Research On Loose Particle Detection System For Aerospace Engines

The aerospace engine is the power plant of launch vehicles and tactical missiles.The reliability of these aerospace engines is vital to the success and safety ofaerospace systems. With the rapid development of aerospace industry, the existingproduction and assemblage techniques can’t guarantee the higher reliability ofaerospace engines. As a result, aerospace accidents caused by loose particle occurfrequently. Therefore, it is critical to investigate loose particle detectiontechnology to ensure that the particle be free. Based on conventional PIND (particleimpact noise detection) method, this dissertation carries out an extensive study onloose particle detection system for aerospace engines.Considering aerospace engines are heavier, larger and more complex internally,traditional detection methods such as visual or auditory detection are low inefficiency and accuracy. Based on PIND theory, overall design of detection systemis proposed. Swivel stand driven by motor and servo control system are designed toprovide mechanical conditions for detection. Amplifier and acquisition circuits withsix channels are designed to amplify and transmit weak signal to the computerreal-timely and rapidly. Detection program is realized with VC++with the functionof automatic detection and identification.Considering that PIND signal is a complex, interference pulses such as inherentmechanical noise and electromagnetic interference result in high rate of false andmissive judgment. Pulse extraction algorithm based on energy threshold is proposed,and characteristics are extracted in time and frequency domain. According to thedifferences of loose particle signal and interference pulses, the existence of looseparticle is identified. On condition that loose particle exists, energy and amplitude ofmulti-channel data are extracted, and the position of particle is located.Due to particle signal with characteristics such as nonlinear and chaotic, dataprocessing method based on time and frequency domain has the disadvantage oflosing high-dimensional information, loose particle sizing algorithm with chaotictheory is investigated. In order to capture the property of particle size adequately,chaotic characteristics such as correlation dimension, Lyapunov exponent andKolmogrov entropy are calculated. Based on these features and clustering algorithm,the size of loose particle is classified to four particle levels.The validity of identification algorithm is verified by means of experimentanalysis. The accuracy of particle existence identification is over90%, and thepresence of particle location and size identification is over80%, achieving the technical specifications of the system.The proposed loose particle detection methods can provide guideline for thecleaning and screening of aerospace engines, which can be extended to looseparticle detection for cabin of spacecrafts and missiles.