Optimization of Pseudo Random Binary Sequence (PRBS) combination for Online Modeling of MIMO Ill-conditioned System

This paper aims to improve the performance of system identification based on optimization of Pseudo Random Binary Sequence (PRBS) excitation signal combination for Multiple-Input Multiple Output (MIMO) Ill-Conditioned system. Ill-conditioned system is defined as system that is formed by various variables and the level of interaction between all the variables is high. It is found that in the case of ill-conditioned system, the design of PRBS combination as excitation signal will affect the performance of system identification. The experimental subject of this paper is the air pilot plant that is located in Universiti Teknologi PETRONAS (UTP). Empirical modeling method is first used to obtain the steady gain matrix of the system, followed by the transfer function based on the time constant of the system. A process will be created on simulation based on the transfer function obtained. High correlated, moderate correlated and un-correlated set of PRBS will be used as excitation signal for system identification. The test signal combination will also be tested in the real plant implementation. The performance of different combination of PRBS will be examined by using Bode plot and fit percentage. The result shows that the lower the correlation, the better the modeling performance for the operation in both simulated and real process environment