Control valve position response identification by Matlab

Modeling and identification of under operation industrial systems could be considered as the first step before making any efforts in order to design improved controller or identify their characteristics and behaviors. There are two specifications that distinguish modeling of operational industrial systems from modeling of non-operational or laboratory cases. The first one is the limited possible input variation range or in the other word, lack of permission to apply desired inputs to the system in order to perform modeling. The second one is that in most cases the real mechanical model is not available due to system complexity and difficulty of its mechanical parts analysis. Moreover, limited available facilities for data sampling could be added as another barrier. In this paper, a black-box SISO model for a control valve, which is installed on a compressor package, is estimated by Matlab identification GUI. The calculations are performed about the three real datasets which are obtained from the valve position and action in sudden full open conditions. It can be concluded that the 3-poles process model is the most appropriate model from the other available models in this GUI. Also the calculated best fit value was more than 87% after considering two employed datasets as Working and Validation datasets respectively.