Engine rotational speed estimation using audio recordings and machine learning algorithms

Abstract

There are many situations in practice when the value of the rotational speed of an engine is needed to be known and a direct instrument for measuring it is not available or, if present, it is decalibrated. One can use instead a non-contact method for estimating the speed. The paper presents such a non-contact method for estimating the rotational speed of a heat engine based on audio recordings and machine learning algorithms. The method principle, experimental validation and a discussion upon the optimal parameters and factors that lead to the best performances are presented.