi-vectored propeller airship. First, the kinematic model and dynamic model of the airship is introduced, and the station-keeping control problem is formulated. Then, a nonlinear model predictive control strategy, combined with radial basis function neural network(RBFNN) approximation is specified to the station-keeping problem, and the stability analysis is implemented. Finally, the simulation results are illustrated to prove the robustness and effectiveness of the proposed NN-NMPC controller. Simulation results show that NN-NMPC method can attenuate the unknown parameters of the airship effectively and drive it to the desired fixed point.

A multi-layer feed forward neural network model for accurate prediction of flue gas sulfuric acid dew points in process industries

Neural predictive controller for closed-loop control of glucose using the subcutaneous route: A simulation study

Multi-vector model predictive control of permanent magnet synchronous motor can effectively overcome the shortcomings of conventional single-vector model predictive control such as large motor current ripples and limited control accuracy, but its control performance is more susceptible to the influence of motor parameter perturbation. To improve the robustness of the multi-vector predictive control algorithm under the motor parameter perturbation, a novel nonparametric three-vector model predictive current control method for permanent magnet synchronous motors is proposed in this paper. Through constructing an ultra-local predictive current model based on the input and output signals of the permanent magnet synchronous motor, the influence of motor parameter perturbation on current prediction is avoided, and a disturbance observer is designed to estimate the non-modeled and perturbed parts of the ultra-local model. In addition, the direct calculation method of vector duty cycles based on current error is introduced to suppress the influence of motor parameter uncertainty on the duty cycle calculation link, which further improves the system robustness. Finally, simulations and experiments of the proposed method are demonstrated to compare with the conventional parametric three-vector model predictive current control, and the results show that the proposed control strategy can effectively suppress the disturbances caused by the motor parameters and ensure the steady-state performance during motor operation.

With Predictive Current Controllers, system behavior or current reference is predicted. In this way, it is aimed to prevent hardware and software related delays. In this study, new Artificial Neural Network (ANN) based Predictive Current Controllers are designed using four different methods for voltage source inverters. The training of the networks is done offline using the data obtained from the simulation results for different parameters in the Matlab environment. In the first proposed method, a static ANN based current controller is designed and trained using the data obtained from the Finite Control Set Model Predictive Control (FCS-MPC) method. Then, a feed-forward reference current predictor ANN (PRefNN) is designed to make sinusoidal reference current prediction in the other three methods. The other proposed predictive ANN methods are trained by taking the data offline from the inverter system in which PRefNN and the classical current controllers (Hysteresis, PR, and PI) are used. In this way, three different predictive current controllers named as Hysteresis based predictive ANN (Hist-PNN), PR based predictive ANN (PR-PNN), and PI based predictive ANN (PI-PNN) are designed. Classical current control methods have been given predictive properties with these three different network structures. And also, it is improved the performance of classical methods against parameter changes and noises. A three phase 5kVA inverter circuit with a 7MBP50RJ120 IPM module in the power stage and STM32f407 as a controller is designed for the experimental setup. The methods are tested in simulation and validated in the experimental setup.

The design and investigation of an intelligent controller for hardware-in-the-loop (HIL) implementation of hybrid electric vehicles (HEVs) are proposed in this article. The proposed intelligent controller is adopted based on the enhancement of a model predictive controller (MPC) by an artificial neural network (ANN) approach. The MPC-based ANN (NNMPC) is proposed to control the speed of HEVs for a simulation system model and experimental HIL test systems. The HIL is established to assess the performance of the NNMPC to control the velocity of HEVs in an experimental environment. The real-time environment of HIL is implemented through a low-cost approach such as the integration of an Arduino Mega 2560 and a host Lenovo PC with a Core i7 @ 3.4 GHz processor. The NNMPC is compared with a proportional–integral (PI) controller, a classical MPC, and two different settings of the ANN methodology to verify the efficiency of the proposed intelligent NNMPC. The obtained results show a distinct behavior of the proposed NNMPC to control the speed of HEVs with good performance based on the distinct transient response, minimum error steady state, and system robustness against parameter perturbation.

This paper proposes an improved constrained networked model predictive tracking control design for the chamber pressure of a coke furnace under uncertainty and packet losses. Unlike conventional constrained model predictive control (MPC) strategies that have a limitation in the consideration of both set-point tracking and the dynamic process responses, the system state variables and output tracking errors are combined and thus can be regulated simultaneously in the new MPC scheme. Based on such advantages, there are more degrees of freedom for the subsequent controller design and improved system performance can then be obtained. Case studies on the regulation of chamber pressure of a coke furnace under process uncertainties and packet losses are investigated to verify the proposed approach in comparison with typical traditional constrained MPC schemes.

The bandsawing as a multi-point cutting operation is the preferred method for cutting off raw materials in industry. Although cutting off with bandsaw is very old process, research efforts are very limited compared to the other cutting process. Appropriate online tool condition monitoring system is essential for sophisticated and automated machine tools to achieve better tool management. Tool wear monitoring models using artificial neural network are developed to predict the tool wear during cutting off the raw materials (American Iron and Steel Institute 1020, 1040 and 4140) by bandsaw. Based on a continuous data acquisition of cutting force signals, it is possible to estimate or to classify certain wear parameters by means of neural networks thanks to reasonably quick data-processing capability. The multi-layered feed forward artificial neural network (ANN) system of a 6 × 9 × 1 structure based on cutting forces was trained using error back-propagation training algorithm to estimate tool wear in bandsawing. The data used for the training and checking of the network were derived from the experiments according to the principles of Taguchi design of experiments planned as L27. The factors considered as input in the experiment were the feed rate, the cutting speed, the engagement length and material hardness. 3D surface plots are generated using ANN model to study the interaction effects of cutting conditions on sawblade. The analysis shows that cutting length, hardness and cutting speed have significant effect on tooth wear, respectively, while feed rate has less effect. In this study, the details of experimentation and ANN application to predict tooth wear have been presented. The system shows that there is close match between the flank wear estimated and measured directly.