ABSTRACT
The step response test is one of the simplest identification experiments to perform. The test involves increasing or decreasing the input variable from one operating point to another in a step fashion, and recording the behaviour of the output variable. Step response tests are often performed in industry in order to determine approximate values for the process gain, time constant and time delay (Ljung, 1987). However, these experiments are widely viewed as only a precursor to the design of further experiments, the collection of more input-output data, and the subsequent analysis of this data using regression-based techniques to obtain a more accurate model. However, the simplicity of the original step response test provides the incentive to fully explore the extent to which an accurate model may be obtained directly from the step response data itself.
