Search results

The conventional pedestrian detection algorithms lack in scale sensitivity. The purpose of this paper is to propose a novel algorithm of self-adaptive scale pedestrian detection, based on deep residual network (DRN), to address such lacks.

First, the “Edge boxes” algorithm is introduced to extract region of interests from pedestrian images. Then, the extracted bounding boxes are incorporated to different DRNs, one is a large-scale DRN and the other one is the small-scale DRN. The height of the bounding boxes is used to classify the results of pedestrians and to regress the bounding boxes to the entity of the pedestrian. At last, a weighted self-adaptive scale function, which combines the large-scale results and small-scale results, is designed for the final pedestrian detection.

To validate the effectiveness and feasibility of the proposed algorithm, some comparison experiments have been done on the common pedestrian detection data sets: Caltech, INRIA, ETH and KITTI. Experimental results show that the proposed algorithm is adapted for the various scales of the pedestrians. For the hard detected small-scale pedestrians, the proposed algorithm has improved the accuracy and robustness of detections.

By applying different models to deal with different scales of pedestrians, the proposed algorithm with the weighted calculation function has improved the accuracy and robustness for different scales of pedestrians.

For the large-scale power grid monitoring system equipment, its working environment is increasingly complex and the probability of fault or failure of the monitoring system is gradually increasing. This paper proposes a fault classification algorithm based on Gaussian mixture model (GMM), which can complete the automatic classification of fault and the elimination of fault sources in the monitoring system.

The algorithm first defines the GMM and obtains the detection value of the fault classification through a method based on the causal Mason Young Tracy (MYT) decomposition under each normal distribution in the GMM. Then, the weight value of GMM is used to calculate weighted classification value of fault detection and separation, and by comparing the actual control limits with the classification result of GMM, the fault classification results are obtained.

The experiment on the defined non-thermostatic continuous stirred-tank reactor model shows that the algorithm proposed in this paper is superior to the traditional algorithm based on the causal MYT decomposition in fault detection and fault separation.

The proposed algorithm fundamentally solves the problem of fault detection and fault separation in large-scale systems and provides support for troubleshooting and identifying fault sources.

In order to solve the current imbalance of academic resources within the discipline, this article builds a three-dimensional talent evaluation model based on the topic–author–citation based on the z index and proposes the ZAS index to evaluate scholars on different research topics within the discipline.

Based on the sample data of the CSSCI journals in the discipline of physical education in the past five years, the keywords were classified into 13 categories of research topics including female sports. The ZAS index of scholars on topic of female sports and so on was calculated, and quantitative indexes such as h index p index and z index were calculated. Comparative analysis of the evaluation effect was performed.

It is found that compared with the h index and p index, the z index achieves a better balance between the quantity, quality and citation distribution of scholars' results and effectively recognizes that the citation quality is higher and the number of citations of each paper is more balanced. In addition, compared to the z index, this article is based on a ZAS index model with an improved three-dimensional topic–author–citation relationship in research fields such as female sports.

It can identify some outstanding scholars who are engaged in small-scale or emerging topic research such as female sports and are excellent in different research areas. Talents create an objective and fair evaluation environment. At the same time, the ranking ability of ZAS indicators in the evaluation of talents is the strongest, and it is expected to be used in practical evaluations.

The purpose of this paper is to provide an exploratory understanding of the impact of social support on job stress of shift working mothers in Macao casinos.

A qualitative research method was adopted in this study. A snowball sampling strategy was used to recruit participants, and in-depth semi-structured interviews were conducted on participants from different casinos of six Macao gaming operators.

The paper found out that emotional and instrumental supports from families and friends are the most important social support for shift working mothers to cope with their job stress in casinos.

A qualitative method may lead to a misinterpretation or ignorance of key information due to the researcher bias. The snowball sampling strategy may lead to a biased sample which may not represent the entire population.

The understanding of what types of social support that shift working mothers appreciate and need the most may provide managerial implications for casino operators for a better support to their female employees.

Aiming at the problems of poor accuracy and limitation in strength assessment of spot welding vehicle body caused by uncertain factors, such as key component size and nugget diameter, the numerical models of strength uncertainty analysis of spot-welded joints were constructed based on evidence theory and fuzzy theory.

Evidence theory and fuzzy theory are used to deal with the uncertainty of design parameter, and differential evolution algorithms are used to calculate the propagation process of uncertainty in this model. Furthermore, efficient relationship between the strength of welded joints and each design parameter is constructed by using response surface proxy model, which effectively avoids the problem of repeated complex finite element analysis in uncertainty analysis.

The results show that the constructed uncertainty numerical model is effective for the multiple uncertainties and give interval results under different probabilities and affiliations, which can more effectively evaluate the strength of the welded body structure to avoid overly conservative estimates for deterministic design.

The evidence theory is improved and combined with differential evolution algorithm and response surface method to effectively improve the computational efficiency. Based on the improved evidence theory and fuzzy algorithm, the numerical models for the uncertainty analysis of solder joint strength of welded structures are constructed and their feasibility is verified.

The purpose of this paper is to discuss the autonomous navigation and guidance scheme for future precise and safe planetary landing.

Autonomous navigation and guidance schemes are proposed based on inertial measurement unit (IMU) and optical navigation sensors for precise and safe landing of spacecrafts on the moon and planetary bodies. First, vision‐aided inertial navigation scheme is suggested to achieve precise relative navigation; second, two autonomous obstacle detection algorithms, based on grey image from optical navigation camera and digital elevation map form light detection and ranging sensor, respectively, are proposed; and third, flowchart of automatic obstacle avoidance maneuver is also given out.

This paper finds that the performance of the proposed scheme precedes the traditional planetary landing navigation and guidance mode based on IMU and deep space network.

The presented schemes need to be further validated by the mathematical simulations and hardware‐in‐loop simulations, and then they can be used in the real flight missions.

The presented schemes are applicable to both future planetary pin‐point landing missions and sample return missions with little modification.

This paper presents the new autonomous navigation and guidance scheme in order to achieve the precise and safe planetary landing.

As a response to the growing operational and disruptive threats to water distribution networks (WDNs), researchers have developed a vast array of methods for the reliability analysis of WDNs. In order to follow this growing number of methods, this paper reviews and documents in one place the historical developments in the reliability analysis of WDN.

A systematic literature review (SLR) is carried out to summarize the state-of-the-art research on reliability analysis of WDNs. In conducting this systemic literature review, the authors adopted an iterative approach to define appropriate keywords, analyze and synthesize data and finalizing the classification results.

First, the hydraulic approach to reliability analysis is currently pervasive, and relatively little academic research has addressed the topological reliability analysis of WDNs. Second, in order to provide a comprehensive picture of the network reliability, a different approach that integrates topological and hydraulic attributes seems a more effective method. Third, the conventional reliability analysis methods are only effective for demonstrating a snapshot of these networks at a given point in time. The availability of methods that enable researchers to evaluate the reliability in response to changes in its variables is still a major challenge.

The present paper facilitates future research in the reliability analysis of WDNs by providing a source of references for researchers and water utilities. Further, this article makes a contribution to the literature by offering a roadmap for future reliability analysis of WDNs by reviewing the evolution of the current reliability analysis methods throughout history.

Assembly sequence optimization is a difficult combinatorial optimization problem having to simultaneously satisfy various feasibility constraints and optimization criteria. Applications of evolutionary algorithms have shown a lot of promise in terms of lower computational cost and time. But there remain challenges like achieving global optimum in least number of iterations with fast convergence speed, robustness/consistency in finding global optimum, etc. With the above challenges in mind, this study aims to propose an improved flower pollination algorithm (FPA) and hybrid genetic algorithm (GA)-FPA.

In view of slower convergence rate and more computational time required by the previous discrete FPA, this paper presents an improved hybrid FPA with different representation scheme, initial population generation strategy and modifications in local and global pollination rules. Different optimization objectives are considered like direction changes, tool changes, assembly stability, base component location and feasibility. The parameter settings of hybrid GA-FPA are also discussed.

The results, when compared with previous discrete FPA and GA, memetic algorithm (MA), harmony search and improved FPA (IFPA), the proposed hybrid GA-FPA gives promising results with respect to higher global best fitness and higher average fitness, faster convergence (especially from the previously developed variant of FPA) and most importantly improved robustness/consistency in generating global optimum solutions.

It is anticipated that using the proposed approach, assembly sequence planning can be accomplished efficiently and consistently with reduced lead time for process planning, making it cost-effective for industrial applications.

Different representation schemes, initial population generation strategy and modifications in local and global pollination rules are introduced in the IFPA. Moreover, hybridization with GA is proposed to improve convergence speed and robustness/consistency in finding globally optimal solutions.

In recent years, centrality measures have been extensively used to analyze real-world complex networks. Water distribution networks (WDNs), as a good example of complex networks, exhibit properties not shared by other networks. This raises concerns about the effectiveness of applying the classical centrality measures to these networks. The purpose of this paper is to generate a new centrality measure in order to stick more closely to WDNs features.

This work refines the traditional betweenness centrality by adding a hydraulic-based weighting factor in order to improve its fit with the WDNs features. Rather than an exclusive focus on the network topology, as does the betweenness centrality, the new centrality measure reflects the importance of each node by taking into account its topological location, its demand value and the demand distribution of other nodes in the network.

Comparative analysis proves that the new centrality measure yields information that cannot be captured by closeness, betweenness and eigenvector centrality and is more accurate at ranking the importance of the nodes in WDNs.

The following practical implications emerge from the centrality analysis proposed in this work. First, the maintenance strategy driven by the new centrality analysis enables practitioners to prioritize the components in the network based on the priority ranking attributed to each node. This allows for least cost decisions to be made for implementing the preventive maintenance strategies. Second, the output of the centrality analysis proposed herein assists water utilities in identifying the effects of components failure on the network performance, which in turn can support the design and deployment of an effective risk management strategy.

The new centrality measure, proposed herein, is distinct from the conventional centrality measures. In contrast to the classical centrality metrics in which the importance of components is assessed based on a pure topological viewpoint, the proposed centrality measure integrates both topological and hydraulic attributes of WDNs and therefore is more accurate at ranking the importance of the nodes.

The purpose of this ontologically qualitative research study was to (a) explore student narratives regarding their educational experiences in at-home internationalization programs; (b) provide an in-depth narrative analysis of student learning challenges and achievements; and (c) add valuable research-based knowledge of student-described experiences for use by program administrators.

Participants were selected with a form of four-stage non-proportional stratified sampling. 29 participants were interviewed using a basic demographic questionnaire and an episodic interview protocol. Data were analyzed in QSR NVivo software through open, axial, and selective coding stages under the framework of grounded theory.

The findings focus on student-identified links between the challenges they encountered and their achievements. In addition, student performance level and gender were associated with the challenges and achievements reported by students. In understanding the results, the student-learning concepts found in the learned optimism, growth mindset, grit and expectancy theory approaches provide potentially fruitful insights.

The findings of this research have instructive implications for program administrators regarding how student challenges can be strategically chosen and shaped to generate specific, positive student achievements.