the pre-stall behavior of a 4-stage transonic compressor and stall monitoring based on artificial neural networks的pre-stall行为四跨声速压气机和失速基于人工神经网络的监控.pdfVIP

International Journal of Rotating Machinery , 10(5): 387–399, 2004 c Copyright Taylor Francis Inc. ISSN: 1023-621X print / 1542-3034 online DOI: 10.1080/10236210490474548 The Pre-Stall Behavior of a 4-Stage Transonic Compressor and Stall Monitoring Based on Artificial Neural Networks Frank-Oliver Methling and Horst Stoff Fluid Energy Machines, Ruhr-University Bochum, Germany Frank Grauer MTU Aero Engines, Munich, Germany both types of instabilities, spike-type stall as well as modal Current research concerned with the aerodynamic insta- waves. bility of compressors aims at an extension of the operat- ing range of the compressor towards decreased massflow. In Keywords Axial compressor, Rotating Stall, Surge, Pre-stall behav- practice, a safety margin is maintained between operating ior, Fourier analysis, Artificial neural networks point and stability limit to prevent the compressor from go- ing into stall and surge. In this article, we analyze the behav- INTRODUCTION ior of a 4-stage transonic axial compressor before entering In a compressor performance map the surge line is the limit the unstable range and present an approach to identifying in- of aerodynamic stability towards low massflow. The different cipient surge and stall using artificial neural networks. This forms of those aerodynamic instabilities are rotating stall and method is based on measurements of the unsteady static wall surge. Rotating stall defines a situation where regions of re- pressure in front of the first rotor. versed flow occur locally. Surge is characterized by a system Analyzing the static pressure signals by using the Fast instability with periodic