Contribution analysis method for vehicle interior noise using independent component analysis

Abstract

For reducing vehicle interior noise efficiently, it is necessary identifying sound sources with high contributions to the noise and countermeasuring them. However, measuring source signals are sometimes difficult depending on the type of sound source such as wind noise. In this case, obtaining the contribution and performing effective countermeasure become difficult. In this study, we considered a contribution separation technique using only response signals by employing independent component analysis (ICA). In order to apply ICA to separate contributions of vehicle interior noise, we added two procedures to frequency domain ICA method. The first is a procedure which could calculate the amplitude of each contribution correctly in each frequency from the obtained source signal by ICA, which amplitude has arbitrary property. Second is a procedure which could solve a permutation problem of frequency domain. Next, we simulated two observation signals by mixing acceleration engine and wind noises measured in c...