I have finally completed my masters thesis on modelling valve distortion in guitar amplifiers. It's been a fascinating project that has introduced me to many interesting modelling techniques. A large portion of the project has been looking at Volterra series representations, and although these have revealed some intruiging behaviour, I have not yet managed to use them to produce acceptable audio output. I had a bit more success using neural networks, particularly using the nnsysid toolbox. You can download the main report below along with the preparation report (supplement with some more of the theory).
Spectrogram showing the response of a valve amplifier to exponentially swept sine signals.