Masakazu Koike

Maitreyee Dutta, Jiayu Chen, Masakazu Koike et al.

In this manuscript, we propose a novel optimal Global Navigation Satellite System (GNSS) time tracking algorithm to collectively steer an ensemble consisting of synchronising miniature atomic clocks towards standard GNSS time. The synchronising miniature atomic clocks generate a common synchronised time which has good short term performance but its accuracy and precision, which is measured by Allan variance, deteriorates in the long run. So, a supervisor designs and periodically broadcasts the proposed GNSS time tracking control to the ensemble miniature atomic clocks that steer the average of ensemble towards the average of GNSS receivers, which are receivers of GNSS time. The tracking control is constructed using a Kalman filter estimation process that estimates the difference in average of GNSS receivers and average of ensemble clocks by using relative clock readings between GNSS receivers and their adjacent ensemble clock. Under the influence of the periodically received tracking control, the stabilised ensemble clocks have better long term accuracy and precision over long averaging periods. Since the tracking control is designed to solely influence the average of the ensemble, the tracking process does not interfere with the synchronisation process and vice versa. The feedback matrix associated with the tracking control is obtained from an optimisation problem that minimises steady-state Allan variance. Numerical results are provided to show the efficacy of the proposed algorithm for enhancing long term performance.

Takayuki Ishizaki, Masakazu Koike, Jun-ichi Imura

In this paper, we propose a method of designing low-dimensional retrofit controllers for interconnected linear systems. In the proposed method, by retrofitting an additional low-dimensional controller to a preexisting control system, we aim at improving transient responses caused by spatially local state deflections, which can be regarded as a local fault occurring at a specific subsystem. It is found that a type of state-space expansion, called hierarchical state-space expansion, is the key to systematically designing a low-dimensional retrofit controller, whose action is specialized to controlling the corresponding subsystem. Furthermore, the state-space expansion enables theoretical clarification of the fact that the performance index of the transient response control is improved by appropriately tuning the retrofit controller. The efficiency of the proposed method is shown through a motivating example of power system control where we clarify the trade-off relation between the dimension of a retrofit controller and its control performance.