The identification process consists of 3 stages:
-
a sound and stability analysis of the control process
-
an initial analysis of the reaction to an actuator pulse, which is shown as the first identification model: a filter is created on the basis of this first estimate; this is used during the last phase
-
a second analysis of the reaction to a second actuator pulse gives more precise information because of the data filter
Finally, a complete process model is created. If the results of the two previous phases are two far apart, the estimate is abandoned and autotuning fails.
After both phases a parameter set is created for the controller being tuned. The resulting control parameters are based on the gain and on the ratio between reaction time and process delay.
The algorithm must be able to withstand the modification of the gain and the time constants in ratio 2 without losing stability. The asymmetrical processes are supported if they fulfill these conditions. If not, an error is displayed during diagnosis diag.