Severe bacterial pneumonia is a serious respiratory disease caused by bacteria, which is mainly transmitted through the respiratory tract. To achieve early recognition of severe pneumonia patients through images, this study collected the CT images of 180 patients diagnosed with bacterial infection in the lungs on the day of emergency admission to a large regional medical center (a Top-Tier (Grade 3 A) hospital). After classification by two deputy chief physicians of the respiratory department, 93 cases of severe bacterial infection were obtained and the rest 87 cases were identified as mild bacterial infection. The CT sequences were then preprocessed and annotated to obtain 599 images with annotated lung infection areas. Together with 107 normal (non-infected) images, these bacterial infection images were randomly divided into a training set of 447 and a test set of 259. In the experiment, four deep learning methods, namely, FCN, PSPNet, deeplabv3, and deeplabv3plus, were used for training and three-class classification (severe bacterial infection, mild bacterial infection, and normal). Deeplabv3plus showed the best performance, with an overall accuracy of 96.91% (including a sensitivity of 95.25%, a specificity of 97.24%, an accuracy of 86.96%, a recall rate of 95.24%, and an F1 score of 90.91%) for severe bacterial infection. Using deep learning technology to diagnose severe pneumonia as early as possible can produce valuable treatment time for patients, thereby significantly reducing mortality and complication rates.

This paper presents the application of both genetic algorithm (GA) and artificial bee colony (ABC) method for parameter identification for the chemical hysteresis model. This model is known to be based on physics approaches, and it is characterized by nine parameters, which describe the reversible and irreversible magnetization mechanisms. Splitting the parameter optimization in two parts using hysteresis curves at various amplitudes offers a more efficient way of solving the optimization problem. Based on the root mean squared error between modeled and experimental B-H loops, it has been shown that GA delivers lower errors in shorter time.

This paper proposes an sensorless control strategy to improve rotor position estimation accuracy and system robustness for permanent magnet synchronous motors (PMSMs) under dynamic conditions. By integrating a surface-mounted PMSM (SMPMSM) model with a super-twisting sliding mode observer (STA-SMO), the study achieves reductions in position estimation errors and enhanced noise attenuation capabilities. The system's performance under saturation and cross-coupling effects was validated through element simulations and experimental testing. Furthermore, the integration of parameter identification and computing models demonstrates the system's adaptability in high-noise and non-stationary environments. Results indicate that the proposed method achieves precision rotor position estimation with superior dynamic response and robustness, laying a solid foundation for subsequent research.

This paper introduces a spatiotemporal encoding method based on metasurface that enables precise frequency control and functional switching of radiation beams. The metasurface is configured with subarrays, and each subarray is designed to reflect a specific frequency, thereby achieving unique multi-target signal diversity. By manipulating the spatiotemporal phase of subarray elements, the metasurface can generate far-field radiation patterns with beam characteristics of consistent beam angle at different distances, or beam characteristics of consistent distance with different beam angles. The radiation energy distribution at harmonic frequencies is verified to remain symmetry under various 1 bit spatiotemporal encoding matrices, while the symmetry is verified to be broken by 2 bit spatiotemporal encoding matrices. An optimization method of genetic algorithm (GA) improved binary particle swarm optimization (BPSO) based on 2-bit-coding is thus developed to optimize the spatiotemporal modulation of the metasurface subarray. The GA with the advantage of the crossover mutation operation is utilized to enhance population diversity and thus prevent the algorithm from falling into local optimality with improved search efficiency in high-dimensional discrete space. The optimization method balances different performance parameters and can achieve unique multi-target signal diversity, thereby improving the metasurface's ability to dynamically control and manipulate energy distribution. Using a 1-bit cross-switching mechanism with a duty cycle of 50%, the metasurface can suppress specific harmonic frequencies on the line of sight to less than -60 dBi while keeping the sidelobes below -20 dBi. The technology can precisely control the harmonic energy distribution while allowing beam at specific harmonic frequencies to be absorbed or reflected, which realize advanced breakthrough for effective selective stealth. Simulation results validate the proposed digital encoding optimization method, and the mainlobe gain of the metasurface harmonics is obtained to be more than 20 dBi. This paper algorithmically improves the beam gain of the metasurface and explores the versatile applications of spatiotemporal metasurfaces.

Permanent magnet-assisted reluctance motors (PMa-SRM) feature high energy efficiency, high power density, and a wide speed regulation range. However, traditional direct instantaneous torque control (DITC) strategies for these motors are limited by issues such as high exciting phase current peaks and large torque ripple, which hinder their development and application. To address this, this paper proposes a novel DITC strategy based on current-torque collaborative control. First, commutation intervals are divided according to inductor curve characteristics, with adaptive hysteresis methods applied in different intervals. Then, to tackle high exciting current peaks, current chopping control is introduced, and an adaptive reference current adjustment algorithm is designed to control exciting phase current at the initial commutation stage based on motor speed and load, suppressing current peaks during commutation. Finally, simulations and prototype experiments are conducted on a three-phase 6/20 PMa-SRM. Results show that the proposed strategy effectively reduces current peaks and enhances torque output capability and dynamic response during commutation.

The use of additive manufacturing for the manufacturing of complex materials requires suitable characterization methods. A free-space measurement method is used for the real permittivity characterization. Depending on the considered printing pattern, the experimental result shows good agreement with theoretical values calculated using mixing laws. The setup gives promising results with characterizations of the permittivity, and it highlights the importance of taking into account the printing pattern used according to the desired effective permittivity.

The antenna array structure represents a pivotal technology for synthetic aperture interferometric radiometers (SAIRs). However, current array optimization metrics often have conflicting relationships among themselves, posing a significant challenge to achieving a harmonious balance. To tackle this issue, this paper introduces the point spread function (PSF) into the array optimization process and proposes a PSF-based antenna array optimization method. As a crucial characterization of the SAIR system, PSF can effectively evaluate the SAIR's comprehensive imaging performance. The mainlobe-sidelobe comprehensive quality (MSCQ) is innovatively proposed is proposed as a system-level metric to evaluate PSF quality and guide array optimization. The MSCQ consists of two parts: the main lobe width and the side lobe energy ratio. The main lobe width can evaluate the spatial resolution of SAIR, and the side lobe energy ratio can evaluate the noise performance. In addition, in order to overcome the defect that the traditional optimization algorithm is prone to fall into the local optimum, this paper adopts the improved velocity-paused particle swarm algorithm (VPPSO) for high-precision optimization. The experimental results show that the PSF-based optimized array can effectively enhance SAIR's comprehensive performance and achieve high-performance imaging.

A notched flower-shaped patch wideband MPA antenna is designed with defected ground (DGS) to realize magnified performance for the various wireless application. The several modes effectively excited in antenna and also higher order modes merging efficiently to enable wider impedance bandwidth. The current distribution is improved with notched flower-shaped patch and efficient stimulation of multiple modes, while the enhanced impedance matching and extension of bandwidth are contributed with defective ground. The antenna has overall size measuring length of 1.14λ, width of 1.14λ and height of 0.04λ. The result exhibits a return loss lower than -10 dB across 5.12 GHz to 8.58 GHz, with 5.58 dB peak gain. This range supports wide range of wireless application, such as Wi-Fi 6/6E, Sub-6 GHz 5G NR, short range automative radar and C-band satellite communication. The compact size makes it appropriate for integration on space constrained device.

The interior permanent magnet synchronous generators (IPMSGs) for range-extended electric vehicles (REEVs) have a complex magnetic circuit distribution, which makes it difficult to analyze their electromagnetic characteristics. This paper investigates the electromagnetic characteristics of a combined pole IPMSG topology. An accurate equivalent magnetic circuit model of the generator, taking into account the influence of leakage flux, is established. The magnetic flux equations and precise calculation methods for the main magnetic circuits and leakage magnetic circuits are provided. The accuracy of the equivalent magnetic circuit model is verified using the finite element method. Based on the analysis results, the permanent magnet end of the generator is designed with magnetization isolation. The multi-parameter multi-objective optimization is performed using the particle swarm method to improve the output performance of the generator. A prototype is manufactured and tested. The generator exhibits excellent no-load and load performance, fulfilling the practical requirements and laying a solid foundation for further optimization of the REEV's power system and further improvement of system-level performance.

A polarization reconfigurable metasurface broadband antenna has been proposed. A 2×2 metasurface was used to achieve circular polarization (CP) characteristics, and four resistors were embedded to achieve linear polarization (LP). Among them, characteristic mode analysis (CMA) was used to discover the CP characteristics of the metasurface. Adding resistors changed the direction of the mode current, which causes CP to switch to LP state. The design results were validated through fabrication and measurement. The measured results show that the impedance bandwidth (IBW) is 21.1%, the axial ratio bandwidth (ARBW) 12.9%, the peak gain (PG) 7.7 dBic at 6.8 GHz in the CP state, its IBW 21.0%, the PG 4.7 dBi at 7.0 GHz in the LP state. The proposed antenna has the characteristics of broadband, polarization reconfigurability, easy processing, and low cost, and its operating frequency can be used in the C-band of wireless communication.

The Substrate Integrated Waveguide (SIW) cavity is one of the cavity-supported slot antennas that use substrate SIW technology and 5G application. The proposed model antenna operates at 26 GHz-46 GHz frequency range. The proposed model resonates at 33.84 GHz and 40.28 GHz frequency range. The use of substrate-integrated waveguide (SIW) cavity offers low cost, easy implementation, high gain, and self-consistent electrical shielding. Existing dipole and monopole antennas have some limitations, such as lack of potential ability, low capability and high mutual coupling. The modified semicircle-shaped rectangular patch is intended on a Rogers RT6002 substrate with the dimension of 26 mm × 26 mm × 0.762 mm and dielectric constant of 2.94. To increase the gain of the antenna, a SIW cavity is placed on top of the patch. A Meta Absorber is added on the left and right sides of the patch to reduce electromagnetic interference, provide excellent absorptivity, and improve antenna performance. A coplanar waveguide (CPW) feed line is added to the bottom of the patch. A coplanar waveguide (CPW) feed line offers high-speed data rates and low latency in a 5G mm wave application. The suggested model obtains enhanced simulated and fabricated performances related to traditional antenna models. The suggested model antenna obtains enhanced simulated and fabricated performance compared to traditional antenna models. The fractional bandwidth of the proposed antenna achieves a fractional bandwidth of 59%.

Reconfigurable Intelligent Surfaces (RISs) have emerged as a transformative solution for enabling energy-efficient and interference-aware wireless communication in Sixth-Generation (6G) networks. This work investigates a novel RIS-assisted Simultaneous Wireless Information and Power Transfer (SWIPT) system where separated Power and Information Transmitters (PTx and ITx) independently serve a Power User (PU) and an Information User (IU). A low-complexity deterministic RIS phase optimization strategy is introduced, combining Semidefinite Relaxation (SDR) and Singular Value Decomposition (SVD), to maximize received power at the PU while minimizing interference at the IU. Extensive simulations under both ideal and practical constraints, including 1-bit phase quantization and 3GPP TR 38.901 Urban Micro (UMi) fading, confirm the method's robustness. Results indicate that the proposed design achieves 57.06 dB average received power at the PU and 13.78 dB signal-to-interference ratio (SIR) at the IU in realistic channels, substantially outperforming Accelerated Particle Swarm Optimization (APSO) and fixed-phase baselines. Moreover, Spectral Efficiency (SE) remains above 4.30 bps/Hz at 60 km/h user mobility, showcasing resilience to Doppler-induced channel variation. The proposed approach requires only 52 ms on MATLAB Online using cloud-based Intel Xeon Platinum hardware, confirming its suitability for near real-time applications. Despite these advantages, the design assumes static RIS configuration and perfect channel knowledge. Future work may extend toward real-time RIS reconfiguration and learning-based control under partial channel state information. These findings highlight the feasibility and adaptability of the proposed RIS-SWIPT approach for next-generation wireless systems.

Hyperspectral images often suffer from various types of noise pollution during acquisition and processing, which can significantly affect their application. However, existing denoising methods have limitation in fully utilizing the spatial and spectral correlation of hyperspectral image. In order to take full advantage of the multiscale spatial features and global spectral correlation of hyperspectral image, a hyperspectral image denoising method based on multiscale spatial-spectral feature fusion in frequency domain is proposed in this paper. The proposed method utilizes the structural decomposition of multiscale wavelet transform to transfer the denoising of hyperspectral image to the frequency domain, not only minimizing information loss, but also decomposing noise into small scales, making it easier to remove in the frequency domain. Moreover, a cross-multiscale fusion attention is designed to improve the model performance by considering multiscale information and cross-space learning. A spectral position-aware self-attention module is proposed to more fully exploit the spectral correlation in hyperspectral image. And a multiscale fusion of spatial-spectral feature module is introduced to merge the different spatial and spectral features, thereby enhancing the denoising performance of the model. The experimental results demonstrate that the proposed method outperforms mainstream denoising methods in terms of performance. In addition, it exhibits better visual quality in texture details and edge protection.

In the field of drilling engineering, innovations in drilling communication(also known as hole-ground communication while drilling) technology are crucial for enhancing exploration efficiency, ensuring operational safety, and optimizing data collection. Extremely Low Frequency electromagnetic (ELF-EM) wave communication transmission technology, with its exceptional penetration capability in formations and low attenuation characteristics, is emerging as a key technology in drilling communications. However, this technology faces challenges such as complex transmission model calculations and difficulty in extracting weak signals from the ground, which hinder its further development. Addressing issues like the inability of conventional models to accurately describe non-uniform media, low frequencies, and near-field open-space conditions in ELF-EM transmission under drilling conditions, as well as numerical dispersion, this paper innovatively conducts a comprehensive and systematic analysis of electromagnetic distribution in extended-reach horizontal wells using the finite element modeling and analysis method. Through software simulations and field tests, the following conclusions are drawn: The induced current on the drill pipe plays a major role in the ground field distribution and the signal received by the system terminal; the horizontal drill pipe in a horizontal well has a certain impact on the ground-received signal, mainly manifesting in that the orientation of the ground-receiving electrode should align with the direction of the horizontal well, and the larger the azimuth difference is from the drilling direction, the smaller the signal reception is; at the surface of the drilling platform, not only can multiple electrodes be used to receive signals, but magnetic sensors can also be employed to receive magnetic component signals. Addressing the issue of extracting communication signals in complex electromagnetic environments during electromagnetic measurement-while-drilling (EM-MWD) operations, a multi-channel intelligent signal extraction method has been designed. This method can improve the in-band signal-to-noise ratio (SNR) by more than 3 to 5 dB and further extend the communication transmission distance compared to single-channel models.

This paper presents the design and implementation of a microstrip patch sensor based on a single complementary split ring resonator (SC−SRR), operating at a resonant frequency of 2.44 GHz for salinity detection. The sensor evaluates liquid under test (LUT) by monitoring variations in resonant frequency, reflection coefficient, and quality factor to extract the complex permittivity. The proposed design was simulated using High Frequency Structure Simulator (HFSS) and fabricated on an FR−4 substrate, incorporating a Teflon container to prevent direct contact with the sensing surface. Simulated and measured results exhibit a significant agreement, validating the sensor's performance. A frequency shift of 104.4 MHz was observed as the salinity concentration varied from 0 to 100 parts per thousand, attributed to dielectric perturbation effects. The proposed sensor demonstrates several advantages, including non-contact and non-destructive measurement, reusability, cost-effectiveness, high sensitivity, minimal fluid volume requirement, and reliable accuracy. These features highlight its potential for applications requiring precise and efficient.

This article presents a dipole antenna with added U- and L-shaped stubs designed with copper plates to support frequencies according to the Digital Video Broadcasting (DVB) standard in the 510-790 MHz range. The dipole antenna was placed on a polyester mylar film substrate with a thickness of 0.3 mm and a dielectric constant of 3.2. The CST program was used for simulating the optimization parameters with a size of 270 × 15.5 × 0.59 mm for the use with a flat-screen television. This research uses the technique of reducing signal wave reflection with a 1 × 41 units I-shaped electromagnetic band gap (EBG) strip, which is made from copper plates, placed beneath the antenna with foam sheets as an intermediary. The distance between the EBG plate and the antenna is appropriately 2 mm, with an impedance bandwidth of 46.01% (502-802 MHz) and a unidirectional pattern, resulting in an average gain of 3.62 dBi. For applications with television structures, installing the antenna and EBG plate at the top position can cover the most suitable frequency range, which is 44.32% (504-791 MHz).

The unbalanced rotor mass of permanent magnet assisted bearingless synchronous reluctance motor (PMa-BSynRM) will cause rotor vibration at the same frequency, which has a great influence on its operating performance at high speed. To solve this problem, a control method of unbalance vibration suppression based on variable step size and variable angle search algorithm is proposed in this paper. Firstly, the causes of rotor vibration are analyzed, and the equations of motion of the rotor in the vibratory state are derived. Secondly, the improved Sigmoid function is used to change the step size and angle of the search algorithm, and a fuzzy inference machine is used to adjust the Sigmoid function weights. The vibration suppression control system is constructed, and vibration suppression simulation is performed. The vibration suppression control of the PMa-BSynRM is realized. Finally, the PMa-BSynRM experimental platform is established, and vibration suppression experiments are carried out under the conditions of equal speed and external interference. Experimental results show that the control algorithm can effectively suppress rotor vibration. The unbalanced vibration suppression control proposed in this paper can achieve stable levitation operation of the PMa-BSynRM.

To address the problems of difficulty in adjusting weight coefficients in model predictive torque control of permanent magnet synchronous motors and the large influence of parameters on the motor control performance, a no weighing factor model predictive torque control based on a composite sliding mode disturbance observer is proposed. Firstly, the parallel structure of torque and magnetic chain is designed. The weighting factors are eliminated by choosing a common optimal voltage vector. Secondly, a composite sliding mode perturbation observer is designed to reduce the dependence on an accurate model of the motor. An improved variable gain approximation rate is introduced to eliminate observer jitter. A power exponential term is added to improve the exponential approximation term and to increase the convergence speed of the system state. Finally, the experimental results show that the proposed strategy not only eliminates the cumbersome tuning work of the weight coefficients but also improves the control performance of the motor under parameter mismatch.

This paper proposes a compact stepped-impedance low-pass filter using coplanar open-circuited stubs. The coplanar open-circuited stubs, which are used to compensate for the capacitive effect of the stepped-impedance low-pass filter, are implemented underneath the stepped-impedance low-pass filter. Consequently, the size of the stepped-impedance low-pass filter can be significantly reduced from 11.1 mm × 23.4 mm to 5.6 mm × 9.4 mm without altering its performance, amounting to a reduction rate of 79.83%. In addition, The transmission coefficient is attenuated below -30 dB, which is less than -20 dB attenuation of the conventional stepped-impedance low-pass filter. To verify the simulation result, the conventional stepped-impedance low-pass filter and the compact stepped-impedance low-pass filter using coplanar open-circuited stubs are fabricated and measured where the measurement results agree well with the simulation ones.

Adaptive control techniques are crucial in optimizing the performance of 2.45 GHz microwave transceivers amidst varying electrical parameters. These transceivers, integral to modern wireless communication systems, often encounter fluctuations in operating conditions that can impact signal quality and reliability. Adaptive control mechanisms enable real-time adjustment of transceiver parameters, ensuring consistent and efficient operation across diverse environments. This study addresses the adaptive control of a 2.45 GHz microwave transceiver in the presence of electrical parameter fluctuations, complemented by a multi-band antenna design aimed at minimizing losses. Electrical parameter fluctuations in transceivers can significantly affect performance and reliability, particularly in dynamic environments. The proposed approach integrates adaptive control algorithms to dynamically adjust transceiver parameters in response to fluctuations, ensuring optimal operational conditions. The integrated approach for adaptive control of a 2.45 GHz microwave transceiver, coupled with a multi-band antenna system optimized to reduce total harmonic distortion (THD). The study addresses the challenges posed by electrical parameter fluctuations in transceiver performance by employing adaptive control algorithms that dynamically adjust operational parameters. The multi-band antenna design, optimized through advanced modeling techniques, achieves a THD reduction of up to 20% across different frequency bands. Experimental validation demonstrates significant improvements in signal purity and transmission efficiency, showcasing the efficacy of this integrated approach in enhancing the reliability and performance of microwave communication systems in dynamic environments.