Training algorithms and learning abilities of three different types of artificial neural networks
Systems Analysis Modelling Simulation
Neural network constitutive model for rate-dependent materials
Computers and Structures
Hi-index | 0.00 |
A novel Bayesian-Gaussian neural network (BGNN) is proposed in this paper for the nonlinear modeling of hydraulic turbine which is difficult to obtain its mathematical model because of its complex and nonlinear characteristics. The topology and connection weights of BGNN can be set immediately when the training samples are available. The threshold matrix parameters of BGNN are updating based an improved E.Coli foraging optimization algorithm (IEFOA) which is an evolutionary optimization algorithm imitating the behaviors of E.Coli bacteria. Simulation results for the nonlinear model of hydraulic turbine generating unit are provided and demonstrate the effectiveness and shorter training time and more effective self-tuning compared with the BP neural network for the identification of hydraulic turbine generating unit.