I wrote this code for a homework problem in an Artificial Intelligence class for Dr. Benito Fernandez in 2010. We were given non-linear spring and damper functions with which to train a neural network. Those two neural networks were then used to simulate a mass spring damper system. The screen shot above shows that the neural networks correctly predict the analytical MSD system.

To complicate matters further, the neural network tool box in Matlab was used initially, but then I wrote my own backpropagation code and used it to train the networks. It was a very rewarding experience. I uploaded it to the Matlab file exchange for the benefit of the Matlab community (embedded website below).

Also below is a very short write up of the homemade backpropagation algorithm. It is included because it contained the matlab code in a pdf.