Knock detection based on SOM

A new method based on self organizing maps (SOM) is proposed for knock detection. The method differs from the other existing methods in that the wavelet coefficients are computed and further, some statistical parameters. The computed parameters are then applied to a SOM in order to classify the engine cycles as affected by knock or not affected cycles. Once a cycle has been detected as a knock containing one, for the next cycle the engine can be protected in order to avoid further appearances of knock.