PARAMETRIC ESTIMATION OF A DC MOTOR

There is a continuing need to develop new experimental techniques for identifying linear models to represent the dynamic behaviour of electric motors. The traditional methods, based upon the application of sinusoidal and step input functions, have yielded limited information due to the presence of noise and other factors. The author has considered the particular case of a dc motor and shown that the experimental determination of the linearised dc motor parameters may be reformulated as a problem of estimating the parameters in a state-space model of known structure. The present paper extends this work to a dc motor when a pseudo-random binary sequence signal used to perturb the rotor. In previous experimental investigation of these motors, it does place restrictions on the form of input signal, whereas here restrictions are focused on the sequence length and clocking interval of these signals. A discrete-time state-space model of the same structure is fitted here to sampled observations from laboratory experiments. The estimated discrete-time parameter matrix is then transformed to obtain estimates of the linearised dc motor parameters. It is shown that this new formulation eradicates possible errors in the interpretation of results which may arise from the steady-states appearing in the measurement. A formula which describes the relation between a system bandwidth and clocking interval, is outlined. The convergence of the obtained estimates is proved. Lastly, experimental results for the estimates of the motor parameters are presented.