Obtaining simple dynamic models

Identifying dynamic systems in order to find an easy way to obtain models that can be used, e.g. in controller tuning, was studied. Three types of models were considered. The first type describes a series connection of identical first-order systems. This type can approximately represent various kinds of systems. The parameters are gain, time constant and order. The second type is the common dead-time and first-order model type. The third type describes a series connection of a pure dead-time system and a second-order system. A specific algorithm for parameter estimation was developed for each model type. The estimation algorithm for the first type model is essentially new. The other algorithms utilize previously known algorithms. The algorithms were tested using simulated data. The tests included cases where the data were obtained from a noisy system controlled by a PI (proportional-integral) controller. The algorithms turned out to be reliable. The feedback and large noise caused some error in the parameter estimates, but the accuracy was good enough for tuning purposes. The methods are applicable to many cases where the graphical analysis of step responses is not