Nonlinear model identification and adaptive model predictive control using neural networks (original) (raw)
Related papers
Using Neural Networks for Identification and Control of Systems
2015
The present work addresses the utilization of Artificial Neural Networks (NN) for the identification and control of systems, in special to control nonlinear dynamic systems or systems with some degree of uncertainty. Because NNs have an inherent ability to approximate functions and to adapt to changes in input and parameters, they can be used to control systems too complex for linear controllers, such as PID controllers. In the present work a mathematical basis for NN is presented, the mathematical representation of a process unit, or neuron, and how they can be put together in order to form nets that can learn from external data. In sequence, it is presented structures of inputs that can be used along with NN to model nonlinear systems. The most common configurations of input vectors for the training of NN are highlighted. Following, a method of control is presented that take advantage of NN, where a NN is used to build a predictive nonlinear controller using a model predictive con...
Application of the RTNN Model for System Identification, Prediction and Control
IFAC Proceedings Volumes, 2001
A parametric Recurrent Neural Network (RNN) model and an improved dynamic Backpropagation (BP) method of its learning are applied for real-time identification and state estimation of nonlinear object. The obtained parameter and state information is used for an adaptive control system design. The paper suggests performing a trajectory tracking state-space fuzzy-rule-based control. The applicability of the proposed adaptive control scheme is confirmed by simulation example of non linear system. Copy right © 200J IFAC
Application of neural networks to adaptive control of nonlinear systems
Automatica, 2000
This book presents a detailed study of supervised learning in feedforward neural network models, when applied to diverse con"gurations for on-line control of unknown nonlinear plants. In very general terms, the content of this monograph gives new insights into the problem of mathematical modelling of unknown and generally nonlinear plants, in order to design adaptive controllers.
Identification and Adaptive Control of Systems Using Recurrent Neural Networks
IFAC Proceedings Volumes, 1997
This work involves the adaptive control of nonlinear systems using a new model of radial base function (RBF) recurrent neural networks (Ciocoiu. 1996). The proposed architecture is made up of two blocks, each with two neural networks: one to identify the physical system (plant)-identifier, and another for control-controller. The identifier, running in parallel with the plant, is designed to obtain the system's Jacobian, which is used to adjust the weights of the controller. Simulations carried out on various types of plant prove that the method works well
Adaptive Control Using Neural Networks and Approximate Models for Nonlinear Dynamic Systems
Adaptive Control Using Neural Networks and Approximate Models for Nonlinear Dynamic Systems, 2020
In this research, a comparative study of two recurrent neural networks, nonlinear autoregressive with exogenous input (NARX) neural network and nonlinear autoregressive moving average (NARMA-L2), and a feedforward neural network (FFNN) is performed for their ability to provide adaptive control of nonlinear systems. Three dynamical nonlinear systems of different complexity are considered. The aim of this work is to make the output of the plant follow the desired reference trajectory. The problem becomes more challenging when the dynamics of the plants are assumed to be unknown, and to tackle this problem, a multilayer neural network-based approximate model is set up which will work in parallel to the plant and the control scheme. The network parameters are updated using the dynamic backpropagation (BP) algorithm.
Neural Network Structures for On-Line Identification and Optimization of Nonlinear Processes
This paper analyzes various formulations for the recursive training of neural networks that can be used for identifying and optimizing nonlinear processes on line. The study considers feedforward type networks (FFNN) adapted by three different methods. The study is completed using two network structures that are linear in the parameters: a radial basis network (RBF) and a principal components (PCA) network, both trained using a recursive least squares algorithm. The corresponding algorithms and a comparative test consisting of the on-line estimation of a reaction rate are detailed. The results indicate that all the structures were capable of converging satisfactorily in a few iteration cycles, FFNN type networks showing better prediction capacity than linear parameter networks, but the computational effort of the recursive algorithms is greater.
Recurrent networks for nonlinear adaptive control
IEE Proceedings - Control Theory and Applications, 1998
Design of control techniques for nonlinear systems, where state measurement is not available, still poses a major challenge. Recent successful applications of static neural networks for control suggest that certain intrinsic properties of neural networks could also be utilised for output feedback control, where the neural network serves as a dynamic model of the system. Some steps have been taken in this direction, most of them of a heuristic nature. An adaptive control technique for nonlinear plants with an unmeasurable state is presented based on a recurrent neural network employed as a dynamical model of the plant. Using this dynamical model a feedback linearising control is computed and applied to the plant, while parameters of the model are updated online to allow for partially unknown and time-varying plant. Stability of the algorithm is proved for the case of constant reference output, and some further insights into convergence issues for the general case of tracking problem are provided. Performance of the proposed control method is illustrated in simulations.
Performance of the new neural network based control structure and learning algorithm
1998 Second International Conference. Knowledge-Based Intelligent Electronic Systems. Proceedings KES'98 (Cat. No.98EX111), 1998
This paper presents a new neural netwodc (NN) based Adaptive Backthrough Control (ABC) & e m for both linear and nodinear dynamic plants. A feedforwanl approach presented here falls into the direct design category. In its simplest form the implementation requires an -on of the process parameters at any sample t . UnWre the other feedforward NN based control schemes the ABC hem comprises of one neural netwodr only which ~t a n m l y acis as bothplant model (emulator) and the contmller (iivers of the emulator). For linear plants, without noise, the resulting feedforwd controller, providing that the order of the plant and plant model are equal, is a perfect adaptive poleszeros canceller. In the case d nonlinear dynamic system, and for the monotonic nonlinearity, the Hoposed ABC control represents the nonlinear predictive controller. The ABC scheme is based on the discrete nonlinear (NARMAX) dynamic model. For such models and for monotonic nonlinearity, the calculation of the desired control si@ js the result of the nonlinear optimization procedure with guaranteed convex searcb function and consequently with an unique solution.
Proceedings of the 2010 American Control Conference, 2010
In many adaptive control applications an accurate model identification process has to be performed in almost every timing instant in which new plant data are monitored. Such an accurate identification process can be based on well trained recurrent neural networks. In this work a new adaptive recurrent neural network training algorithm (ARNNTA) based on supervised learning with a new trust region strategy is developed. The ARNNTA is applied to two highly multivariable nonlinear systems that is, a wastewater treatment plant and the F-16 fighter aircraft. Comparison of model validation results with the back propagation and recursive incremental back-propagation algorithms show the superiority of the ARNNTA.