An Automated Aero-Engine Thrust Detecting Method Based on Sound Recognition

摘要

In order to confirm the aero-engine thrust before a carrier aircraft is launched, an automated aero-engine thrust detecting method based on sound recognition has been presented. Using the flight simulation software, it is possible to obtain an aero-engine's sound signal which can be divided into sections and put into the frequency domain by using FFT (Fast Fourier Transform). A 3-layer BP (Back Propagation) neural network is introduced to classify the spectrum of sound signal sections. In training the network, the L-BFGS algorithm is used to optimize the network parameters to make the accuracy of the network up to 99% in terms of test data. The result proves the algorithm to be effective. In practical application, the decision logic for 3 continuous sections is used to make the misjudgment rate for the launching thrust less than one millionth. The automated aero-engine thrust detecting method based on sound recognition can be adopted for a real-time detection of the aero-engine thrust, thereby replacing the current artificial confirmation methods of engine thrust to increase the efficiency and reliability of carrier aircraft launch.