Structure optimization of artificial neural networks using pruning methods

The presented paper deals with two pruning algorithms used for structure optimization of artificial neural networks. These networks will be used for modeling of nonlinear dynamic systems. The quality of proposed algorithms will be verified and compared on the real data of the steam turbine in thermal power plant. Totally, four different models will be created to cover the behavior of selected real system: a neural network model, an adaptive neuro-fuzzy model and two models of optimized neural networks using Optimal Brain Damage and Optimal Brain Surgeon algorithms.