对采样数据序列进行时频分解法的改进

Improvement on Time-Frequency Decomposition Method for Sampled Data Sequences

经验模态分解法(EMD变换),能够从采样数据序列中提取不同时间尺度波动曲线,以达到时频分解的目的。文章对经验模态分解法的重大改进,主要体现在零平均线的求取上,它无需求上、下包络线,而直接把N点原始数据按时间尺度分段压缩,再用3次样条函数光滑插值到N点,并以此作零平均线,从而得出分层模态分解,改进了Harr小波分解。

Empirical mode decomposition method （ EMD transform ） can extract wave fluctuation at different time scales from sampled data sequences , therefore can achieve the purpose of time -frequency decomposition .This paper proposes a major improvement for EMD transform , which is mainly reflected in the calculation of the zero average line.The method has no demand for upper and lower envelope calculation .Instead , it piecewisely compresses the origi-nal N-point data at different time scales , and then uses cubic spline function to smoothly interpolate to N points .The result is set as zero average line , and by this way the hierarchical mode decomposition is achieved .This paper also im -proves the Harr wavelet decomposition .