P. R. China needs to develop its own method and experimental facility for assessing pressure distortion of a certain turbofan engine. Section-1 describes the features of the facility, among which we mention fluid-driven movable flashboard distortion generator and the pressure transducers that can make transient-state measurements. Section-2 explains the four steps of our experimental procedure. By changed the depth of movable flashboard and engine speed, the distortion of complicated inlet flow field and distortion index were investigated. The results show that the circumferential turbulence level is an axial symmetry distribution with center of the flashboard, the distribution curve almost likes the letter ‘M’. The turbulence level of each testing point is in direct proportion to flashboard depth, so is the radial turbulence level. And the distribution curve just likes the letter ‘V’ for single harrow. By way of analysis above, we then gained several testing point positions with high amplitude and sensitive to un-stability state from 30 testing points in air inlet duct. Section-3 discusses how to identify the inlet pressure distortion signal based on wavelet and neural network. While the surge was happening, the dynamic pressure signals firstly jumped then dived, and surge signals were not lasting more than 0.5s. In order to surveillance watch engine’s running, we first applied wavelet theory to process distortion signal’s de-noising and extracted the frequency characterization of distortion signal. Because of only processing a certain frequency bands, calculation workload is little with anti-jamming and convenient to process data in real time. Then we designed and trained a perceptron as neural network to identify the distortion signal automatically to help operators detect surge in time.
Discussion(0)
No comments yet. Be the first to comment.