In the changeover section of high speed railway, the frequent operation of surge arresters has been observed. In our previous study, it was confirmed that such frequent discharges by the changeover switchgears do not deteriorate the arresters. However, occasionally the surge arrester could develop troubles. Therefore, we developed a new surge counter to measure not only the amount of total discharge current, but also the resistive current of the leakage current in order to find out the deterioration trends of the surge arrester. In this paper, the history of our research related to the development of the new surge counter is reviewed.

When train traffic is disrupted by accidents or technical problems, train dispatchers need to make a rescheduling plan taking train cancellation and/or changes of rolling stock allocation into account based on prior experiences. However, unpredictable events such as a temporal train congestion can cause additional delays owing to which train traffic does not recover as expected. Although we need to analyze various types of factors that cause unpredictable delays in order to prevent or reduce their impact, it is very difficult to analyze all of them in detail because there are several delays in actual situations where train traffic is disrupted. In this paper, we present an algorithm for extraction of unpredictable delays with original and resultant train timetable data using train operation simulation. The simulation model is based on PERT and critical path method. We apply the proposed algorithm to several actual disruption cases and are able to extract unpredictable delays that need to be analyzed further, for the determination of causative factors.

In recent years, much research has been conducted on the use of wireless sensor networks (WSN) for monitoring the conditions of railway vehicles. In WSNs for vehicle condition monitoring, the changes in the network configuration caused by both the coupling and decoupling of vehicles and communication environment changes resulting from vehicular motion make the power consumption of the sensor node unpredictable based on the existing approaches. In this paper, we propose a method of predicting the sensor node power consumption for WSNs used for vehicle condition monitoring based on the time-series Monte Carlo method.

This paper proposes and demonstrates a novel slip stabilizing control method for electric railway vehicles to achieve both high acceleration and riding comfort. The method stabilizes wheel velocity during wheel slips and avoids diverging of wheel slipping that can cause vibrations, noises or wheel and rail damage. The method also achieves slower re-adhesion with less longitudinal shock. The test run results show that the method achieves both higher acceleration and better riding comfort than the conventional re-adhesion control method.

The equivalent 12-pulse rectifier consists of a pair of rectifier units having different transformer winding configurations (e. g. Δ-Y and Δ-Δ). In some cases, the output currents of these rectifiers are different, though these units have the same rated capacity. Moreover, there are some cases in which despite using a 12-pulse rectifier, the relative 6th harmonic content of the DC-output voltage from the rectifier still remains high. This paper reports the effect of harmonics contained included in the AC-input voltage of the rectifier on these phenomena.

The breaking of contact wires occurs due to the arc and large current flow that are generated between the pantograph and the contact wires when a train comes to a standstill at an insulated overlap. Especially, when the arc occurs, it melts the contact wire, and subsequently, when it touches the additional wire above the contact wire, the catenary could break. This study first shows the potential differences and current at an insulated overlap, and then shows the maximum arc length through experiments. Using these results, the effectiveness of the countermeasures against the breaking of contact wires at insulated overlaps when a train comes to a standstill is demonstrated there.

Three-phase voltage-fed converters, which are based on the two-phase PWM scheme offer higher ac voltage, lower switching times, and lower switching losses compared with those based on the three-phase PWM scheme. However, the harmonics of the ac-side voltages and currents produced using a two-phase PWM scheme are distributed over a wide frequency range, and this can be a significant drawback for certain applications. To overcome this problem, the author proposes a novel two-phase PWM scheme using pulse-voltage injection method. With the proposed PWM scheme, the harmonics appears in a narrow frequency range, similar to that obtained with the conventional three-phase PWM scheme. In this paper, the author presents experimental results that validate the proposed PWM scheme.

A gradient Si steel sheet shows low iron loss in the high frequency range and lower iron loss deterioration under compressive stresses, therefore, it is suitable for use as the core material of high-speed motors, which have seen considerable downsizing in recent years. By performing shearing and shrink-fitting, the advantages of gradient Si steel as a motor material are investigated in this study. In both shearing and compressive stress experiments, gradient Si steel showed lower iron loss compared to conventional non-oriented electrical steel (0.2mm thick). By numerical simulation, the efficiencies of the shrink-fitted motors using gradient Si steel and 3%Si steel were calculated. The motor using the gradient Si steel showed significantly higher efficiency than that using the 3%Si steel. Moreover, even under city driving conditions, gradient Si steel showed good efficiency, because of lower iron loss deterioration. Therefore, gradient Si steel can be used for a wide variety of motors in addition to high- frequency devices.

This paper proposes an online-type decentralized controller tuning method for the decentralized control system of a web transport system, in which mutual interactions are significant. The controllers are updated online by employing the modified fictitious reference iterative tuning (FRIT) and online particle swarm optimization (OPSO) methods based on the structure of the overlapping decomposition. With the proposed tuning method, the computation time required for the control is significantly reduced while maintaining a good control performance. The effectiveness of the constructed system is verified by the experimental results.

Recently, the development of power devices, such as SiC and GaN devices has lead to the increase in research on the influence of parasitic parameters, such as wiring inductance, parasitic capacitance, and packaging inductance in power converter circuits. In general, the converter circuit has several circuit loops, including higher-voltage circuit, gate-drive, and low voltage control circuits, on one printed circuit board. However, studies have discussed the influence of a mutual inductance between higher-voltage and gate-drive circuits. Mutual inductances between the circuits could influence the switching characteristics, such as surge voltage and switching loss. This paper describes an influence of mutual inductance on the higher-voltage and gate-drive circuits, and presents the simulation and experimental results rated at 500V and 20A using SiC-MOSFET and SBD. Moreover, this paper proposes an implementation of a gate drive circuit by using mutual inductance for suppression of both surge voltage and switching loss.

This paper proposes a new diode-clamped linear amplifier (DCLA) topology using only n-channel MOSFETs. Although the conventional DCLA consists of n and p-channel MOSFETs, the proposed DCLA is composed of n-channel MOSFETs. Therefore, the proposed DCLA is easy to develop, because selection of complementary pair MOSFETs is not needed. As with the conventional DCLA, noise level of output voltage of the proposed DCLA is lower than general pulse width modulation (PWM) inverters, because the DCLA uses multilevel topology and active state of MOSFETs. Additionally, this paper also proposes a method of voltage feed-back control, which is capable of compensating crossover distortion of the DCLA. The output voltage of the DCLA has crossover distortion that causes the increase of voltage error and total harmonic distortion (THD). In order to compensate the crossover distortion, a voltage compensator including a plant model for the proposed DCLA is implemented. The proposed method reduces THD of output voltage by 90 percent through the simulations and experimental results.

This paper presents a new sensorless vector control method for permanent-magnet synchronous motors (PMSMs) that employs discrete-time high-frequency voltage injection. The frequency range of the injected voltage is almost the same as that of the pulse width modulation (PWM) carrier of the voltage command. This allows high-speed phase estimation and high-performance drive, and possibly low acoustic noise caused by the injected voltage. The current associated with the injected voltage, which has a frequency almost equal to the carrier frequency, must be analyzed in a discrete-time manner. This is because, a simple approximation of the continuous-time analysis of the high-frequency current cannot be applied in the range of the carrier frequency. Based on discrete-time analysis, this paper reconstructs the conventional rotor phase estimation methods using positive- and negative-phase high-frequency current correlation signals. The usefulness of the proposed method is verified through extensive experiments.

Magnetic-geared motors have attracted considerable attention because of their high torque density. However, the gear ratio of a magnetic-geared motor depends on the number of poles in the rotor, which is fixed during operation. In this paper, a magnetic-geared motor with a continuously variable transmission gear ratio and its control method are proposed. The structure and operational principle of the proposed motor are described. The N-T curve and efficiency are computed and measured using a prototype, and the computed and measured characteristics are compared.

Our laboratory focuses on intelligent cooperative systems for future human-robot co-existence. The cooperative system is required to elucidate their interaction that is difficult for modelling. Therefore, the machine learning is utilized for the interaction modeling among human-beings, robots and sensors.