Recurrent Neural Network based Internal Model Controller Design for pH Neutralization Process

DOI: 10.5176/2251-2446_CAR20

Authors: K.Srinivasan, T.K.Radhakrishnan and S.Devi

Abstract: In this paper, authors have developed a neural network (NN) based internal model controller (IMC) design for a nonlinear pH neutralization process. In this NN based IMC design, the forward neuro model is developed with either a global neuro model or multiple local neuro models and the inverse model is developed based on the inverse of model gain at each operating point cascading with a filter transfer function. The advantage of this design is the simpler inverse model building for a highly nonlinear process. The dynamic performance of the designed controller is assured by the proper selection of filter tuning parameter. The performance of multi NN model based internal model control is compared with a single global NN IMC design, a multiple transfer function IMC design and a single local NN IMC design.
Keywords: Internal Model Controller, Neural Network, pH process, Multi model, nonlinear controller design.