Design of neuro-fuzzy controller for water-level tank utilizing genetic algorithms

الملخص EN

Fusion of Artificial Neural networks (ANNs) and Fuzzy Inference Systems (FIS) have attracted the growing interest of researchers in various scientific and engineering areas due to the ^wingneed$ of adaptive intelligent systems to Oveal world problems.

In this paper.

Fuzzy Logic (FL), Neural Network (NN), and Genetic Algorithms (GAs) were combined to design and tune a neuro-fuzzy controller (NFC).

This design is based on multi-layer neuro-fuzzy network.

The adaptation of neuro-fuzzy rules consequents and tuning of NFC weights is accomplished utilizing genetic algorithms.

Then ineffective rule$ are removed from the rule-base of the controller.

A real code representation is used to er^ ode the GA chromosome.

Two selection methods are used, namely.

Roulette wheel and Tournament selection methods.

The steps of building, tuning, and the removal of the ineffective rules are accomplished in an off-line phase.

In on-line phase, the resulting NFC is operated and it is noticed that the response of the system is not robust enough.

So, to avoid any shortage that my happen in the system performance, an adaptive fuzzy controller is added to the NFC.

Application of the NFC to a water-level tank is investigated in this paper.