Abstract:

Abstract: Functional electrical stimulation (FES) was an important rehabilitation treatment for stroke-induced disabilities. However, existing FES devices primarily used single-pair surface electrodes to stimulate muscles, making it difficult to precisely locate the optimal stimulation position during limb movements. Additionally, the accuracy of rehabilitation movement control could be affected by internal and external disturbances. To achieve precise control of upper limb movements. A multi-electrode coordinated control method based on active disturbance rejection control (ADRC) was proposed in this study. First, a multi-electrode functional electrical stimulation system for upper limb rehabilitation was designed. Next, an electrode switching strategy was determined by analyzing the dynamics model of the upper limb, dynamically adjusting the electrode combination according to the angle range during angle tracking to track the optimal stimulation position. Subsequently, ADRC was introduced as the controller to estimate and compensate for system disturbances in real-time. Finally, healthy subjects were recruited for simulations and experimental validation. The results showed that the proposed control method achieved precise control of upper limb movements. Compared with the fixed electrode strategy, the average tracking error was reduced by approximately 50.41%, and the root mean square error was reduced by approximately 43.30%. Furthermore, analysis of the subjects′ electromyographic signals indicated that the electrode switching strategy reduced mean absolute value MAV and median frequency MF by approximately 44.21% and 17.97%, respectively, while mitigating muscle fatigue.

Abstract: To address the issues of the lack of three-dimensional depth information and slow convergence speed in monocular vision-based manipulator aiming, a fast aiming algorithm based on monocular visual feedforward approach for a 6D of robotic arm was proposed. First of all, the CSPDarknet of YOLOv4 was cropped through reducing the number of the Cov2D_BN_Mish units and simplifying the complexity of the backbone network so as to form a Lite YOLOv4 algorithm to accelerate object detection speed. Then, the pixel coordinates of the target object detected from monocular images were inversely projected to a 3D ray emitted from the camera center. The 3D ray was exactly the target ray for the aiming laser. What was more, a multiple common points based calibration approach was proposed to obtain the extrinsic parameters of the monocular camera and the aiming laser, and the expected joint poses of the robotic arm were then deducted based on the expected 3D ray coordinates. Finally, direct position control was used to replace visual servoing control based on the expected joint poses to accelerate convergence speed, and realize the monocular visual feedforward based fast 3D aiming for robotic arms. Experiments verified that the average monocular aiming time was 0.611 second. The aiming success rate was 95.238%, which was 4.762 percentage points higher than traditional image based visual servoing.

Abstract: In order to solve the problems of HybridA∗ algorithm′s, such as low search efficiency and long time consumption in high-resolution map of complex scenes, J-HybridA∗ algorithm was proposed after analyzing the factors affecting the search efficiency of traditional HybridA∗ algorithm. Firstly, before HybridA∗ algorithm expanded its nodes, JPS algorithm was used to search the path from the starting point to the end point, and the path was straightened and processed as the basis for calculating the node heuristic value, which reduced the time required for HybridA∗ to calculate the heuristic value when expanding the nodes. Second, a new function is created to calculate all of the node heuristic values before the HybridA∗ algorithm is expanded. This reduces the time needed to calculate the heuristic values when HybridA∗ expands the nodes. Finally, the change of the RS curve from minimum turning radius search to variable radius RS curve search enabled the RS curve to search for a collision-free path earlier, which further accelerated the search of the HybridA∗ algorithm. The results of the simulation demonstrate that the enhanced J-HybridA∗ algorithm reduced the search time by 68% and 21% in comparison to the conventional HybridA∗ algorithm and the reverse HybridA∗ algorithm in straightforward environments. In complex environments, the reduction in search time was 59% and 27%. Furthermore, the planning efficiency is markedly enhanced with the increase of map resolution in diverse resolution map scenarios. Experiments conducted on actual vehicles demonstrated that the J-HybridA∗ algorithm, reduced the search time by 88% and 82% in comparison to the traditional HybridA∗ algorithm and the reverse HybridA∗ algorithm. This effectively enhanced the search efficiency of the HybridA∗ algorithm and reduced the time required for path planning.

Abstract: An improved A∗ algorithm was proposed to address the problems of excessive path turning points, redundant nodes, and susceptibility to local optimum in AGV path planning when the traditional A∗ algorithm was applied in obstacle-dense scenarios. The environment model was constructed using the grid method. Firstly, an obstacle density function K(n) was introduced into the heuristic function to improve the cost function, enabling a more accurate estimation of the actual cost from the current node to the target node. Secondly, a dynamic neighborhood search strategy was adopted to enhance the search efficiency and operational performance of the algorithm. Finally, a redundant node processing strategy was implemented to reduce path turning points and remove redundant nodes, yielding a path that contained only the starting point, turning points, and the endpoint. Simulation experiments were conducted on grid maps with varying sizes and complexities. The results demonstrated that, compared to the traditional A∗ algorithm and other improved A∗ algorithm, the proposed algorithm achieved path length reductions of 4.71% and 2.07%, turning point reductions of 45.45% and 20.54%, and node reductions of 84.24% and 62.45%, respectively.

Abstract: To cope with the issues of low convergence speed and low success rate in the realization of segregating multiple coupled subgroups, a swarm intelligence optimization inspired method was proposed to provide a decentralized navigation method for the robots. The designed navigation algorithm enabled each robot to utilize information of the other robots within a limited sensing range, and calculated the preferred navigation speed for its aggregation with other robots of the same subgroup by following principles of the grey wolf optimization (GWO) algorithm. Based on the inclination of information sharing among different subgroups, the applicable information for each robot was determined separately for the cases of being cooperative or being independent among different subgroups, so that the original centralized GWO algorithm could be applied in a decentralized manner. Furthermore, the original GWO algorithm was improved by using a nonlinearly convergence factor which could improved the exploration ability of the robots. In this way, the chance of encountering with group members for each robot was increased, and eventually improved the success rate of group aggregation. In order to avoid collisions between robots during the moving process, in this study the decentralized optimal reciprocal collision avoidance (ORCA) algorithm was used to adjust the preferred navigation speed of each robot. Simulation results demonstrated the effectiveness of the designed group aggregation navigation algorithm, showing higher success rates, faster convergence speed, and greater stability compared to an existing particle swarm optimization (PSO) algorithm-based method.

Abstract: In order to improve the current situation that the existing gesture recognition algorithms required a large amount of training data, aiming at the drawbacks of low accuracy and complex recognition process, a gesture recognition method for human-machine interaction based on improved GAN model was proposed through taking the generative confrontation networks combined with the variational self-encoder and the label information. Firstly, the improved InceptionV2 and InceptionV2-trans structures were added to the encoder and decoder respectively to enhance the feature recovery ability of the model. Secondly, conditional batch normalization (CBN) was carried out in each component network to improve overfitting, and Mish activation function was used to replace ReLU to improve the network performance. Finally, the experimental results indicated that the proposed method could obtain 100% classification accuracy with fewer samples and short convergence time, which verified the reliability of the method.

Abstract: In order to solve the problem of low management efficiency of classification surrogate-assisted evolutionary algorithms based on fuzzy classification and how to effectively reduce the number of real function evaluations, in this study a fuzzy classification surrogate-assisted evolutionary algorithm was proposed based on an adaptive sampling strategy. Firstly, in the algorithm the agent model was constructed by screening samples through the Pareto dominance relationship. Then, the selection pressure was improved by using a transfer-based density estimation strategy, which balanced convergence and diversity. At the same time, ten-fold cross-validation was used to obtain accuracy information to divide the states. Finally, an adaptive model management strategy was designed. Considering the convergence, diversity, and uncertainty of the current population, targeted sampling methods was adopted according to different accuracy states. The algorithm could ensure overall performance while rationally reducing the number of real evaluations. To verify the performance of the proposed algorithm, it was compared with four other algorithms on the MaF and WFG test sets and real-world engineering problems of automotive side impact design and driving cabin design. The experimental results showed that the proposed algorithm, which could prove that the algorithm had good competitiveness in solving expensive multi-objective optimization problems with the limited number of real evaluations.

Abstract: In view of the fact that the self-supervised remote sensing image retrieval model by using the sample pair for learning due to the lack of labels, resulted in sampling bias and affecting the accuracy of image representation, a self-supervised remote sensing image retrieval model based on cross-quantization and sample correction (CQSC) was proposed. Firstly, in order to reduce the load of data storage and processing, the mapping layer and product quantization in traditional contrast learning were combined to compress the high-dimensional image data and improve the retrieval efficiency. Secondly, the cross-learning strategy was used to maximize the cross-similarity before and after the feature mapping in the retrieval model, and the feature generation ability and retrieval accuracy of the model were enhanced. Finally, design adaptive correction labels to annotate training samples, correct erroneous negative samples during training and address sampling bias caused by missing labels in self-supervised remote sensing image retrieval. Experiments on UCMerced and EuroSAT datasets showed that compared with PLSH, mAP@20 of CQSC improved by 2.52 percentage points on average on UCMerced, and mAP@100 of CQSC improved by 3.83 percentage points on EuroSAT with 64bits.

Abstract: In order to enhance the spectrum efficiency of the RIS-UAV relay communication system formed by integrating UAV and RIS in complex urban environments, research was conducted to explore the optimization of maximizing system downlink sum rate of RIS-UAV relay communication system with fixed trajectory, and a multi-variable non-convex optimization problem of joint active and passive beamforming was constructed. To address this, a lowcomplexity AO dual-loop algorithm based on FP and manifold optimization was designed. Firstly, it was simplified by using the FP algorithm. Secondly, followed by the design of the base station′s active beamforming by using standard convex optimization algorithms, and the design of the RIS′s passive beamforming by using manifold optimization algorithms. Finally, the alternating iterative optimization through the dual internal and external loops continues until convergence is achieved. Simulation results demonstrated that the proposed algorithm had better convergence and lower complexity compared to classical schemes. Moreover, at a maximum transmission power of 20 dBm, the sum rate achieved approximately 6 dB gain over the random phase scheme. Additionally, when the RIS employed 3 bit discrete phase shift, the system′s sum rate performance was nearly equivalent to that with continuous phase shift.

Abstract: To accurately and quickly evaluate the performance of cascade systems and address the shortcomings of the translation invariance in entropy metrics, new performance indicators were proposed based on knowledge related to image quality evaluation in image processing. Additionally, a hybrid sparrow algorithm was introduced to tackle the inaccuracy and slowness of traditional system identification algorithms. Firstly, the Diophantine equation decomposition was applied to the cascade system based on the minimum variance theory to obtain system feedback invariants, thereby establishing the system evaluation benchmark. Secondly, a hybrid sparrow algorithm incorporating sine and cosine concepts was used to identify the system, resulting in the model parameters and the noise probability density functions of the primary and secondary loops. Finally, the new indicators were mixed with entropy metrics to create a more applicable hybrid metric. Simulations were conducted on cascade systems in different noise conditions. The simulation results showed a significant improvement in the accuracy of the algorithm, with a speed increase of 11.98%, and the evaluation results of the new indicators were more reasonable than those of the entropy metrics.

Abstract: To alleviate the congestion problem and improve the reliability of the urban road traffic networks, the cascading failures based on complex network theory were studied in this study. Firstly, a double-layer weighted urbon road traflic networks network model was constructed. Secondly, a failure threshold was integrated into the nonlinear load-capacity model to quantify the probability of failure. Thus, the nodes were also classified into three states: normal, failure and congestion. Additionally, a hybrid load redistribution strategy was proposed to distribute the failure load to the neighboring and sub-neighboring nodes of the failure node considering the spatio-temporal correlation. Finally, the proposed model was simulated by a case study of Nanjing to analyze the impact of different attack strategies and failure thresholds on the reliability of the urbon road traflic networks. The results showed that the urbon road traflic networks had the ability to resist small-scale node and link failures. Attacking high-importance nodes could accelerate congestion diffusion, leading to a decrease in network reliability. During the early stages of congestion propagation (t=40 to 65), increasing the failure threshold δ was able to delay the spread of cascading failure to some extent. However, after congestion reached the critical time, a higher failure threshold (δ= 1.6 compared to δ=1.8) triggered more severe cascading failure.

Abstract: In order to solve the problem of inconsistent state of charge of single cells in battery pack, the active equalization control technology was studied with series connected battery pack as the research object. The research content included the improvement of the balancing topology and the design of the balancing control strategy. Firstly, a new topology was proposed and verified. Secondly, the mathematical model of equalization circuit was established, and the effects of voltage difference and switching frequency on equalization performance were analyzed. According to the results of voltage difference analysis, a multi-cell-to-multi-cell balancing control strategy based on variable duty cycle is designed to improve the equalization speed and consistency of battery pack. Finally, the joint simulation of equalization topology and equalization strategy was carried out in MATLABR2021b/Simulink. The results showed that, compared with the fixed group balancing control strategy, the proposed balancing topology and control strategy could improve the balancing speed and consistency of the battery pack, the time efficiency was increased by 29.71%, the battery SOC variance was reduced by 16.13% and the number of energy transfers was reduced by 52.5%.

Abstract: In order to avoid the phenomena of combustion instability and knock in the process of power generation of large bore low concentration coalbed methane engine, the promoting effect on combustion of hydrogen jet ignition was studied. The influence of ignition-hydrogen injection interval on jet flame structure and combustion-supporting efficiency of engine were investigated. Based on the three-dimensional fluid mechanics software, the effects of different ignition-hydrogen injection matching strategies on the combustion performance of the engine were studied by changing the ignition time with fixed hydrogen injection time and hydrogen injection pulse width. The results showed that the obvious plume jet flame could be formed in the two working conditions of ignition before hydrogen injection (ignition in 5 ℃A before hydrogen injection) and ignition during hydrogen injection, which could improve the combustion rate of the engine. The ignition delay period and combustion duration were shortened by 64% and 68% respectively compared with the original engine. The peak pressure and the peak heat release rate were increased by 2.1 MPa and 1.05 kJ/℃A. Fixing the hydrogen injection time and changing the ignition time, the indicated thermal efficiency of the two working conditions of the ignition before hydrogen injection and the ignition during hydrogen injection was not much different, both of which are about 40%. The efficiency of the two groups of cases that ignition after hydrogen injection were 38.3% and 36.7%, respectively. The indicated thermal efficiency of the engine could be increased by about 4% compared with the original engine by reasonably controlling the ignitionhydrogen injection time.

Abstract: Aiming at the problem of low entrainment ratio of the traditional air ejector, the topology optimization method was used to optimize its structure. And an air ejector with airfoil-like spoiler element was proposed based on the topology optimization results and grag reduction mechanism analysis. The internal flow field of the ejector was studied by numerical simulation, and the influence of the height of the airfoil-like spoiler element and the distance between the element and the nozzle throat on the ejector performance was analyzed. The results showed that the secondary nozzle effect would occur after adding the spoiler element and the primary flow accelerated after flowing out of the nozzle, and the pressure was further reduced. The driving pressure difference of the primary flow was increased, the velocity difference between the two flows was reduced, and the entrainment ratio was significantly improved. The entrainment ratio increased as the height of the spoiler element increased, and the entrainment ratio increased first then decreased with the increase of the distance between the spoiler element and the nozzle throat. The entrainment ratio of the airfoil-like spoiler element ejector could increase by 146%~224% by changing the height of the airfoil-like spoiler element and the distance between the element and the nozzle throat. The entrainment ratio reached the optimal value of 2.07 when the height of the airfoil-like spoiler element was 3.3 mm and the distance between the spoiler element and the nozzle throat was 0.8 mm.

Abstract: In order to improve the imputation accuracy of power load missing value data and guarantee the efficiency of subsequent data analysis and application, firstly, a imputation model based on dynamic fusion of attention mechanism dynamic fusion of attention mechanism imputation model (DFAIM) was proposed. The model consisted of an attention mechanism module and a dynamic weighted fusion module, where the deep correlation between features and timestamps was mined through the two different attention mechanisms of the attention mechanism module. Secondly, feature representations were obtained by assigning learnable weights to the two outputs of the attention mechanism module through the dynamic weighted fusion module. Finally, replacing the values of the missing locations with the obtained feature representations to obtain the imputed values. The proposed model was validated using the meteorological and load dataset and the UCI electric load dataset for an area in New York City, and the experimental results showed that DFAIM had certain advantages in evaluating metrics such as MAE, RMSE, and MRE compared to statistics, machine learning, and deep learning models imputation models.

Abstract: In order to improve the operational efficiency of photovoltaic (PV) power stations a proposal of multi-site forecasting model (MSFM) was proposed to addressing the multi-site location selection problem. In the proposed model, spatiotemporal correlation, event data, and meteorological factors were leveraged to predict power output across multiple sites. A three-dimensional tensor was introduced to represent spatiotemporal data, and tensor decomposition techniques were utilized to recover missing entries. Additionally, the spatiotemporal adjacency, trends, event text data, and meteorological impacts were modeled using both the three-dimensional tensor and the ResNet model. An experimental dataset was established using operational and meteorological data from 1,155 PV power stations in Jinan, Shandong Province. The performance of the proposed method was validated through mean absolute error, relative absolute error, root mean square error, and relative root mean square error, with at least 2.3%, 0.9%, 2.6%, and 2.5% reductions, respectively, in these four evaluation metrics, the experimental results demonstrated that the proposed method was applicable to multi-site location selection problems.

Abstract: The subgrade along the river was affected by unilateral water level changes for a long time, which could increase the risk of overall instability of the subgrade. In order to analyze the seepage characteristics and stability of the flooded widened subgrade with the condition of seasonal change of water level, the seepage characteristics of the subgrade were calculated in real time based on the secondary development of FLAC3D software, and the soil strength parameters were corrected on this basis, and the influence of seasonal change of water level on the stability of the widened subgrade was studied. The results showed that the distribution of pore water pressure and saturation curves of subgrade with horizontal distance was different at the same water level at different times in a certain area near the water surface of the subgrade. During the high water level period, the curve formed a hysteresis circle-like shape, and the pore water pressure and saturation in the middle region were greater than those in the water level rise period. In the low water level period, the curves converge at a certain depth, but the pore water pressure and saturation in the near water area were greater than those in the water level rise period. The safety factor of the frontal slope varied greatly with the seasonal change of water level, with a maximum of 22%, while the backwater surface was less affected by it, with a variation of only 4.5%. The change trend of safety factor of frontal slope was positively correlated with the seasonal trend of water level, while that of backwater surface was diametrically opposite. In the process of seasonal fluctuations of water level, the most unfavorable condition occurred when the water level reached its peak, and the potential sliding surface appeared at the junction of the old and new subgrade on the backwater.

Abstract: The land use pattern with uncertainty influences was deeply studied based on land use suitability. The geographic information system (GIS) and interval probability planning (IPP) technology was combined to build a land use planning management (IPP-LUPM) model. Firstly, land use suitability evaluation was carried out based on GIS data. Secondly, taking the maximum area of different land use types and the constraints as input, the IPPLUPM model was used to optimize land use area and to obtain land use optimum schemes with uncertainties. Taking Henan Province as an example, it was shown by the study results that when the land use suitability i=1, uncertainty probability p=0.01 level,[4.78, 5.55] trillion Yuan could be obtained for the land use benefit, when i=1, p=0.5,[9.66, 10.44] trillion Yuan could be obtained for the land use benefit. The greater the suitability level, the higher the land proposed benefit could be, and the greater the ecological environmental risks would be. It was shown by the results that the model could provide an effective way to solve the land use optimization under uncertainty conditions. It was of important reference value to optimize the land use pattern and management scheme.