Detecting charakteristics of a signal and cluster them - Self organizing map?

Hello there

I'm trying to find a way to find characteristics of a signal and cluster these according to similarities. In advance I want to use the clusters as a submission for other things.

I was thinking to take the mean values and then take into account the standard deviation. There would also be an opportunity to look at the high peaks (or Lowpeaks) and take the distances from each other.

Well and then I want to cluster these signals. I want to do that in such a way that I can use it later as an submission for additional signals. So train my file to know when the signal does a specific "movement" and cluster it in "movements".

I have read a little bit on the internet and I think that culstering can be done with a self-organizing map. But I'm not sure if this is really the right tool. And I have tried the examples that are avaliable in the toolbox( I am using the Statistics and the neuronal network toolbox) .. but I can not understand the plots. So I was wondering...

Could you tell me if I am totaly wrong with my ideas?? Or am I just not understanding the plot of the self organizing maps...

Many thanks in advance and best wishes

Florentine