With the rapid advancement of smart manufacturing, automated detection of welding defects plays a crucial role in ensuring product quality and enhancing the efficiency and stability of production processes. Traditional manual detection methods struggle to meet modern production needs due to low accuracy, poor efficiency, and subjectivity. This paper utilizes a deformable convolutional neural network based on YOLOv5m to improve the accuracy of weld seam defect detection in X-ray images. It introduces deformable convolution kernels to identify irregular welding defects and employs decision and memory modules, proposing a feature repetition unit structure to optimize the network by reducing parameters and enhancing learning for small samples. Through comparative analysis with the original network, the improved deformable convolutional neural network shows significant improvements in loss, precision, and mAP metrics on small samples.

The application of magnetorheological equipment and its sensors in high-speed trains is restricted to a certain extent due to the demand for external power supply. Usually, there are different degrees of vibration in the operation of high-speed trains. If the vibration energy is converted into electric energy, it will be beneficial to the wide application of magnetorheological equipment in high-speed trains. In this paper, based on the load characteristics of the anti-yaw damper in high-speed trains, a ring-iron resonant power generation device based on the spring oscillator system is designed by using the amplification effect of the spring oscillator system on the amplitude. In this paper, a vibration test was carried out by an experimental prototype and two control prototypes, and the power generation capacity of the device was verified. The test results show that the maximum power generation of the device designed in this paper is 20.18 watts, which is 98% higher than that of the control group. The average power generation in random vibration is 6 watts, which can meet the power supply demand of magnetorheological damping equipment in high-speed trains.

This letter presents a millimeter-wave bandpass filter using 3D-micromachined air core recta-coax line (ARCL). The ARCL structure was fabricated through a high-precision micromachining technique which has advantages of low dielectric and radiation loss, high power capacity, and easy integration. The grounded and open stub lines, as well as impedance transformation lines contribute to the controllable transmission poles (TPs) and transmission zeros (TZs), is finally proposed to realize 87-97-GHz bandpass characteristics. Excellent in-band performance and out-of-band rejection can be obtained by control six TPs and two TZs. Detailed design method is given to guide the ARCL-based filter design. The 87-97-GHz passband filter with ARCL-to-semi-rectangular coaxial transmission line (SCL) ground-signal-ground (GSG) transition is fabricated and measured, which has a low insertion loss (IL) of 1dB, and return loss (RL) is better than -20dB in the passband.

The performance of photovoltaic cells is critically influenced by temperature conditions. This study introduces a novel energy conversion system integrating Photovoltaic, Phase Change Material, and Thermoelectric modules (PV-PCM-TE), designed to leverage phase change thermal storage capabilities. The system dynamically monitors the PCM temperature and adaptively switches the thermoelectric module’s operational state to mitigate the photovoltaic cell temperature increase, thereby enhancing the overall energy conversion efficiency. Experimental validation of the proposed system has yielded a peak energy conversion efficiency of 33.56%, presenting an innovative approach for the development of smart, energy-efficient power systems.

The electronic system is located in the shielded chamber, but the pulse field environment in the chamber is complicated, and the test environment is difficult to simulate accurately. Therefore, in order to study the high-altitude electromagnetic pulse coupling field environment inside the screened cavity with holes, the pulse field inside the screened cavity with holes is calculated based on the CST electromagnetic simulation software, and the waveforms in the time domain and frequency domain of the internal monitoring points are analyzed. The results show that there is a main pulse in the pulse field near the hole, and the high frequency component of the high-altitude electromagnetic pulse can basically be coupled into the cavity, and the pulse field environment of the coupled cavity is affected by the aperture size, the incidence Angle and the number of holes, etc. The pulse field environment of the coupled into the shielded cavity is different under different backgrounds.

In comparison to silicon steel sheet motors, utilizing amorphous alloys for the motor core can enhance motor efficiency; however, their significant magnetostriction must be taken into account. Given the high magnetostriction of core materials, it is crucial to consider the impact of magnetostriction on motor vibrations. This paper initially derives the equivalent magnetostriction force and electromagnetic force based on an analytical model. By analyzing the magnetization curve and hysteresis strain curve, the radial magnetostriction force and electromagnetic force at the stator teeth of the amorphous alloy motor are examined using Ansys software. Subsequently, a broadband “electromagnetic-force-structure” multiphysics model of the amorphous alloy prototype is established to obtain the vibration characteristics of the prototype, and the spectral characteristics of magnetostrictive and electromagnetic force vibrations in the amorphous alloy motor are analyzed. Finally, the vibration characteristics of an 18-slot 4-pole amorphous alloy experimental prototype are experimentally measured and numerically analyzed. The study reveals that the magnetostriction effect exacerbates motor vibrations, although the primary source of vibration remains the electromagnetic force.

Conventional three-dimensional packaging chips are based the through-silicon via (TSV) technology. Compared with TSV, inductively coupled interconnect (ICI) ensures reduced costs and increased flexibility. However, some limitations of ICI include high transmission delays and wire-bonding costs. Moreover, it requires the determination of the chip ID during testing. To address these issues, an automatic chip-ID-determining (Auto ID) circuit was combined with a chip edge connect (CEC) technology based on the inter-integrated circuit (IIC) protocol. The experimental results revealed that the CEC technology generated conductive channels at the chip edge, and the Auto ID circuit obtained the chip ID without additional processes. The transmission delay of IIC was one-sixth that of ICI when data were transmitted across 16-layer chips.

Software-defined networking (SDN), with its decoupled control and data planes, offers greater flexibility compared to traditional networks, making it ideal for modern, dynamic network environments. This paper proposes a consistent, protocol-independent, and flexible flow configuration system within a 100Gbps reconfigurable match tables (RMT) pipeline enhancing the capability of the open-source NIC framework Corundum for SDN on Field-Programmable Gate Array (FPGA). Our prototype supports concurrent Write, Read, and Delete operations for exact, wildcard, and stateful matching, with a single thread in our server handling up to 719.42K, 751.88K, and 1.45M flows per second for Write, Read, and Delete operations, respectively, which is sufficient for production networks. Additionally, managing a single flow table requires less than 0.1% area overhead on the Xilinx AU200 platform, and each entry is handled in a single clock cycle without blocking the processing pipeline ensuring timely and consistent flow configurations. In conclusion, our prototype is highly scalable, resource-efficient, and efficient.

This paper introduces a multi-channel programmable gamma voltage generator. Based on the characteristics of the gamma curve, the gamma correction system applies a two-stage adjustment and a three-region output structure. The proposed programmable preset circuit allows the chip to deliver a programmable initial gamma. By integrating a transient enhancement circuit into the Class AB buffer, the amplifier achieves improvements of 27.8%/30.7% in undershoot/overshoot voltages, respectively. It also features a slew rate of 13.4V/µs, a 0.1% settling time of 0.62µs, and a DC gain of 138dB.

A balanced mixer module for operation in the J-band was developed based on Fermi-level managed barrier (FMB) diodes using epi-layer-transfer technique on a Si substrate. Deep dry etching of the Si substrate enabled an arbitrary chip shape to monolithically integrate FMB diodes, waveguide couplers, a 90-degree hybrid circuit, and low-pass filters with better design flexibility. The fabricated module integrated with a transimpedance amplifier, exhibited a double-side intermediate frequency bandwidth of about 43GHz. The lowest noise-equivalent-power obtained was as low as 5×10^-20W/Hz for a local oscillator power of 400µW at a signal frequency of 300GHz.

In this letter, a broadband power amplifier (PA) monolithic microwave integrated circuit (MMIC) based on the 0.25µm gallium nitride (GaN) high electron mobility transistor (HEMT) technology is presented for radar communication. The proposed PA is designed based on the optimal impedance area of the fundamental and second harmonic within the design frequency band acquired by multi-harmonic bilateral pull technique. It can be observed that good efficiency performance can still be expected when the second harmonic impedance includes the resistive-reactive with complex load impedances. As a verification, a wideband PA operating in the S-C band is implemented and measured. The results of measurement indicate that from 2-6GHz, a saturated output power of 46.2-47.1dBm, a power added efficiency (PAE) of 36.3%-47%, and a gain of 21.2-22.1dB can be achieved under a drain voltage of 28V.

Alternating Current (AC) motors play important roles across various fields. Traditional processors utilizing Sine Pulse Width Modulation (SPWM) to control the speed of AC motors face challenges such as high computational complexity and limited flexibility. To address these problems in the variable frequency speed control process of AC motors, this paper proposes the design of a System-on-Chip (SoC) capable of generating SPWM waves. Our proposal is based on an open-source Reduced Instruction Set Computer fifth-generation (RISC-V) processor. The proposal is verified on the Field Programmable Gate Array (FPGA) platform. The results demonstrate that, with the sine wave frequency set to 3200Hz and the triangular wave frequency set to 15KHz, our proposal achieves a 413× acceleration compared to the traditional software method. Compared to the previous accelerator, our design reduces (Look-Up-Table) LUT consumption by 41.10% and (Total Harmonic Distortion) THD by 17.79%.

A two-step fault diagnosis (FD) strategy is proposed for a single-switch open-circuit fault (OCF) of an input-series-output-parallel dual active bridge (DAB) converter in this paper. The strategy localizes the faulty switch through the primary AC voltage and current of each DAB module. The primary fault diagnosis can locate the secondary faulty switch and limit the primary faulty switch in the range of diagonal switches based on OCF characteristics. By actively controlling the relevant switches, the secondary fault diagnosis is carried out to locate the primary faulty switch. Finally, an experimental prototype is built to verify the proposed fault diagnosis strategy.

This paper presents the LED driver with optimized output voltage control technique. The characteristics of LEDs may be affected by usage time, environmental factors, and process variations. Thus, the driving voltage for each LED is not the same, which results in poor efficiency of the LED driver. The proposed LED driver with optimized output voltage control technique can achieve the design target. This work is implemented using TSMC 0.35µm CMOS process with a chip area of about 0.28mm². The maximum driving current of 200mA, and the operating frequency of 100 kHz. The simulation results demonstrate that the output voltage of the LED driver is optimized, which improves the efficiency of the system by about 7%.

This letter presents an integrated high-voltage assisted high-linearity wideband front end for high-speed analog-to-digital converters (ADCs). The on-chip high-voltage generation provides high-swing current injection, which effectively improves flatness within the bandwidth and power supply fluctuation suppression for input buffer. Additionally, modeling and analysis of the front end matching network, along with the inclusion of inductive compensation, increased the bandwidth by 3.1GHz compared to traditional structures. Using 28nm CMOS technology, the front end achieves an SNDR of 56.21dB and an SFDR of 67.48dBc with a FoMs of 163.7dB at a 6GHz input.