Showing papers by "Ivan Petrović published in 2000"

Decoupling multivariable GPC with reference observation

Abstract: This paper deals with an algorithm for decoupling multivariable systems based on generalized predictive control (GPC). The proposed control algorithm uses adaptive error weighting in the cost function and reference change observation. From the filtered reference it auto-tunes weight coefficients in order to reduce the coupling between control loops. The algorithm is applied to greenhouse climate control; the obtained simulation results show its efficiency.

Model structure selection for nonlinear system identification using feedforward neural networks

Abstract: The majority of nonlinear models based on neural networks are of the black-box structure. A nonlinear system can be nonlinear in many different ways, thus the nonlinear black-box model structure must be very flexible. This means that it must have many parameters. A model offering many parameters usually creates problems, and the variance contribution to the error might be high. For a particular identification problem, only a subset of the parameters may be necessary, and the main topic in nonlinear system identification is how to select a model structure that describes the system dynamics with the minimum number of parameters. This paper discusses nonlinear input-output models that are suitable for implementation of feedforward neural networks. The proposed model structures were tested and compared using the identification procedure of a pH process. The results indicated that a simplest model structure can satisfactorily represent the investigated process.

Regularization and validation of neural network models of nonlinear systems

Abstract: A characteristic feature of the neural network models is the large number of parameters. A model offering many parameters usually gives rise to problems, and the variance contribution to the modeling error might be very high. Therefore, it is crucial to find the model with the optimal number of parameters. In this paper two techniques of selection of the optimal number of model parameters are described and compared: explicit and implicit regularization techniques. Model validation forms the final stage of an identification procedure with the aim of assessing objectively whether the identified model agrees sufficiently well with the observed data. In this paper the reliability of the correlation-based validation tests and the χ2-test is analyzed.

Modification and application of autotuning PID controller

Abstract: This contribution presents a modified autotuning algorithm of the PID controller. The motivation for the modification of the basic autotuning algorithm is to enlarge the class of processes to which it can be applied. The basic autotuning algorithm introduced by Astrom and Hagglund is extended by the preliminary identification procedure and through the usage of the dead time compensating controller. These modifications are detailed through the description of the algorithms' functioning. The proposed algorithm has been implemented in the programmable logic controller (PLC) Siemens SIMATIC S7-300. The experimental results confirm the good robustness properties of the proposed algorithm, which were demonstrated in a simulation study.

Application of a neural predictive controller in boost converter input current control

Abstract: The application of a nonlinear predictive controller in the cascade control structure of a boost converter is investigated. The neuro-predictive controller is realized as a nonlinear optimizer, using the Levenberg-Marquardt unconstrained optimization procedure. For the prediction of future process responses, two MLP neural networks are used. One network is used for modeling the converter dynamics and the other for online estimation of the converter's input voltage. This structure ensures good system performance in all operating regions and inherent compensation of ripples in the converter's input current caused by variations of the input voltage. The advantages of the proposed control structure are demonstrated through experimental comparison with a linear GPC with manually adjusted feedforward compensator.