When driving the piezoelectric transducer (PT: piezo transducer), which is a key device, it is important for the ultrasonic system (using ultrasonic waves of 20 kHz or higher) to operate at a resonant frequency that can maximize the conversion of mechanical energy (vibration) from electrical energy. The resonant frequency of the PT changes during the actual operation according to the load fluctuations and environmental conditions. Therefore, to maintain a stable output in an ultrasonic system, it is essential to track the resonant frequency in a short time. In particular, fast resonant frequency tracking (RFT: resonant frequency tracking) is an important factor in the medical ultrasonic system, i.e., the system applied in this thesis. The reason is that in the case of a medical ultrasonic system, heat-induced skin necrosis, etc., may cause the procedure to be completed within a short period of time. Therefore, tracking the RFT time for maximum power transfer is an important factor; in this thesis, we propose a new high-speed RFT method. The proposed method finds the whole system resonance frequency by using the transient phenomenon (underdamped response characteristic) that appears in an impedance system, such as an ultrasonic generator, and uses this to derive the mechanical resonance frequency of the PT. To increase the accuracy of the proposed method, parameter fluctuations of the pressure of the PT, the equivalent circuit impedance analysis of the PT, and a MATLAB simulation were performed. Through this, the correlation between the resonance frequency of the ultrasonic system, including the $LC$ filter with nonlinear characteristics and the mechanical resonance frequency of the PT, was analyzed. Based on the analyzed results, a method for tracking the mechanical resonance frequency that can transfer the maximum output to the PT is proposed in this thesis. Experiments show that using the proposed high-speed RFT method, the ultrasonic system can track the mechanical resonance frequency of the PT with high accuracy in a short time.

As the demand for digital systems that require dc power increases rapidly, the importance of stable dc power supply methods increases. Various elements, such as renewable energy and load, are configured in the power grid that supplies dc, and they are connected in various forms and power patterns. These elements have different starting and operating characteristics, which can be disadvantageous to the design of controllers for power converters to maintain a stable bus voltage in the dc power grid. Therefore, in a dc power system, an energy storage system with active bidirectional power control technology that can minimize transient conditions, such as overshoot, which can occur when a load is connected or power accelerates rapidly, is significantly important. This article proposes a novel power control method capable of compensating for the instability of the dc bus voltage owing to the transient state in the dc power supply system. This method minimizes overshoot and ensures fast responsiveness and system stability by calculating the flux linkage of an inductor in a bidirectional dc–dc converter and applying an excitation voltage without pulsewidth modulation switching. The performance and validity of the proposed transient high-speed controller are verified through simulations and experiments.

Multi-speed transmissions for electric vehicles (EVs) can achieve superior economic and dynamic performances than single-speed transmissions. Since gear shifting causes an equivalent inertia variation in multi-speed transmissions, the optimal shifting pattern should be determined by considering the inertia variation effect to maximize EV performances. To consider the dynamic inertia variation effect owing to gear shifting, the equivalent inertia for each speed gear and dynamic inertia efficiency are mathematically derived. An EV analysis model is constructed to evaluate the EV performances, and energy efficiency and acceleration ability are adopted as quantification measures for economic and dynamic performances, respectively. The result comparison of the optimal shifting patterns when considering and not considering the dynamic inertia efficiency exhibits the importance of the optimal shifting pattern considering the dynamic inertia efficiency for the superior transmission design of EVs. A multi-objective optimization problem is formulated that includes the design variables as gear ratios and shifting patterns and the objective functions as energy efficiency and acceleration ability. As an alternative to the excessive calculation burden of conducting multi-objective optimization, an artificial neural network (ANN)-based multi-objective optimization process is utilized. To verify the importance of the dynamic inertia efficiency on economic and dynamic performances, the gear ratios and shifting patterns are optimized by considering the dynamic inertia efficiency and none. The different optimum solutions and objective function values demonstrate the necessity of considering the dynamic inertia efficiency owing to gear shifting; the economic and dynamic performances are improved from 2.7% to 7.8% and 2.8% to 3.0%, respectively.

Considerable efforts are being made to reduce CO₂ emissions and thereby solve the problems of environmental pollution and global warming. Technologies for environmentally friendly transportation are being developed using batteries. In particular, with the increase in urbanization and one-person households, e-mobility products are drawing increasing attention as short-distance transportation devices. Among these vehicles, personal mobility devices (PMDs) are receiving attention as new transportation devices that are simple to operate. This paper proposes a new multilevel charging system that is advantageous in responding to the charging voltage specifications of various mobile devices with a single charging system while ensuring a low charging current ripple. The proposed diode-parallel multilevel converter consists of an independent Buck converter in series. The switch of the buck converter is configured at the negative terminal of the input power source so that the gate amplifier voltage is used as the power supply voltage; it can therefore be simply configured without a separate gate amplifier power supply. In addition, it is improved so as to have a wider charging voltage range in a low output voltage region and a better efficiency than the existing diode series multilevel converter. To verify the feasibility of the proposed system, simulations were performed using the software PowerSIM(PSIM), and, in order to verify the validity, a prototype charging system was fabricated to compare and analyze losses according to operating conditions.

Photovoltaic systems are a core component of eco-friendly energy technologies and are now widely utilized across the world for power generation. However, solar modules that are continuously exposed to the external environment experience gradual performance degradation, which results in significant power loss and operational problems. Existing aging diagnostic methods such as current–voltage curve analysis and electroluminescence/photoluminescence testing have limitations in terms of real-time monitoring, quantitative evaluation, and applicability to large-scale power plants. To address these challenges, this study proposes a novel degradation detection method that utilizes the first-order derivative of the voltage–power curve of solar modules to extract key features. This method can estimate the number of degraded solar modules within a string and the degree of degradation, enabling early detection of subtle changes in electrical characteristics. In this study, we developed an AI model based on long short-term memory to classify normal and abnormal states and predict aging status, thereby supporting monitoring and early diagnosis. The model architecture was designed to reflect the characteristics of solar power systems, adopting a relatively shallow network due to the time-series data not being excessively long and the feature changes being clear. This design effectively mitigates the issues of overfitting and gradient vanishing, thereby positively contributing to the stability of model training. The training and validation results of the proposed long short-term memory model were verified through MATLAB simulations, confirming its effectiveness in learning and convergence.