Decision fusion methodologies in Structural Health Monitoring systems

Abstract

Structural Health Monitoring (SHM) is a process of continuous monitoring of the physical condition of a structure for purpose of ensuring the integrity of the structure. SHM techniques have been employed to reduce maintenance and repair costs while maintaining safety and reliability of aircrafts. In this paper we have investigated the benefits provided by integrating decision fusion algorithms to SHM systems. The decisions made by classifiers acting on sensory data are combined using decision fusion algorithms to arrive at unified final decisions regarding the status of the monitored structure. First, synthetic decisions were generated and used for testing and performance evaluation of the different decision fusion algorithms. Second, several different decision fusion algorithms were developed and tested on the synthetic decisions. The Dempster-Shafer theory of evidence, fuzzy logic type-1, and fuzzy logic-type2 were used for development of the decision-fusion algorithms. Finally, the fusion algorithms were tested on decisions extracted from experimental data to validate their performances. The testing and evaluation results showed significant improvement due to fusion process integrated at the end of the feature classification process. The development of the fusion algorithms, their testing results on the synthetic decisions and decisions extracted from real experiment are reported in this paper. Also, performance analysis of decision fusion algorithms is provided in the paper.