Examining thyroid fine-needle aspiration (FNA) can grade cancer risks, derive prognostic information, and guide follow-up care or surgery. The digitization of biopsy and deep learning techniques has recently enabled computational pathology. However, there is still lack of systematic diagnostic system for the complicated gigapixel cytopathology images, which can match physician-level basic perception. In this study, we design a deep learning framework, thyroid segmentation and hierarchy fine-needle aspiration (TshFNA)-Examiner to quantitatively profile the cancer risk of a thyroid FNA image. In the TshFNA-Examiner, cellular-intensive areas strongly correlated with diagnostic medical information are detected by a nuclei segmentation neural network; cell-level image patches are catalogued following The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) system, by a classification neural network which is further enhanced by leveraging unlabeled data. A cohort of 333 thyroid FNA cases collected from 2019 to 2022 from I to VI is studied, with pixel-wise and image-wise image patches annotated. Empirically, TshFNA-Examiner is evaluated with comprehensive metrics and multiple tasks to demonstrate its superiority to state-of-the-art deep learning approaches. The average performance of cellular area segmentation achieves a Dice of 0.931 and Jaccard index of 0.871. The cancer risk classifier achieves a macro-F1-score of 0.959, macro-AUC of 0.998, and accuracy of 0.959 following TBSRTC. The corresponding metrics can be enhanced to a macro-F1-score of 0.970, macro-AUC of 0.999, and accuracy of 0.970 by leveraging informative unlabeled data. In clinical practice, TshFNA-Examiner can help cytologists to visualize the output of deep learning networks in a convenient way to facilitate making the final decision.

Deep learning has been applied for motor imagery electroencephalogram (MI-EEG) classification in brain-computer system to help people who suffer from serious neuromotor disorders. The inefficiency network and data shortage are the primary issues that the researchers face and need to solve. A novel MI-EEG classification method is proposed in this paper. A plain convolutional neural network (pCNN), which contains two convolution layers, is designed to extract the temporal-spatial information of MI-EEG, and a linear interpolation-based data augmentation (LIDA) method is introduced, by which any two unrepeated trials are randomly selected to generate a new data. Based on two publicly available brain-computer interface competition datasets, the experiments are conducted to confirm the structure of pCNN and optimize the parameters of pCNN and LIDA as well. The average classification accuracy values achieve 90.27% and 98.23%, and the average Kappa values are 0.805 and 0.965 respectively. The experiment results show the advantage of the proposed classification method in both accuracy and statistical consistency, compared with the existing methods.

In recent years, significant progress has been made through impedance pneumography (IP) in diagnosing pulmonary function. Since there is no need to measure inhalation and exhalation air flow through a pipeline, IP does not increase respiratory resistance and poses no risk of cross-infection, which makes it superior to existing gas flowmeter-based spirometers in clinics. However, the changes in thoracic impedance caused by pulmonary ventilation present significant individual variability. The ratio between pulmonary ventilation volume change (ΔV ) and thoracic impedance change (ΔZ), noted as kΔV/ΔZ , differs among people. IP has to be calibrated for each person by flowmeter-type spirometer before it can be used for quantitative diagnosis. This study aimed to develop a universal model for kΔV/ΔZ using individual parameters such as body height, body mass, body mass index, body fat rate, and chest circumference. The experimental procedure, the way to identify factors for multiple regression via significance analysis and the comparison among different models are presented. This paper demonstrates the possibility of establishing a universal regression model for kΔV/ΔZ , to lay the foundation for the clinical application of IP-based pulmonary function test.

This study aims to investigate the short-term effects of ambient air pollutants on outpatient visits for childhood allergic diseases. Daily data on ambient air pollutants (NO2, SO2, CO and PM2.5) and outpatient visits for childhood allergic diseases (asthma, atopic dermatitis and allergic rhinitis) were obtained in Shanghai, China from 2013 to 2014. The effects of ambient air pollutants were estimated for total outpatient visits for childhood allergic diseases, gender and age stratification and disease classification by using distributed lag non-linear model (DLNM). We found positive associations between short-term exposure to air pollutants and childhood allergic diseases. Girls and children aged  7 years old were more likely to be sensitive to ambient air pollutants. NO2 and SO2 showed stronger effects on asthma and atopic dermatitis, respectively. This study provides evidence that short-term exposure to ambient air pollutants can increase the risk of childhood allergic diseases.

Carpooling is a sustainable, economical, and environmentally friendly solution to reduce air pollution and ease traffic congestion in urban areas. However, existing regret theories lack consideration of the heterogeneity of attribute perception in different ways and the psychological factors that affect regret, so they cannot accurately portray urban residents’ carpool travel decisions and cannot provide a correct explanation of the actual carpool choice behavior. In this paper, based on the analysis of classical random regret minimization models and random regret minimization models considering heterogeneity, the concept of psychological distance is introduced to address shortcomings of the existing models and construct an improved random regret minimization model considering heterogeneity and psychological distance. The results show that the fit and explanatory effect of the improved model proposed in this paper is better than that of the other two models. The psychological distance of travel residents during the Corona Virus Disease 2019 (COVID-19) affects the anticipated regret value and the willingness to carpool. The model can better describe the carpool travel choice mechanism of travelers and effectively explain the carpool travel choice behavior of travelers.

Kneeling seat is an ergonomic chair that can help the human body’s spine in a sitting posture to be closer to the natural state. In this study, we used non-contact camera method to measure visual distance. Using surface electromyography (sEMG) combined with subjective evaluation, we studied the obvious effects of seat angle and leg support angle in kneeling sitting posture on the ride comfort of healthy female school-age children without myopia. Using three experiment seat angles (10◦, 20◦ and 30◦), we found that as the sitting angle increased, the absolute value of the slope of the erector spinae linearity curve, MPF-t, gradually decreased. At 30◦, the slope of MPF-t was −0.26, the descent speed was the slowest, the activity of erector spinae was relatively lowest, and the comfort of children’s waist was also improved, while the comfort of calf gastrocnemius decreased, just the opposite. At the same time, leg support angles of 20◦, 30◦ and 40◦ were used. And in the study we found that the elevation of the leg support angle had no significant effect on the erector spinae muscle, but had a significant effect on the gastrocnemius muscle. When the leg support angle was 30◦, the slope of MPF-t was −0.42, and the gastrocnemius comfort reached its peak.

In this paper, a classification method based on convolutional neural network (CNN) and received signal strength (RSS) is proposed to solve the problem of non-cooperative emitter beam state sensing in electrical situational awareness. RSS, sensor coordinates, and received signal frequency are taken as the input features of CNN, while real state is taken as the output of CNN. To increase the RSS gradient contained in the eigenvector, a multi-layer sensor array is proposed to measure RSS. Simulation results show that the proposed method is robust to array location disturbance, and has the ability to generalize the mismatches in target location and main lobe beam width between first nulls.

This paper presents a distributed arithmetic (DA) architecture that can efficiently implement finite impulse response (FIR) filters for biomedical signal processor applications. FIR filter design is more efficient when it uses a look-up table (LUT)-based technique rather than a serial one. The design’s performance and efficiency can be improved by using segmented memory banks as well as memory lookup for multiply operation. Verilog HDL is used to model the proposed design, and Synopsys Design Compiler tool is used for synthesis. The FIR filter architecture utilizing DA results in a 24.82% reduction in total power compared with the serial FIR structure.

Model predictive control (MPC) is a model-based optimal control strategy widely used in robot systems.In this work, the MPC controller tuning problem for the path tracking of the wheeled mobile robot is studied and a novel self-tuning approach is developed. First, two novel path tracking performance indices, i.e., steadystate time ratio and steady-state distance ratio are proposed to more accurately reflect the control performance.Second, the mapping relationship between the proposed indices and the MPC parameters is established based on machine learning technique, and then a novel controller structure which can automatically tune the control parameters online is further designed. Finally, experimental verification with an actual wheeled mobile robot is conducted, which shows that the proposed method could outperform the existing method via achieving significant improvement in the rapidity, accuracy and adaptability of the robot path tracking.

One of the most important problems in complex networks is to identify the influential vertices for understanding and controlling of information diffusion and disease spreading. Most of the current centrality algorithms focus on single feature or manually extract the attributes, which occasionally results in the failure to fully capture the vertex’s importance. A new vertex centrality approach based on symmetric nonnegative matrix factorization (SNMF), called VCSNMF, is proposed in this paper. For highlight the characteristics of a network,the adjacency matrix and the degree matrix are fused to represent original data of the network via a weighted linear combination. First, SNMF automatically extracts the latent characteristics of vertices by factorizing the established original data matrix. Then we prove that each vertex’s composite feature which is constructed with one-dimensional factor matrix can be approximated as the term of eigenvector associated with the spectral radius of the network, otherwise obtained by the factor matrix on the hyperspace. Finally, VCSNMF integrates the composite feature and the topological structure to evaluate the performance of vertices. To verify the effectiveness of the VCSNMF criterion, eight existing centrality approaches are used as comparison measures to rank influential vertices in ten real-world networks. The experimental results assert the superiority of the method.

Technical advances and sustainable development tendency accelerate the implementation of electric trucks. However, the penetration of dynamic charging tariff policy poses a huge challenge to the cost-optimal operation of the electric truck fleet. To this end, a two-stage stochastic electric vehicle routing model is formulated to support cost-efficient routing and charging decisions. Furthermore, an experimental study based on a real-world distribution network is conducted to evaluate impacts of dynamic charging tariffs on logistics planning. The results show that the daily operation cost can reduce by 3.57% to 5.55% as the number of dynamic charging stations increases. The value of stochastic solution confirms the benefits of implementing stochastic programming model,which will ensure a lower operation cost in the long-term through robust route planning.

At present, most of the studies on autonomous vehicles mainly focus on improving driving safety and efficiency, while less consideration is given to the comfort of passengers. Therefore, in order to gain and optimize quantitative indices for the ride experience of autonomous vehicles, this paper proposes an evaluation method for the correlation between driving behavior and passenger comfort with bidirectional long short-term memory network and attention mechanism. By collecting subjective feeling scores of passengers under different driving styles, and measuring the pressure level with skin conductance response and heart rate variability, the comprehensive quantitative indices of passenger comfort caused by driving behavior are evaluated. Based on this, a personalized comfort evaluation model for passengers with different driving style preferences is established. The results obtained from experiments in open road and closed test areas have validated the effectiveness and feasibility of the method proposed in this paper.

For effectively improving the overall performance of fire truck frame structure, and solving the complexity of previous methods in the frame optimization design process, the traditional grey relational grade ranking needs to be improved. First, the first-order modal test was conducted to verify the validity of the initial frame model. Then, based on this model, a high-strength steel frame was designed to reduce deformation, maximum stress, and frame mass, and increase the fatigue life and the frequencies of the first bending modal and first torsional modal. Sixty groups of sample points were generated through Hammersley method. Subsequently, improved grey relational analysis with principal component analysis was proposed to realize the optimal design of the frame structure. Finally, the optimal combination of design parameters for the frame was obtained using the proposed method. Meanwhile, the optimized frame structure is found by comparing the models before and after optimization, and the mass is reduced by 14.8%. Moreover, the computational cost can be reduced by 135% when the proposed method is compared with the previous algorithm. Therefore, the proposed method can effectively improve the performance of the frame and improve the computational efficiency.

The vehicles and pedestrians ranging is one of the basic functions of advanced driving assistance system. However, most of the ranging systems can only work on workstations with high computing power. To solve this problem, a lightweight algorithm is proposed to be packaged into Android application package, and be installed in Android smartphones for vehicles and pedestrians ranging. The proposed ranging system is based on the images obtained by smartphone’s monocular camera. To achieve real-time ranging, an 8-bit integer (int8) quantization algorithm is proposed to accelerate the inference of convolutional neural networks. To increase the detection precision, a zoom-in algorithm is further proposed to detect small targets in the distance. After having detected the 2D bounding boxes of vehicles and pedestrians, a pinhole ranging method is applied to estimate the distance. In order to verify the proposed algorithm, the mean average precision (mAP) and the frame per second (FPS) are first tested by using COCO dataset on Huawei P40Pro, then, the ranging precision on the real road.The experimental results show that this algorithm can successfully perform real-time ranging (15 FPS) with high precision (34.8 mAP) onto the tested smartphones. Finally, a possible mobile application based on the ranging algorithm, i.e., distance keeping warning, is also provided.

: Simulation of pedestrians’ behavior in the hub can help decision-makers to formulate better evacuation strategies. With this aim, this study develops an improved cellular automata model considering pedestrian’s mass-following psychology and competitive awareness, and based on this model, pedestrian’s evacuation process from the channel of the hub with two exits is simulated. Moreover, dynamic guidance information, e.g., the realtime congestion situation of the evacuation routes, plays an important role during pedestrian evacuation processes in a hub, as the evaluation routes can be adjusted based on this information. That is, the congestion situation during the evaluation can be improved. Thus, dynamic signs are incorporated into the proposed model to study the influence of dynamic guidance information on pedestrian evacuation behavior. In simulation experiments, the influence of two parameters, namely the proportion of pedestrians unfamiliar with the hub and update interval of dynamic signs, on pedestrian evacuation behavior is studied. Results show that dynamic guidance information can improve the efficiency of pedestrian evacuation. In particular, the higher the proportion of pedestrians unfamiliar with the hub is, the more obvious the effect of dynamic guidance information is. Besides, different proportions of pedestrians unfamiliar with the hub lead to different update intervals of dynamic signs. Finally, the results of this study can provide some implications to the practical hub operation and evacuation, e.g., to standardize the order of evacuation routes and improve the information service level in the hub.

In this study, a half-space 13-degree-of-freedom vehicle model, a double track model, and a train-bridge interaction model were integrated to form a combined people-train-rail-bridge interaction model to analyze the vertical Sperling index of the train body and passengers as realistically as possible. In this bigger, more complete, and novel model, the separation between the vehicle and bridge is considered. By comparing measured data and simulated results obtained using the proposed model with the Newmark-Beta algorithm, the effectiveness of the model was verified, and the results demonstrated that these two values were very close. Upon further numerical analysis, the dynamic responses of the train and the three equivalent human bodies at different train speeds were computed using the developed vehicle-structure dynamic analysis program with different abruptness values in the random rail irregularities. The results of these four dynamic responses revealed that the rail irregularities affected the vertical acceleration of the three equivalent human bodies and train, and the best Sperling index evaluation standard for the train was not fixed (as assumed when only considering the train body) but varied with the passenger position as the train traveled over irregularities.

Traffic issues have always received enthusiastic attention from the society. To better simulate the traffic environment, we use the well-known snowdrift game (SDG). Punishment has been regarded as a significant method to promote cooperation. We propose a novel punishment mechanism and discuss its influence on the cooperation of the SDG. Considering that the snowball causes traffic jam, we add the role of the traffic police in the SDG. When the traffic police choose to cooperate, they have the right to punish the defectors. The scope of jurisdiction, the record of punishment and the method of deployment are decisive factors in deciding whether or not to punish the defectors and the severity of the punishment. Whether to sanction the defector and the severity of the punishment is jointly determined by the traffic police’s punishment record, jurisdiction, and deployment method. Through extensive simulation, we found that the difference between the two distribution methods becomes smaller as the jurisdiction becomes smaller. We need to choose the dominant distribution method based on the jurisdiction and the neighbor pattern. The results demonstrate that the punitive record, jurisdiction and distribution method all have an important impact on the SDG and traffic governance.

In this paper, a novel method is proposed for structural damage detection and localization. The key of the proposed method is the use of the response difference transmissibility (RDT) for damage localization. The RDT is defined to relate response differences of the system to be evaluated and those of the original system. The invariance properties of RDT are also investigated. Based on this, a damage indicator is proposed, and the benefits are verified by a numerical example. Finally, the proposed method is illustrated and validated by finite element simulations and experimental case studies.

In this paper, a design method of moving-mass stratospheric airship with constant total mass is presented, and the general dynamics equation based on Newton-Euler method is derived. Considering the timedelay of the slider command response and the dynamic coupling to the airship’s state parameters, a position tracking controller with input and state constraints was designed to make the dynamic response system of the slider have critical damping characteristics. By taking the longitudinal attitude motion of moving-mass stratospheric airship as the research object, parametric modeling and attitude control simulation were carried out, and the attitude control ability of moving-mass control under different mass ratios was analyzed. The simulation results show that the attitude control ability is not affected by airspeed, and the mass ratio of slider is the main factor affecting the attitude control ability. The parameters of the slider controller have a direct influence on the dynamic performance of attitude control and also determine the dynamic coupling level of the airship. Compared with the attitude control based on the aerodynamic control surface, moving-mass control can make the airspeed and attack angle converged to the initial state at the steady state, and keep a good aerodynamic shape.

The trench casing often occurs in axial compressors due to the casing rubbing or casing treatment. However, the effect of the trench casing on the aerodynamic performance of axial compressors has not been fully understood, especially at different rotating speeds. Therefore, we numerically investigate the effect of the trench casing on a transonic compressor at two rotating speeds. A detailed comparison of overall performance and flow characteristics has been performed. The results show that the trench configurations slightly increase the total pressure ratio and mass flow rate near the choking condition but reduce the total pressure ratio and adiabatic efficiency at small mass flow rates. The largest efficiency reduction of the parallel trench (PT) and within trench (IT) cases is more than 1%, which is located at the mid blade passage near 90% span. The effect of the trench configurations on the stall margin is different for the two rotating speeds. At 100% rotating speed, the outside trench (OT) and IT cases improve the stall margin by 2.8% and 1.1%, respectively, but the PT case decreases the stall margin by 1.3% due to the increased blockage in the core region of the tip leakage vortex. At 80% rotating speed, the stall margin of the trench configurations becomes worse. Because of the increased blockage of the mid blade passage, the PT and IT cases decrease the stall margin by 2.9% and 2.1%, respectively. Though there are some differences in the flow characteristics of the trench configurations at the two rotating speeds, the change of the stall margin always depends on the blockage near the tip region. This work can contribute to further understanding the impact of the trench casing on axial compressors.

Vehicle data is one of the important sources of traffic accident digital forensics. We propose a novel method using long short-term memory-deep belief network by binary encoding (LSTM-BiDBN) controller area network identifier (CAN ID) to extract the event sequence of CAN IDs and the semantic of CAN IDs themselves. Instead of detecting attacks only aimed at a specific CAN ID, the proposed method fully considers the potential interaction between electronic control units. By this means, we can detect whether the vehicle has been invaded by the outside, to online determine the responsible party of the accident. We use our LSTM-BiDBN to distinguish attack-free and abnormal situations on CAN-intrusion-dataset. Experimental results show that our proposed method is more effective in identifying anomalies caused by denial of service attack, fuzzy attack and impersonation attack with an accuracy value of 97.02%, a false-positive rate of 6.09%, and a false-negative rate of 1.94% compared with traditional methods.

Automatic extraction of key data from design specifications is an important means to assist in engineering design automation. Considering the characteristics of diverse data types, small scale, insufficient character information content and strong contextual relevance of design specification, a named entity recognition model integrated with high-quality topic and attention mechanism, namely Quality Topic-Char Embedding-BiLSTMAttention-CRF, was proposed to automatically identify entities in design specification. Based on the topic model,an improved algorithm for high-quality topic extraction was proposed first, and then the high-quality topic information obtained was added into the distributed representation of Chinese characters to better enrich character features. Next, the attention mechanism was used in parallel on the basis of the BiLSTM-CRF model to fully mine the contextual semantic information. Finally, the experiment was performed on the collected corpus of Chinese ship design specification, and the model was compared with multiple sets of models. The results show that F-score (harmonic mean of precision and recall) of the model is 80.24%. The model performs better than other models in design specification, and is expected to provide an automatic means for engineering design.

In order to improve the performance of non-orthogonal multiple access (NOMA) system, a downlink NOMA system with two symmetric users is investigated. By noting that the requirement of reliability for these two users is different in some practical transmission scenarios, a performance-controlled max-min relay selection scheme for near-user and far-user is proposed. According to the practical transmission requirements, the outage performance of these two users can be controlled by different settings in the considered system. Considering the influence of the imperfect successive interference cancellation (SIC) in the practical scenarios, the exact outage performance of the far and near users in the system is analyzed, and the approximation results in high signalto-noise ratio are also obtained. Based on the analysis of the approximate results, the optimal power allocation factor for the outage performance of these two users is obtained. Finally, the correctness of the theoretical results is verified by simulation analysis.

In the case of a medium-long baseline, for real-time kinematic (RTK) positioning, the fixed rate of integer ambiguity is low due to the distance between the base station and the observation station. Moreover, the atmospheric delay after differential processing cannot be ignored. For correcting the residual atmospheric errors, we proposed a GPS/BDS/Galileo/GLONASS four-system fusion RTK positioning algorithm, which is based on the extended Kalman filter (EKF) algorithm. After realizing the spatio-temporal unification of multiple global navigation satellite systems (GNSSs), we introduced a parameter estimation of atmospheric errors based on the EKF model, using the least-squares integer ambiguity decorrelation adjustment (LAMBDA) to calculate the integer ambiguity. After conducting experiments for different baselines, the proposed RTK positioning algorithm can achieve centimeter-level positioning accuracy in the case of medium-long baselines. In addition, the time required to solve the fixed solution is shorter than that of the traditional RTK positioning algorithm.

The sixth-generation fighter has superior stealth performance, but for the traditional kernel density estimation (KDE), precision requirements are difficult to satisfy when dealing with the fluctuation characteristics of complex radar cross section (RCS). To solve this problem, this paper studies the KDE algorithm for F/AXX stealth fighter. By considering the accuracy lack of existing fixed bandwidth algorithms, a novel adaptive kernel density estimation (AKDE) algorithm equipped with least square cross validation and integrated squared error criterion is proposed to optimize the bandwidth. Meanwhile, an adaptive RCS density estimation can be obtained according to the optimized bandwidth. Finally, simulations verify that the estimation accuracy of the adaptive bandwidth RCS density estimation algorithm is more than 50% higher than that of the traditional algorithm. Based on the proposed algorithm (i.e., AKDE), statistical characteristics of the considered fighter are more accurately acquired, and then the significant advantages of the AKDE algorithm in solving cumulative distribution function estimation of RCS less than 1 m2 are analyzed.