Abstract:In fields such as precision measurement and industrial testing, due to the extremely small changes in the capacitance to be measured, which may reach the order of picofarads or even lower, capacitance detection circuits are needed to capture these small capacitance changes. This article analyzes the basic principles of weak capacitance detection, distinguishing five types of capacitance detection circuits: diode type、 frequency modulation type、 AC type、 charge discharge type and resonant type. Through comparative analysis of measurement accuracy、 power consumption and other aspects, the characteristics, advantages, disadvantages and applicable scenarios of each scheme are summarized, and typical application cases are listed. On the basis of summarizing current capacitance detection methods, this paper explores the limitations, directions for further research and development trends of weak capacitance detection circuits, hoping to promote the further development of capacitance detection technology.

Abstract:Failures in vulnerable branches can lead to a decline in the stability of regional integrated energy systems and even trigger large-scale system instability. Therefore, identifying the vulnerable branches of the system becomes a crucial step in ensuring safe and stable operation as well as prevention. To address this, a method for identifying vulnerable branches in electric-gas-thermal regional integrated energy systems is proposed. First, based on the unified energy route theory, unified energy flow modeling and calculations are performed for the electric-gas-thermal network to obtain the initial operating state of the network. At the same time, a set of hypothetical accidents for the regional integrated energy system is established, considering the impact of N-1 contingencies in the thermal and electric networks on the internal network. The post-fault operational state of the branches is then determined through N-1 analysis. Next, vulnerability analysis is conducted based on the operational characteristics of energy flow entropy changes and the structural characteristics of local concentration, and a comprehensive vulnerability index for the branches is proposed. Particle swarm optimization (PSO) is used to determine the weights of operational and structural vulnerabilities, ensuring the objectivity and accuracy of the evaluation process. The branches are then ranked according to the vulnerability index, allowing the identification of the system′s vulnerable branches. Finally, the method is applied to an IEEE 14-bus electric system, a 14-node gas network, and a 14-node thermal network to identify the vulnerable branches of the system. The experimental results show that the proposed method can reasonably reflect the actual conditions of the system, and the weights determined by the PSO algorithm can more effectively identify the vulnerable branches, improving the accuracy and reliability of the identification process.

Abstract:Tire internal defect detection can effectively identify potential issues during the manufacturing process, providing strong support for process adjustments and ensuring driving safety. Defect targets in tire X-ray images are characterized by multi-scale features, extreme aspect ratios, diverse and irregular shapes, a large number of small targets, and an imbalance between positive and negative samples, which results in low detection accuracy. To address these challenges, we propose a tire defect detection method based on an efficient encoder and multi-scale feature fusion. First, an efficient encoder is designed by combining deformable attention mechanisms and channel attention mechanisms to enhance feature extraction and representation capabilities. Then, a multi-scale feature extraction and fusion module is constructed to integrate shallow and deep feature information, preserving critical contextual information and improving feature representation diversity. Finally, an adaptive bounding box regression method is employed during model training to dynamically allocate weights to samples based on difficulty, reducing the impact of invalid samples and achieving faster model convergence while enhancing generalization. Experimental results demonstrate that the proposed model achieves a mean average precision (mAP) of 95.5% on the tire defect dataset, a 3.6 percentage point improvement over the baseline network, thus laying a solid foundation for the practical application of tire defect detection.

Abstract:In order to solve the problem of inaccurate identification of noise source of vehicle steering motor. In this paper, a complete empirical mode decomposition based on adaptive noise and an improved independent analysis method for Salps populations are proposed. Firstly, an independent analysis method for improved Salp population was proposed. The method used improved Tent chaotic mapping to initialize the population, and Logistic chaotic mapping and dynamic learning were used to update the leader and follower, respectively. The simulation results show that the separation efficiency of the proposed method is 4.38% and 1.01% higher than that of FastICA and SSAICA, respectively. Finally, the combined algorithm is used to separate and identify the single channel noise signal of the vehicle steering motor. The results show that the combined algorithm can effectively separate the characteristic signals of different frequencies in the vibration noise signal of the motor. The main reasons for the motor noise under stable working conditions are rotor unbalance and electromagnetic noise.

Abstract:To improve the complex parameter design of traditional virtual synchronous generator (VSG) system and deal with the dynamic oscillation of grid-connected active power during the step of active power instruction, an adaptive optimization strategy on parameters of energy storage VSG based on BP neural network is proposed. Firstly, the working principle and characteristics of the energy storage VSG are described, and the influence of virtual inertia and virtual damping on dynamic response characteristics of the grid-connected active power is analyzed to determine the range of parameters. Secondly, BP neural network with nonlinear mapping performance is introduced and applied to the adaptive design of energy storage VSG parameters to realize real-time dynamic adjustment of virtual inertia and virtual damping parameters, and then optimize the dynamic response performance of active power. Finally, the experimental results show that the active power overshot of the control strategy is reduced from 45% to 2.5%, the adjustment time is reduced by 0.89 s, and the amplitude of the output frequency is reduced by 0.1 Hz compared with the fixed parameter VSG control strategy under the condition of sudden change of active power instruction, which fully reflects the effectiveness and superiority of the strategy.

Abstract:This paper introduces the design of a recyclable pregnancy test circuit based on photoelectric detection technology. The design uses the gold label test paper as the primary sensor and the photoelectric sensor as the secondary sensor, uses the gold label to detect the sensitivity of the color region of the test paper to the specific wavelength of light, and carefully selects the matching light source wavelength. And through the person of different pregnancy chorionic gonadotropin human chorionic gonadotropin (HCG), the concentration of standard sample for calibration, and then realize the gold test color fast and accurate testing results. The circuit parameters are optimized by orthogonal experiment, and the optimal combination is determined by variance analysis and range analysis. The experimental results show that the circuit system designed in this paper is stable and reliable. Compared with traditional pregnancy test paper and disposable electronic pregnancy test products, the detection method proposed in this paper can not only realize non-contact measurement and repeated use, but also has high measurement accuracy and portability, and has broad application prospects.

Abstract:Radome is an important telecommunication and structural component of radar system. If carbon fibers filaments are mixed into the radome accidentally during manufacturing process, it may cause local ablation failure of the radome. Now the heating mechanism of carbon fibers in electromagnetic fields and their impact on the radome are not clear. Therefore, the article intends to explore and study it through finite element analysis (FEA) method. Firstly, the finite element model construction technology of composite plates containing carbon fiber filaments were researched. Secondly, the heating mechanism of composite plates, which had different specifications of carbon fiber filaments, under electromagnetic fields was analyzed. Finally, the corresponding ralationship between the heat generation of carbon fiber filaments and failure modes of composite plate, such as discoloration and ablation, were researched through experiments. The relevant research results have a guiding role in the structural design, production process and fault analysis of radome.

Abstract:In order to develop a new type of power system scheduling model and method under the dual-carbon background, a multi-energy complementary scheduling model with Watershed-type integration of hydro-wind-photovoltaic (WHWP) is constructed by considering stepped carbon trading. In order to improve the solution efficiency and adaptability of such high-dimensional non-convex optimization problems of the WHWP-containing multi-energy complementary scheduling model, this paper proposes an enhanced golden jackal optimization algorithm (EGJO) based on Logistic chaotic mapping, quasi-reflective learning strategy, Gaussian random wandering strategy, and Optimal Individual local search mechanism combined with differential variational perturbation strategy. First, the initialized population is generated using Logistic chaos mapping, which enhances the spatial diversity of the algorithm. Second, by introducing quasi-reflective learning strategy and Gaussian random wandering strategy to update the jackal pair positions in the search, encircle-and-attack phases of the golden jackal algorithm, respectively, the algorithm′s global optimization capability as well as convergence speed are strengthened. Finally, the optimal Individual local search mechanism combined with the differential variational perturbation strategy is introduced after merging the updated positions to improve the solution accuracy. The analysis of the algorithm is carried out in the extended IEEE3 machine 9 node and a simplified power system in a provincial area. The results show that the comprehensive operating costs of the WHWP-containing multi-energy complementary dispatch model considering stepwise carbon trading are reduced by 8.55% and 10.78%, and the carbon emissions are reduced by 178.26 t and 17 841.68 t, respectively; compared with the mainstream seven optimization algorithms, the cost of the EGJO solution is reduced by at least 11 080 yuan and 14.01 million yuan, and the standard deviation of the cost is reduced by 1.598 and 0.004, respectively; fully verifying the effectiveness and superiority of the model and method proposed in this paper. 1.598 and 0.004, respectively; fully verified the effectiveness and superiority of the model and method proposed in this paper.

Abstract:Autism spectrum disorder is a group of complex neurological disorders that usually appear in early childhood. At present, the diagnosis of autistic children mainly relies on behavioral observation and diagnostic scales. However, some behavioral symptoms of children may not be obvious, the diagnosis results are general subjective. In order to improve the accuracy of early diagnosis and identification of autistic children, the paper proposes the diagnosis method based on temporal-spectral-spatial three-domain features and improved fast correlation based filter. Firstly, the complementarity between temporal-spectral-spatial features of EEG signals is used to analyze the brain functional network. Secondly, the improved fast correlation based filter algorithm is used to optimize the features and screen out the relevant but non-redundant features. Finally, BP-Adaboost classifier is used for identification and diagnosis. Through comparative analysis of experiments, it is found that the model has excellent effect, and the BP-Adaboost classifier has a higher identification accuracy, with an average diagnostic accuracy of 98.72%. The model can be used as an auxiliary tool to assist neurologists in diagnosing autism.

Abstract:The UFSA-LD algorithm faces challenges in extracting the thin and long structural features of lane lines, such as information loss, difficulty in capturing long-distance context, and insensitivity to boundary detail recognition. This paper proposes a lane line detection algorithm based on multi-scale atrous feature fusion attention: an MDFA module is added to the UFSA-LD auxiliary segmentation branch, and the receptive field of the network is expanded through atrous spatial pyramid pooling (ASPP) to capture lane features at multiple scales; a fusion channel and spatial attention mechanism (FCBAM) is used to filter out interfering information from channel and spatial dimensions, enhancing the representation of key features. The introduction of the Dice Loss loss function focuses more on the edges and local structural information of the lane lines. Experimental results show that the detection accuracy of the improved model on the TuSimple dataset has been increased from 95.81% to 96.03%; the F1 metric on the CULane dataset has improved by 1.8 compared to the original, validating the effectiveness of the model improvement.

Abstract:This paper addresses the issues of low convergence accuracy, imbalance between global and local search, and the tendency to get stuck in local optima in the Sea-horse Optimizer. An Improved Sea-horse Optimizer based on a hybrid strategy, denoted as ISHO, is proposed. Firstly, the search characteristics of the Grey Wolf Optimizer are integrated to improve the movement behavior of the SHO, enabling more effective global and local searches within the search space. Then, an elitism and reverse learning strategy is incorporated to refine the search process and enhance convergence accuracy. Finally, adjustments are made to the parameters of the predation phase of the SHO to give it stronger adaptability, avoiding premature convergence to local optima. The ISHO is compared with six other intelligent optimization algorithms on eight test functions. Experimental results show that the proposed algorithm has better convergence speed, accuracy, and stability compared to the other algorithms. Applying the improved seahorse optimization algorithm to solve engineering constraint problems further proves the practicality of the improved algorithm.

Abstract:To better meet the diverse needs of industrial production and automation control, and to address the limitations of poor portability and limited functionality in existing general-purpose data acquisition systems, this paper presents the design and implementation of a multifunctional USB data acquisition system based on the STM32F723ZET6 microcontroller. Through the USB interface, the system can be controlled via host computer software to perform 16-channel analog-to-digital conversion data acquisition, dual-channel digital-to-analog conversion output and PWM signal detection and generation. Meanwhile, the acquired or output data is displayed in real-time on the host computer. Practical measurements indicate that the system can achieve a maximum analog input sampling rate and analog output update rate of 1 MS/s. Additionally, the system can detect and generate signal with a frequency up to 1 MHz and a minimum duty cycle of 1%, which ensuring high precision signal sampling and reliable data transmission. This system offers high portability, cost-effectiveness, comprehensive functionality, and meets the requirements of practical applications.

Abstract:Functional near-infrared spectroscopy (fNIRS) is an emerging optical neuroimaging technology that offers a non-invasive, portable, and cost-effective method for monitoring brain activity. Aiming at the motion artifacts caused by the subjects′ head movement, an improved Schr-dinger filter algorithm for removing motion artifacts from fNIRS signals was proposed in combination with mathematical morphology method. The algorithm was applied to the optical density signals obtained from simulation and real experiments, reflecting the changes in hemoglobin concentration of the subjects, and its performance was compared with artifact removal algorithms such as time derivative distribution repair and kurtosis wavelet. The results show that the proposed algorithm can improve the signal-to-noise ratio of the uncorrected signal by 28.66 dB, reduce the root mean square error to 0.06, increase the square of the Pearson correlation coefficient to 0.83, and reduce the peak-to-peak error to 0.05. Compared with other algorithms, it can remove motion artifacts more effectively.

Abstract:Most traditional visual simultaneous localization and mapping (SLAM) systems typically assume a static environment; however, real-world environments often contain moving objects and obstacles, leading to a significant number of mismatched and dynamic points which can degrade localization accuracy. This paper proposes a semantic vSLAM system based on the ORB-SLAM3 framework and deep learning techniques, integrating object detection and optical flow methods to improve localization accuracy in dynamic environments. Firstly, an enhanced YOLOX-S object detection algorithm is utilized to identify potential dynamic targets. Subsequently, a combination of geometric and optical flow methods is employed to precisely detect outliers, with continuous adjustments to dynamic bounding box thresholds based on the motion states of objects and humans. Ultimately, points within static bounding boxes retained in dynamic frames are preserved, while others within dynamic frames are eliminated. The system′s accuracy is evaluated using the TUM and KITTI datasets. Experimental results demonstrate that under highly dynamic sequences, the proposed system achieves an average reduction of 69.26% and 16% in root mean square error of absolute trajectories compared to ORB-SLAM3 and Crowd-SLAM, respectively, and a 15% average improvement in localization accuracy in dynamic scenes when compared to DynaSLAM, thereby validating the enhanced system performance in dynamic environments.Moreover, the results of real-world scene tests demonstrate that the system performs well in various complex environments.

Abstract:Alzheimer′s disease (AD) is a neurodegenerative disease that is a significant contributor to dementia. Accurate diagnosis of Alzheimer′s disease (AD) is of great significance. The integration of multimodal data from Fluorodeoxyglucose positron emission tomography (FDG-PET) and structural magnetic resonance imaging (sMRI) of the brain provides a comprehensive information of lesions from multiple perspectives and enhancing diagnostic accuracy. However, the image data is highly redundant, and the features of the various modes are also significantly disparate. Traditional convolutional neural networks and simple feature concatenation methods are unable to effectively utilize the complementary information of multi-modal data, consequently, this limits the diagnostic performance of AD. To solve this problem, we propose a multimodal image AD classification network combining sMRI and FDG-PET. The network incorporates coordinate attention mechanisms and spatial-channel reconstruction convolution to capture specific regions in images and limit redundant information. A parallel interaction network is also designed, which not only enhances each modality′s own features, but also adaptively adjusts itself according to the features of other modalities, thus realizing effective interaction between modalities. The classification performance of the proposed network is evaluated on the ADNI dataset, and the accuracy, sensitivity, and specificity reach 93.66%、91.67% and 95.41%, respectively, the experimental results show that the proposed network in this paper has a superior performance compared to the existing AD classification networks.

Abstract:Traditional denoising methods for frequency-hopping (FH) signals parameter estimation often fail to effectively preserve the boundaries of FH signals in the time-frequency graph, resulting in low accuracy in estimating the time parameters of FH signals. To address this, a denoising method for FH signals time-frequency graphs based on local energy thresholding is proposed. Firstly, to increase the energy proportion of FH signals in the time-frequency graph after short-time Fourier transform, instantaneous frequency operators are used to mark and remove time-frequency coefficients that do not match the frequency of the FH signals as noise. Then, to avoid losing the energy of the FH signals during denoising, a search window is set to locate the area with the highest energy density in the time-frequency graph, and thresholds are adaptively set for denoising based on the energy distribution in different areas. Finally, a synchronous compression method is used to compress the time-frequency coefficients to the position of the local energy centroid, making the boundaries of the FH signals in the time-frequency graph clearer. Experimental results show that this method can simultaneously improve the accuracy of time and frequency parameter estimation of FH signals when the signal-to-noise ratio is greater than -5 dB, with normalized mean square errors below 0.1 and 0.2, respectively.

Abstract:In view of the challenges of dense small targets, multi-scale variations, and complex weather background interference in roadside perspective target detection tasks, a multi-scale feature fusion and interaction-based target detection algorithm, MF-YOLO, is proposed. Design C2f-CAST, interact and transform features from different subspaces through star operations, and introduce MLCA to capture local, global, channel, and spatial features between distant pixels. Multi-scale information aggregation enhances attention to significant semantic information of occluded objects and eliminates background influence; to address the problem of low efficiency in context information fusion for the neck layer, we add lightweight convolution GSConv to optimize traditional convolution, and design a cross-level partial network module VoV-GSCSP to reduce model complexity and parameter count. Construct a cross-level fusion module SDFM to perform self-calibration on shallow feature maps and fuse semantic information from deep feature maps to solve the problem of missed detection of small targets; finally, the design is based on an adaptive penalty factor, a gradient adjustment function for anchor box quality combined with a dynamic clustering mechanism to improve the WPIoU loss function, enhancing the performance of bounding box regression and detection robustness. The experimental results show that MF-YOLO achieves mAP@0.5 of 85.1% and 92.3% on DAIR-V2X-I and UA-DETRAC datasets, respectively, which is 4.4% and 1.8% higher than the original YOLOv8s, with a reduction of 19.8% in computational complexity and 8.18% in parameter count. The detection speed reaches 152 fps, meeting the real-time requirements.

Abstract:Aiming at the problem that infrared images of substation power equipment are generally blurred and low signal-to-noise ratio, and obvious stitching marks or glistening are easy to appear in the stitching process, an infrared image stitching method of substation equipment based on ROI and improved optimal seam-line was proposed. Firstly, the 50% adjacent area of the two images to be stitching was divided as the ROI to improve the accuracy and efficiency of feature point extraction and matching. Then, the energy difference function of the best suture line was improved by introducing the mean-cosine similarity, so that the best seam-line could accurately traverse the position with the smallest energy difference in the overlapping area. Finally, the experimental verification is carried out to compare the Image stitching effects of three image stitching software Autostitch、ICE and Autopano Giga with the image stitching algorithm based on ROI and improved optimal seam-line in this paper. The results show that the proposed method can effectively improve the efficiency of the overall process of image stitching, effectively reduce stitching traces and glowy phenomena, and achieve high quality infrared image stitching of power equipment.

Abstract:In order to solve the problem of limited model generalization ability caused by imbalanced distribution of target semantic categories in some scarce medical image segmentation tasks, this paper proposes a semi-supervised learning medical image segmentation model CDCL-SSLNet, which achieves feature complementarity through cross-learning of two segmentation submodels with different attributes, namely, UNet and Swin-UNet. The introduction of class distribution fairness factor and class learning fairness factor reasonably weights the loss function, dynamically guides the model to learn the unbalanced data of semantic categories, effectively reduces the learning bias, and then improves the model generalization ability. In the experiment, 5% and 10% of the data in Synapse multi-organ segmentation dataset are selected to simulate labeled data to train the model. When only 5% and 10% of the label data participated in the training, the Dice coefficients of CDCL-SSLNet reached 65.71% and 77.49%, respectively, and the performance of HD95 was 28.97 and 22.07, respectively, and the performance of these two indicators was improved by 17%. The results show that CDCL-SSLNet is able to take into account the accurate segmentation of large and small targets, effectively solves the problem of insufficient model generalization ability caused by the imbalance of class distribution in scarce data, and effectively improves the efficiency and accuracy of medical image segmentation.

Abstract:To address the issues of low detection accuracy, high computational load, large parameter size, and complex variable backgrounds in existing distributed photovoltaic panel defect detection, we propose an improved lightweight YOLOv8 defect detection algorithm for distributed photovoltaic panels. We adopt the efficient lightweight StarNet architecture as the feature extraction network to reduce computational costs and parameter size, achieving a balance between high efficiency and high performance. The SPPF-AM module is designed to enhance the model′s spatial information perception capability, effectively handling targets of different scales. We incorporate the Triplet attention mechanism to effectively extract multi-scale target features, improving the model′s representation ability and task performance. The C2f_DSConv2D, which combines deformable convolution, replaces the original C2f in the network, improving defect detection efficiency with lower storage and higher computation speed. A spatial context-aware module (SCAM) is introduced in the feature fusion network to reduce noise impact and effectively suppress irrelevant background interference. We design ECIoU to replace CIoU, enhancing the fitting ability of the bounding box loss and accelerating the network′s convergence speed. Experimental results show that the improved YOLOv8 model reduces parameter size by 35% and computational load by 29.6%, achieving a detection accuracy of 90.1%, with mAP@50 increasing from 85.9% to 89.7%, an improvement of 4.2%. The improved model demonstrates a certain enhancement in detection accuracy while reducing parameter size and computational load. The proposed improved algorithm demonstrated good performance in defect detection tasks, effectively enhancing the detection capability of the photovoltaic panel defect detection model.