Adaptive artificial neural network for robust estimation of parameters of asynchronous state

In many applications, very fast methods are required for estimating of amplitudes of current signals distorted by noise. In this paper new parallel algorithms are proposed, which can be implemented by analogue adaptive circuits employing selected neural networks principles. Algorithms based on the total least-squares (TLS) and the robust total least-squares (RTLS) criteria are developed and used to detection of asynchronous state of synchronous machines. The problems are formulated as optimization tasks and solved using the steepest descent continuous-time optimization algorithm. The corresponding architectures of analogue neuron-like adaptive processors are also shown. The developed networks are more robust against noise in the measured signal than other known neural network algorithms. The network based on the TLS criterion realizes the optimization process under the assumption that the signal model can also be deteriorated (frequency or sampling interval fluctuation and so forth). The TLS estimation effect is better and more reliable than the corresponding LS structure, when higher sampling frequency and a wider sampling window is applied. An asynchronous state of a synchronous machine may be identified through determining the amplitudes of particular components of stator´s current provided that a constant slip value is assumed. Following a synchronism loss, this adopted value is assumed to be achieved and, for sure, exceeded. This work provides a description of artificial neural networks realising mentioned task of asynchronous state detection. Extensive computer simulations confirm the validity and performance of the proposed algorithms.