Abstract
Using sound features to complete the monitoring of automobile engine condition is not only less in operation process and cost, but also less affected by weather and other factors. However, the collected signals will inevitably be polluted by other sounds and become noisy signals, which will affect the output of the research results. Therefore, the purpose of this paper is to extract and identify all kinds of signals in the sound signal, so as to provide effective data for a signal to be used in the acoustic monitoring of engine state. In this paper, VMD algorithm is used to decompose two kinds of recorded mixed sound signals. The number of decomposition layers of the two signals is determined by calculating the mean of center frequency and instantaneous frequency respectively, and multiple intrinsic modal components are obtained. The type of each component is determined by calculating the correlation coefficient method. Then, MPE value was calculated by selecting the appropriate embedded dimension and scale factor by combining MPE. Observe the result of feature extraction. Finally, each component is classified and recognized by SVM, and the experimental results show that the method mentioned in this paper is accurate and effective for separating and recognizing mixed sound event signals.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have