Three-level boost (TLB) converters are a promising interface of bipolar DC microgrids. With different power of the positive and negative DC buses, the conventional modulation method used in TLB converters fails to have zero voltage switching (ZVS) for all switches, and the peak current is also high, leading to the risk of inductor saturation. In this article, an optimized modulation (OM) method is proposed to achieve ZVS for all switches even under unbalanced loads. The OM method features the same valley current at both the high and low output power sides, contributing to wider ZVS range as well as smaller peak-to-peak current. The modulation degree of freedom in TLB converters under unbalanced-load conditions is analyzed in detail for the first time, and the advantages of the proposed OM method are compared with its counterparts in terms of the ZVS range, the peak and valley current, and the efficiency. The effectiveness and advantages of the proposed OM method are finally validated by experiment.

This paper presents a novel constant power (CP) control method for series-parallel resonant converters in the context of repetitive-frequency (RF) pulsed power supplies. The objective is to isolate pulsating power generated by RF operation from the power grid and enhance power supply utilization (PSU). The proposed CP control is based on the time-domain model, especially under discontinuous current mode (DCM). First, the CP model is established from the analytical correlation between the output voltage and the initial voltage of resonant capacitors. Second, real-time calculation of the charging current and power is achieved using the developed CP model, thereby improving dynamic response. Additionally, the switching frequency and duty cycle are updated based on the timely calculated output power, ensuring the desired CP charging and full isolation of pulsating power from the power grid. Finally, experimental results show the superiority of the proposed CP control method over conventional proportion-integration (PI) control methods in terms of the RF, the AC input-side power factor (PF), and the PSU.

As more renewable energy resources are injected into the power system, it is an increasingly challenging task to retain the dependability and stability of the power grid. In this situation, installing an energy storage system (ESS) is an effective approach to improve the transient stability of the power system. Among the ESSs, the flywheel ESS (FESS) is an excellent source for the power system because of its several advantages. Therefore, in this paper, we propose an adaptive neural network (NN) error-constrained control algorithm that can accomplish the speedy charging of the FESS with guaranteed steady state performance and transient performance. First, as opposed to previous studies based on error transformation, we apply the error transformation to each subsystem of FESS, and the original constrained system is converted to a new unconstrained system. Because of this, we can guarantee predefined performance of each subsystem and further improve the system performance. Then, the new unconstrained system is stabilized by using the backstepping method and the unknown dynamics are tackled by NN approximators. Finally, we guarantee the boundness of all error signals via Lyapunov analysis and verify the effectiveness of theoretical results via numerical analysis.

In this communication, a triband AMC-backed microstrip antenna is designed. It is implemented by placing a triband microwave antenna which uses coplanar-waveguide (CPW) over a novel artificial magnetic conductor (AMC) with three zero-phases in the reflection coefficient. The AMC unit cell is composed of three homocentric copper rings. The simulated S-parameters generally corresponds to the performance of original antenna without AMC. The return loss of the frequency range 2.4GHz which can be used for WLAN is enhanced. By adjusting the distance between the antenna and the AMC plane, the performance near 1.8GHz can be optimized.

Transcranial magnetic stimulation (TMS) is a noninvasive technique that stimulates the brain via electric fields induced by pulsed currents in a coil. Beyond coil design, the pulse waveform also affects neuronal activation, though specific mechanisms remain unclear. A waveform optimization method is presented to improve stimulation selectivity, in which a multiscale model with realistic neuronal morphology quantifies the selectivity index, the induced electric field is parameterized for a pulse parameter controllable TMS discharge circuit, and particle swarm optimization is applied to refine the waveform parameters. Optimized waveforms show the potential to achieve higher selectivity than monophasic waveforms, which are known for relatively high selectivity. The results further show that waveform polarity, defined as the ratio of the positive to negative integrated areas of the induced electric field waveform, plays a key role in selectivity across cortical layers. Other waveform parameters, such as the relative order and amplitude of induced electric field levels, may also affect selectivity, with their influence varying according to neuronal morphology and the local electric field distribution. The neuronal time constant also influences how waveform parameters affect stimulation selectivity. Furthermore, stronger polarity-selectivity correlations are often associated with greater alignment between selectivity and coil heating. The proposed method can enhance TMS selectivity by optimizing stimulation waveforms. The identified patterns may help explain variability in TMS outcomes across different waveforms and between individuals, and support the development of personalized designs that enhance stimulation efficacy while minimizing coil heating.

With the development of wide-area measurement system (WAMS), the prospect of state perception and faulty line detection of power grid is more and more broad by using phasor measurement units (PMUs) to obtain the information of the whole power grid. In order to realize the accurate detection of power grid faulty lines, aiming at the difficulty of artificial feature selection in automatic detection of power grid fault, this paper proposes a power grid faulty line detection method based on PMU data and capsule network. PMU data has the advantages of high sampling frequency and good real-time performance, which can avoid the problems of protection, circuit breaker rejection and misoperation, and is conducive to the rapid and accurate identification of faulty lines. Capsule network can reflect the internal spatial relationship between the local and the whole, and can retain the connection and natural order between contexts in the original data, which is conducive to faulty line detection based on PMU data. According to the measured data and simulation model data, the detection efficiency and accuracy of the faulty line detection model are verified.