Iterative optimization of parameterized trajectories for complex mechatronic systems

Many mechatronic systems are controlled with predefined input trajectories, either in open loop or using feedback. This research considers the optimization of these trajectories, for mechatronic applications whose dynamics are difficult to model, and vary in a significant and unpredictable manner over time. For these applications, solving the trajectories with an optimal control problem requires a considerable modelling effort. In many practical applications this is avoided by parameterizing the trajectories, and estimating the optimal parameter values during an experimental calibration. Repeating the calibration at regular time intervals can compensate for system variation, but requires the machine to be taken out of production. In this research a new methodology is developed to iteratively optimize these parameterized trajectories during normal operation, hence avoiding calibrations and limiting required modelling effort.