Process identification

To be able to calculate the parameters of the PID controller, a simple or complex process model is required, depending on the calculation method chosen for the controller and on the type of process experiment that can be carried out, in order to identify its dynamics. The basic idea behind identification is to obtain a process model of a pre-determined type, from an experiment on the actual process, by applying an excitation signal to the process input u (t ) and measuring the output y (t ).

The excitation signals typically used at the process input are: step, sinusoid and pseudo-random binary sequence (SBPA), and consequently methods based on index response, frequency response or parametric identification itself are used. Parametric identification is very suitable for finding a sampled model of the process, which will be used for the direct synthesis of a digital...