Modelling of a nonlinear multivariable boiler plant using Hammerstein model: A nonparametric approach

Of the many model structures that can represent a nonlinear process effectively, the Hammerstein model is one such model which has attracted a lot of attention. This paper considers a real industrial problem of modelling a nonlinear multivariable steam generating plant using the methods of system identification. The work uses Hammerstein model to model the plant from sampled data collected at Abbott Power Plant in Campaign, IL. Neural networks and statespace model are used to model the nonlinearities and the dynamics of the system respectively. A recursive algorithm is developed which makes use of Particle Swarm Optimisation (PSO) and Subspace Identification Method (SIM) to estimate the parameters of the nonlinear and linear parts respectively. Validation results using computer simulation are included at the end to demonstrate the good fit and concordance of predicted outputs with actual data.