Hi

This might be a bit too much of a difficult question for this forum but have a go anyway.

I have some time series data from a sensor. It doesn't matter what the sensor is but it records electrical data containing multiple sine waves, very noisy and the Fourier transform is very complex. I would like to try to recognize repetitive patterns in that data. I do not know what those patterns are, how long they might be, when they will occur, whether they have a specific periodicity. So really the only constraint is that they will repeat themselves to some degree (not identically) and that they will not overlap. To give a guide the time series may be 10 minutes of data recorded at 600hz and the patterns may last around 100-500ms, so plenty of room for repetition.

The question is, does anyone know of a technique or algorithm that would effectively search for repetitive patterns of any kind without any prior knowledge or clue as to what those patterns might be?

Any Ideas? If this is not the right place for this question could anyone tell me where is?

Thanks for your time it is much appreciated.

Thanks

Darren