Modelling and state observation of Simulated Moving Bed processes

State observation of distributed parameter systems pose the problem of coping with the infinite state dimension of the process models. In addition to this difficulty, the dynamics of Simulated Moving Bed (SMB) processes is described by partial differential equations which are governed by switching initial conditions or switching boundary conditions, respectively. Furthermore, due to the setup of SMB plants only limited measurement information of the process behaviour is available. To overcome these difficulties, a model of the wave fronts of the SMB concentration profiles is derived which comprises a linear time-variant discrete-time state-space model describing the form, position and propagation of the wave fronts. A new approach to the solution of the state observation problem is proposed which is based on this model. Simulations of a closed-loop controlled SMB process are presented to show the applicability of the developed observer in the surrounding of the stationary operating point.