To evaluate the working state of a diesel valve by vibration response signal analysis and carry out fault diagnosis, a method is proposed to resolve the problem of how to effectively extract the vibrational characteristic parameters of the valve seat from a nonstationary vibration signal on the surface of a diesel engine by using local wave decomposition and reconstruction technology, continuous wavelet transform (CWT) time spectrum images, and image processing technology. The difference in the energy distribution characteristics of the vibration signal time and frequency domains of the diesel engine is enhanced by the local wave decomposition-reconstruction analysis method, the time spectrum map of the reconstructed signal is drawn by using the CWT, the vibration characteristics and corresponding features are extracted by image segmentation technology, and the intrinsic relationship between the gas valve gap and the gas valve landing shock vibration characteristics is discussed. Gas valve clearance status identification and the diagnostic basis function are established, and the quantitative monitoring of the air valve clearance of the diesel engine and the fault diagnosis of abnormal gas valve clearance are realized.