Hardware Implementation and Evaluation of a Knowledge-Based Tuner for a Servo Motor

Abstract A commercially available servo motor system consisting of a DC motor, optical encoder, power driver/amplifier, and a digital feedback controller is considered. The controller contains a lead compensator and an integrator. A knowledge-based system is developed and implemented for tuning the digital controller so as to meet a set of performance requirements. A reference model is utilized for several purposes. First the parameters of this model are adjusted so as to meet the specifications. This will provide some buffer against unreasonable specifications. Then, during tuning, the deviation of the response of the actual servo system from the reference model, for the same input signal, is used to establish the performance conditions. The knowledge-based tuner uses these conditions to infer the necessary tuning actions for the controller parameters which include the pole and zero locations of the lead compensator and the control gain. A model of the control system may be used to gain knowledge on the nature of the response of the system to various changes in tuning parameters and tuning levels. Additional knowledge and experience is obtained through actual operation of the servo system under different tuning conditions. The knowledge base obtained in this manner is fuzzy in general. Furthermore, the relationships of controller attributes such as the maximum phase lead, the crossover frequency, the compensator gain at the crossover frequency, and the low frequency where this gain equals that of the integral controller, with the actual tuning parameters, have to be established as well. Tuning inferences are determined using the compositional rule of inference. The performance of the digital servo system is evaluated through experimentation.