Induction Machine Speed Neural Estimator Implemented in Programmable Architecture

Abstract

This work presents a state observer based on Artificial Neural Networks (ANN's) and a Programmable Architecture to estimate an Induction Machine (IM) speed. Indirect measurement system is the observer's physical accomplishment formed by hardware components, voltage/current meters and digital circuit, by the software and the estimation algorithm. Speed obtained by simulation based on mathematical models and by estimation algorithm implementation in PSoC programmable circuit is compared to machine's speed measured values by a tachometer.

A Robust ANN For Rotor Speed Indirect Measurement Of The Induction Machine

“…The weight w 2 depends on the machine speed and is variable. The synaptic weights (w1, w2, w3) are adjusted to minimize the energy function [14], [18], [19]. The instantaneous value of the error energy for these neurons is defined as…”

“…The essential difference between this work and the published works in [18], [19] and [22] was the rotor speed indirect measurement via ANN, included the parametric uncertainties in the induction machine model. In the works mentioned previously was not considered of the robustness of the induction machine.…”