Increasing the artificial neural network based model building speed for supporting efficient technical diagnostics

Zsolt János Viharos, Krisztián Balázs Kis

Abstract

The paper introduces an algorithm for increasing the speed for building up the general system model on Artificial Neural Network (ANN) basis. It finds that input/output configuration of the field analysed which realizes the most accurate estimation and explores the maximal number of dependencies among the related system parameters. The availability and the estimation capabilities of the needed system models are especially important for technical diagnostics in order to be able in differentiating between conform and nonconform situations. The performance of the novel solution is tested and evaluated under a field specific analysis applying the classical equations from the cutting theory. Experiments were done also for cutting diagnosis based on real measured parameters under varying conditions. These validations showed good empirical performance and practical applicability of the algorithm introduced. As result, the proposed algorithm increased significantly the speed of the model building stage.

Download

IMEKO-TC10-2013-012.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings