This paper proposes a method to rank air-conditioning systems in a building based on drift anomalies in sensor measurements. The proposed method employs an approach to estimate the drift anomalies from changes in the sensor measurement data of the systems. However, the changes depend on the contexts in which the systems work as well as the drift anomalies. Thus, we propose feature selection and drift anomaly estimation that depend on the contexts. Experimental results imply that the proposed method will be benefical even in a case where the drift anomalies arise over the course of a year.

This paper considers an independent-four-wheel-drive EV to pursue the limit of maneuverability. An effective method of tire force control and maximization of tire workload by controlling slip ratio and tire slip angle is proposed. First, the vehicle model of an independent-four-wheel-drive EV is defined. Next, a simulation of the tire force control is conducted, which shows that the proposed method can maintain the tire workload at 1 for an arbitrary tire force direction. Finally, the effectiveness of the proposed method is verified by simulation and experimental validation using a real EV.

Recently, from the viewpoint of vibration and noise reductions, torque ripple reduction of motor-equipped electrical machines is the need of the hour. The authors have already proposed to improve torque characteristics of a synchronous reluctance motor using the topology optimization method. The evaluated rotor structure succeeded in substantial torque ripple reduction, but the cause for this reduction was unknown. In this paper, we investigate the reason for torque ripple reduction using the improved rotor structure of the motor by a topology optimization method known as R-SLP, by comparing it to the reference model using finite element analysis. A calculation method is proposed to analyze the circumferential magnetic force density of spatial and time harmonic orders using the magnetic density in the air gap. By focusing on the time order at which the torque waveform was evaluated, we estimate which circumferential magnetic force density orders affect torque ripple reduction. Then, we clarify the reason for torque ripple reduction to examine the influence of local shape of rotor structure using 12^th time-harmonic magnetic fields. As a result, the modified flux barrier rotor structure is obtained, which further reduces the torque ripple in comparison to the previous structure.

This paper proposes a T-type dual active bridge (DAB) converter with switching operation mode to improve the converter efficiency over a wide range of load and voltage. The T-type DAB converter outputs two waveforms: a square wave and a square waveform with half of the input voltage. By applying the switching operation mode, the converter efficiency is improved. In addition, voltage control for the T-type DAB converter is also proposed. Owing to the proposed voltage control, the slow voltage response is avoided because the operation modes are seamlessly switched by matching the transferred power of each mode. In an experiment conducted using a 1.5-kW prototype, the validity of proposed method is confirmed. A maximum efficiency of 97.9% is achieved at the nominal voltage condition. At the half-output-voltage condition, the converter loss is reduced by up to 62.3%, when the switching mode is applied. Moreover, the output voltage control is effective, regardless of switching operation mode.

After it was reported that an AI player scored higher in Atari2600 games than skilled human players by using deep reinforcement learning techniques, many researchers were inspired to apply deep reinforcement leaning in various fields. This paper focuses on the autonomous ground leveling work by a bulldozer, which is expected to optimize the action of the bulldozer. In a previous work, we implemented a deep Q learning method by giving the images as the input data for the network. However, when learning the image using the convolution layer as the input using deep reinforcement learning, it requires a large computational cost for the learning process. If the size of the neural network is shrunken by contriving the data to be supplied to the input, the learning time (duration) will be reduced. This paper describes the comparison results for different orders of input data. the transition of the learning sequence is also evaluated.

In this paper, we propose a high frequency power inductor for an automatic matching unit of a plasma generator. For the power inductor, the change in the inductance with respect to frequency, temperature, and current value must be prevented as much as possible. Moreover, it is required to have low loss, small size, light weight and low cost. In the conventional research, the inductance and the loss per unit weight were evaluated in terms of coil shape, bundle shape of winding and laminated structure; then, their effectiveness was confirmed. However, the methods for dividing the flux and loss generation mechanism caused by the proximity effect have not been investigated. Therefore, the purpose of this paper is to clarify these points and to derive a structure showing high characteristics. The proposed structure is a flat wise type coil with wide window area. A rectangular wire with thickness equal to or less than the skin depth is used as the winding wire. Furthermore, the experimental results suggested the practical effectiveness of the proposed structure.

The increase in temperature of a traction battery installed on a battery-powered electric multiple-unit train (EMU) can be easily estimated using the equivalent thermal circuit model. However, the accuracy of the estimated temperature depends on the parameters (thermal capacities and resistances) used in the model. To overcome this problem, we developed a parameter fitting method to automatically identify the optimal parameters for the model. The method requires only the measured temperature and current curves of the battery system, which can be obtained by performing a test run. The accuracy of the estimated temperatures and the effect of the cooling fans were quantitatively evaluated by applying the proposed method to the test-run results of the battery-powered EMU.

In recent years, biology researchers have elucidated biological and behavioral mechanisms from the measurement of observation data. The measurement of the trajectory of a creature from observation data based on human eyes could incur significant time costs and may result in misrecognitions. Therefore, biology researchers require automated measurement and recording support systems. In this paper, we introduce a new method for generating the swimming trajectory of medaka to analyze its behavior. Then, in order to recognize the behavior of medaka, a method that combines orientation code matching by rotation invariant multiple-templates and particle filter based on multiple-likelihood functions is introduced. Through experimentations, we prove the effectiveness of our method.

A modular equalization system integrating series-resonant voltage multiplier (SRVM)-based cell equalizers and switched capacitor converter (SCC)-based module equalizers has been proposed for energy storage systems consisting of multiple cells/modules connected in series. In this system, however, cell/module equalization currents cannot be limited within desired levels, likely resulting in excessive currents as well as associated losses when cell/module voltages are severely imbalanced. This paper proposes a modular equalization system integrating tapped-inductor-based SRVM (TI-SRVM) and phase-shift SCC (PS-SCC) to limit equalization currents to desired values. The derivation procedure of the proposed equalization system is explained, followed by separate detailed analyses for the TI-SRVM and PS-SCC. The mathematical analysis revealed that both the TI-SRVM and PS-SCC offer inherent constant-current characteristics even without feedback control. An equalization test was performed using electric double-layer capacitors (EDLCs) connected in series, and the results demonstrated the efficacy of the proposed modular equalization system.

A new control method for multi-level PFC converters is proposed in this paper. Multi-level AC/DC comprises a PFC reactor, sub converter, and main converter. With cooperative control, the sub converter and the main converter switch each other, and they cooperate to control input current and output voltage. The proposed control can reduce the capacity of the sub converter capacitor to 1/200 of the conventional control. As a result, high power factor and constant voltage controls in the buck-boost modes are verified experimentally with a 3-kW prototype using film capacitors. Moreover, the same efficiency as that of a conventional one is achieved.

Our research focuses on switch-mode power supply (SMPS), especially new control method for DC-DC converters that require high responsiveness and high robustness.