Mechanical Parameter Tuning Based on Iterative Learning Mechatronics Approach

In most mechatronics applications, the best control performance cannot be obtained by only shaping a control input signal because, in practice, such control is effective within only the performance range realizable by the actuator and control system. Therefore, to obtain the best control performance, the mechanical parameters should be optimally selected such that the desired control performance can be achieved with minimal control effort. However, it is difficult to accurately predict the control performance without conducting an actual experiment because the control performance is dependent on not only the mechanical design parameters, but also on various practical factors, such as the input and output saturation of the actuator, the heat problem, and sensor limitations. For these reasons, a recursive mechanical parameter tuning process based on control experiments is proposed in this paper. Based on a set of control signals (e.g., a control input and a tracking error), the proposed mechanical parameter tuning method seeks a better mechanical design parameter for improving the control performance (i.e., to reduce the control input power). For verification of the proposed method, the method was applied to case studies including simulations and experiments.