Kristina Vassiljeva

A novel taylor series based approach for control computation in NN-ANARX structure based control of nonlinear systems

This paper presents an alternative approach for control computation in a closed loop of discrete-time nonlinear system and NN-ANARX based dynamic output feedback. Proposed technique is based on an application of Taylor series expansion for computation of control directly from neural network based model. Two modifications of the algorithm are proposed for both single-input single-output and multi-input multi-output nonlinear systems. The effectiveness of the proposed approach is demonstrated on numerical examples.

State-space control of nonlinear systems identified by ANARX and Neural Network based SANARX models

A state-space technique for control of nonlinear SISO systems identified by an Additive Nonlinear Autoregressive eXogenous (ANARX) model is presented. Two cases are shown. In the first case system model is given explicitly in the form of ANARX structure. In the second case controlled system is identified by Neural Network based Simplified Additive NARX (NN-SANARX) model linearized by dynamic feedback. The neural network based model is represented in the discrete-time state-space form. The effectiveness of the approach proposed in the paper is demonstrated on numerical examples with SISO and MIMO systems.

Identification of industrial water boiler for model predictive control of district heat plant

Widely used PID controller has number of limitations that do not allow using it effectively to solve complicated control issues. The framework of the solution is presented in the paper. A nonlinear model of a district heat plant boiler is identified by training an artificial neural network. The model is used to predict the behavior of real plant.

Application of MPC to industrial water boiler control system in district heat plant

The main goal of this paper is to identify an industrial water boiler model and design a model predictive controller (MPC). The boiler model was identified from the real process data collected during a heating season. Controller was designed and tested in virtual environment and its performance was compared with performance of classical PI control algorithm that is currently used to control a boiler system. Use of the designed controller leads to significant improvement of accumulated output error.

GA based optimization of NN-SANARX model for adaptive control of nonlinear systems

This paper discusses application of dynamic state feedback algorithm for adaptive control of nonlinear MIMO systems. Neural Network based Simplified Additive Nonlinear AutoRegressive eXogenous (NN-SANARX) structure is used for identification of nonlinear MIMO systems. For better and faster adaptation it is important to minimize the number of parameters to be tuned. Therefore, structural identification of the neural network is done by the genetic algorithm. To avoid some of the complications caused by on-line adaptation the model is divided into adaptable and nonadaptable parts.

Neural networks based minimal or reduced model representation for control of nonlinear MIMO systems

This paper raises the issue of finding reduced/ minimal state-space form for MIMO systems based on neural networks. Two cases are studied: when system is given as a “black-box” model and when order of the controlled system is known a priori. Modified structure of the standard NN-ANARX (Additive Nonlinear AutoRegressive with eXogenous inputs based on Neural Networks ) allows to eliminate all reduced interconnections between neurons and thus to get the minimal state-space representation in second case. If we deal with unknown dynamical system then we reduce model and find optimal structure of the neural network automatically using genetic algorithm. After the model was found parameters of the NN can be used to design a state controller for the control of nonlinear MIMO systems using the linearization feedback.

Structure identification of NN-ANARX model by genetic algorithm with combined cross-correlation-test based evaluation function

Application of genetic algorithm to determine structure of Neural Networks based Additive Nonlinear eXogenous (NN-ANARX) model and if possible to simplify the architecture of corresponding neural network constitutes subject of present paper. In this paper, we construct a specific fitness function, which depends on mean square error, certain cross correlation coefficients and an order of the model.

Application of Neural Networks Based SANARX Model for Identification and Control Liquid Level Tank System

This paper is devoted to application of artificial Neural Network based Simplified Additive Autoregressive exogenous model for identification and control of a liquid level tank system consisting of three water reservoirs. A specific restricted connectivity structure of the neural network is trained on input-output data set to identify a nonlinear dynamic single-input single-output model of the liquid level tank system. Parameters of the identified neural network based model can be used to design a dynamic controller for the system. The designed neural network based controller is verified on mathematical model inMATLAB/Simulink environment and applied to the real-time control of the plant. The goal of the control algorithm is to track the desired level of liquid in the upper tank. Experimental result have shown a very good performance of the proposed technique. The designed nonlinear controller is capable of tracking the desired water level for all set points with high degree of accuracy, maximally fast and without significant overshoot.

Evolutionary design of the closed loop control on the basis of NN-ANARX model using genetic algorithm

Genetic algorithm is used in the present paper to perform evolutionary design of the closed loop control for the given process. Main distinctive feature of the proposed approach is that arguments of the fitness function describe model, and therefore controller quality, both in the open and closed loops. Namely model validity with cross-correlation functions determined in the open loop and mean square error is measured for the performance in the closed loop with a controller, which equations analytically derived from from the equations of the model.

Neural network based minimal state-space representation of nonlinear MIMO systems for feedback control

A state-space technique for control of nonlinear multi-input multi-output (MIMO) systems identified by an Additive Nonlinear Autoregressive eXogenous (ANARX) model is presented. Controlled system is identified by Neural Network based Simplified Additive NARX (NN-SANARX) model linearized by dynamic feedback. The neural network based model is represented in the discrete-time state-space form. The problem of finding the minimal state-space representation is considered.