Abstract:

Abstract: Aiming to address the problem of multi-scale feature and texture detail information loss in the process of multimodal medical image fusion, a novel image fusion algorithm based on generative adversarial network (GAN) and multi-scale spatial attention mechanism was proposed. Firstly, the generator adopted an autoencoder structure to extract, fuse, and reconstructed the input images using an encoder and a decoder, generating the fused image. Secondly, the entire GAN framework employed a dual discriminator structure, enabling the generator to preserve salient features from multiple modal images in the fused image. Finally, a multi-scale spatial attention mechanism was constructed as a fundamental module for feature extraction in the encoder. It could effectively capture and retain multi-scale features from the source images, and incorporate spatial attention mechanism to better preserve the structures and details of the source images. Experimental results on the Whole Brain Atlas database demonstrated that the fused images generated by the proposed algorithm could exhibit richer texture details, enhancing human visual observation. Furthermore, the algorithm outperformed other advanced algorithms in such objective evaluation metrics as average gradient, peak signal-to-noise ratio, mutual information, and visual information fidelity for three different types of medical image fusion tasks, with average values of 0.302 3, 20.720 7, 1.441 4, and 0.649 8, respectively. Thus, the proposed algorithm demonstrated a certain advantage over other advanced algorithms.

Abstract: The tortuous and complex environment of the tracheobronchial tree poses significant challenges for force sensing at the distal end of flexible robots used in pulmonary interventional surgeries, with conventional force sensors often falling short of the required performance. To address the specific needs of these procedures and to equip flexible robots with precise and real-time force feedback, a segmented three-dimensional force fiber Bragg grating (FBG) sensor was designed. The sensor′s structure was meticulously analyzed and optimized through finite element simulations, and a decoupling algorithm was devised to separate the wavelength shifts of the four internal gratings with three-dimensional forces, facilitating independent detection of lateral and axial forces. In order to mitigate the effects of temperature variations on sensing accuracy, temperature-compensating gratings were incorporated into the design, accompanied by a carefully tailored compensation strategy. Following this, the designed FBG force sensor underwent calibration and experimental validation. The outcomes revealed that the sensor exhibited lateral force sensitivities of 431.3 pm/N and 517.6 pm/N, an axial force sensitivity of 153.5 pm/N, and root mean square errors of 0.026 N, 0.025 N, and 0.041 N for the respective force dimensions.

Abstract: To solve the hybrid flowline energy-saving scheduling problem with two intermediate buffer constraints, infinite buffer and blocking, between production stages, a mathematical model was formulated by considering the uncorrelated parallel machines and multiple time constraints. Taking into account the characteristics of the problem, an improved multi-objective memetic algorithm (IMOMA) was proposed to minimize simultaneously makespan and total energy consumption of the machines. The algorithm adopted a matrix encoding method based on uncorrelated machine assignment. Using a hybrid initialization strategy based on Tent chaotic map to generate the initial cell array, an non-dominated sorting genetic algorithm improved by parameter-based adaptive genetic strategy was applied for the global optimization operator, and a search strategy integrating adaptive selection neighborhood search and multi-objective simulated annealing was designed for the locally enhanced search operator to improve the algorithm′s search capability. The effectiveness and superiority of the proposed algorithm were verified through case experiments with 24 problem scales. The experimental results showed that the average IGD value of 47.89 and the average SP value of 857.25 obtained by IMOMA within the average running time of 241.26 s were lower than the other three comparison algorithms. So the solution set obtained by IMOMA had better convergence, diversity and distributivity.

Abstract: Speed and area are two basic indexes to evaluate the performance of multiplier unit. Aiming at the problem of balancing layout area and transmission delay in current multiplier design, a new 16-bit signed fixed-point multiplier was designed by using Radix-4 Booth algorithm. In the process of partial product generation, firstly, the complementary code circuit for multiplier was improved, and then the modified Booth encoder and decoder with Radix-4 were optimized. The structure used less logic gate resources and was easy to parallelize the input bits. In the Wallace compression circuit, the symbol extension bits were preprocessed and a new compressor structure was designed to optimize the whole Wallace compression module. In the second stage of compression, the ripple carry adder structure was used to calculate the high bits in advance, which reduced the sum bits of multi-bit pseudo-sums. In the summation circuit, a two-stage carry look-ahead adder structure was used to shorten the transmission delay of the critical path and avoid increasing the chip area, thus improved the running speed of the multiplier. Compared with the existing multiplier structure, the new fixed-point multiplier reduced the area by 12.0% and the delay by 20.5%.

Abstract: Aiming at the problem that disguised botnet hosts are difficult to detect, a botnet detection method based on graph reconstruction and subgraph mining (GR-SGM) was proposed. Firstly, network data was converted into graph data which was reconstructed to enhance the host node feature representation. Then, based on the topological structure, node characteristics, and position changes in the reconstructed graph, a botnet subgraph scoring function was designed. In this way, the camouflaged features were captured, the botnet subgraph was extracted, and the original and reconstructed graphs were pre-detected to improve detection accuracy and efficiency reducing reconstruction errors. Finally, the pre-detection results and botnet subgraphs were comprehensively scored to obtain complete botnet information. Experimental results on the ISCX2014 botnet dataset and CICIDS2017 botnet dataset showed that the detection accuracy of GR-SGM was 99.98% and 99.91%, respectively, and the F1 reached 99.94% and 99.65%, respectively. Compared with other botnet detection models, GR-SGM could identify botnet nodes more efficiently and accurately, while having a lower false alarm rate.

Abstract: The k-center problem is fundamental in facility siting, and is also an NP-hard problem with practical applications in the fields of distribution, emergency services, etc. However, with the expansion of the problem scale, the original algorithms were no longer applicable and should be further optimized or improved. In order to find an efficient algorithm to solve the problem, the existing algorithms were studied. The algorithms for the k-center problem were classified into exact algorithms, heuristic algorithms, meta-heuristic algorithms, approximation algorithms, etc. The algorithms were compared in terms of their principles, ideas for improvement, performance, accuracy, etc. The exact algorithm obtained the optimal solution in polynomial time when solving small-scale k-center problems, but was inefficient and not applicable to large-scale problems. The heuristic algorithm could give the relative optimal solution in polynomial time, but there was no theoretical guarantee to measure the relationship with the most optimal solution. The meta-heuristic algorithm could be improved according to the existing intelligent optimization algorithms to give the relative optimal solution, but the quality of the solution could not be guaranteed. The solution obtained by using the approximation algorithm had the guarantee of the approximation ratio, which was of greater theoretical research value, but the practical value was weaker. At present, the meta-heuristic algorithm for solving the k-center problem achieved certain research results, but there were still breakthroughs in the solution time, solution scale and algorithm efficiency, which could be the focus of the future research on the k-center problem.

Abstract: It is crucial to studying the propagation patterns of urban traffic congestion in alleviating traffic problems. A propagation model of urban traffic congestion based on the SIR to analyze the propagation process of urban traffic congestion was established. Firstly, the road dual topology network was constructed according to the actual urban road network. The traffic congestion propagation model was developed based on the SIR. Considering the complex network characteristics of road networks and the relevant attributes of roads themselves, the random forest algorithm was introduced to compute the corresponding weights, determining key parameters such as the propagation rate in the congestion model. Subsequently, using a specific area road network in Qinhuai District, Nanjing City as an example, a city road dual topology network with 69 nodes and 163 links was constructed for simulation experiments. The results indicated that road node degree and road saturation were key factors influencing the spread of road congestion. The impact of road node degree was relatively small, with the propagation range increasing by within 5% and the recovery time being affected by around 10%. In contrast, the impact of road saturation was relatively large. As saturation increased, the propagation range could grow by up to 40%, and the recovery time could be affected by around 20%.

Abstract: For the problems of gradual aging of reservoir dams and increasing risks and operation costs, a model to quantify the comprehensive costs and benefits of reservoir dam operation considering potential dam breach losses was constructed. Firstly, the impact of potential dam breach was considered comprehensively, the interaction relationship between the cost and benefit influencing factors of reservoir dam operation according to the ISM theory was clarified, and the quantitative indicators of the comprehensive cost and benefit of reservoir dam operation were determined. Secondly, the degree of interaction between the factors was analyzed based on the DEMATEL method, which clarified the relative importance between each cost and benefit indicator. Finally, the comprehensive cost and benefit of reservoir dam operation with potential dam breach losses were calculated in combination with the technoeconomic theory. The model was applied to a reservoir in China. The comprehensive cost and benefit of the reservoir operation were calculated, and its economic benefits were analyzed. The results showed that the comprehensive cost of operating this reservoir was 7 436 200 Yuan while considering of potential dam breach losses, the comprehensive benefit was 109 118 300 Yuan, and the benefit-cost ratio was 14.67. The comprehensive benefit of operation was much larger than the comprehensive cost of operation, the revenue-generating capacity was good, and the economic benefits were significant.

Abstract: In order to study the influence of pipe jacking method on surface settlement of subway contact channel in water-rich silty sand layer, the surface settlement of contact channel with 15 m depth and 7.5 m length in 08 work area of Zhengzhou Metro Line 12 during pipe jacking construction was monitored, and the spatial and temporal evolution law of surface settlement during pipe jacking construction was analyzed in combination with 3D numerical simulation of "stratum-main tunnel-inner support-contact channel " system. The surface settlement increased gradually with the pushing of pipe jacking machine, and reached the maximum settlement value after pushing in place, in which the settlement increment produced in the stage of trolley positioning, starting hole breaking and receiving hole breaking was obviously larger than that in other excavation stages. After the equipment was dismantled, the stratum would rebound a certain amount due to unloading effect, and the settlement value would decrease. The surface settlement curve was funnel-shaped, symmetrical along the direction perpendicular to the center line of the contact channel, asymmetrical along the direction parallel to the center line, and the maximum settlement value was located 3 m away from the center of the contact channel and close to the starting end in the longitudinal direction. The horizontal influence range of the contact channel pipe jacking method on the surface was larger, and the concentrated area of the surface settlement was approximately elliptical, and its long axis was perpendicular to the axis of the contact channel.

Abstract: In order to accurately predict the surface settlement of the shield tunnel in the silty clay and sand gravel composite formation in Luoyang, a linear regression fitting method was proposed. To fit a certain section of Luoyang subway by the monitoring data of surface settlement during shield tunneling, maximum surface settlement correction factor α and settlement tank width correction factor β were brought out modifying the single-line Peck formula. The results indicated that for single-line tunnels, the linear regression method used to fit the surface settlement curve of the tunnel cross-section was in highly consistent with the actual monitoring data. When the correction factor α within the range of 0.4 to 0.9, and β within the range of 0.5 to 0.7, the corrected Peck formula obtained the highest accuracy. For double-line tunnels, the surface settlement under the formation was basically in a bimodal "W" shape. Double-line tunnel’s soil loss rate η and the distribution range of the width coefficient of the settlement tank K was 0.2% to 0.8% and 0.2 to 0.5, respectively and the mean values were 0.45% and 0.37, respectively. The average soil loss rate of the leading line and the average width coefficient of settlement tank was 1.4 times and 1.3 times of the rear line, respectively. The research results provided data support for the subsequent construction of the shield tunnel in the typical silty clay and sand gravel composite formation in Luoyang.

Abstract: Urban flood monitoring is crucial for disaster management and emergency response. Single data sources often have their own shortcomings, which are not conducive to the multi-dimensional and all-round analysis of complex urban flooding disasters. The GF-3 SAR images were combined, which were not affected by clouds and rain and had a large coverage, with real-time social media data to establish a flood monitoring method for the downtown area of Zhengzhou City during the "7·20" heavy rainstorm. Threshold segmentation and random forest were used to extract water bodies from GF-3 SAR images before and during the flood, the most accurate water body extraction results and flood monitoring results were obtained, and the effect of water body extraction in typical areas was analyzed. Using Python tools,the obtained social media data about urban waterlogging were processed, visualized and spatially analyzed. GF-3 SAR images and social media data flood monitoring results were combined to explore the complementary advantages of the two. The results showed that the overall extraction accuracy of SAR water bodies was in the order of random forest, Otsu threshold method, and water body index method. However, in some typical regional analyses, the extraction effect of random forest was lower than that of the other methods.The flood inundation range extracted based on SAR images was mainly concentrated in the urban fringe areas outside the Third Ring Road and around large water bodies.The flood information extracted based on social media data was mainly concentrated in the Third Ring Road, where the urban population was densely populated with buildings.

Abstract: The design and simulation of a vehicle-mounted air-cooled lithium battery pack for new energy vehicles were presented. Finite element simulation was employed to analyze the thermal management system of the air-cooled lithium battery pack. The results indicated that with an inlet air velocity of 2 m/s, the maximum temperature differences for the parallel and serpentine structures were 6.96 ℃ and 6.29 ℃, respectively, with a reduction in the maximum temperature difference for the serpentine structure. Introducing flow baffles to create turbulent structures reduced the maximum temperature difference to 5.69 ℃ compared to the serpentine structure, representing a decrease of 0.60 ℃ in the maximum temperature difference. When the inlet air velocity exceeded 4 m/s, the maximum temperature difference was below 5 ℃, meeting the optimal discharge efficiency of lithium batteries. The battery pack exhibited optimal cooling performance when the flow baffles were arranged in a symmetric structure, with significantly lower maximum temperature and temperature differences compared to other layout arrangements. Furthermore, the inlet air velocity was positively correlated with the cooling performance of the battery pack. The optimal cooling effect was achieved when the inlet air velocity reached 6 m/s, after which the change in cooling performance diminished with further increases in inlet air velocity. In conclusion, the designed 18650 lithium battery pack demonstrated optimal flow disturbance cooling performance with symmetric layout spacing, and controlling the inlet air velocity above 4 m/s enabled the battery to operate at optimal discharge efficiency.

Abstract: Even with the advantages of small size, light weight, and easy matrixing, flexible ultrasonic sensors based on polyvinylidene fluoride (PVDF) piezoelectric film are susceptible to back clutter and impedance mismatching, resulting in a complex frequency component of the return signal and a low signal-to-noise ratio. In order to effectively absorb the stray waves at the back of the sensor and improve the ability of unidirectional radiation of ultrasonic tungsten powder and silica gel were used to design the backing layer for flexible ultrasonic sensors, which could increase the signal voltage amplitude of the sensor by 200% while maintaining the overall flexibility of the transducer. Secondly, according to the electrical characteristics of the flexible ultrasonic sensor, an impedance matching method for flexible ultrasonic sensors was proposed, and the best matching parameters were obtained through experiments, so that the signal-to-noise ratio of the sensor was increased from 2 dB to 30 dB. The optimized flexible ultrasonic sensors had the excellent performance of wide bandwidth, high frequency, narrow pulse. The experimental results of the measurement of the thickness of pipe test specimens with different curvatures and wall thicknesses showed that it could reach the accuracy of 0.01 mm.

Abstract: In response to the issue of artifacts in the maximum amplitude images in laser ultrasonic (LU) defect detection, principal component analysis (PCA) was integrated with two unsupervised machine learning algorithms including local outlier factor (LOF) and isolated forest (IF) to perform unsupervised anomaly detection on LU data. Firstly, the PCA algorithm was used to reduce the dimensionality of the LU data, simplifying its complexity. Secondly, the LOF and IF algorithms were employed to identify outliers in the data, and the thresholds for these outliers were determined using the cumulative distribution function and kernel density estimation. Finally, a comparison of the detection results from the LOF, IF algorithms, and the maximum amplitude images revealed that the LOF algorithm offered superior defect detection precision and a lower false positive rate.

Abstract: To reflect the influence of intense momentum exchange of the fluid in the annular jet pump suction chamber accurately, a comparison calculation using different turbulence models and wall functions was conducted, and hydraulic experiments were carried out to verify the results. The results showed that different wall functions had little effect on the performance and wall pressure coefficient calculation of the annular jet pump.Compared to the RNG k-ε (RNG), Standard k-ω (SKW), SST k-ω (SST)models, the results calculated using the Realizable k-ε (RKE) model or Reynolds stress model (RSM) combined with the scalable wall function (ScWF) was more consistent with the measurements.By comparing the simulated flow field, it was found that the performance prediction results of different turbulence models with low flow ratio conditions were related to the distribution range of the recirculation zone simulated. The larger the range of the recirculation zone simulated, the lower the predicted performance of the annular jet pump had. Based on the analysis of entropy production theory, it was believed that the energy loss of the annular jet pump was mainly caused by the turbulent entropy production distributed in the wall and the mixed shear layer. With high flow rate conditions, the high turbulent entropy production rate calculated by the RSM model was the main reason for its lower prediction performance compared to the RKE model.

Abstract: Based on the influence of focality on transcranial magnetic coil, a single-channel double-trapezoidal and double-layer coil with high focality was designed to address the insufficient commercial use. Firstly, the ball-head model was used to analyze the coils with same structure but different sizes, and the better coil size was obtained. In order to further improve the focality, the differences of medium-size coil between five different structures of doublelayer coils and two commercial figure-8 coils were analyzed, and the optimal coil structure was obtained. Finally, 50 groups of real brain models with individual differences were used for simulation verification. The simulation results indicated that when using the ball-head model, the optimized coil improved the focality by 69.48% and reduced the depth by 27.18% compared with the 70 mm figure-8 coil, the focality was improved by 44.78%, and the depth was reduced by 8.5% compared with 25 mm figure-8 coil. When using 50 groups of real brain models, the focality of the optimized coil was improved by 62.07% and the depth was reduced by 25.71% compared with the 70 mm figure-8 coil, the focality was improved by 39.49%, and the depth was reduced by 9.5% compared with 25 mm figure-8 coil. Experiment results of two models tended to be consistent, which confirmed the reliability of the simulation and also proved that the optimized coil had stronger stimulation intensity and focality, which could greatly improve the safety of TMS treatment and reduce discomfort. At the same time, the single-channel design was easy to implement and had high performance.

Abstract: Aiming at the problem of larger torque ripple of switched reluctance motors, a compensation cosine torque sharing optimization control strategy for switched reluctance motors was proposed by improving the traditional cosine torque sharing function. Based on the steady-state deviation between the actual torque and the desired torque, the proposed method introduced a Fourier approximation curve to compensate the distributed torque. A power conversion circuit with double-off-angle drive control was used to achieve zero-voltage freewheeling and improve the tracking ability of the actual torque to the distributed torque. To obtain better control performance, a genetic algorithm was further used to globally optimize the control parameters with the goal of reducing torque ripple and copper loss. A four-phase 8/6 pole switched reluctance motor was modeled and simulated to verify the effectiveness of the proposed strategy. The results showed that compared with the traditional cosine torque sharing control, the proposed control strategy could improve the performance of the motor at different speeds, with torque ripple decreased by up to 68.89% and root mean square current reduced by up to 11.74%.

Abstract: In order to solve the computing energy efficiency optimization problem of the wireless powered mobile edge computing(MEC) system, a computing resource allocation strategy of wireless powered MEC system based on non-orthogonal multiple access(NOMA) was proposed, which applied a nonlinear energy harvesting model to mobile devices. By jointly optimizing the calculation frequency, execution time, base station transmission power, equipment transmission power, offloading time and energy collection time of MEC server and mobile equipment, this strategy could fully utilize the available computing resources of mobile devices and MEC servers, improve device throughput and computing bits, and thus maximize system computing energy efficiency. Then the joint optimization problem was transformed into a non-convex fractional programming problem, and an iterative algorithm based on Dinkelbach was designed to obtain the optimal resource allocation scheme. The comparative simulation results showed that the resource allocation strategy could achieve higher computing energy efficiency and better performance gains.