Search

In this paper, we have developed a new method of accurate detection of retinal blood vessels based on a deep convolutional neural network (CNN) model. This method plays an important role in the observation of many eye diseases. Retinal Images have many issues that make the process of vessels segmentation very hard. We treat each issue of the retina image with the greatest observation to obtain a well-segmented image. The first step is to apply a pre-processing method based on fuzzy logic and image processing tactics. In a second step, in order to generate the segmented images, we propose a strided encoder-decoder CNN model. This network is trained and optimized using the Dice Loss function that supports the class imbalance problem that is in the database. The proposed model has a U-Net shape, but it is deeper and the pooling layers are replaced with strided convolutional layers in the encoder. This modification allows for a more precise segmentation of vessels and accelerates the training process. The last step is post-processing for removing the noisy pixels as well as the shadow of the optic disc. The performance of the proposed method was evaluated on DRIVE and STARE databases. The proposed method gives a sensitivity of 0.802 and 0.801 respectively on DRIVE and STARE, with an accuracy of 0.959 and 0.961 respectively. We focused on sensitivity and accuracy measurements that represent the accuracy of the model, especially tiny vessels. According to the results, the model outperforms many other proposed methods, especially in the above-mentioned measures.

We demonstrate a broadband and polarization selective terahertz modulator with ultrafast modulation speed, based on the Si grating metasurface on the Sapphire substrate. The Si layer is implanted with B+ and subsequently annealed, which significantly reduces the lifetime of carriers and increases the carrier density, and thus realizes ultrafast and efficient modulation. Our modulator achieves modulation depth up to 60% for the x-polarized THz wave and less than 1% for the y-polarized THz wave in the frequency ranging from 0.5 to 2.5 THz, with switch-on and switch-off time of modulation in 20 ps and around 300 ps, respectively. The carrier dynamics of the fabricated Si grating is also studied to illustrate the ultrafast modulated mechanism. Our results pave the way for ultrafast manipulation of terahertz spectrum and shows great potential in terahertz communication.

Based on the suitable artificial intelligence model, this paper establishes the short-term passenger flow forecasting model of subway, studies the correlation characteristics that affect the short-term passenger flow forecasting of subway, and combines with the existing results, selects five related forecasting variables through data analysis, and uses copula correlation coefficient and Granger causality correlation analysis to process a large number of swipe card data. The weather factors, holidays, peak, the number of relevant stations and the number of outbound factors, are analyzed. By extracting the features that affect the accuracy of passenger flow forecast, using multiple constraints and taking the 15 minute interval as the time granularity, a passenger flow forecast method based on the combination of inbound passenger flow feature fusion and neural network prediction model (LSTM) is proposed. It not only improves the accuracy of passenger flow prediction, but also provides the basis for management decision-making, and provides reference information for people's travel.

This paper studies a distributed heuristic algorithm for the generation and adjustment of the formation of homogeneous multi-rotor UAVs to improve the applicability and flexibility of multi-rotor UAV dense formation. Each drone makes an independent decision using a multi-agent framework based on the received satellite positioning, communication information, and obstacle avoidance sensor information. For the motion mode of a multi-rotor UAV, the action decision method in a dynamic environment is selected, avoiding complicated problems such as sensor handover and prejudging the moving trajectory. The Hungarian algorithm is used to match the target positions to improve the formation efficiency. The distance between each drone and each target position as the element in the cost matrix can ensure that the path does not cross, which is used for the first match when forming the team; the cost matrix element is set to the square of the distance, which can be used for formation transformation and dynamic adjustment. Through simulation, it is verified that the algorithm with action decision and target position matching can be used for dense formation of homogeneous multi-rotor UAVs and that the process is safe and efficient.

There are frequent reports of cable rupture accidents on the transmission line crossing high-speed railways. A simulation method for dynamic response and fatigue life of transmission lines considering natural wind and train induced wind is presented and verified by field tests. Based on a case study, the displacement time history and fatigue damage of conductors under wind loads are analyzed, and the effects of different parameters are discussed. The results show that the fatigue damage caused by train induced wind can reach 127% of natural wind, and its influence on the fatigue life of transmission lines cannot be ignored. Raising height or avoiding the most unfavorable span length can effectively reduce the influence of train induced wind. The proposed method can be used for the structural design and safety assessment of transmission lines crossing the high-speed railway, and it can provide references to the design of damping devices.

In recent years, the influence of dust on the external insulation of power system can not be ignored, but there are not many related researches at home and abroad. In order to study the discharge characteristics under dust environment, this paper selects the rod-plate gap as the research object and builds the dust simulation experiment. The platform is used to conduct experimental research on AC discharge characteristics, and the influence of factors such as gap distance and dust volume fraction on the AC-discharge of the rod-plate gap is obtained. When dust volume fraction is relatively small, AC breakdown voltage of the tip-plate gap decreases with increase of dust volume fraction And the voltage increases with increase of dust volume fraction when sand dust volume fraction is larger. It can be considered that there is a certain dust volume fraction (between 0.5% and 4%), so that the AC gap breakdown voltage is the smallest, and the volume fraction is independent of the gap distance. It can be considered that there is a fixed dust volume fraction (between 0.5% and 4%) that minimizes the gap AC breakdown voltage, which is independent of the gap distance.

The modeling and control problem of folding aircraft performing morphing-aided maneuver is studied in this paper. The relationship between the aircraft folding angle and the aerodynamic parameters is analyzed and modeled. The wing folding angle is viewed as an extra control input and the dynamic model of the morphing-aided maneuver is constructed. Considering the high nonlinearity and the hybrid disturbances of the system, a novel feedback linearization-based non-singular dynamic terminal sliding mode control (NDTSMC) approach is proposed. The convergence of the aircraft states within a finite time is proved to be guaranteed by NDTSMC, and chattering is eliminated by the integration of the switching control signals. Simulation results show that, the control precision and the robustness of the morphing aircraft are improved by NDTSMC, with a better chattering elimination. Higher maneuverability and disturbance rejection capability compared to conventional aircraft are also demonstrated by simulation.

Previous work on image clustering focused on seeking a low-dimensional structure from the high-dimensional image data by a shallow linear model, such as sparse subspace clustering (SSC) or low-rank representation (LRR). The recent advance of deep learning shows its superiority via handling data with nonlinear structure, i.e., sparse auto-encoder and independent subspace analysis(ISA), etc. However, most of this type of methods may ignore lots of useful information embedded in the original data. To this end, we propose a novel unsuper-vised learning algorithm via ISA incorporating the subspace structure within data. Specifically, we adopt the ISA to learn local translation invariant feature from data and integrate a prior subspace information into the output of the network simultaneously. This method performs an impressive powerful ability to learn the nature of data. By evaluating on public databases, CMU-PIE and ORL, the experimental results show that the proposed approach achieves better clustering results compared with the state-of-the-art ones.

Fan-out wafer level packaging (FOWLP) not only provides simplified supply chain management and lower cost structure, but also enables thinner profile and heterogeneous system integration. FOWLP is becoming increasingly significant and is projected to drive growth in advanced packaging for the foreseeable future. There are many different processing technologies for fabricating FOWLP. One common key practice, which is very different from fan-in wafer level packaging, is the use of a temporary carrier to support wafer-level fabrication. The redistribution-layer (RDL) first approach is one of two mainstream processing technologies for FOWLP at present. One benefit is that the RDL is fabricated with direct support of a flat, rigid carrier prior to the occurrence of molding warpage and die shift. However, the RDL-first approach requires a carrier sacrificial layer that can withstand high-temperature/high-vacuum RDL build-up fabrication. Determined by adhesive chemistry's availability, the present forms of RDL-first FOWLP processing require carrier release by laser ablation, thus further limiting the choice of carrier to glass. At present, laser debonding for RDL-first FOWLP is a very costly and lengthy process. This paper presents a design for optimizing and processing a carrier-sacrificial layer that is compatible not only with current RDL-first FOWLP fabrication, but also enables the carrier's instant release by air jetting at room temperature. This air-assisted mechanical release of the carrier minimizes debonding stress on the wafer surface without localized heating and burning, and provides even more stress relief for larger carriers. A fan-out WLP process flow with air jetting carrier release is presented and evaluated. Key material properties of the current sacrificial-layer design are also analyzed.

To find the relation between conflict and employee satisfaction in geographically distributed software development (GDD) team, study the characters of GDD team and put forward four types of team conflict: emotional conflict, task conflict, cognitive conflict, and process conflict. Besides, bring in team interaction as an influencing factor of team conflict. Then come up with hypotheses about the impact of team conflict on employee satisfaction. And use collected data to carry out regression analyze. The results show that emotional, task and cognitive conflict all have significant influence on employee satisfaction, instead of process conflict. And emotional conflict and task conflict can change with team interaction. The results can provide some help with handling conflict more efficient and improving the employee satisfaction in GDD team.

An 80Kb logic Embedded Multi-Time Programmable Memory (MTPM) employs charge trapping and de-trapping behavior in 32nm/22nm High-K transistor, resulting in no added process complexity. Multi-step verification with overwrite protection employs block-write and signal margin degradation (∼30%) to satisfy 10 year retention at 105° C.

Depending on the material, the currently available wafer debonding mechanisms include thermal slide, peeling, or laser ablation. These all require physical exposure to the thinned device wafer. On the other hand, typical batch debonding requires chemical diffusion through a thin gap or perforated holes in the carrier, which is usually considered high cost and with extremely low throughput. In this paper we present the AirDebond technology, which enables instant wafer debonding at room temperature with extremely low-stress air jetting. With an isolation film layer of precisely controlled low peel adhesion on its surface, the carrier stacks the device wafer with a bonding adhesive to support fabrication processing at a later time. By breaking the isolation film sealing around the carrier, we inject a stream of airflow into the interface between the film and the carrier to separate the two, thus lifting the carrier from the stack. A full process evaluation on a 200-mm wafer is presented, which includes the isolation film properties and coating specifications. To provide good mechanical support while enabling the ease of separation with air jetting, the material and adhesion properties of the isolation film were optimized and are thoroughly discussed in this paper.

The scientific cost drive Parameters identification and screening can improve the civil airplane development cost estimation accuracy, and then realize control and management of the development cost effectively. According to the estimation of the conceptual design stage of civil aircraft cost is studied, considering the lack of the information of civil aircraft cost parameters, parameters have been collected exist the problem of Multicollinearity, firstly, through qualitative identification method of civil aircraft cost parameter system and construct a generalized grey relative correlation model and relational clustering model combination method of screening cost drive parameters, algorithm steps are given; and collected to verify the model at home and abroad of 9 different kinds of civil aircraft information, the result shows that the proposed model can effectively screening exit key cost drive parameters. The results show that the prediction method established in this paper has improved significantly on the accuracy, illustrating the effectiveness and scientific of this method.

This paper puts forward a method of fault prognostic based on least squares support vector machine for electronic equipment, because fault information of the electronic equipment is insufficient, high incidence of failure. Firstly it introduces the basic principle and the fault prognostic algorithm of the LSSVM process to fault prognostic. Then, fault prognostic research for the electronic equipment can be from a similar analog band-pass filter circuit. Compared to the least square method and support vector machine method, LSSVM method may apply to the different fault condition of the circuit, which obtained the different results. It shows that the proposed method can achieve graded fault prognostic for the analog circuit, and has the better prediction effect.

Sequential data are ubiquitous in data analysis. For example hyperspectral data taken from a drill hole in geology, high throughput X-ray diffraction measurements in materials research and EEG brain wave signals in neuroscience. The common feature of sequential data is that they are all acquired subject to one external variable such as location, time or temperature. The data evolve along the direction of that variable through several patterns and the “neighboring” data are very likely to share similar features. The purpose of the segmentation for sequential data is then to identify those sequentially continuous segments/patterns. We approach this problem by adopting the subspace clustering method and propose a novel algorithm called low rank sequential subspace clustering (LRSSC), inspired by another method called spatial subspace clustering (SpatSC). SpatSC finds the subspaces by data self-reconstruction with a sparsity constraint on reconstruction weights and promotes the spatial smoothness of the weights by fusion, the essential part in the fused LASSO. However, the subspace identification capability is limited due to the indeterminacy of the sparse regression in finding suitable samples to linearly reconstruct a given sample. This confuses the graph cut algorithm that produces the final clustering results on the weights. To overcome this drawback, we propose to use the low rank penalty instead of sparsity in learning phase to separate subspaces. This improves the subspace identification as well as the robustness to noise. To demonstrate its effectiveness, we test LRSSC on both simulated and real world data compared with SpatSC and other methods. The proposed algorithm is superior to others when noise level is very high.

This paper presents the industry's smallest Embedded Dynamic Random Access Memory (eDRAM) implemented in IBM's 22nm SOI technology. The bit cell area of 0.026µm2 achieves ∼60% scaling over the previous generation with deep trench (DT) capacitance optimized for performance and retention requirements. We report, for the first time, the asymmetric embedded stressor, cavity implant, through gate implant, and substrate n-band innovations to maintain aggressive cell scaling for the 22nm eDRAM technology.

This paper presents the modeling and analysis of the longitudinal dynamic of variable-sweep aircraft. The variable-sweep aircraft have complex nonlinear and time-varying characteristics. To capture the complicated characteristics of variable-sweep aircraft, a systematic modeling method of the longitudinal dynamics is investigated. By using Digital Datcom software, aerodynamic data of the morphing aircraft is obtained; then the nonlinear dynamic models of the variable-sweep aircraft is established. To facilitate future control design, a T-S fuzzy model was constructed based on the nonlinear model. Both models are simulated and analyzed. Comparison verifies the effectiveness of the T-S fuzzy model. The characteristics of the longitudinal dynamic are analyzed based on simulation results.

In this paper, a novel image blocky artifact removal scheme based on low-rank matrix recovery is proposed. The problem of suppressing blocky artifacts is formulated as recovering a low-rank matrix from corrupted observations. During the deblocking processing, we do not directly recover the whole clean image but only its high-frequency component and then synthesize the clean image by incorporating the low-frequency component of blocky image. To take advantage of the low-rank matrix recovery paradigm, we first cluster the similar patches of the high-frequency component of image via local pixel clustering, then the clean high-frequency component of image is recovered by formulating an optimization problem of the nuclear norm and l 1 -norm. The experimental results show that the proposed algorithm can achieve competitive performance in terms of both quantitative and subjective quality.

An entirely wireless neural and behavior recording platform has been developed for the purpose of studying cortical circuits during behavior of a freely moving animal with minimal observer intrusion. Our platform consists of (a) a 100-element microelectrode, fully implantable, wireless, broadband neural recording device [1], and (b) a 2-channel, wireless, event tracking system, both of which simultaneously stream asynchronous data to a custom programmed FPGA (field-programmable gate array). All data is packetized and transmitted over Ethernet. The data are received by custom software (in MATLAB) which simultaneously animates neural data and behavioral data for the observer to draw correlations and witness patterns. This platform has been implemented in a 10-month old female Yucatan mini-pig swine model (with the chronic, wireless, neural implant for over 9 months), extracting broadband neural data recorded simultaneously with time stamped behavioral recordings.

Future brain-computer interfaces will benefit from technologies that enable efficient, reliable and high-speed transmission of multichannel neural data. As one approach, we present a miniature multichannel neural recording device that utilizes infrared (IR) fiber optics for dual purpose of powering active electronic circuits in the compact head-mounted module and extracting of broadband multichannel neural data. We demonstrate the microsystem's chronic performance in awake, behaving rats with intracortical neural sensors. In a prototype for small animal model, the microelectronic device architecture consisted of (i) an implantable `front-end' carrying an ultra-low power 16-channel preamplifier and a multiplexer chip integrated directly onto a 4×4 cortical microelectrode array to access neural activity, and (ii) an ultra-light `back-end' housing an analog-to-digital converter, a low power digital controller chip, an IR photovoltaic device and an IR microlaser, the latter to convert multiplexed multichannel neural data to a stream of IR pulses. The implanted device was able to monitor motor cortical activity in rats during free behavior for more than a month without degradation of device performance.

Compressive Sensing (CS) theory witnesses great breakthrough in signal acquisition and processing. In this paper we focus on the application of direction of arrival (DoA) and proposed a compressive sensing based direct DoA estimation framework (CS-DDoA) for wireless sensor array network. Unlike the former signal reconstruction-processing scheme, CS-DDoA utilize the spatial and spectral sparsity of target, then reformulates them into a global problems. It eliminates the error propagation between this two independent procedure and provides a enhanced DoA performance. Under this framework, array data transmission volume can be greatly reduce without distinct performance decline, which has great potential for wireless sensor array network with limited resources. At last theoretical analysis and prototype system are provided to validate this framework.

We present AWSAN, a adjustable wireless sensor array network based target monitoring system. It is universal for different scenarios and convenient for deploymen- t. A compressed sampling scheme is introduced to greatly reduce the data exchange volume, moving most processing load to fusion center. It provides similar performance to the traditional wireless or fixed array while using low-cost & low-power nodes.

Image super-resolution means forming high-resolution images from low-resolution images. In this paper, we develop a new approach based on the deep Restricted Boltzmann Machines (RBM) for image super-resolution. The RBM architecture has ability of learning a set of visual patterns, called dictionary elements from a set of training images. The learned dictionary will be then used to synthesize high resolution images. We test the proposed algorithm on both benchmark and natural images, comparing with several other techniques. The visual quality of the results has also been assessed by both human evaluation and quantitative measurement.

Feature extraction is one of the most fundamental problems in face recognition tasks. In this paper, motivated by low-rank representation (LRR) model on exploring the multiple subspace structures of observation data, we propose a double low-rank matrix recovery method to learn low-rank subspaces from face images, where it takes into account the recovery of row space and column space information simultaneously. Applying Augmented Lagrangian Multiplier (ALM), the optimization problem on minimization of nuclear norm is resolved efficiently. By evaluating on public face databases, experimental results show that our proposed method works much better than existing face recognition methods based on feature extraction. It is more robust to outliers, varying illumination and occlusion.

Compressive sensing (CS) is applied to enable real time direction of arrival (DoA) estimation in a wireless array sensor network by significantly reduce the amount of sensor data that needs to be transmitted over wireless channels. This is accomplished by (i) random sampling of data samples collected at sensor arrays, and (ii) transmitting only the sign-bit of the data samples over wireless channels. It is shown that CS based DoA estimation is capable of delivering similar performance as conventional DoA methods while requiring less than 10% of wireless bandwidth to transmit data. This proposed scheme is validated using a prototype sensor array test bed. Preliminary experimental results clearly validate the superior performance of this proposed scheme.

This paper presents the recent results of utilization of micro-Raman spectroscopy to measure and characterize residual stress in polysilicon doubly-clamped MEMS resonators with small lateral size. Due to imprecise prediction of the magnitude of intrinsic residual stress, detrimental effect of the residual stress severely shifts the resonant frequency of MEMS resonator from the analytical pre-designed value. The stress is not only determined by the fabrication process but also related to the structural dimensions of resonators. In this work, microRaman spectroscopy was used to measure the residual stress of resonators with widths down to 2μm. The results show that the optimized resonator with length shorter than 50μm and width between 3.2μm and 4.1μm exhibits minimum residual stress.

Using single chip microcomputer as the system control center, compose of controllable amplifier, programmable filter and amplitude-frequency characteristics of the test parts. Realizing the input signal gain control by voltage gain adjustable, adopt switched capacitor technology realize the programmable filtration, the choice of DDS technology generate sweep signal; all the basic requirements and play part of the requirements in the title are achieved. In addition, we have expanded the band-pass filter, realized the storage and display of amplitude-frequency characteristic curves. It is using the menu function selection mode, and user-friendly human-computer interaction. A circuit of high integration is helpful to reduce cost and power consumption.

A 100-ch fully head-implantable wireless broadband neural recording system was developed. It features 100 parallel wideband (0.1Hz ∼ 7.8 kHz) neural recording channels, a medical grade 200mAh Li-ion battery recharged inductively at 2 MHz, and data telemetry using both 3.2/3.8GHz FSK wireless and 850nm IR laser data link. All active electronics are hermetically sealed in a titanium enclosure with a sapphire window. Custom 100 individual hermetic feedthroughs enable connection to the microelectrode intracortical array. This wireless implant has been validated both on-bench and in freely moving pig and monkey models.

Based on the current situation which the bucket wheel excavator systems can't been arranged according to the previous design scheme at Heidaigou Open-pit Coal Mine, three feasible alternatives to arrange bucket wheel excavator systems are put forward. Then their reliability is calculated, and the optimal alternative is recommended according to the calculated reliability. Finally the economic benefit is estimated if the recommend alternative were put in practice.

A new model has been established in the domestic pig for neural prosthetic device development and testing. To this end, we report on a complete neural prosthetic developmental system using a wireless sensor as the implant, a pig as the animal model, and a novel data acquisition paradigm for actuator control. A new type of stereotactic frame with clinically-inspired fixations pins that place the pig brain in standard surgical plane was developed and tested with success during the implantation of the microsystem. The microsystem implanted was an ultra-low power (12.5 mW) 16-channel intracortical/epicranial device transmitting broadband (40 kS/s) data over a wireless infrared telemetric link. Pigs were implanted and neural data was collected over a period of 5 weeks, clearly showing single unit spiking activity.

Multi-audio frequency signal have been applied in vary case. In this paper, a novel method for weak multi-audio frequency signal detection have been proposed. We aim at the multi-audio frequency signal detection in very strong noise circumstance based on the chaotic oscillator. At the end of paper, we exploit our approach to a typical multi-audio signal, TVM430 railway signal system, which have been widely used in railway. The experiment results demonstrate that the proposed method has preferable performance.

Image denoising is an important problem and widely studied in image processing. Denoising is a crucial step to increase image conspicuity and to improve the performances of all the processing needed for quantitative imaging analysis. This work presents a modified nonlocal-mean denoising method based on the texture collection and edge information. At last, experiments result illustrate that the technique can be successfully used to the classical case of additive Gaussian noise. The proposed algorithm seems to improve on the state of the art performance.

An in-depth understanding of large-scale wind power integration (LSWI) in China is conducted in the paper. Firstly, the characteristics of China's LSWI are analyzed. Secondly, major problems associated with large-scale wind power integration in China are emphasized, namely balancing capability, grid interconnection, transmission costs and technology trend. For China, its energy mix and forecasting technology highlight the balancing capability of the system. The current grid interconnection can hardly adapt to the large-scale wind power integration. The transmission costs caused by China's wind power development mode are likely to one of the most demanding problems that it has to face and solve. Besides, as the strategic planning for constructing the strong and smart grid was formally kicked off in May 2009, the technology trend becomes an important factor impacting large-scale wind power integration because of the introduction of new technologies.

A synthesized theory model which describes the relationships between specific asset investment, supply chain collaboration trust, and cooperation intention is proposed. Forty three supply chain companies with different scales in different fields are chosen to be the objects of experimental research and a sample data of 318 actual members is finally collected. Factor analysis and structural equation modeling are implemented to validate the hypothesized relationship between specific asset investment, supply chain collaboration trust, and cooperation intention hypothesized in model. The result shows that partner's perceived specific asset investment has positive effect on credibility and cooperation intention, respondent's perceived specific asset investment has negative effect on credibility and cooperation intention, credibility and benevolence both has positive effect on cooperation intention.

On the basis of dynamic trust theory, the paper proposes a synthesized theory model which describes the key influential factors of dynamic collaboration trust in supply chain from the point of organization relationship, which studies the dynamic dissimilar role of factors on collaboration trust in supply chain in differential phase. Forty three supply chain companies with different scales in different fields are chosen to be the objects of experimental research and a sample data of 318 actual members is collected. Factor analysis and structural equation modeling are implemented to validate hypothesizes which are put forward in model. The result shows institution based trust, information sharing, and specific asset investment has dissimilar effect on supply chain cooperation trust in the differential phase of dynamic process, which would provide theory foundation for trust practice of supply chain companies.

A Logarithmic Current Mode Amplifier (LGCMA) is designed and verified by 0.5µm process aiming at electrochemical signal detection in the miniaturized polymerase chain reaction (µPCR) chip for the quantitative analysis of deoxyribonucleic acid (DNA) concentration. Fully electronic integrated front end detection system is concepted to readout electrochemical signal using the LGCMA as a pre-amplifier. This paper has a insight to the stability, logarithmic property, bandwidth and dynamic range (DR) and noise level of the LGCMA. The post simulation result shows the operation current ranges six orders of magnitude from 5pA to 5µA. Considering that noise presented in the circuits the minimum current resolution is 1nA.

A novel design of NEMS resonator based LC oscillation circuit for detection of tiny vibration is proposed. The relationship between the tiny vibration of the NEMS resonator beam and the frequency of LC oscillation is analyzed. With the resonator as the part of the LC oscillation, IC compatible processes based oscillation circuit are conceived and simulated. The simulation result demonstrates the oscillation bandwidth is 1.75MHz under the 2nm NEMS beam vibration amplitude.