Search

Network representation learning has attracted more and more attention, whether in academia or industry. It can convert large scale network data into low dimensional node embedding for various tasks, such as node classification, link prediction, etc. Recently, representation learning on dynamic networks has emerged, which is more similar to the situation in real life. Inspired by the interaction between entities and entities in real life, we propose a model to embedding temporal network via interest vectors (ETIV). The interest vector of an entity can be generated according to the historical entities of the interaction. Interest vector can infer the possibility of interaction between entities. Then, we can obtain the node representation by optimizing the possibility of interaction. For calculating the interest vector of a node, we use a multi-head attention mechanism to capture the information of the historical interaction nodes from different aspects. Moreover, according to the interaction time of historical nodes, we introduce a learnable time parameter to simulate the forgetting of historical information. We conduct experiments on three real data sets and find that our model performs better than state-of-the-art methods in various tasks.

A model based internal short circuit fault diagnosis method for a series-connected battery pack under varying temperature is proposed in this paper. Systematic experiments are conducted to study the relationship between battery capacity, internal resistance and temperature. Internal short circuit fault can be diagnosed based on the decay of dischargeable capacity. To reduce the computational complexity of estimating state of charge(SOC) and diagnosing faults online, a dual time-scale approach is proposed to select standard cell and updating cell. Based on the approach, unscented kalman filter and extended kalman filter are utilized to estimate battery pack SOC. With the accurate SOC, recursive least squares algorithm is utilized to quantitatively diagnose the fault. Finally, dynamic stress test on a series-connected battery pack is conducted to evaluate the proposed method. The results indicate that the method can accurately diagnose the fault in a reasonable time.

Contemporary urban living environment design and construction process are facing lots of ecological issues. It needs us to find a new perspective and efficient design approach which can become the bridge or link between special form and optimized function structure, natural and human resources protection, traffic and physical environment improvement, land-use saving and energy sustainability, landscape environment and behavior system design, etc. Based on Urban Design subject, Ecological Urban Design regards special environment system as research object, uses Ecology principles for reference, harmonizes relationship of natural and built environment and urban inhabitant, and becomes an important method and theory to adjust urban special environment ecology. For promoting Ecology Urban Design theory developing, it is necessary to create a research and practice platform for Urban Design and Ecology. Meanwhile, we need not only absorb foreign experiences, but also recognize our country objective condition and demand, explore Chinese Ecological Urban Design theory and practice. It has an important meaning to build up our country living ecological environment and develop local Urban Design theories.

In practice, such a problem is often encountered that the proportion of target to background in image segmentation is very small. For this, a thresholding method used for the image segmentation of infrared small target is proposed in this paper, which is based on the area difference between background and target and intra-cluster variance. This method points out the existing methods of image threshold segmentation can not effectively extract infrared small targets, and offers the thresholding formula of the area difference between background and target and intra-cluster variance based on 2-D histogram of vertical and more effective oblique. It puts forward the niche chaotic mutation particle swarm optimization algorithm based on 2-D histogram of vertical and oblique. Finally the result and running time of this paper are given out in the result and analysis of experimental, which is compared with the NCPSO of 2-D Otsu, maximum and Fisher. The results show that the method in this paper can accurately extract infrared small targets, and has short running time and strong anti-noise performance.

The marine power plant traditional design method was popular introduced when the computer design and application had not yet appeared. Based on the analysis of the problems of the traditional design method of the marine power plant, the issue that applying digital design method to the design process of the marine power plant was studied in this paper. The digital design scheme for the marine power plant was proposed, and the key problems was studied. The research in the paper could help to improve the design of the marine power plant.

From the aspect of methods, technique and data, bringing forward that the spatial organization of FNLCR has spatialized and digitalized characteristics. In the research, the main requirement of data includes the data of 2D vector graphics, RS image, the data of digital elevation models, the data of various statistics, and the data of 3D urban graphics and texture, etc.. Bringing forward the basic logic of the spatial organization research of FNLCR, and discuss the division of the factors.

In terms of methodology, technology and data demand, the study on spatial organization of functional network in low-carbon region (FNLCR) is identified by the spatialized and digitalized characteristics. Therefore, digital city technology (DCT) becomes important in the study of spatial organization of FNLCR. The research components and ideas are provided, and the research approach named digital low-carbon region (DLCR) with broad connections with DCT is proposed. An initiative framework of spatial organization of FNLCR is presented based on the platform of DCT, and it will help to establish a complete routine of digital low-carbon planning (DLCP) on regional scale.