摘要 : Vehicle detection and tracking from a UAV perspective often encounters omission and misdetection due to the small targets, complex scenes and target occlusion, which finally influences hugely on detection accuracy and target tracking stability. Additionally, the number of parameters of current model is large that makes it is hard to be deployed on mobile devices. Therefore, this paper proposes a YOLO-LMP and NGCTrack-based target detection and tracking algorithm to address these issues. Firstly, the performance of detecting small targets in occluded scenes is enhanced by adding a MODConv to the small-target detection head and increasing its size; In addition, excessive deletion of prediction boxes is prevented by utilizing LSKAttention mechanism to adaptively adjust the target sensing field at the downsampling stage and combining it with the Soft-NMS strategy. Furthermore, the C2f module is replaced by the FPW to reduce the pointless computation and memory utilization of the model. At the target tracking stage, the so-called NGCTrack in our algorithm replaces IOU with GIOU and employs a modified NSA Kalman filter to adjust the state-space aspect ratio for width prediction. Finally, the camera adjustment mechanism was introduced to improve the precision and consistency of tracking. The experimental results show that, compared to YOLOv8, the YOLO-LMP model improves map50 and map50:95 metrics by 10.3 and 12.2%, respectively and the number of parameters is decreased by 47.7%. After combined it with the improved NGCTrack, the number of IDSW reduced by 73.6% compared to the ByteTrack method, while the MOTA and IDF1 increase by 5.2 and 9.8%, respectively.

摘要 : For better explore the relations of inter-modal and inner-modal, even in deep learning fusion framework, the concept of decomposition plays a crucial role. However, the previous decomposition strategies (base & detail or low-frequency & high-frequency) are too rough to present the common features and the unique features of source modalities, which leads to a decline in the quality of the fused images. The existing strategies treat these relations as a binary system, which may not be suitable for the complex generation task (e.g. image fusion). To address this issue, a continuous decomposition-based fusion framework (Conti-Fuse) is proposed. Conti-Fuse treats the decomposition results as few samples along the feature variation trajectory of the source images, extending this concept to a more general state to achieve continuous decomposition. This novel continuous decomposition strategy enhances the representation of complementary information of inter-modal by increasing the number of decomposition samples, thus reducing the loss of critical information. To facilitate this process, the continuous decomposition module (CDM) is introduced to decompose the input into a series continuous components. The core module of CDM, State Transformer (ST), is utilized to efficiently capture the complementary information from source modalities. Furthermore, a novel decomposition loss function is also designed which ensures the smooth progression of the decomposition process while maintaining linear growth in time complexity with respect to the number of decomposition samples. Extensive experiments demonstrate that our proposed Conti-Fuse achieves superior performance compared to the state-of-the-art fusion methods.

摘要 : This paper investigates the problem of parameter estimation for fractional-order Hammerstein nonlinear systems. To handle the identification difficulty of the parameters of the system and the order, the maximum likelihood and hierarchical identification principles are combined to derive a maximum likelihood gradient-based iterative algorithm. Moreover, to achieve the higher estimation accuracy, the multi-innovation identification theory is introduced, based on which the residual can be formulated as a linear combination of the innovation. Then, a multi-innovation maximum likelihood gradient-based iterative algorithm is proposed, which further improves the innovation utilization. Meanwhile, the computational cost of the proposed algorithm is assessed through the use of flops, which is less than those of its peers. Finally, the convergence analysis and simulation examples demonstrate the efficacy and robustness of the proposed algorithms.

摘要 : In commercial buildings, implementing precooling measures before office hours in summer can effectively meet the thermal comfort needs of employees. However, in multi-zone environments, differences in the cooling rates between regions often exacerbate the heat transfer interference between zones, increasing the complexity of the precooling system and leading to energy waste with limited cooling capacity. To overcome these challenges, we have developed a novel multi-zone precooling control method, which integrates deep reinforcement learning (DRL) to optimize the heat transfer process by adjusting the Air Handling Units (AHUs) valve openings, thus achieving uniform precooling across the building. Comparisons with traditional precooling control methods demonstrate the effectiveness of the proposed method. The results show that, under conventional conditions, compared with the rule-based control (RBC) and proportional integral derivative (PID) methods, the precooling time is reduced by 11.4% and 5.8%, respectively, the complexity of heat transfer is reduced by 77.6% and 64.1%, and energy consumption is reduced by 14.5% and 9.3%. In addition, the study analyzes the influence of environmental parameters on precooling optimization. The findings indicate that weather conditions have the most substantial impact on short-term precooling performance, followed by building thermal performance and cooling conditions.

摘要 : The finite-time cluster synchronization (FTCS) of fractional-order complex-valued (FOCV) neural network has attracted wide attention. It is inconvenient and difficult to decompose complex-valued neural networks into real parts and imaginary parts. This paper addresses the FTCS of coupled memristive neural networks (CMNNs), which are FOCV systems with a time delay. A controller is designed with a complex-valued sign function to achieve FTCS using a non-decomposition approach, which eliminates the need to separate the complex-valued system into its real and imaginary components. By applying fractional-order stability theory, some conditions are derived for FTCS based on the proposed controller. The settling time, related to the system's initial values, can be computed using the Mittag–Leffler function. We further investigate the optimization of control parameters by formulating an optimization model, which is solved using particle swarm optimization (PSO) to determine the optimal control parameters. Finally, a numerical example and a comparative experiment are both provided to verify the theoretical results and optimization method.

摘要 : This paper focuses on the problem of dynamic event-triggered predefined-time adaptive attitude control of quadrotor unmanned aerial vehicle (QUAV) suffering from unknown deception attacks. The command filter is utilized to avoid the "explosion of complexity" problem, while concurrently eliminating the effect of filtered error by constructing the fractional power error compensation signals. By using the Nussbaum gain technique, the unknown control coefficients generated by unknown deception attacks have been resisted. A dynamic event-triggered predefined-time attitude control scheme is proposed by introducing internal dynamic variables, which reduces the data transmissions and avoids the Zeno behavior. It is proved that the closed-loop system is practically predefined-time stable, and the attitude of QUAV is driven into a small region near the origin in a predefined time. Finally, a simulation example is provided to show the effectiveness and superiority of the developed predefined-time attitude control algorithm.

摘要 : Background: Excessive alcohol consumption poses a significant threat to human health, leading to cellular dehydration, degeneration, and necrosis. Alcohol-induced cellular damage is closely linked to alterations in cellular mechanical properties. However, characterizing these changes following alcohol-related injury remains challenging. Moreover, current research on single-cell mechanics often struggles to culture and measure cells within a controlled microenvironment, leading to complex experimental procedures and imprecise results. (63). Results: In this study, we developed a novel single cell measurement method that combines cell microculture in alcohol-containing solutions with cytomechanics assessment within microdroplets. This approach integrates key operations, including single-cell encapsulation and culture in droplets, droplet reinjection, and cell deformation analysis within droplets, enabling high-throughput and multi-parameter quantification of single-cell mechanical properties. The use of droplets provides a precisely regulated microculture environment, effectively avoiding channel clogging issues. Additionally, the integration of cytomechanics measurement simplifies the analytical process by eliminating the need for complex techniques within the droplets. Gastric mucosal epithelial cells (GES-1) and human umbilical vein endothelial cells (HUVECs) were selected as models for ethanol-induced injury to validate the proposed technique. The results demonstrate a bidirectional response in cellular deformability following ethanol treatment, with cells becoming stiffer at lower ethanol concentrations and softer at higher concentrations. (136). Significance: The integration of droplet microfluidics and cell mechanics offers a powerful platform for investigating the underlying mechanisms of ethanol-induced cellular damage. This approach is also applicable for studying changes in cellular mechanical properties by precisely modulating the microculture environment, providing a reliable tool for drug screening and disease modeling in biochemistry and biomedical engineering. (54).

摘要 : Document-level Relation Extraction (DocRE) aims to extract semantic relations between entity pairs, spanning multiple sentences, paragraphs or even the entire document. These relations can often be predicted by partial sentences within the document, the evidence sentence. However, the relation derived only from sentence information is incomplete, because it ignores the case of multiple relations between entity pairs. Therefore, how to select effective evidence sentences and how to predict multiple relations more accurately have become challenges for the existing DocRE models. In response to these challenges, we introduce Reinforcement Learning (RL) to select more effective evidence sentences, while using heuristic rules to narrow down the search space of RL. Secondly, we utilize GAT to acquire the features of co-occurrence relations, which can greatly improve multiple relations prediction performance. Moreover, the combination of the features of co-occurrence relations and the evidence sentence information enables our method to achieve both high effectiveness and precision. The experimental results show that, compared with other advanced methods, our method achieves an F 1 score of 66.56 and the E v i F 1 score of 56.69, which attains the state-of-the-art performance on public datasets.

摘要 : In order to regulate the load peak of households and achieve energy conservation, this study proposes a household energy management system (HEMS). The proposed HEMS embeds the Self-attention mechanism in the LSTM network to predict the load demand accurately for the next time step. Based on the prediction information, the HEMS optimize the control of household energy storage devices by deep reinforcement learning (DRL) in real time. According to the experimental results during two testing periods, the HEMS reduces peak load by 19.85 % and 26.38 %, and reduces energy consuming by 26.28 % and 22.08 %, outperforming other predictive control frameworks. Additionally, it achieves 31.9 % reduction in electricity costs. It can be seen that the optimal control of energy storage devices by the proposed HEMS through the predictive control framework is effective for achieving household load regulation and energy conservation.

摘要 : This paper presents a path planning method based on an improved simulated annealing (SA) for multi-robot navigation in a 2D plane. The method can achieve collision-free and efficient movement in environments where dynamic obstacles exist. To address the problem of considerable computational effort of general heuristic algorithms, this study improves the running process of the algorithm so that it can lock the optimal path in the process of searching for a path at a very fast speed. In addition, a prioritisation strategy is proposed for the problem of difficult coordination among multiple robots.The method has a large improvement in the coordinated operation between individual robots. Simulation tests show that the proposed method can coordinate multiple robots to avoid collisions, whilst effectively avoiding local minima and completing the task in the shortest possible time. Compared with other algorithms, the advantages of the improved SA are more obvious, and the path length obtained is about 10% shorter than other dynamic path planning algorithms, and the success rate can reach 100%.