Cockpit Alarm Detection and Identification Algorithm for Helicopters

In recent years, the National Transportation Safety Board (NTSB) has emphasized the importance of analyzing flight data such as cockpit voice recordings as an effective method to improve the safety of helicopter operations. Cockpit voice recordings contain the sounds of engines, crew conversations, alarms, switch activations, and others within a cockpit. Thus, analyzing cockpit voice recordings can contribute to identifying the causes of an accident or incident. Among various types of the sounds in cockpit voice recordings, this paper focuses on cockpit alarm sounds as an object of analysis. Identifying the cockpit alarm sound which is activated when a helicopter enters an atypical state of flying could help identify the state and timing of the incident. Nonetheless, alarm sound analysis presents challenges due to the corruption of the alarm sounds by various noises from the engine and wind. In order to assist in resolving such a problem, this paper proposes an alarm sound analysis algorithm as a way to identify types of alarm sounds and detect the occurrence times of an abnormal flight. For this purpose, the algorithm finds the highest correlation with the Short Time Fourier Transform (STFT) and the Cumulative Sum Control Chart (CUSUM) using a database of the characteristic features of the alarm sounds. The proposed algorithm is successfully applied to a set of simulated audio data which was generated by the X-plane flight simulator in order to demonstrate its desired performance and utility in enhancing helicopter safety.