In the course to improve the safety of helicopter operations, the National Transportation Safety Board (NTSB) has stressed the benefits of flight data analysis including cockpit audio recordings. Cockpit audio recordings contain sounds of the crew conversations, alarms, engine noise, etc.; analysis of which may assist in the investigations of incidents and accidents. This paper focuses on the analysis of the engine sounds that may be logged by any recording within the cockpit. The analysis of engine sounds can serve as a method to evaluate the state of the engine from cockpit audio. However, this is a challenging task as cockpit audio does not providea clean engine audio and the noise profiles of different engines vary greatly. This paper proposes an engine noise analysis algorithm to estimate engine related flight parameters and identify anomalous engine behavior. This is done using signal analysis, data mining, and statistical modeling techniques to identify specific frequencies associated to engine operations. The proposed algorithm is successfully applied to real cockpit audio data to demonstrate and validate its performance.
Vaidya, A., Shin, S. and Hwang, I., “Helicopter Engine State Estimation via Cockpit Audio Analysis,” AIAA SciTech Forum, Jan 2017, https://doi.org/10.2514/6.2017-0792