The invention provides a diesel engine combustion resonant frequency extraction method based on empirical wavelet and Hilbert transform. The method comprises the following steps that: 1: carrying outFourier transform on a cylinder pressure signal to obtain a cylinder pressure frequency spectrum, utilizing a scale space algorithm in an empirical wavelet algorithm to identify characteristic sub-modality in the cylinder pressure frequency spectrum, determining the position of the sub-modality, and calculating the boundary frequency of the sub-modality; 2: constructing an empirical wavelet filter, utilizing the constructed wavelet filter to carry out band-pass filtering on the cylinder pressure signal, extracting a sub-modality signal corresponding to each piece of combustion fluctuation in the cylinder pressure frequency spectrum, and selecting the sub-modality with maximum oscillation energy as a combustion oscillation signal; and 3: carrying out the Hilbert transform on the combustionoscillation signal obtained in S2, drawing a Hilbert spectrum, and obtaining oscillation frequency and an oscillation amplitude corresponding to the oscillation frequency. The method has the advantages of good astringency, high adaptivity and high accuracy, and the obtained oscillation frequency can be used for the low-noise optimal design of the diesel engine and temperature and gas quality estimation in a cylinder.