Empirical mode decomposition

Empirical mode decomposition is like sorting a bunch of different sounds by their pitch. Imagine you're listening to a bunch of different instruments playing together in a band, but you want to hear each instrument separately. You might start by listening for the highest pitched sound, like a flute. Once you hear that, you can separate it from the other sounds and listen to it all by itself. Then you might listen for the next highest pitched sound, like a violin, and separate that from the rest of the sounds. You keep doing this until you've listened for and separated all the different pitches from the original sounds.

It's the same with empirical mode decomposition (EMD), only instead of sounds, it's signals. Signals can be things like sounds, or they can be data from a sensor measuring temperature or pressure. EMD separates a complex signal into a bunch of simpler signals, each with a different frequency. It does this by finding a series of "modes," each of which represents a different frequency of the original signal. Each mode is like a separate layer of an onion, with the highest frequency being the outermost layer, and the lowest frequency being the innermost layer.

EMD works by repeatedly finding the "envelope" of the signal and subtracting it from the original signal. The envelope is like the shape that the signal follows, and it changes over time. By finding the envelope at each point in time and subtracting it from the original signal, EMD isolates the different frequencies present in the original signal. Each time EMD finds an envelope, it creates a new mode at a different frequency.

Once EMD has found all the modes, you can examine each one separately and see how it contributes to the original signal. This can be useful for things like analyzing the different frequencies of a sound, or identifying patterns in sensor data. By breaking down a complex signal into its simpler components, EMD makes it easier to work with and understand.