Online aircraft damage case identification and classification for database information retrieval

Abstract

This paper reports the latest development in database-driven safe flight envelope prediction systems. By using a damage assessment system based on identification and pattern classification methods, structural damage to in-flight aircraft can be detected and estimated online. This paper focuses on aircraft structural integrity assessment after sudden damage based on online aerodynamic model identification. Considering the fact that the modeled damage cases may not cover all the conditions that may happen in real flight, a classifier that can identify points in between the training classes is needed. In this paper, two nonliner classification methods, support vector machines and neural networks are evaluated and compared in damage severity estimation. It is concluded that support vector machines outperform neural networks in covering more data points in between the training classes with a broader generalization region. In the end, the proposed damage assessment system is used to detect and estimate damage severity in a simulated multi-damage scenario.