Search

In this work a novel approach to the training of recurrent neural nets is presented. The algorithm exploits the separability of each neuron into its linear and nonlinear part. Each Iteration of the learning consists of two steps: first the descent of the error functional in the space of the linear outputs of the neurons is performed (descent in the neuron space);then the weights are updated by solving a linear system with a Recursive Least Squares technique.

In this paper a novel approach to learning in Recurrent Neural Networks (RNN) is introduced and applied to the problem of digital adaptive equalization. The proposed method extends to RNN a technique applied with success to feedforward NN and is based on the descent of the error functional in the space of the linear combinations of the neurons (the neuron space); it exploits the principle of discriminative learning, based on the minimization of an error functional which is a direct measure of the classification error considered in equalization problems. Main features of the new approach are higher speed of convergence and better numerical conditioning w.r.t. gradient-based approaches, while numerical stability is assured by the use of robust Least Squares solvers. Preliminary experiments regarding the equalization of PAM signals in different transmission channels are described, which demonstrate the effectiveness of the proposed approach.

In this paper, a neural network-based method is proposed for monitoring the online voltage security of electric power systems. Using a dynamic model of the system, voltage stability is measured totally, considering a suitable stability index for the whole system, and locally, by defining appropriate voltage-margins for detecting the area of the system where the instability phenomenon arises. A three-layer feedforward neural network is trained to give, as outputs to a pre-defined set of input variables, the expected values of the above defined indices. The neural network is designed by using a fast learning strategy that allows the optimal number of hidden neurons to be easily determined. Moreover, it is shown that, in the operation mode, the system power-margin and the bus power-margins can be easily evaluated using the value of the voltage stability index given by the designed NN. The effectiveness of the proposed approach has been demonstrated on the IEEE 118-bus test system.

The multilayer perception is one of the most commonly used types of feed-forward neural networks and it is used in a large number of applications. Its strength resides in its capacity of mapping arbitrarily complex non-linear functions by a convenient number of layers of sigmoidal non-linearities (Rumelhart et al., 1986). The Backpropagation algorithm is still the most used learning algorithm; it consists in the minimization of the Mean-Squared Error (MSE) at the network output performed by means of a gradient descent on the error surface in the space of weights.