EMD Theory And Its Application In Crack Damage Detection

As the theoretical basis of traditional signal analysis, Fourier transform replaces a signal with a series of sinusoidal signals, which has no beginning or end. When used to nonlinear or non-stationary signal analysis, Fourier transform may cause false mode. Based on the characteristic scale, Empirical Mode Decomposition(EMD) can decompose a signal into a series of Intrinsic Mode Functions(IMFs). EMD breaks through the limitations of basis function, which exists extensively in the traditional signal analysis. For owningmuch more advantages in nonlinear and non-stationary signal analysis, EMD has been widely applied in different fields. However, as lacking a theoretical foundation, EMD still has some weaknesses. This paper studies these weaknesses and applied the research results to detection of crack damage.Firstly, this paper describes the development history, basic algorithm theory and characteristics of EMD. Some common weaknesses, include sifting stop criteria, sampling frequency, extremum interpolation, mode mixing, has been preliminarily discussed. The advantages and disadvantages of different solutions for these weaknesses has been compared with examples, and solutions for these weaknesses has been provided.Secondly, this paper presents a detailed review on end effects, and analyses the mechanism, suppression principle, and impact of end effects. Through improving the existing waveform extrapolation, this paper advances an improved method for end effects, which is based on intrinsic waveform matching. At the same time, a comprehensive index for end effects has also been introduced by integrating the existing evaluation index. The method proposed by this paper and other existing methods for end effects has been validated and evaluated by the comprehensive index with simulation and project examples. The results shows that the proposed method can effectively suppress the end effects, and the precision of decomposition is much higher than the existing methods.Thirdly, this paper details the theory of crack damage detection, and validates this theory by simulation. Based on the vibration experiment platform, this paper designs 4 beams with different crack depths, and tests the modal frequencies and modal displacement of each beam. Then the method proposed by this paper and modal analysis are compared in the crack detection. The result shows that EMD has more advantages in the weak damage detection, and EMD can detect the position of crack damage and measure the extent of crack damage more accurately.