2016 年 82 巻 837 号 p. 15-00552
This paper proposes an automatic tuning method for a motor speed control system. Parameters of a PI controller and a low-pass filter (LPF) for motor speed signal are automatically derived satisfying required specifications of a time constant of step response and a tolerable steady-state error of ramp response. In addition, cut-off frequency of the LPF is set as low as possible, which solves the problem of a tradeoff between response speed and noise reduction. The method helps people to tune a speed controller without trial and error and reduces man-hour for developments. In the derivation of the proposed method, the coefficient diagram method (CDM) is utilized. In this paper, the stability index, the equivalent time constant, and the representation of characteristic polynomial defined in the CDM are theoretically linked to the specifications of the step and ramp responses. This enables the proposed method to directly incorporate the required performance for tuning the PI controller and the LPF. Simulations results show that the proposed method derive a desirable controller which satisfies the specifications as theoretically expected. Effectiveness is also verified in experiments in the presence of sensor noise of a resolver and the controller achieves desired performance without trial and error tuning of the parameters.
