Loading...
3 results
Search Results
Now showing 1 - 3 of 3
- Automatic tuning of PID controllers using a neuro-genetic systemPublication . Ruano, Antonio; Lima, João; Azevedo, Ana Beatriz da Piedade de; Duarte, N. M.; Fleming, P. J.Neural networks and genetic algorithms have been in the past successfully applied, separately, to controller turning problems. In this paper we propose to combine its joint use, by exploiting the nonlinear mapping capabilites of neural networks to model objective functions, and to use them to supply their values to a genetic algorithm which performs on-line minimization.
- Comparison of off-line and on-line performance of alternative neural network modelsPublication . Lima, João; Ruano, AntonioThe Proportional Integral and Derivative (PID) controller is often used in industrial applications due to its functional simplicity and robust performance. Autotuning methods for these simple controllers are economically important. In order to accomplish this auto tuning in real time, without perturbing the closed-loop operation, models of criteria that are intended to be optimised are needed. In this paper, the ITAE criterion will be employed, as responses obtained with this criterion are well damped. In this paper neural networks are proposed as tools that allow these kinds of mappings. To improve the autotuner performance in a continuous operation, these models should be updated online. This way, the corresponding neural networks, after being trained off-line should be adapted on-line in real time. In the present work, the off-line and on-line performances of Multi-layer Perceptrons (MLPs), Radial Basis Function (RBFs) and Basis-Spline neural networks (B-splines), are analysed and compared.
- A novel technique for controller tuningPublication . Ruano, Antonio; Lima, João; Azevedo, Ana Beatriz da Piedade de; Duarte, N. M.; Fleming, P. J.Neural networks and genetic algorithms have been in the past successfully applied, separately, to controller tuning problems. In this paper we purpose to combine its joint use, by exploiting the nonlinear mapping capabilities of neural networks to model objective functions, and use them to supply their values to a genetic algorithm which performs on-line minimization. Simulation results show that this is a valid approach, offering desired properties for on-line use such as a dramatic reduction in computation time and avoiding the need of perturbing the closed-loop operation.