Abstract
The System we developed detects knocking by sound signals, which are captured by a microphone. Feature values are calculated with the spectrum analysis from sound data near the TDC. Threshold values for detecting knocking are decided by anomaly detection techniques. Thereby realizing to detect knocking even under a circumference, where mechanical noises exist. Further, the knock index calculated from the measurement results and knocking intensity judged by experts showed the similar trend. From the above results, we conclude that the system is effective for detection and objective rating of knocking.
