In this paper we present a neural network control approach
of a temperature system. The controller uses a serie of
optimization training methods for improve the speed of its
convergence. The training is made totally on-line and the
neural network architecture is the Multi-layer Perceptron.
We present the details of the temperature system and the
control method, as well the optimization methods which
permits the on-line training of the neural network.
The showed results prove the capacity of the neural
controllers in problems which involve difficulties like
non-linearities and changes on its structure.
