Regresar

An Adaptive Neuro-Fuzzy PID Controller Approach for thermal Systems: An Experimental Validation

Abstract:

The current work includes an experimental evaluation using a device known as TCLAB as a benchmark. A NARX neural network is used to identify the thermal process in a broad operation region and can be considered the model of reference. The control scheme employs a neuro-fuzzy adaptive control. The controller systems work by comparing the neural netwok model to the actual process, and from this comparison a model error can be obtained; if the modeling error is different from zero, a fuzzy system adjusts the PID tuning parameters to get the best performance in the operating point. The proposed approach implementation is compared to a typical PID controller are noticeable the improving of settling time for disturbances rejection.