Search

In this article we investigates the practical fixed-time consensus tracking problem for multiple Euler–Lagrange systems, which is subject to, concurrently, stochastic packet losses in directed communication networks (DCNs) and input/output constraints. In practice, the packet losses in the communication network may lead to communication failure and control accuracy degradation. To address this impact, a novel fully distributed observer is established to estimate the leader’s states within fixed time. Subsequently, the state constraint function is introduced so that the designed adaptive control protocol can be used in a uniform structure with unconstrained or asymmetric/symmetric output constraints without changing the adaptive structure. Combining the backstepping technique with fuzzy-logic systems, an adaptive fixed-time local control protocol is constructed to drive each agent to track the leader’s states estimated by the observer. Besides, a novel auxiliary system is introduced to solve the constraint problem of the control input. Furthermore, through rigorous theoretical analysis, it is verified that all signals converge to the compact set close to zero with guaranteed fixed-time convergence rate. At last, a representative simulation is carried out to demonstrate the practicability of the proposed method.

This article addresses the three-dimensional (3-D) coordinated control problem of directed networked aircraft-like vehicles, that is to track a set of given orbits on a sphere and achieve a lateral formation flight. Different from the case of Newton particles, a nonholonomic dynamics with unknown disturbances is considered. A novel method to decouple the spherical orbit tracking subsystem and the lateral formation flying subsystem is proposed. By overlooking the control of the vehicle's surge velocity, a nonsmooth spherical orbit tracking algorithm is designed by backstepping. Without considering the spherical orbit tracking errors and using any global information of topologies, a distributed, nonsmooth formation protocol is designed. The input-to-state stability (ISS) theory is used to analyze the converge property of the interconnected system consisting of these two subsystems. Simulation results are given to verify the theoretical analysis.

The requirements for green energy have become more critical with the increasing risk from the global warming effect. In this study, previous study results were applied to obtain an optimal small surface boat configuration design and estimated its power requirement. The Proton Exchange Membrane Fuel Cells (PEMFC) are wildly used on vehicles. Its high working temperature (80 °C) make it a safer power source for small surface boat. This research used PEMFC as the test platform with test range from 0 to 2.4 kW. In order to obtain the optimum performance, Taguchi’s orthogonal array (L8 and L27) method was applied, with three factors: flow rate of hydrogen, humidification conditions, and temperature of the fuel cell stack, as the control parameters. According to the case study results, a dimensionless analysis were performed to acquire the ship model’s resistance–power curve and to evaluate the required power for a real ship (with a size that is three times that of the ship model). The speed was approximately 5 knots, and the required power was about 250 W. Finally, the best factor-level combination of Taguchi’s method is A2B3C2 and that the SNR is 57.81 dB. This is better than the value (approximately 0.43 dB) for the original design (A1B1C1). In addition, both the main effect and interaction of the three factors have significant influence, especially toward the fuel cell stack temperature. The robust design was obtained with the intelligence parameter method matching with the results of the Taguchi’s optimal prediction.

Alternating direction method of multipliers (ADMM) and adaptive moment estimation (ADAM) are two optimizers of paramount importance in convex optimization (CO) and deep learning (DL), respectively. Numerous state-of-the-art algorithms for solving inverse problems are achieved by carefully designing a convex criterion, typically composed of a data-fitting term and a regularizer. Even when the regularizer is convex, its mathematical form is often sophisticated, hence inducing a math-heavy optimization procedure and making the algorithm design a daunting task for software engineers. Probably for this reason, people turn to solve the inverse problems via DL, but this requires big data collection, quite time-consuming if not impossible. Motivated by these facts, we propose a new framework, termed as ADMM-ADAM, for solving inverse problems. As the key contribution, even just with small/single data, the proposed ADMM-ADAM is able to exploit DL to obtain a convex regularizer of very simple math form, followed by solving the regularized criterion using simple CO algorithm. As a side contribution, a state-of-the-art hyperspectral inpainting algorithm is designed under ADMM-ADAM, demonstrating its superiority even without the aid of big data or sophisticated mathematical regularization.

Online handwritten diagram recognition (OHDR) has attracted considerable attention for its potential applications in many areas, but it is a challenging task due to the complex 2D structure, writing style variation, and lack of annotated data. Existing OHDR methods often have limitations in modeling and learning complex contextual relationships. To overcome these challenges, we propose an OHDR method based on graph neural networks (GNNs) in this paper. In particular, we formulate symbol segmentation and symbol recognition as node clustering and node classification problems on stroke graphs and solve the problems jointly under a unified learning framework with a GNN model. This GNN model is denoted as Instance GNN since it gives the symbol instance label as well as the semantic label. Extensive experiments on two flowchart datasets and a finite automata dataset show that our method consistently outperforms previous methods with large margins and achieves state-of-the-art performance. In addition, we release a large-scale annotated online handwritten flowchart dataset, CASIA-OHFC, and provide initial experimental results as a baseline.

Data representation learning is one of the most important problems in machine learning. Unsupervised representation learning becomes meritorious as it has no necessity of label information with observed data. Due to the highly time-consuming learning of deep-learning models, there are many machine-learning models directly adapting well-trained deep models that are obtained in a supervised and end-to-end manner as feature abstractors to distinct problems. However, it is obvious that different machine-learning tasks require disparate representation of original input data. Taking human action recognition as an example, it is well known that human actions in a video sequence are 3-D signals containing both visual appearance and motion dynamics of humans and objects. Therefore, the data representation approaches with the capabilities to capture both spatial and temporal correlations in videos are meaningful. Most of the existing human motion recognition models build classifiers based on deep-learning structures such as deep convolutional networks. These models require a large quantity of training videos with annotations. Meanwhile, these supervised models cannot recognize samples from the distinct dataset without retraining. In this article, we propose a new 3-D deconvolutional network (3DDN) for representation learning of high-dimensional video data, in which the high-level features are obtained through the optimization approach. The proposed 3DDN decomposes the video frames into spatiotemporal features under a sparse constraint in an unsupervised way. In addition, it also can be regarded as a building block to develop deep architectures by stacking. The high-level representation of input sequential data can be used in multiple downstream machine-learning tasks, we evaluate the proposed 3DDN and its deep models in human action recognition. The experimental results from three datasets: 1) KTH data; 2) HMDB-51; and 3) UCF-101, demonstrate that the proposed 3DDN is an alternative approach to feedforward convolutional neural networks (CNNs), that attains comparable results.

Although text detection methods have addressed several challenges in the past, there is a dearth of effective methods for text detection in deformable images, such as images containing text embedded on cloth, banners, rubber, sports jerseys, uniforms, etc. This is because deformable regions contain surfaces of arbitrarily shapes, which lead to poor text quality. This paper presents a new method for deformable text detection in natural scene images. It is observed that although the shapes of characters change in a deformable region, the pixel values and spatial relationship between the pixels do not change. This motivated us to explore extraction of Maximally Stable Extremal Regions (MSER) in an image in which pixels that share common features are grouped into components. The unique character shape variations led us to explore harmonic features to represent the component shape variations, using which a classifier classifies text and non-text components from the output of the MSER step. Additionally, the objective of developing a lightweight method with low computational cost motivated us to introduce a modified Pixel Aggression Network (PAN) for text deformable text detection at a component level. Comprehensive experiments which include experiments on our Deformable Text Dataset (DTD) and standard natural scene text datasets, namely, MSRATD-500, ICDAR 2019 MLT, Total-Text, CTW1500, ICDAR 2019 ArT and DSTA1500 datasets show that the proposed model outperforms the existing methods for our dataset as well as the standard datasets.

Most of the existing ubiquitous clinic recommendation (UCR) systems adopt linear mechanisms to aggregate the attribute-level performances of a clinic to evaluate the overall performance. However, such linear mechanisms may not be able to explain the choices of all patients. To solve this problem, the modified mixed binary nonlinear programming (MMBNLP)–feedforward neural network (FNN) approach is proposed in this study. In the proposed methodology, first, the existing MBNLP model is modified to improve the successful recommendation rate using a linear recommendation mechanism. Subsequently, an FNN is constructed to fit the relationship between the attribute-level performances of a clinic and its overall performance, thereby providing possible ways to further enhance the recommendation performance. The results of a regional experiment showed that the MMBNLP–FNN approach improved the successful recommendation rate by 30%.

Speech emotion recognition (SER) is an important research area, with direct impacts in applications of our daily lives, spanning education, health care, security and defense, entertainment, and human–computer interaction. The advances in many other speech signal modeling tasks, such as automatic speech recognition, text-to-speech synthesis, and speaker identification, have led to the current proliferation of speech-based technology. Incorporating SER solutions into existing and future systems can take these voice-based solutions to the next level. Speech is a highly nonstationary signal, with dynamically evolving spatial-temporal patterns. It often requires a sophisticated representation modeling framework to develop algorithms capable of handling real-life complexities.

The success of mediation is affected by many factors, such as the context of the quarrel, the personality of both parties, and the negotiation skill of the mediator, which lead to uncertainty for the work of prediction. This article takes a different approach from that of previous legal prediction research. It analyzes and predicts whether two parties in a dispute can reach an agreement peacefully through the conciliation of mediation. With the inference result, we can know whether mediation is a more practical and time-saving method to solve the dispute. Existing works about legal case prediction mostly focus on prosecution or criminal cases. In this work, we propose a long short-term memory (LSTM)–based framework, called LSTMEnsembler, to predict mediation results by assembling multiple classifiers. Among these classifiers, some are powerful for modeling the numerical and categorical features of case information, for example, XGBoost and LightGBM. Some are effective for dealing with textual data, for example, TextCNN and BERT. The proposed LSTMEnsembler aims to not only combine the effectiveness of different classifiers intelligently but also to capture temporal dependencies from previous cases to boost the performance of mediation prediction. Our experimental results show that our proposed LSTMEnsembler can achieve 85.6% for F-measure on real-world mediation data.

Requirements of compressed sensing techniques targeted at miniaturized hyperspectral satellite applications include lightweight onboard hardware, high-speed sensing, low sampling rate for compressing the massive volume of typical hyperspectral data, and noise robustness for reliable data transmission to the ground station. We achieve all these aims via deep learning, and neural networks resulted from which can be implemented on-chip, thereby allowing light hardware implementation. Our neural networks were trained from small-scaled data, but, even so, the resulting encoder achieves a very low sampling rate and very high speed. Unlike typical network training, the input–output pairs are not square but stripe-like images, partly because compressed acquisition does not allow performing compression after obtaining complete data cube and partly because stripe-like acquisition well matches the popular pushbroom hyperspectral sensing schemes. Even with such hard restriction caused by nontraditional training, the resulting decoder still reconstructs the image with high accuracy. To match the requirement of pushbroom sensing, a lightweight encoder is proposed to compress the stripe-like images immediately. Meanwhile, multiscale feature fusion block (MFB) and aggregation (MFA) modules are proposed to form our decoder for enhancing the feature representation of the compressed acquisitions. Furthermore, we achieve joint spatial/spectral super-resolution (SR) progressively, ensuring accurate hyperspectral reconstruction via a low-rank-driven decoder. The encoder and decoder are trained in an end-to-end manner, where noise robustness is forced during the training stage. Comprehensive experiments demonstrate the superiority of the proposed hyperspectral compressed sensing method, as well as its one-shot transfer learning (OTL)-based extension, both quantitatively and qualitatively.

The belief rule-base (BRB) model is a new intelligent expert system with the characteristics of both expert system and data-driven model. In BRB there are many if-then rules which use belief degrees to express various types of uncertain information, including fuzziness, randomness, and ignorance. As a semi-quantitative modeling tool for complex systems, BRB has the superiorities of dealing both numerical quantitative data and linguistic qualitative knowledge that are derived from heterogeneous sources. Moreover, it is also a white box approach which can provide direct access and transparency to decision makers and stakeholders. Currently, BRB has been widely applied in many fields, such as decision making, reliability evaluation, network security situation awareness, fault diagnosis, and so on. To fully demonstrate the progress of BRB, the original BRB, and some evolution forms are introduced in this article.

The safety of lithium-ion batteries under mechanical abuse has become one of the major obstacles affecting the development of electric vehicles. In this paper, the lithium-ion battery safety model under mechanical abuse conditions is proposed by the Back Propagation Artificial Neural Network (BP-ANN) optimized by the Genetic Algorithm (GA). By experimental and simulation results, the proposed method can effectively predict battery mechanical properties. The corresponding correlation coefficient is greater than 0.99, the failure warning model has more safety margin, and the average security margin is greater than 29%. The multi-source warning weight shows that the mechanical soft short-circuit has the greatest warning margin, followed by that of the soft short-circuit of the electrical signal. The thermal soft short-circuit has the lowest warning margin because of the low thermal conductivity. The qualitative simulation results of the battery module reveal that, when electric vehicles are subjected to mechanical abuse conditions, the rapid reduction in the state-of-charge (SOC) of the rear batteries can effectively increase the reliability of the battery module. The proposed safety model is important to protect the safety and stability of lithium-ion batteries, which is conducive to promoting new energy vehicles and protecting the environment.

Under the background of further promoting the construction of "golden class" in universities, how to use teaching innovation and promote the cultivation of learners is the focus of educational researchers. Guided by the theory of deep learning, based on MOOC resource platform and teaching management platform, this study constructs a "3-stage-2 platform" blended learning mode for deep learning in three stages: before class, in class and after class, puts forward teaching strategies to promote deep learning, and based on this mode, takes the computer network course as an example to carry out the blended teaching design and practice. The results show that the model has a certain role in promoting learners' knowledge mastery and ability training, and good emotional experience is conducive to learners' deep learning. Keywords: Deep learning, mixed teaching, teaching design.

An accurate driving cycle prediction is a vital function of an onboard energy management strategy (EMS) for a battery/ultracapacitor hybrid energy storage system (HESS) in electric vehicles. In this paper, we address the requirements to achieve better EMS performances for a HESS. First, a long short-term memory-based method is proposed to predict driving cycles under the framework of a model predictive control (MPC) algorithm. Secondly, the performances of three EMSs based on fuzzy logic, MPC, and dynamic programming are systematically evaluated and analyzed. For online implementation, the MPC-based EMS can alleviate the stress on the battery in the HESS and significantly reduce energy dissipation by up to 15.3% in comparison with the fuzzy logic-based EMS. Thirdly, the impact of battery aging on EMS performances is explored; it indicates that battery aging consciousness can slightly extend battery life. Finally, a hardware-in-the-loop test platform is established to verify the effectiveness of the MPC-based EMS for the power allocation of a HESS in electric vehicles.

In this paper, we address the fixed-time consensus tracking problem of second-order leader-follower multi-agent systems with nonlinear dynamics under directed topology. The consensus tracking algorithm consists of distributed observer and observer-based decentralized controller. The fixed-time distributed observer guarantees that each follower estimates the leader’s state under directed topology within a fixed time, where the upper bound of convergence time is independent on the initial conditions. The fixed-time decentralized controller makes each follower converge to the leader’s state in fixed time via tracking the distributed observer’s state and overcome the nonlinear dynamics without adding linear control terms. Finally, the numerical example is provided to illustrate the effectiveness of the results.

Most existing methods for forecasting the productivity of a factory cannot estimate the range of productivity reliably, especially when future conditions are distinct from those in the past. To address this issue, a fuzzified feedforward neural network (FFNN) approach is proposed in this study. The FFNN approach improves the forecasting precision after generating accurate fuzzy productivity forecasts. In addition, the acceptable range of a fuzzy productivity forecast is specified, based on which the sum of the memberships of actual values is maximized. In this way, the range of productivity can be precisely estimated. After applying the FFNN approach to a real case, the experimental results revealed the superiority of the FFNN approach by improving the forecasting precision, in terms of the hit rate, by 25%. Such an improvement also contributed to a better forecasting accuracy. The superiority of the FFNN approach is in the context that the accuracy of forecasting productivity is optimized only after the range of productivity has been precisely estimated. In contrast, most state-of-the-art methods focus on optimizing the forecasting accuracy, but may be ineffective without information about the range of productivity when future conditions are distinct from the past.

The high-level penetration of intermittent renewable power generation may limit power system inertia, resulting in system frequency instability in increasing power converter-based energy sources. To resolve this problem, virtual inertia control using distributed gray wolf optimization (DGWO) method in a synchronous generator is simulated under a distinct output fluctuation condition. First, the DGWO algorithm was established to achieve a local and global balance solution, and standard test functions were employed to verify the model convergence. Second, the key parameters that determine the effect of the virtual inertia controller in the power grid were analyzed. A DGWO-based optimization strategy to stabilize inertia was also developed. Finally, simulation results using step and random loads under a high permeability level are provided to verify the effectiveness of the proposed model. In the step load disturbance, the system can recover from the disturbance point to the stable point after 3 s under the regulation of the proposed control strategy, which is reduced by 18 s compared with the traditional control method. In the random load test, it takes only 12 s, 63 s less than the traditional one. Accordingly, the power system frequency can be stabilized more quickly from a disturbance state to a stable stage.

In large-scale agriculture, insufficient irrigation water may lead to overpumping of groundwater, increasing the risk of land subsidence. Growing dryland crops can effectively decrease the demand for irrigation water. However, the previous works on annual crop planning (ACP) focused on maximizing the profit through growing wetland crops and consuming much water. For sustainability, in this article, we propose a mathematical programming model for an ACP that allocates a land area for growing dryland and wetland crops to maximize the total profit and minimize the total irrigation water used for multiple cropping, under practical constraints. The simplified swarm optimization (SSO) improves the particle swarm optimization with four probabilities to determine the operations of updating solutions. We further propose dynamic SSO (DSSO) to solve the concerned ACP in which the four probabilities are adjusted dynamically according to the performance of the operations executed. Through simulation on a case study, the proposed DSSO demonstrates high performance over some classical approaches.

The rest if this article proposes a field-weakening control strategy along with two characteristic curves of IPMSM under electric vehicle driving conditions, where the operation speed, load torque, and target torque change rapidly and continuously. Two characteristic curves, the maximum torque curve, and the switching torque curve are introduced to obtain a strategy that can generate the optimal operation point with minimum stator current at any machine speed. The reference current of the optimal operation points can be derived from both calculation method and conventional calibration method. However, sometimes it is difficult to get solutions directly by the calculation method, not to mention it relies on precise motor parameters that vary with the temperature and magnetic saturation. Additionally, conventional calibration method is really a time-consuming process. In this article, a much simpler approach to obtain reference current based on polynomial fitting is presented, which has a great performance on fitting precision and much less workload with only one additional operation point to be calibrated except for the two characteristic curves. With this fitting method, the reference current generation procedure becomes completely parameter independent. The proposed strategy is demonstrated through simulations and test bench results.

Stationary average-tracking problems of unmatched constant reference signals are investigated for first-order multi-agent systems, and an integral-type average-tracking algorithm with delayed-state...

This paper investigates the event-triggered consensus problem of second-order nonlinear multi-agent systems subject to denial-of-service (DoS) attacks, which make the communication network unable to provide normal services. Considering a general class of DoS attack with limited duration, a novel distributed event-triggered consensus protocol accompanied with first-order hold is adopted to guarantee the globally bounded consensus convergence under directed network topology. Based on the linear matrix inequality approach and Lyapunov stability method, consensus converging properties are analysed and sufficient criteria are obtained. Furthermore, Zeno-free triggering of our proposed protocol is demonstrated. Finally, a numerical simulation is given to show the effectiveness of our theoretical results.

In this paper, consensus problem is studied for the first-order nonlinear multi-agent systems under directed topology, and a novel event-triggered consensus protocol is proposed. Different from the existing results with general zero-order hold, the first-order hold is adopted to construct the event-triggered control signal so as to reduce the triggering rate. By using Lyapunov stability method and matrix theory, a sufficient consensus condition is derived for the agents to reach the asymptotic consensus, and Zeno-behavior can be avoided. Finally, a numerical example is presented to demonstrate the effectiveness of our proposed protocol.

The lyrics generator should consider the context and the singability of the songs because every song expresses a story through the context of lyrics, and the lyrics should sound with the music well. Therefore, this study proposes a framework to generate the singable lyrics, and the context of lyrics should fit the given musical style. For the context, this study adopts the GPT-2 model which is powerful for text generation. The conditional GPT-2 model can be used to generate lyrics according to the given style. For suitable for singing, this study adjusts the structure and rhyme of lyrics through the use of a syntactic parser and a rhyme modification module. With automatic and human evaluations, the experimental results show that the proposed method can generate lyrics with high structural consistency, rhyme consistency, and originality according to the given music style.

Reading irregular text of arbitrary shape or low quality in natural scene images is a challenging task. Existing irregular scene text recognition methods mainly focus on irregular text with arbitrary shape, but rarely focus on irregular text of low quality. In this work, we propose a simple but effective method for recognizing irregular texts with arbitrary shape and low quality simultaneously. The proposed Multi-Branch guided Attention Network (MBAN) makes mutual guidance among multi-branch data in training, so as to learn invariant semantic representation between regular text images and the corresponding irregular images. Compared with the standard attention framework for text recognition, MBAN can significantly improve the performance of irregular text recognition while preserving similar performance for regular text recognition. In addition, regarding the attention drift problem encountered in standard attention network, MBAN can significantly improve the accuracy of alignment factors at each time step. We verify the effectiveness of MBAN in irregular text recognition and attention drift problem through extensive experiments. The performance of MBAN is shown to be comparable on regular datasets and superior on some irregular datasets with state-of-the-art methods.

Segmenting the region of interest (ROI) from medical images is a fundamental but challenging task due to the illumination change and imaging devices. Although many models based on the local region‐based active contour model (LR‐ACM) are proposed to deal with intensity inhomogeneity, it is still difficult for the global energy‐based ACM with local image information to accurately extract the ROI from medical images. To solve this problem, this study proposes a novel localised ACM by constructing the gradient information based on the probability scores from the fuzzy k‐nearest neighbour classifier. Different from the traditional LR‐ACMs, our model is based on local rather than global image statistics, where a probability score‐based edge detector is directly used for gradient information. The energy functional consists of localised region energy and an edge energy. By introducing a local characteristics function, the localised region energy with the probability‐score‐based edge information is formulated, which can make the evolution curve stop on the exact boundaries of ROI. The edge energy including the regularisation and the penalty terms is used to avoid the reinitialisation process and smooth the evolution curve during evolution. To maintain the signed distance property of the evolution curve, a novel potential function in the penalty term is designed, which can consistently control the diffusion direction of the evolution curve. Experiments on the medical images including the cardiac magnetic resonance imaging and the 3DIRCADb databases demonstrate that the proposed model is more robust and accurate to extract the ROI than the popular localised and region‐based ACMs. The code is available at: https://github.com/HuaxiangLiu/Localized_ERAC/.

This paper investigates the multi-objective problem of deploying wireless sensor networks with cooperative distance-based sensing coverage. This problem considers deploying a number of sensor nodes to cove multiple target points on a deployment area. Based on the locations of target points and the sensor nodes with their own inner and outer coverage radii, the distance-based sensing coverages of target nodes by sensor nodes are divided into three categories: full coverage (i.e., within the inner coverage radius), no coverage (i.e., outside the outer coverage radius), and partial coverage (i.e., between the inner and outer radii). Furthermore, this paper additionally considers the cooperative sensing coverage in which the sensing coverage of a target point is provided by more than one sensor node. The decision of sensor deployment in this paper is to select sensor nodes from potential sensor node positions so as to simultaneously maximize the collective sensing coverage of all target points and minimize the total distances between each target point and the selected sensor node(s). This paper first formulates this problem as a multi-objective optimization model, and then develops a solution procedure to determine the best non-dominated solution set for the problem model. Numerical experiments for the concerned problem by the proposed solution approach are demonstrated.

The customer’s order (CO) issues have received growing attention in the scheduling research community. In the CO design, a customer’s order consists of several components, and the processed orders are assigned to m parallel machines. The completion time of an order is assumed at the time at which all components in a customer’s order are finished. In the published articles, the processing times of all components were considered as fixed and known numbers in customer order scheduling problems. This assumption is indeed at odds in many real practical productions in which there exist a lot of relevant uncertainty factors; for example, machine may break down, the working situation may change, some operator’s performance may become instability, and so on. In such a situation, the processing times of given jobs are impropriate assumed as fixed and constant numbers. Thus, this article introduces a customer’s order scheduling problem on m parallel machines along with scenario-dependent processing times of all components and scenario-dependent due dates. Based on the worse case, the measurement rule is to find an appropriate policy to minimize the maximum total tardiness of given n customer’s orders across the possible scenarios among all possible schedules. In this study we derive several dominant rules and lower bounds used in a branch-and-bound method searching for exact solutions. Subsequently, we propose several heuristics and an iterated greedy algorithm with and without a population-based for finding approximate solutions. Finally, the computational results of all proposed algorithms are tested and reported.

This paper proposes a novel fault classification method with application to induction motors, which is based on integrating and combining with receiver operating characteristic (ROC) curve and t-distribution stochastic neighbor embedding (t-SNE). According to the feature selection methods of ReliefF, symmetrical uncertainty (SU), and fast correlation-based filter (FCBF), the significant features were verified. Additionally, support vector machine (SVM), k-nearest neighbor (KNN), and decision tree (DT) are also considered as classifiers to identify the simulation results. To begin with, the current signals obtained from distinctive four topologies of working conditions of the motor, which includes healthy, bearing damage, broken rotor bar, and short circuit in stator windings, respectively. The potential feature set is extracted by using Hilbert-Huang transform (HHT) technique. Then, three feature selection methods are adopted to select three optimal feature subsets from the original feature set. Finally, the classification accuracy (ACC) and ROC curve are used to demonstrate the capability of classifiers’ recognition. The results showed that the optimal feature subsets significantly reduce the number of selected features and improve the classification ACC and area under the curve (AUC) compared with the original feature set. In conclusion, the proposed method can downgrade the data, demonstrate the scatter plot more intuitively, and identify various types of faults, unlike with other fault diagnosis literature.

This study applied the information processing view and the concept of fit to improve understanding of how attachment behaviour and continued use in mobile social network game (M-SNG) players is affected by the alignment between their motivation and technological capability. A quantitative method and a cross-sectional study design were used to evaluate the research model. Data were collected for 427 active M-SNG players in Taiwan. The proposed model was tested with fit as a covariation and with fit as a propensity score in matching method. The results of this study verified that (1) coalignment between motivation and technological capability is positively associated with attachment behaviour in M-SNG players; (2) M-SNG players who have a high coalignment between motivation and technological capability have stronger attachment behaviour compared to players with low coalignment; and (3) attachment behaviour is positively associated with continued use of M-SNGs. This study contributes to the literature by improving understanding of the antecedents of attachment and continued use in M-SNG players. For providers and developers, this study also provides insights into how to improve the business performance of M-SNGs.

For mass rapid transit (MRT) projects, the characteristics of MRT engineering include large scopes, long durations, complex interfaces, significant requirements for management, and complicated coordination and cooperation. Owing to the complexity of these projects, the use of building information modeling (BIM) technology will be necessary and helpful to improve construction management. The increasing application of BIM in the architecture, engineering, and construction (AEC) industries has strengthened the need for effective construction management for general contractors (GC) for BIM implementation during the construction phase. Collaboration-based BIM model development for GCs typically involves relevant participants from different fields during the construction phase of MRT projects. However, collaboration-based BIM model development tasks often create numerous practical problems (e.g., reworks and errors regarding BIM model development). As the BIM manager and engineers for the general contractor who handle BIM model development are located in different departments and different office locations in MRT construction projects, a centralized platform is needed to share the latest real-time BIM model developments for related participants. Effective collaboration-based BIM model development management (CBMDM) is therefore essential for GCs to obtain accurate BIM models before the implementation of BIM-related applications. While the latest information for collaboration-based BIM model development should be updated and tracked effectively for CBMDM for related participants during the development process, in practice, GCs lack systematic approaches and information management platforms for use in CBMDM tasks. Thus, this study proposes a novel approach and develops an information system to manage and track all CBMDM tasks for GCs during the BIM development process. The purpose of this study is to develop a CBMDM system to improve the efficiency of sharing and tracking information for CBMDM activities. The research scope of this study is to propose an approach and system for CBMDM work in large-scale construction and infrastructure projects. The CBMDM system assists users by providing workflow management of CBMDM tasks for BIM engineers and managers so users can manage all BIM model development tasks effectively. The proposed system can also be utilized for other types of large-scale construction projects for CBMDM tasks. Furthermore, to understand its effectiveness, the proposed CBMDM system is applied to a case study of an MRT project in Taiwan. Finally, the major practical advantages, disadvantages, and limitations of the system are identified for infrastructure projects.

This paper presents a smart detection technology for personal Electrocardiography (ECG) monitoring based on data integral-transform of chaotic system. First of all, a set of data-feeding system is developed, ECG data is technically converted into multiple-dimensional phase space, i.e., the dynamics of ECG data in time domain has been mapped into chaotic domain. Further, some effective and potential features in different sub-dimensional phase plane of the data, such as Euclidean Feature Values (EFV), Central Point Distribution (CPD), are captured, which indicates key biomarkers for different ECG states. In the final stage, following the key biomarkers, explicit boundary thresholds are defined for classification of different ECG states. Three ECG states given via open database-PhysioNet are validated, including normal sinus rhythm (NSR), congestive heart failure (CHF) and sleep apnea (SA). The experimental results show that the developed smart detection technology is effective and feasible for detecting and monitoring the states of such personal ECG states.

In this study, the authors address the fixed-time connectivity-preserving consensus problem for second-order multi-agent systems with external disturbances under state-dependent communication topology, where each agent has a limited sensing range. A novel global integral sliding-mode consensus protocol is proposed to achieve fixed-time consensus for leaderless multi-agent systems, meanwhile maintaining the connectivity of initial communication topology and suppress external disturbances. By Lyapunov theory and homogeneity theory, it is proved that multi-agent systems achieve fixed-time consensus convergence and the sufficient condition of topology connectivity preservation is obtained. In addition, the results are extended to the leader–follower multi-agent systems. The numerical examples are provided to illustrate the effectiveness of the results.

The ability to accurately judge another person's emotional states with a short duration of observations is a unique perceptual mechanism of humans, termed as the thin-sliced judgment. In this work, we propose a computational framework based on mutual information to identify the thin-sliced emotion-rich behavior segments within each session and further use these segments to train the session-level affect regressors. Our proposed thin-sliced framework obtains regression accuracies measured in Spearman correlations of 0.605, 0.633, and 0.672 on session-level attributes of activation, dominance, and valence, respectively. It outperforms framework using data of the entire session as baseline. The significant improvement in the regression correlations reinforces the thin-sliced nature of human emotion perception. By properly extracting these emotion-rich behavior segments, we obtain not only an improved overall accuracy but also bring additional insights. Specifically, our detailed analyses indicate that this thin-sliced nature in emotion perception is more evident for attributes of activation and valence, and the within-session time distribution of emotion-salient behavior is located more toward the ending portion. Lastly, we observe that there indeed exists a certain set of behavior types that carry high emotion-related content, and this is especially apparent in the extreme emotion levels.

With the advancement of AI and the widespread application of IoT and cloud computing, the pick-and-pass system (PKPS) can potentially be transformed into a cyber-physical system (CPS) based intelli...

Recent empirical works on graph drawing have investigated visual properties of graph drawings created by users based on adjacency lists of graphs as well as drawing behaviors. This is mainly done by asking participants to sketch these graphs on a tablet computer so that they can freely express their interpretation. However, previous works did not consider weighted vertices, i.e., assigning a weight to a vertex to reflect its importance. Therefore, we conducted an empirical study on graphs with weighted vertices. More specifically, this work conducts an experiment and analyzes characteristics of the final graph layouts, participants’ drawing processes and strategies and their drawing preferences. Results indicated that minimizing the number of edge crossings was still the most important aesthetic for participants, and that participants preferred the aesthetic of creating grid-like drawings in the condition with weighted vertices. Hence, this work suggested that aesthetics of minimizing number of edge crossings and creating grid-like patterns should be the main consideration for designing a graph drawing software application.

Bioprinting has been applied to fabricate biomedical parts such as skin, scaffolds, tissues, and muscles. However, most previous studies in this field compared the advantages and/or disadvantages of a bioprinter subjectively. A systematic and objective method for comparing various bioprinters, so as to recommend the most suitable one to a decision maker, is lacking. To address this issue, a fuzzy analytic hierarchy process (FAHP) and fuzzy technique for order preference by similarity to ideal solution (FTOPSIS) approach is proposed in this study. In the proposed FAHP-FTOPSIS approach, FAHP is applied to derive the fuzzy weights of factors critical to the suitability of a bioprinter, based on a decision maker’s subjective judgments. Subsequently, the derived fuzzy weights are fed into the FTOPSIS approach to evaluate the overall performance of a bioprinter. The FAHP-FTOPSIS approach has been successfully applied to the case of making a choice from nine bioprinters. Parametric analyses have also been conducted to show the robustness of the proposed methodology.

Employing the concept and function of tangency with similarity measures and counterpart distances for reliable medical consultations has been extensively studied in the past decades and results in lots of isomorphic measures for application. We compared the majority of such isomorphic measures proposed by various researchers and classified them into (a) maximum norm and (b) one-norm categories. Moreover, we found that previous researchers used monotonic functions to transform an identity function and resulted in complicated expressions. In this study, we provide a theoretical foundation to explain the isomorphic nature of a newer measure proposed by the following research paper against its studied existing one in deriving the same pattern recognition results. Specifically, this study initially proposes two similarity measures using maximum norm, arithmetic mean, and aggregation operators and followed by a detailed discussion on their mathematical characteristics. Subsequently, a simplified version of such measures is presented for easy application. This study completely covers two previous methods to point out that the complex approaches used were unnecessary. The findings will help physicians, patients, and their family members to obtain a proper medical diagnosis during multiple examinations.

Despite some interesting results from different research groups, a public database for Uyghur online handwriting recognition and a baseline study are not yet available for comparison purpose. In order to fill this void, we present a database of Uyghur online handwritten words and carry out the first benchmark experiments using it. This database contains 125,020 samples of 2030 words collected from 393 writers. According to Uyghur lexicon characteristics, two out-of-vocabulary datasets are especially provided for evaluation. We carry out some unconstrained handwritten word recognition experiments on the database using recurrent neural networks as base model. Recognition results are acquired using connectionist temporal classification without lexicon search and external language model. Concatenated and averaged bidirectional recurrent layers are compared for better generalization. Based on Uyghur unicode representation, we are interested in comparing the models using different alphabets, based both on character types and character forms. To improve generalization, we propose 1D convolutional model which implements 1D convolutional layers for sequence feature extraction. In our experiments, the proposed 1D convolutional model and its variations surpassed the base recurrent layered model on the out-of-vocabulary words by clear margin. 83.23% CAR (character accurate rate) was resulted when out-of-vocabulary samples are used for testing. The highest recognition rate is as high as 94.95% CAR when the test set shares the same lexicon to the training set. The experiments in this paper can be the baseline references for the future study using this database.

Online semi-supervised learning (OSSL) is a learning paradigm simulating human learning, in which the data appear in a sequential manner with a mixture of both labeled and unlabeled samples. Despite the recent advances, there are still many unsolved problems in this area. In this paper, we propose a novel OSSL method based on learning vector quantization (LVQ). LVQ classifiers, which represent the data of each class by a set of prototypes, have found their usage in a wide range of pattern recognition problems and can be naturally adapted to the online scenario by updating the prototypes with stochastic gradient optimization. However, most of the existing LVQ algorithms were designed for supervised classification. To extract useful information from unlabeled data, we propose two simple and computationally efficient methods based on clustering assumption. To be specific, we use the maximum conditional likelihood criterion for updating prototypes when data sample is labeled, and the Gaussian mixture clustering criterion or neural gas clustering criterion for adjusting prototypes when data sample is unlabeled. These two criteria are utilized alternatively according to the availability of label information to make full use of both supervised and unsupervised data to boost the performance. By extensive experiments, we show that the proposed method exhibits higher accuracy compared with the baseline methods and graph-based methods and is much more efficient than graph-based methods in both training and test time.

This study presents a measurement method for a bike frame to solve a problem that usually relies on the digimatic callipers or mechanical molds handled by a human, therefore avoiding a long time process and measurement errors. In such a mechanism, the measured data is based on the handwritten record, and thus a difficulty may occur in further data analysis using the computer. For this reason, this study aims to develop a mathematical model of three-dimensional geometry especially applied for robotic arm-based bike frame measurement. The proposed mathematical geometric model effectively integrates the sphere formula with the inner product of normal vector to find four parameters in the sphere formula using only three measured points. Accordingly, the centre coordinate of the check point and its diameter can be calculated accurately and simply. The practical model performance also prooves that all crucial quality checking items such as centre plane offset, inside diameter, axis point and parallelism in different rotating shafts of bike frame can be achieved in term of rapid, robustness and precision. Moreover, the measurement accuracy in the estimated standard deviation and measurement uncertainty using two different bike frames is presented to verify the effectiveness of the proposed model.

This work investigates the dynamical weighted deployment of mobile fog computing devices to support a mobile edge computing environment, in which each edge device is associated with a weight to reflect its importance based on the application. Since edge devices are mobile and could be switched off, it is challenging to dynamically optimize the deployment to adapt to dynamic change. This work further models the problem mathematically and solves it by a bat-inspired algorithm (BA), which searches the optimal solutions by simulating the food-searching behavior of bats via echolocation. Furthermore, three local search methods designed specifically for this problem are integrated into the BA, and a dynamic local search selection mechanism is proposed to adjust the probabilities of choosing the three local search methods iteratively in the BA main loop. Simulation results show outperformance of the proposed BA over the BA without local search and the previous approach.

The technique of simultaneous wireless information and power transmission (SWIPT) has been applied to wireless sensor networks, which employ static or mobile base stations (BSs) such as drones and ships to charge passively powered devices. SWIPT can be strongly expanded by solar power satellites (SPSs), which collect solar energy and transmit it to the earth through microwaves to alleviate the power shortage problem. Furthermore, multi-beam SPSs can serve a broader range than terrestrial BSs for information transmission.In 6G networks, satellites are core devices in space-terrestrial integrated networks (STINs) supporting super Internet-of-Things. However, when discussing 6G wireless systems, previous works did not consider SWIPT applied in STINs through multi-beam SPSs. Therefore, this work proposes a novel resource allocation problem for SWIPT performed by multi-beam SPSs in the STIN while optimizing the following two objectives: minimizing deficit or excess of information transmission rate and maximizing power transmission based on two receiving architectures of terrestrial devices for information decoding and energy harvesting. Different from previous works, this problem considers not only assigning power to one of multiple satellite beams but also further allocating power in each beam into two parts for information and power transmission. This problem is NP-hard as it includes an NP-hard problem. Artificial intelligence (AI) algorithms can be used to optimize the network resource management. Hence, this problem with continuous decision variables is further solved by a classical and two recent AI algorithms specially designed for continuous variables, i.e., particle swarm optimization, improved harmony search algorithm, and monkey algorithm. Through simulation, the most appropriate AI algorithms to the concerned problem are analyzed, and the results show that for the two special designed receiving architectures of the terrestrial devices, the power splitting architecture generally outperforms the time switching architecture.

A noninvasive approach for high-voltage (HV) circuit breaker (CB) diagnosis, based on the time–frequency analysis of the transient electric fields (E-fields) radiated due to switching operations outside the enclosure [switching transient E-fields (STEFs)], is presented in this paper. By use of a synthetic test platform, this paper measures the STEFs generated by CBs working in both normal and fault status and analyses their primary parameters, such as the amplitude, the center frequency, and the number of pulses. The correlation between the different working status of the CBs and the time–frequency characteristics of their STEFs is established. Based on the comparison and analysis of the main features, the STEFs could reflect the operating states and serve as an indicator of the early insulation defect of the HV CBs. The proposed defect detection method has been validated experimentally, which makes the most of the high changing rate of the induced current on the CB enclosure that itself is a weak signal and hard to capture by traditionally adopted means. This method can further be a useful reference of defect or fault diagnosis for the switchgear in operation.

PurposeTo improve the efficiency, accuracy and adaptivity of the parameter inversion analysis method of a rockfill dam, this study aims to establish an adaptive model based on a harmony search algorithm (HS) and a mixed multi-output relevance vector machine (MMRVM).Design/methodology/approachBy introducing the mixed kernel function, the MMRVM can accurately simulate the nonlinear relationship between the material parameters and dam settlement. Therefore, the finite element method with time consumption can be replaced by the MMRVM. Because of its excellent global search capability, the HS is used to optimize the kernel parameters of the MMRVM and the material parameters of a rockfill dam.FindingsBecause the parameters of the HS and the variation range of the MMRVM parameters are relatively fixed, the HS-MMRVM can imbue the inversion analysis with adaptivity; the number of observation points required and the robustness of the HS-MMRVM are analyzed. An application example involving a concrete-faced rockfill dam shows that the HS-MMRVM exhibits high accuracy and high speed in the parameter inversion analysis of static and creep constitutive models.Practical implicationsThe applicability of the HS-MMRVM in hydraulic engineering is proved in this paper, which should further validate in inversion problems of other fields.Originality/valueAn adaptive inversion analysis model is established to avoid the parameters of traditional methods that need to be set by humans, which strongly affect the inversion analysis results. By introducing the mixed kernel function, the MMRVM can accurately simulate the nonlinear relationship between the material parameters and dam settlement. To reduce the data dimensions and verify the model’s robustness, the number of observation points required for inversion analysis and the acceptable degree of noise are determined. The confidence interval is built to monitor dam settlement and provide the foundation for dam monitoring and reservoir operation management.

The high efficiency video coding (HEVC) standard provides superior efficiency for encoding and improves the compression ratio by almost 50% compared with previous video coding standards, such as advanced video coding (AVC). However, more intensive computation complexity is introduced by implementing the flexible quad tree-structured coding model. In a typical HEVC encoder, coding units (CUs) in a coding tree unit (CTU) that is built as a quad-tree structure are recursively traversed each depth level (CU size) to select the optimal coding configuration. Therefore, most of the encoding time is spent searching for the optimal coding configuration. In this paper, an efficient and fast CU size decision algorithm is proposed to reduce HEVC encoder complexity by the spatiotemporal features. First, an adaptive depth-range prediction method minimizes the possible range depth level by observing previous frames and proximal CTUs. Second, an early termination method based on the boundary examination from the de-blocking filter (DBF) prevents unnecessary calculation on small CU sizes. Furthermore, according to the sum of absolution difference (SAD), a smooth area detection mechanism is triggered when the predictive depth range excludes the largest CU size. This mechanism increase the bitrate of the CU, which contains static objects with complex textures. Compared with the HM 16, the experimental results revealed that the proposed algorithm can achieve an average 59.73% and 64.98% reduction in encoding time along with a 0.68% and 1.27% Bjontegaard Delta bitrate (BDBR) penalty for various test videos under low-delay P and random-access conditions, respectively.

Deep neural networks have gained state-of-the-art results in many applications, such as pattern recognition and computer vision. However, most of the deep neural networks are designed manually by researchers. This architecture design process is generally time consuming and needs much expertise. Hence, automatic neural network design becomes an important issue. In this paper, we propose a novel method, called DNA computing inspired networks design (DNAND), to automatically learn high performance deep networks. In DNAND, we use DNA strands to represent blocks of a model, and these DNA strands are reacted to construct the overall networks according to the base pairing principle. We also present the killing strategy, with which we stop training “bad” models if they fail to reach the specific accuracy threshold on the validation set, so as to reduce the computational cost and accelerate the learning process. Extensive experiments on image classification and detection data sets demonstrate the effectiveness of the proposed method.

In this paper, adaptive quantized state estimation fusion is deeply studied. To approach the model mismatching problem induced by random quantization, some quantized Kalman filters have been presented in the previous work, such as the quantized Kalman filter with strong tracking filtering (QKF-STF), the variational Bayesian adaptive quantized Kalman filter (VB-AQKF), and a centralized fusion frame-based complex quantized filter called variational Bayesian adaptive QKF-STF (VB-AQKF-STF). Based on the previous work for the single sensor system, a distributed complex quantized filter is designed in this paper. A novel quantized Kalman filter based on multiple-method fusion scheme (QKF-MMF) is proposed. Similar to the VB-AQKF-STF, the QKF-MMF can also realize joint estimation on the state and the quantization error covariance under the distributed fusion frame. Furthermore, it extends the single sensor results to multisensor tracking systems by using centralized and distributed fusion frames. Two multisensor quantized fusion estimators are proposed for a parallel structure with main-secondary processors in the fusion center. The weighted fusion and embedded integration ways are deeply applied to design the multisensor quantized fusion methods. The proposed work can perfect the quantized estimation algorithms and provide different choices for practical engineering applications.

This paper proposes a new distributed formation flight protocol for unmanned aerial vehicles &#x0028 UAVs &#x0029 to perform coordinated circular tracking around a set of circles on a target sphere. Different from the previous results limited in bidirectional networks and disturbance-free motions, this paper handles the circular formation flight control problem with both directed network and spatiotemporal disturbance with the knowledge of its upper bound. Distinguishing from the design of a common Lyapunov function for bidirectional cases, we separately design the control for the circular tracking subsystem and the formation keeping subsystem with the circular tracking error as input. Then the whole control system is regarded as a cascade connection of these two subsystems, which is proved to be stable by input-to-state stability &#x0028 ISS &#x0029 theory. For the purpose of encountering the external disturbance, the backstepping technology is introduced to design the control inputs of each UAV pointing to North and Down along the special sphere &#x0028 say, the circular tracking control algorithm &#x0029 with the help of the switching function. Meanwhile, the distributed linear consensus protocol integrated with anther switching anti-interference item is developed to construct the control input of each UAV pointing to east along the special sphere &#x0028 say, the formation keeping control law &#x0029 for formation keeping. The validity of the proposed control law is proved both in the rigorous theory and through numerical simulations.