APPLICATION OF RANDOM FOREST TO THE FAULT DETECTION IN ANALOG CIRCUITS

Piotr Bilski

Abstract

The paper presents the application of the Random Forest (RF) to the diagnostics of analog circuits. The modified structure of the forest proposed for the task and its generation algorithm is presented. The approach is used to detect parametric faults in the fifth order lowpass filter. Multiple versions of the forest are tested. The proposed method is also confronted against the decision tree, being superior over the traditional classifier. The future prospects of the approach are considered.

Keywords

decision tree, random forest, fault detection, analog circuits

Download

IMEKO-WC-2015-TC10-231.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings