Abstract:Infrared cameras are suitable for complex environments, and the use of infrared images to detect black-flying UAV targets has important application value. Aiming at the problems such as small size of UAV target, few pixels in the image, weak texture detail information, and the difficulty of the algorithm to effectively extract the infrared UAV target features resulting in low detection accuracy, this paper proposes a target detection algorithm with multi-scale learning. By constructing a multi-scale feature fusion structure in the neck network of the model, introducing a multi-scale feature learning module, cascading the features of the deep network and the shallow network, acquiring the features of the target at multiple scales, enriching the semantic and feature information of the feature map, the algorithm significantly improves the accuracy of the detection of the target of small UAVs. The SIoU is used instead of the CIoU loss function in the training process, which minimizes the loss of the network model in the training process and improves the regression accuracy. The experimental results show that compared with other infrared small targets and mainstream detection algorithms, the method proposed in this paper can effectively improve the detection accuracy of UAV targets, and can meet the detection accuracy requirements for detecting UAV targets in practical applications.

Abstract:Linear motor traction requires real-time and accurate feedback of rotor position and speed information for closed-loop control, in order to achieve functions such as constant speed cruise and fixed-point parking. However, electromagnetic sleds run in a long distance and non-contact way under the outdoor complex magnetic field environment, making their positioning and speed measuring technology unable to inherit from traditional wheel rail transportation and industrial machine tools. Presently, the main position and speed measuring methods that suit on maglev trains are faced with limitation from their basic principle or high cost. Aiming to solve the above problem, inspired by the structure of vernier calipers, this work proposes and designs a new position and speed measuring system. The principles of improved positioning accuracy, motion direction discrimination and position measurement are elucidated through theoretical analysis. proving the feasibility of the positioning and speed measuring scheme. The position prediction algorithm and Kalman filter algorithm are designed to further improve the accuracy and real-time performance. The corresponding hardware circuit and software program are designed and the positioning and speed measuring system is implemented. And, a synchronous belt guide rail verification platform is built to verify the designed system. Test results show that the system can achieve the designed positioning accuracy, and performs well in terms of real-time performance, accuracy, and engineering application value. Finally, the positioning and speed measuring system is applied to electromagnetic levitation propulsion platform.

Abstract:Fiber hybridization is one of the effective means to improve the toughness and ductility of fiber-reinforced polymer matrix composites, hence avoiding their catastrophic brittle failure. The interlaminar fracture toughness is an important factor affecting the mechanical behavior of fiber hybrid composites. In this paper, two kinds of epoxy resins with different toughness, 7901 and 9A16, were used as the matrix. Interlayer carbon/glass hybrid composites with different numbers of carbon fiber layers were designed and manufactured. The effects of mode Ⅱ interlaminar fracture toughness (GⅡC) on the failure mode and mechanical properties of carbon/glass hybrid composites were investigated through both theoretical and experimental investigation. The results showed that, the higher mode Ⅱ interlaminar fracture toughness was, the more the carbon layer tended to fail in fragmentation, which was beneficial for achieving pseudo-ductility. In addition, the GⅡC on the modulus and strength of hybrid composites was marginal

Abstract:To meet the requirements of lightweight and low error sensitivity in the optimization design of stiffened panels, the optimization design of stiffened panels is carried out considering the twist angle error of stringers. The finite element model of post-buckling instability of stiffened panels under axial compression was established, and the sensitivity of the load-carrying capacity to the twist angle error on stringers and the distribution characterisation of the torsional stringer was analyzed. At the same time, a sequential approximate optimization method based on surrogate model was proposed by using parallel sequential sampling strategy, and the lightweight design of stiffened panel was carried out under the influence of twist angle error of stringers. The optimal results show that, compared with the optimization design scheme without error influence, the optimization scheme considering the twist angle error of stringers has lower sensitivity to the twist angle error when the weight is reduced by more than 32%, which can effectively improve the reliability and engineering application value of the optimized structure.

Abstract:To address the challenges of intelligent diagnosis for open-circuit faults in power electronic inverters, such as the lack of actual fault samples and the issue of varying characteristic adaptability, a set of optimization methods was proposed from two key intelligent elements: data and algorithm, to support the practical applications of intelligent diagnosis for open-circuit faults in power electronic inverters. For the data element, a fault sample amplification method based on inverters′ characteristics was proposed, which finds out the minimum number of practical samples required for model training. For the algorithm element, an attention-enhanced method and a frequency points adaptive training method for the diagnosis model were proposed, which significantly improve model training effectiveness and diagnosis accuracy under wide-frequency inverter operation. The effectiveness of the proposed optimization methods for the intelligent elements was validated by experiments.

Abstract:Permanent magnet synchronous motors are widely used in the field of propulsion motors due to their high efficiency, high torque density, and other advantages. This article focuses on the research of fault diagnosis methods for common stator turn short circuits and rotor eccentricity in surface mounted permanent magnet synchronous motors. Most existing methods are based on stator port voltage and current to extract fault features, but the aggregated parameter information obtained from the motor ports is easily affected by winding structure, pole slot coordination, and other factors, resulting in effective fault signals being overwhelmed and small information dimensions, leading to low signal-to-noise ratio and poor dynamic performance in detection. In order to obtain internal magnetic field information that can directly characterize the state of the motor, and considering the compact structure of high-power density motors, this paper uses a flexible printed circuit board (FPCB) with small space occupation and a large number of turns to make a detection coil, and arranges it in the stator slot to capture magnetic field information. A fast fault diagnosis method based on fault feature extraction is proposed for inter turn short circuit and eccentricity faults. For mixed faults, decoupling of fault diagnosis cannot be achieved through simple coefficient correction. A fault discrimination scheme based on convolutional neural networks is proposed, and the performance of different types of learning methods is compared. The experimental results show that under mixed fault conditions, an accuracy evaluation of about 98% for inter turn short circuits is achieved, and the eccentricity detection error of AlexNet is only about 5% when the training data proportion is 60%.

Abstract:To achieve accurate and stable online identification of inductance parameters for permanent magnet synchronous motors (PMSMs), an online inductance observation method based on virtual voltage vector excitation and current differential response is proposed. Firstly, by introducing the concept of a virtual voltage vector–oriented coordinate system, it is analytically derived and proven that the d- and q-axis inductances of a PMSM can be observed independently of the angular position in the conventional d-q synchronous reference frame. The implementation procedure for extracting virtual voltage vectors and current differential information is discussed in detail, enabling non-intrusive inductance identification without any signal injection. The effectiveness and accuracy of the proposed method are validated by comparison with offline test procedures in IEEE standards.

Abstract:To further optimize the hardware offloading of collective communication based on the NIC(network interface card ) in the "Tianhe" network, and to support more types of collective communication algorithms and larger message sizes, this study investigated the order-preserving triggering mechanism and data buffering methods for collective communication hardware offloading was investigated. An order-preserving triggering mechanism for concurrent multitasking was proposed, which meets the desired semantics of collective communication and ensures the reproducibility of floating-point computation results. A dynamic network data buffering method based on hash tables and pulsed credit flow control was proposed to alleviate the contradiction between limited hardware buffering resources and the high demand for buffering a large amount of network data from concurrent multitasking. Experimental results show that compared with software-based collective communication operations, this workmethod can support the hardware offloading of various algorithms for several typical collective communication operations, with significant performance improvement. Meanwhile, the hardware implementation cost is low, especially with high utilization of buffering resources.

Abstract:To address the demand for base-station traffic forecasting in ultra-dense 5G/6G deployments, this study proposes an Enhanced Randomly Mixed Kernel K-Nearest Neighbors algorithm (ER-MKKNN). By fusing a radial basis function kernel with a white-noise kernel into a hybrid kernel function, it overcomes the trade-off bottleneck between nonlinear relationship modeling and noise suppression inherent in single-kernel approaches. Innovatively introducing dual random subsampling on both samples and features, together with a randomized hyperparameter‐interval strategy, markedly improves generalization stability in high-dimensional, sparse scenarios. A dynamic weight‐allocation mechanism based on out-of-bag (OOB) error inversion enhances the algorithm’s robustness to abrupt traffic fluctuations. The accompanying multi-level parallel architecture offers a scalable prediction solution for ultra-dense network topologies. Experimental results show that ER-MKKNN outperforms the best deep-learning models by 4.6%, 63.5%, and 8.6% on RMSE, MAPE, and MAE, respectively, charting a new technical pathway for intelligent network operations and maintenance.

Abstract:Modular multilevel converters (MMC) experience notable capacitor voltage ripple under low-speed, high-torque operating conditions. While high-frequency injection methods can mitigate the ripple, they can intensify device current stress, losses, and overmodulation risks. Furthermore, current parameter optimization strategies lack dynamic adaptability across all operating conditions. This paper proposes a multi-constrained adaptive optimization strategy for high-frequency injection parameters. First, a reference parameter table was generated using a variable-step gradient descent algorithm based on system characteristics and steady-state models, minimizing injected current while satisfying both capacitor voltage ripple and modulation wave amplitude constraints. Second, an online adaptive correction mechanism was designed to dynamically adjust injection parameters through real-time acquisition of capacitor voltage ripple and modulation information, compensating for parameter deviations and operational condition variations. This hierarchical architecture integrated offline global optimization with online local refinement. Simulation and experimental results confirm that the proposed strategy maintains voltage ripple suppression capability while significantly reducing high-frequency circulating currents, demonstrating dynamic tracking capability for optimization objectives.