Fig.2 An example of a standard EMD for the 1s-MMG signal (500 sample points).

The standard EMD decomposes a signal x(𝑡) into N sets of frequency- and amplitude-modulated signal components by employing an iterative process called the sifting algorithm, such asx(𝑡)=Σ𝑐𝑖(𝑡)𝑁𝑖=1+𝑟(𝑡),𝑡=1,…,𝑛,𝑖=1,…,𝑁 (1)where the series {𝑐𝑖(𝑡)}𝑖=1𝑁 is called the intrinsic mode functions (IMFs), which represent the intrinsic oscillations present in the raw signal, and 𝑟(𝑡) is a residual trend of the lower frequency (Fig. 2). In brief, the IMF is extracted from a signal by subtracting the average of the upper and lower envelopes, where the envelopes are respectively generated by concatenating all local maxima and minima.