名取 賢二

This paper proposes a quasi-MPPT (maximum power point tracking) control for integrating multiple energy harvesters of different characteristics. Energy harvesting (EH) technology is a promising technology and is expected to be widely used in various fields. However, the amount of the harvested power is inherently limited and strongly depends on the weather and other conditions. Therefore, multiple energy harvesters are often integrated to increase total harvested energy. In the case of integrating multiple energy harvesters of different characteristics, the control tends to be complicated due to the different characteristics of the energy harvesters. The proposed quasi-MPPT control enables the multiple energy harvesters to work near their MPP voltages by referring to the command voltages derived from preliminary identification of the energy harvesters and measurement of the open-circuit voltages. It achieves harvesting of approximately maximum powers of the multiple energy harvesters by a simple control strategy using a power management circuit. The effectiveness is validated by using an experimental system that consists of photovoltaic (PV) and thermoelectric (TE) energy harvesters.

The voltage drop in GaN HEMTs during reverse conduction is larger than that of Si-MOSFETs, causing an increase in dead-Time loss and a rise in the bootstrap capacitor voltage when a bootstrap circuit is applied. We propose connecting a low-side device in parallel with a Schottky barrier diode as a solution. This allows the bootstrap capacitor to be charged at the appropriate voltage because the voltage drop is lower. This also reduces dead-Time losses but increases the losses due to capacitance across the connected diode terminals. These losses increase with frequency and are not negligible in ultrahigh frequency switching operations such as 1MHz. This study examines the effect of using anti-parallel diodes for GaN devices.

Carrier frequency of N-level flying capacitor converter (FCMLC) is equivalently (N-1) times higher than that of the 2-level converter with the same carrier frequency. In consequence, its control bandwidth is extended. However, updating the command value in that equivalent carrier period causes capacitor voltages to deviate from predetermined voltages. It is caused by mismatches between charging and discharging times of flying capacitors. Imbalanced flying capacitor voltages cause problems such as increase of harmonics in the output current and applied voltages to the devices. In this study, we derive the discrete-time state-space equation of FCMLC considering flying capacitor voltages and analyze stabilities of flying capacitors. Depending on the analysis, a compensation method for capacitor voltages is proposed, and the effectiveness of it is verified by the experiments.

In motion control systems, detailed characteristics of current control systems are usually not considered. However, as the performance of the overall system advances, higher-performance current control systems are required. This paper proposes a novel PI current control method for realizing deadbeat control characteristics. Conventional deadbeat control is complicated because it is necessary to calculate the pulse width for each sampling cycle. On the other hand, proposed method achieves deadbeat characteristics easier than conventional method by simply determining the PI gains. It can also be applied to multi-level inverters. By simulations and experiments, we confirm that the deadbeat characteristic is obtained. The robustness of the system against disturbances and modeling errors is also confirmed.

This paper proposes a control method for a bidirectional isolated three-phase AC/DC Dual Active Bridge (DAB) converter based on matrix converter (MC) that realizes a wide output voltage range. This method achieves the reduction of excessive high frequency link current and soft switching with online calculation by changing two modulation methods according to the input and output voltage ratio. The proposed method realizes higher efficiency in expanded output voltage range with reduced online calculation burden. The feasibility and validity of the proposed method are verified by experimental demonstration.