Abstract
This paper proposes an algorithm for parameter estimation of continuous SISO systems purely using a digital computer. The sampled input and output signals are passed through the low-pass IIR filters which are designed in continuous-time domain and then discretized by bilinear transformation. The low-pass IIR filters are introduced to avoid taking direct derivatives and to reduce the effects of measurement noise. This approach does not invole the problem of the initial conditions in contrast with the well known multiple integration approach. It is shown that the IIR filters are more convenient to be designed than the FIR filters of the linear integral filter approach.
If the pass-band is appropriately selected, the LS method is efficient for the case where measurement noise level is low. The application of an IV method in a ‘bootstrap’ fashion to the high measurement noise case is also considered to yield consistent estimates. Simulation results show that if the cut-off frequency of the filters is selected in a relatively wide range near that of the continuous system under study, the results of the estimation algorithm are quite satisfactory.
