Search

The optical network-on-chip (ONoC) has the advantages of high bandwidth, low latency and high performance, so it has gradually become a research hotspot of on-chip communication systems. The in-depth research on wavelength division multiplexing (WDM) technology has further improved the bandwidth of on-chip communication. What follows is the influence of the crosstalk noise which is inevitable because of the structure and material of the optical device itself. Crosstalk noise has become an important factor restricting the on-chip technology of the optical interconnect. A five-port WDM-based optical router named Nerx with high optical signal-to-noise ratio (OSNR) is proposed in this paper for reducing the insertion loss and crosstalk noise. Nerx uses fewer optical switch elements and guarantees the connectivity between any two ports. Compared with Crux and Crossbar optical routers, experiments show that Nerx can achieve 20 optical channels and has a 5.6 dB improvement in average OSNR.

With the development of Internet of things (IoT) technology, deep learning algorithm has been widely used in IoT devices. Convolutional neural network (CNN) and recurrent neural network (RNN) play a significant role in image field and sequence data respectively. In order to enable the IoT terminal SoC with limited computing power and resources to support CNN and RNN algorithms, this paper proposes a small area configurable deep neural network accelerator with fixed-point accuracy. The main computing components of CNN and RNN are implemented by hardware. Each computing module completes the calculation of complex neural network through parameter configuration and combination. In order to verify the performance of the accelerator, a SoC verification system based on Cortex-M3 was constructed. The lenet-5 network and the Long Short-Term Memory (LSTM) network with 2 layers and 128 hidden units are implemented on this accelerator, and the execution time of the two networks on Intel i5 7500, Cortex-A53 and Cortex-A7 processors is compared. The comparison results show that the CNN and RNN computing power of the accelerator exceeds that of Cortex-A53 and Cortex-A7 at the main frequency of 50MHz.

The trans-media aircraft refers to a new concept special aircraft that can both fly through the air and dive underwater, which has a broad prospect of military application. To make sure the aircraft can move smoothly in both media with low resistance, the design of the aircraft should pay attention to the needs of both air flying and underwater prowling in terms of shape and structure. This paper designs a kind of morphing wing, borrowing from bionics design philosophy, which could fold when the aircraft is diving underwater and then unfold when it is flying through the air.

The aim of this study is to improve the effect of Eulerian video magnification in non-contact heart rate measurement. Extracting the heart rate information from subtle head movements caused by blood circulation, we overcome the adverse condition that the person in the measurement should be completely still. Firstly, the temporal acceleration filter is used to preprocess the video and filter out the large low-frequency motion in the video. Secondly, face recognition is performed on each frame of the pre-processing video and face regions are cut out. The tracked face regions are reorganized to further reduce the impact of large movements. Heart rate information could be extracted from this reconstituted video. Experiment results indicate that the proposed method is more accurate in dynamic heart rate monitoring.

In this paper, we investigate the problem of verifying initial-state opacity (ISO) for partially-observed discrete-event systems (DES) by using the semi-tensor product (STP) theory. First, four types of ISO are introduced to extend the study of privacy in the context of DES. Second, under the framework of the STP theory, the matrix-based initial-state estimator is constructed. Accordingly, several necessary and sufficient criteria for verifying the different types of ISO are given, respectively. Furthermore, we analyze the complexity of using the proposed approaches to verify these four types of ISO. Finally, several examples are provided to illustrate the theoretical results.

Curiosity plays an important role in the development of biological cognition, which drives animals to explore unknown environments. Inspired by the research of psychology, this paper proposes a curiosity-based mobile robot path planning method with a mental energy function, which is suitable for robot state evaluation. In our proposed method, a novel mental energy function is designed to guide the learning direction of the robot in navigation task. The experimental results show that the novel method we proposed can effectively guide robot to learn environment information. By changing the simulation environment, robot can reach the end point through re-learning the environment, that reflects the autonomous learning ability and the environmental adaptability of robots.

Establishing correspondences between two images requires both local and global spatial context. Given putative correspondences of feature points in two views, in this paper, we propose Order-Aware Network, which infers the probabilities of correspondences being inliers and regresses the relative pose encoded by the essential matrix. Specifically, this proposed network is built hierarchically and comprises three novel operations. First, to capture the local context of sparse correspondences, the network clusters unordered input correspondences by learning a soft assignment matrix. These clusters are in a canonical order and invariant to input permutations. Next, the clusters are spatially correlated to form the global context of correspondences. After that, the context-encoded clusters are recovered back to the original size through a proposed upsampling operator. We intensively experiment on both outdoor and indoor datasets. The accuracy of the two-view geometry and correspondences are significantly improved over the state-of-the-arts. Code will be available at https://github.com/zjhthu/OANet.git., Comment: Accepted to ICCV 2019, and Winner solution to both tracks of CVPR IMW 2019 Challenge. Code will be available soon at https://github.com/zjhthu/OANet.git

This paper evaluates cost efficiency on listed Chinese water firms from 2010 to 2015 by using the stochastic frontier production analysis. Their accounting data is selected as sample. The finding suggests that the cost efficiency on Chinese water industry is improved from 0.5581 to 0.5790 due to the reform of property rights. Besides, there is still a 35% cost inefficiency among the sample enterprises. Since inefficiency exists, there is an opportunity to improve cost efficiency on Chinese water utilities. In addition, the empirical results explicate that the debt asset ratio, independent director size and per capita GDP will significantly and positively affect cost efficiency. The ratio of state owned assets and market concentration are negatively correlated with cost efficiency.

Gastric cancer is a malignant neoplasm with a high mortality rate in the world. Nearly one million new cases occur each year. The most important measure to diagnose gastric cancer is the detection and treatment of diseases early. Gastric cancer detection is currently performed by pathologists reviewing large expanses of biological tissues, but this process is labor intensive and error-prone. In this paper, a framework for automatically detection of tumors in gastric pathology image (slide) has been proposed based on deep learning. A deep residual network with 50 layers is built by identity mapping on a dataset of pathology images. The proposed method makes the training of models easier and improves the generalization performance. Finally, the experimental results show that the F-score of our method achieves 96%. The research in auto-classification of gastric pathology images has great value for gastric cancer detection in clinical medicine.

This paper addresses the optimal operation issues for virtual power plant (VPP). Two conflicting objectives are considered and a bi-objective problem is formulated mathematically. The first objective (economical performance) contains cost of fuel consumption and cost of power purchased from grid. The second objective (environmental performance) takes account of several pollutant emissions. Multi-objective particle swarm optimization (MOPSO) is applied to derive the Pareto frontier which is a set of noninferior solutions. A realistic case study is performed on Hongfeng Eco-town as a model of VPP, which includes various distributed energy resources (DER) such as hydropower units, wind turbines, photovoltaic panels, combined cooling heating and power units (CCHP) and storage devices. Simulation concentrates on three strategies: 1) money-oriented, 2) environment-only, 3) Pareto-optimal. Comparison of simulation results reveals that Pareto-optimal strategy can better compromise costs and emissions. Further, system operators could choose noninferior solutions from Pareto frontier according to their specific preference.

Virtual power plant (VPP) is a main method of solving grid-integration of distributed generators (DG). This paper proposes an hour-ahead cooperation model of VPPs, including wind turbines, photovoltaic panels, controllable load and energy storage system (ESS), and focuses on hour-ahead cooperative space of VPPs, considering uncertainties of wind and photovoltaic power. In order to obtain the day-ahead results of VPPs, a day-ahead independent scheduling model is built as well. Typical scenarios with probability information are used to describe the uncertainties of wind and photovoltaic output. Roth-Erevalgorithm and clearing mechanism are applied to simulate hour-ahead cooperation process. The case study compares the results before cooperation with that after cooperation, which proves that the cooperation can reduce day-ahead bidding deviations from the actual output and increase profits. Besides, the number of VPPs and the involvement of demand response (DR) and ESS also have an impact on the results, which provides a reference for the cooperation of VPPs in the hour-ahead market.

Inverts, especially multi-level inverters are widely used in many fields, such as industrial production, transportation, aviation and so on. So great significance should be attached to the diagnosis and fault tolerance of inverters to keep the stability of systems. Data-driven approaches make full use of the process data to monitor the systems, so the voltage signals are collected firstly and then preprocessed and processed by specific strategy, fault labels will be produced hereafter. When the fault labels from data-driven fault detection and diagnosis system are generated, relevant fault tolerant control method will be activated in fault tolerant control system. Some measurements are necessary to achieve the higher utilization ratio of healthy IGBTs and sinusoidal output voltage. Based on above consideration, a group isolation and reconfiguration fault tolerant control method based on data-driven methodology for cascaded seven-level inverter is proposed here to reconfigure the SPWM, in which every H-bridge is divided into two groups. The simulation of cascaded seven-level inverter is built and the result indicates that the utilization of healthy IGBTs is improved.

In the process of comprehensive evaluation for guided and control law performance of aircraft, index weight value directly affects the evaluation result. In this article, the multiple indexes system for guided and control law performance of aircraft is established. An optimized method is put forward for weight determination of indexes, which is based on analytic hierarchy process and entropy weight method, in condition of multple criteria. And a case study shows that the method takes account of more factors and makes it more reasonable.

This paper focuses on the day-ahead active power dispatch for interconnected power systems with wind power integration via HVDC tie-line. Considering the complex operating characteristics of HVDC tie-line, a decentralized dispatch model based on analytical target cascading (ATC) technique is proposed. The day-ahead power dispatch is divided into the master problem of the upper-level coordination dispatch and the subproblem of the lower-level regional dispatch. The upper-level dispatch center is responsible for coordinating the HVDC tie-line power flow between the interconnected areas, and the lower-level dispatch centers solve the security-constrained unit commitment (SCUC) problem of each area independently. Through the decomposition and coordination mechanisms, the economy of the entire interconnected system is achieved. A 12-bus two area system interconnected via HVDC tie-line is utilized to verify the effectiveness of the proposed model.

A potential application of LiDAR data in street-level greenery assessment has been explored in this study. Taking full advantage of the fine-scale tree structure modeling of LiDAR data, we propose a new green view index calculation method to assess street greenery visualization. Results indicate that LiDAR data is capable of differentiating the amount of greenness being perceived from different sites and the calculation method gives an objective measure of street-level greenery.

An analysis of oil thermosyphon circulation and winding hot-spot temperature in Oil Naturally Cooled (ON) transformer system has been conducted in this study. The oil circulation is analyzed as a flow network of multiple thermosyphon loops, through winding columns, iron cores, oil tank, radiators and other transformer units. A network-based non-isothermal hydraulic model has been developed to predict oil flow, pressure, and temperature distribution over the entire oil circulation loops, in which, the hydraulic and thermal fields at each of transformer components are simulated with individual sub-models. The secant numerical method is employed to guarantee a convergent solution. The predicted oil temperatures have a good agreement with actual heat-run testing data. With the determination of the global oil distribution, winding temperature can be obtained from winding sub-model developed by authors in previous studies. Taking oil pump as a pressure source, the global model can also be applied in the analysis and optimization of forced oil transformer, as well as ON transformer.