作者: M.D. Brown , G. Lightbody , G.W. Irwin
DOI: 10.1016/S1474-6670(17)43132-6
关键词:
摘要: Abstract Local Model Networks represent a nonlinear dynamical system by set of locally valid submodels across the operating range. Training such feedforward structures involves combined estimation submodel parameters and those interpolation functions. The paper describes new hybrid learning approach for local model networks that uses combination singular value decomposition second order gradient optimization. A Internal Control scheme is proposed which has important property controller can be derived analytically. Simulation studies pH neutralization process confirm excellent modelling control performance using approach.