Fault measurement for SISO system using the chaotic excitation

摘要

This work presents an efficient Procedure for fault measurement of the single-input-single-output (SISO) system by driving the system directly with the output of a chaotic oscillator. Firstly, we propose a method to adjust the Lyapunov exponents (LEs) of the chaotic excitation by multiplying each equation with a different acceleration factor. These acceleration factors constitute an acceleration factor field (AFF). Secondly, in order to evaluate the degree of activation to the SISO system, which is excited by the chaotic signal with different acceleration factors, we partition the AFF into fully matched, partially matched and non-matched areas by using the criteria derived from Kaplan-Yorke conjecture. It is proved that chaotic excitation with acceleration factors chosen from the fully matched AFF can sufficiently excite the system, i.e. every change in the eigen-structure of the system will be reflected in the output. Thirdly, attractor based fault feature "prediction error (PE)" and evolutionary algorithm "Backtracking Search Optimization Algorithm (BSA)" is used to find the optimal acceleration factor's location in the matched area, then the fault feature is signified by using the chaotic excitation with optimized acceleration factors. Finally, two linear analog circuits with different chaotic excitations are tested to validate the power of the approach. Experiment results show that fully matched chaotic excitation can be used to detect faults in a SISO system, while, non-matched chaotic excitation may be capable of detecting some faults in the system, but there is a risk to make the wrong decision. (C) 2015 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.