Multi input multi output frequency sampling filters for real system applications

Real world system applications are mostly involved with multivariable process. Thus, the observation and study over single input single output system only, may no longer sufficient. In addition, data acquisition process from real system typically yields large amounts of discrete-time data. This may lead to unacceptable computational time during the identification process. To overcome the problem, the multivariable frequency sampling filter approach is introduced in this paper. Here, the multi input multi output (MIMO) raw data will be analysed in order to obtain only an important and meaningful parameter that describes the empirical model of the analysed data. By performing this procedure, the compressed, cleaned and unbiased data will be obtained, in which, it can be further used to develop a model of the system, conduct time and frequency response analysis and also plan for next control system design. The efficacy of the developed MIMO frequency sampling filters is demonstrated by compressing simulation data of a pH neutralization process in a stirring tank and data from model of a steam generator at Abbott Power Plant. Result shows that the amount of raw data is successfully compressed and the outcome is ready for modelling and identification purposes with less computational time.