• Current Issue
  • Online First
  • Archive
  • Most Downloaded
  • Special Issue
  • Special Column
    Select AllDeselectExport14
    Display Method:
    2025,17(1):1-12, DOI: 10.13878/j.cnki.jnuist.20240425002
    Abstract:
    With the deep integration of artificial intelligence technology and wireless communication,semantic communication technology emerges as a vital mode focusing on semantic-level information transmission and interaction,thereby significantly enhancing communication accuracy and reliability.In the scenarios of low latency and high traffic density communication applications,semantic communication technology surpasses traditional syntactic-level communication grounded in classical information theory,presenting a new paradigm in wireless communication and expanding the application scope of modern communication technology.However,the development of semantic communication technology is still in its infancy,and the security issues it faces in the application process have not been thoroughly researched and comprehensively analyzed.To advance the development and implementation of semantic communication technology,this paper first provides an overview of various security threats in semantic communication systems;then,it details the research status of model security and data security in semantic communication systems;finally,it summarizes the challenges faced by semantic communication security research and outlooks the future trends.
    2025,17(1):13-21, DOI: 10.13878/j.cnki.jnuist.20240705001
    Abstract:
    This study examined the response tendency of the human brain to a standardized corpus of speech by measuring changes in speech-related electroencephalographic (EEG) signals.Sixteen participants listened to 120 standardized speech items,each lasting 8 seconds,with intervals of 1 to 2 seconds between them and played in a random order.During the listening process,the EEG signals were extracted from the participants,and the signals within the frequency band of 1-40 Hz were preprocessed and analyzed in comparison with the speech signals.The results showed that participants exhibited similar EEG response trends when exposed to the same standardized speech.Furthermore,phase difference analysis between EEG and speech signals was conducted using the phase locking value method,which demonstrated the functional connectivity between EEG signals and speech quality.Notably,the EEG signals achieved a 99.62% accuracy in distinguishing speech quality.
    2025,17(1):22-30, DOI: 10.13878/j.cnki.jnuist.20240705002
    Abstract:
    To address the issues inherent in existing image and video restoration and enhancement techniques,this paper proposes a neural network model approach rooted in semantic feature extraction.Firstly,an image restoration and enhancement framework centered on semantic feature is introduced,followed by the joint optimization of degradation and reconstruction models.The proposed model is validated on a publicly accessible dataset and compared with existing algorithms.The results indicate that the proposed approach achieves a 50% improvement in RankIQA (Rank Image Quality Assessment) scores compared to the state-of-the-art super-resolution algorithm PULSE (Photo Upsampling via Latent Space Exploration).Furthermore,the quality scores of the enhanced images and videos are comparable to those of the original HD ones.In terms of user evaluation,81% of the reconstructed results are considered to be superior to those produced by the comparison algorithms,demonstrating that the proposed approach offers higher quality in reconstructed images and videos.
    2025,17(1):31-41, DOI: 10.13878/j.cnki.jnuist.20240410002
    Abstract:
    To enable harvesting robots to quickly and accurately detect apples of varying maturity levels in complex orchard environments (including different lighting conditions,leaf occlusion,dense apple clusters,and ultra-long-range vision scenarios),we propose an apple detection model based on improved YOLOv8.First,the Efficient Multi-scale Attention (EMA) module is integrated into the YOLOv8 to enable the model to focus on the region of interest for fruit detection and suppress general feature information such as background and foliage occlusion,thus improving the detection accuracy of occluded fruits.Second,the original C2f module is replaced by a more efficient three-branch Dilation-Wise Residual (DWR) module for feature extraction,which enhances the detection capability for small objects through multi-scale feature fusion.Simultaneously,inspired by the DAMO-YOLO concept,the original YOLOv8 neck is reconstructed to achieve efficient fusion of high-level semantics and low-level spatial features.Finally,the model is optimized using the Inner-SIoU loss function to improve the recognition accuracy.In complex orchard environments with apples as the detection target,experimental results show that the proposed algorithm achieves Precision,Recall,mAP0.5,mAP0.5-0.95,and F1 score of 86.1%,89.2%,94.0%,64.4%,and 87.6%,respectively on the test set.The improved algorithm outperforms the original model in most indicators,and demonstrates excellent robustness through comparative experiments with varying fruit counts,offering practical value for applications in addressing the precise identification challenge faced by fruit harvesting robots in complex environments.
    2025,17(1):42-52, DOI: 10.13878/j.cnki.jnuist.20230722001
    Abstract:
    To address the low efficiency and accuracy of manual observation in recognition of crop development stages,a recognition approach based on I_CBAM-DenseNet model is proposed.The approach utilizes a densely connected convolutional network (DenseNet) as the backbone extraction network and incorporates a Convolutional Block Attention Module (CBAM).The Spatial Attention Module (SAM) and Channel Attention Module (CAM) in CBAM are modified from traditional serial connection to parallel connection,and the Improved CBAM (I_CBAM) is inserted into the last dense block of DenseNet to construct the I_CBAM-DenseNet model.Seven important development periods of wheat are selected for automatic identification.To maximize wheat feature extraction,the Excess Green (ExG) feature factor and the maximum inter-class variance method of Otsu are combined to segment the acquired wheat images.The accuracy and loss values of models including I_CBAM-DenseNet,AlexNet,ResNet,DenseNet,CBAM-DenseNet and VGG are compared and analyzed.The results show that the proposed I_CBAM-DenseNet model outperforms other models with a high accuracy of 99.64%.
    2025,17(1):53-62, DOI: 10.13878/j.cnki.jnuist.20240426001
    Abstract:
    To address the issue of inefficient processing of machine vision tasks in industrial environments caused by non-uniform blur in images captured in moving scenes,this paper proposes a motion blurred image restoration approach based on multi-weight adaptive interaction.Firstly,a multi-strategy feature extraction module is employed to extract shallow and critical texture information from blurred images while smoothing noise.Meanwhile,a residual semantic block is constructed to deeply mine the deep semantic information of the images.Secondly,a dual-channel adaptive weight extraction module is introduced to capture spatial and pixel weight information from degraded images and gradually incorporate these information into the network.Finally,a weighted feature fusion module is designed to fuse the multi-spatial weighted features extracted by the network,and multiple loss functions are combined to further improve image quality.The subjective,objective and ablation experimental results of the proposed approach on standard datasets show that the SSIM and PSNR indices reach 0.93 and 31.89,respectively.The modules work well in coordination,exhibiting significant advantages in restoring non-uniform blurred images in moving scenes.
    2025,17(1):63-73, DOI: 10.13878/j.cnki.jnuist.20240515002
    Abstract:
    Artificial Intelligence Generated Content (AIGC) technology offers a wide range of information generation services.However,the accurate assessment of AIGC quality is a critical issue that needs to be addressed.This study delves into the quality of images generated by large models and their evaluation metrics.First,it summarizes common methods for evaluating AIGC from a technical perspective,such as deep learning and computer vision approaches.The study introduces the metrics used in these evaluation methods,including accuracy,relevance,consistency,and interpretability,and examines their performance in evaluating diverse generated content.Then,to demonstrate the practical application of these evaluation metrics,this study conducts an evaluation experiment using images generated by ERNIE Bot as an example.Objective evaluation of the generated images is carried out through quantitative metrics like histograms and noise counts,while subjective evaluation focuses on the overall coordination and aesthetic appeal of the images.Finally,by comparing the results of objective and subjective evaluations,this study identifies highly reliable metrics for evaluating the quality of AIGC images,including color bias,noise count,and psychological expectations.This research provides a theoretical foundation for evaluating the AIGC quality and verifies the effectiveness and reliability of a combined approach using both objective and subjective metrics for AIGC product evaluation through experimental results.
    2025,17(1):74-87, DOI: 10.13878/j.cnki.jnuist.20240404002
    Abstract:
    To address the issues of strong fluctuation of wind power and low prediction accuracy,this paper proposes a hybrid ultra-short-term wind power prediction model that utilizes an improved Dung Beetle Optimizer,namely Logistic-T-Dung Beetle Optimizer (LTDBO),to optimize both the parameters of Variational Mode Decomposition (VMD) and the hyperparameters of Long Short-Term Memory (LSTM) network.Firstly,with the average envelope spectral kurtosis serving as the fitness function,the LTDBO algorithm is employed to optimize the decomposition layers and penalty factors of VMD.Subsequently,the cleaned wind power sequences are decomposed via VMD to obtain the stationary Intrinsic Mode Functions (IMFs) of varying frequencies.Each IMF is then input into the LSTM network,whose hyperparameters have been optimized by LTDBO,for prediction.Finally,the predicted values of all IMFs are superimposed and reconstructed to obtain the final prediction.Experimental results show that the LTDBO algorithm can effectively identify the optimal combination of VMD and LSTM hyperparameters,and the combined model of LTDBO-VMD-LTDBO-LSTM exhibits superior prediction accuracy and robustness in the field of wind power prediction.
    2025,17(1):88-97, DOI: 10.13878/j.cnki.jnuist.20240326003
    Abstract:
    Path tracking precision is fundamental for safe and autonomous driving of intelligent vehicles.To address the problem of chattering in sliding mode control Systems and enhance the control precision of path tracking controllers,a novel PID Integral Sliding Mode Control strategy with Activation Function (PIDSM-AF) is proposed.Firstly,based on a two-degree-of-freedom vehicle model,the vehicle dynamics model is decomposed into a lateral deviation one to establish the lateral control model.Subsequently,an integral sliding mode surface incorporating both heading angle deviation and lateral deviation is constructed employing the extremum method.Considering the system chattering that is difficult to eliminate by general exponential approaching rate,a nonlinear activation function is introduced to adjust the rate when the system state is close to the sliding surface on the basis of an improved exponential approaching rate.This leads to the design of a lateral path tracking controller based on sliding mode control with an activation function.Finally,the improved sliding mode controller is subjected to double lane change tests through Carsim/Simulink co-simulation.The results show that,compared with traditional terminal sliding mode controller,the maximum lateral deviation of the optimized integrated sliding mode controller is reduced by about 64% and 34.9%,and the average lateral deviation is reduced by about 68.4% and 59.7%,under the conditions of low-speed low-adhesion and high-speed high-adhesion conditions,respectively.Furthermore,the optimized controller effectively suppresses the chattering and overshoot changes of vehicle heading angle and front wheel rotation angle,demonstrating strong robustness.
    2025,17(1):98-107, DOI: 10.13878/j.cnki.jnuist.20230907002
    Abstract:
    Here,an event-triggered γ sliding mode control scheme is proposed for the second-order leader-follower multi-agent systems with unknown nonlinear functions.First,a novel sliding mode reaching law based on inverse hyperbolic sine function is selected to ensure that the multi-agent system achieves consensus in limited time.Second,a sliding mode function with gain scaling factor γ is designed and the event triggering mechanism is introduced.Through Lyapunov stability analysis,it is proved that the proposed control scheme is effective,which can not only eliminate the system chattering but also reduce the sampling frequency of the control action.In addition,the minimum lower bound of the triggering time interval is proved,which excludes Zeno phenomenon.Finally,the effectiveness of the proposed scheme is verified by Matlab/Simulink simulation results.
    2025,17(1):108-116, DOI: 10.13878/j.cnki.jnuist.20230915004
    Abstract:
    This paper addresses the current sensorless finite time control for the buck-boost converter with unknown constant power load.The low frequency oscillation caused by negative impedance of constant power loads can adversely affect the stability of buck-boost converters.First,to reconstruct unavailable inductor current and unknown power load,a reduced-order generalized parameter estimation based observer with finite time convergence is designed on the basis of dynamic regression extension and mixing techniques,which is able to reformulate the state observation as parameter estimation.Second,the nonlinear system is converted into a linear one via a feedback linearization approach,and a Fast Terminal Sliding Mode Controller (FTSMC) is designed to stabilize the system.Subsequently,a current sensorless finite time controller is proposed by combining the FTSMC with the generalized parameter estimation based observer.Then the finite time stability of the closed-loop system is proved by the finite time stability result of the cascaded system.Finally,the effectiveness of the proposed current sensorless finite time control scheme is verified by simulation and experiment results.

DownloadsMore+

    Authors CornerMore+

      Search

      • Search by:
      • Search term:
      • from to

      External Links

      Address:No. 219, Ningliu Road, Nanjing, Jiangsu Province

      Postcode:210044

      Phone:025-58731025