Search

Aiming at the problem of poor image contrast and loss of visual information on electrowetting electronic paper display， this paper proposes an image hierarchical mapping technique based on human visual system. Firstly, the image is decomposed into the base layer and the detail layer, and the brightness of the base layer is enhanced. Then, the enhanced image is partitioned by Weber-Fechner law to realize the partition logarithmic mapping, and weighted and merged with the detail layer as the output image. At the same time, in order to achieve precise modulation of the electrowetting display, the brightness is local linearly stretched. The experimental results show that compared with traditional mapping methods, the entropy value of the algorithm in this paper is closer to the original map, the PSNR,SSIM and UIQI values of image evaluation index are respectively improved by 0.2~0.6，0.07~0.4 and 0.1~0.6. The method makes the relationship between the corrected voltage and brightness more linear, increases the image details on the electrowetting electronic paper, improves the contrast, and the overall visual effect is better.

Tilt-rotor UAV(unmanned aerial vehicle), integrating the VTOL(vertical take-off and landing) and hovering of the four-rotor UAV and long-distance and high-speed navigation capabilities of the fixed-wing UAV, is a new model of UAV. In this paper, the dynamic model of tilt-rotor UAV is established by analyzing and calculating the pulling force of such UAV blades. And this paper measures the response of pulling force coefficient and steering gear motor in the model. At the same time, a control circuit is designed for multiple PWM outputs. The program is designed for the tilt angle and tilt process to complete the transition process of the two modes. The test result shows that the control system can complete the movement of the tilting process and realize self-stabilization.

L-band radio frequency is in the range of 1 GHz~2 GHz，which is widely used in rader, communication and navigation. The main research content of this paper is to design an L-band broadband, high dynamic RF digital receiver using RF digital receiver demodulation technology. In this paper, the design of L-band RF digital receiver is research at RF sampling, digital channel separation, receiver dynamic expansion and high-speed signal demodulation, and give a specific implement scheme of digital channel separation, high-speed signal demodulation based on FPGA digital signal process platform.

This paper analyzed the differentiated requirements of different business systems for communication network in three domains including urban rail transit safety production, internal management and external services, and studied the key technologies of 5G virtual private network construction, such as 5G network slicing and multi-access edge computing. Based on 5G network slicing and multi-access edge computing technology, this paper presented a technical scheme of constructing virtual 5G private network of urban rail transit for different demand scenarios, which provided a strong support for the development of urban rail transit to convenience, efficiency and intelligence.

With the development of science and technology, the design of integrated circuit becomes more and more complex.In the design process, the division of hardware and software also produces more choices. In order to meet the complex design requirements for integrated circuits, the development and design of SystemC, which inherits the C + + extension library, came into being. SytemC can meet the requirements of hardware and software codesign, and can be quickly simulated and verified in the design process. In this paper, the communication protocol(ISO/IEC 18000-6C) is modeled and designed at a higher level. The protocol communication between interrogator and tag is realized by function call. For UHF RFID(Ultra High Frequency Radio Frequency Identification)communication protocol, from system level design to behavior level design, the top-down design of the protocol is realized, and the design of tag identification layer and baseband communication link layer is completed. Finally, the source code of this systemc is transformed into Verilog and other hardware description languages by using Vivado HLS synthesis software, and the software description of system level is synthesized into RTL code. This paper focuses on label baseband link encode_T module, after FM0 / Miller coding simulation of data "1001001001011001" in Microsoft Visual C++ 6.0, the waveform verification is completed in Modelsim, and it is transformed into RTL level circuit by Vivado HLS integrated software.

In the information-based war era, OODA command and control loop is widely used in military simulation. The flexibility and efficient parallelism of distributed simulation promote the development of combat simulation. According to the requirement of combat simulation, this paper studies the core technology of distributed simulation system, which is time management algorithm. Based on the characteristics of OODA command and control loop, this paper proposes an adaptive moving time window algorithm. The algorithm adaptively adjusts the size of time window according to the model, which can avoid frequent synchronization and clear the memory accurately. In order to further optimize the time management algorithm, the author also proposes an improved global virtual time algorithm to improve the blocking process in global virtual time synchronization calculation. Through detailed analysis of the steps and principles of the algorithm, and verification by an example, the results show that the algorithm can effectively guarantee the running of distributed simulation, and the running speed is better than other traditional time management algorithms.

The paper adopts the architecture of "ADC+FPGA", designs and develops a 1 GS/s, 16-bit high-speed and high-precision data acquisition system(DAS), which realized the purpose of using a single channel to measure large dynamic range(DR>1 000) signals. The study uses two sysref modes, which are periodic sysref mode and pulse sysref mode, to establishe a stable JESD204B link with a deterministic delay. The differences in the sampled data spectrum between the two modes is presented, and the considerations of hardware design and firmware design are given. The paper recommends that the periodic sysref mode be used to establish JESD204B link. By analyzing the time-domain waveform and frequency spectrum of the sampled data, the conclusion that the ADC chip contains 4 on-chip ADC channels is verified.

Aiming at the problem of how to implement effective countermeasures against frequency-stepping millimeter-wave precision guided weapons, this paper analyzes the working principle of two-dimensional imaging of frequency-stepping millimeter-wave guided seekers, constructs the two-dimensional imaging scene of frequency-stepping millimeter-wave guided seekers, and carries out systematic and comprehensive simulation on imaging processing of frequency-stepping millimeter-wave guided seekers and various jamming techniques. The simulation results show that no matter whether the jamming source is located in the range of antenna beam coverage of the seeker or not, both incoherent and coherent jamming can jam frequency-stepping millimeter-wave guided seekers. The energy of incoherent jamming is more dispersed, and the energy of coherent jamming is more concentrated. Therefore, a compound jamming strategy of inverse gain and active jamming is proposed in this paper, which can realize jamming of frequency-stepping millimeter-wave guided seeker and simultaneously realize self -protection of the jamming source.

Based on PowerPC architecture, this paper proposes an optimization strategy of beam pointing algorithm, which is realized from the trigonometric function calculation optimization, floating point arithmetic optimization, loop nesting optimization, and PowerPC instruction optimization. Through the optimization algorithm proposed in this paper, the processing time of the algorithm is reduced to one tenth of the original. At the same time, the optimization strategy proposed in this paper also has certain guidance and reference for the algorithm development and optimization of other platforms.

For the problem of computing offloading decision in mobile edge computing, this paper proposes an offloading decision algorithm based on enhanced learning in multiuser MEC system. According to the task model, calculation model and channel state, ODRL algorithm makes the task unloading decision and uses reinforcement learning method to solve the optimal computing unloading strategy. Simulation results show that the ODRL algorithm proposed in this paper has lower total system cost compared with the baseline strategy.

It′s meaningful for electricity power company to get the outage information on time. This paper proposed a new voluntary report method: Electricity energy meter generated outage event log, communication module transferred it. In order to solve the compatibility and expansibility disadvantages of current scheme, this paper proposed a power recovery voluntary report and broadcast timing method, realized closed loop management of outage maintenance and repair, and solved the wrong clock problem caused by battery low voltage after outage. This paper introduced the design thoughts of next generation smart meter，and put forward an outage and recovery voluntary report and timing strategy implementation plan in next generation smart meter. This paper provides a solve plan for next generation smart meter popularization, and it′s helpful for construction of electricity energy Internet.

This paper introduces the design of low noise amplifier, using for ultra-low frequency. There′s a new loop in the paper, formed by a low-pass transconductance and a self-cascode MOSFET, proposed to stabilize the output bias voltage of the amplifier. By using the loop with the constant transconductance bias circuit, the open-loop gain of the amplifier is constant at about 40 dB. It does not vary with the supply voltage, process and temperature. In the amplifier, a self-cascode MOSFET is used as a gain unit to enable the amplifier output range to reach rail to rail, with the use of full differential current bias. This amplifier has a remarkable in-band high-power supply rejection ratio, up to 101.4 dB. The power ripple has almost no effect on the output signal, when the wearable devices are powered by high impedance sources.

In order to make the spectrum access of communication users more effective and enhance the flexibility of spectrum utilization in time domain and space domain, this paper firstly introduces the process of BP neural network learning with the steepest descent method, and makes mathematical modeling for the spectrum prediction algorithm of BP neural network. By learning and training the electromagnetic spectrum state in a period of time, this paper makes the algorithm model establish the cognitive relationship between the input data and the output results by adjusting the parameters. Then, the structure of BP neural network algorithm was changed, and the weights and thresholds were optimized, so that the accuracy of spectrum prediction data was closer to the actual value and the prediction error was reduced.

Aiming at the problem of target track offset that easily occurs in multi-radar multi-target detection and tracking when the target distance is close, this paper proposes a new type of multi-radar multi-target particle filter tracking before detection algorithm(NM-PF-TBD). The algorithm uses a two-layer particle filter structure. When tracking a target in the target tracking layer, the sensor detection information is recalculated based on other target points in the previous cycle. Secondly, the particle weight is calculated based on the corrected sensor detection information, and then. Based on the distribution of particles with different weights in the particle swarm, it deletes harmful particles, supplements beneficial particles, and avoids the shift of the center of the tracking particle swarm after re-sampling, so as to ensure that the target point trace does not shift. Simulation results show that, compared with other algorithms, the algorithm proposed in this paper can accurately correlate target traces and improve target accuracy when observing multiple closer targets at a long distance.

This paper deeply analyzes the shortcomings of the current reversing image system in the actual use scene, introduces the limitations of radar in radiation surface, angle blind area and object type recognition, and the millimeter wave radar is difficult to popularize due to high cost. This paper puts forward the design of automobile reverse anti-collision system based on ADAS, makes an in-depth analysis of the principle of the system′s reverse function, and introduces the ADAS algorithm based on the image analysis of the reverse camera in detail its technical key points and the innovative application based on the fusion features of the reverse scene. The algorithm recognition effect is accurate and effective. According to the differences of the installation position and orientation of different vehicle cameras, the innovation design is carried out lexible early warning and alarm area adjustment function, combined with TTS voice reminder and graphic display to remind the driver to take appropriate operation to ensure driving safety, greatly improved the user experience and practical role of the reversing anti-collision system, and provided strong protection for the driver′s safe driving.

Wi-Fi6 standard will be major Wi-Fi technology dominated for deployment after 2020. But to support increasing bandwidth requirements and number of access devices per year, Wi-Fi technology and throughput still need to be further developed and improved. This paper will introduce the key technology involved for latest IEEE802.11be standard discussion, describing the major changes on physical and MAC layer, particularly providing the analysis for multi-access point coordination, hybrid automatic repeat request and Wi-Fi full duplex mode. Finally, this paper will introduce the granted progress of latest 6 GHz applied by both new 802.11ax version and 802.11be with related Wi-Fi function impact.

With the wide application of reconfigurable design technology in the field of cryptography, the existing reconfigurable cryptography system can hardly meet the needs of high-speed communication and equipment miniaturization.Therefore, it is the goal of this paper to design a high performance reconfigurable processing architecture that can flexibly implement block cipher algorithms.Based on the reconfigurable array structure and the multi-launch mechanism, this paper proposes a reconfigurable multi-launch pipeline processing architecture to solve the problems of low throughput and low resource utilization of the reconfigurable cryptography system.This architecture is able to transmit multiple packets in parallel and pipeline processing, and has higher throughput and unit utilization, which greatly improves the algorithm performance of the reconfigurable system.Experiments show that the throughput of AES can reach 3.19 Gb/s under the operating frequency of 350 MHz.Compared with the single-launch structure, its performance improved by about 1.1 times.

The solar photovoltaic industry has developed rapidly in recent years. Accurate diagnosis of the location and type of PV module faults can improve the efficiency of operation and maintenance personnel. In this paper, a deep learning diagnostic model based on convolutional neural networks-long short term memory(CNN-LSTM) is proposed, which can be used to complete the detection task. In this paper, a fault classification method based on current performance is established. The algorithm firstly designs a feature extraction algorithm based on the layout characteristics of the PV array, and extracts the lateral and vertical features of the PV array current to obtain the spatial and temporal characteristics. The CNN network further extracts the lateral features and compresses the vertical features to solve the problem of single feature types and slow training. Finally, the LSTM neural network is used to complete the fault diagnosis of the PV modules.

Because the designed frequency band of Geosynchronous synthetic aperture radar(GEO SAR) satellite signal possibly shares or overlaps with the frequency band of various Global Navigation Satellite System(GNSS) signals which have been put into use, the ground GNSS receiver may be subject to the pulse radio frequency interference generated by GEO SAR signal. In order to evaluate the influence of GEO SAR signal on the performance of GNSS ground receiver, a parametric theoretical model based on International Telecommunication Union(ITU) related standards is proposed in this paper. The model uses the "source-path-receiver" analysis method to extract parameters from the three aspects of GEO SAR signal, signal propagation path and receiver model for interference evaluation. Then, according to ITU standards, the influence of high-orbit SAR signal as additive pulse radio frequency interference on the performance of GNSS receiver is evaluated. Taking the representative B3I signal as an example, this paper evaluates the performance of different types of B3I receivers for GEO SAR signal under the condition of the assumed design parameters. The results show that the performance of the receiver decreases slightly in the presence of high-orbit SAR signal interference, but it can still work normally.

In this paper we introduce an electronic seal base on 13.56 MHz RFID technology. The seal includes national commercial encryption algorithm SM7 to ensure the reliability and security. The radio frequency identification also allows the seal to be easily managed. In this paper we also introduce architecture of this kind of seal which include the antenna structure, function module, state machine working flow, 3 pass authentication and so on. Finally we realize a chip on HJEE110nm standard CMOS technology, the chip core only occupies 0.5 mm2 and the power consumptions are less than 100 μA. The chip packaged in ID1 completely matches the standard of ISO14443 and ISO10373-6, and chip packaged in seals in three specifications works perfect with the encrypt reader.

This paper proposes a three-dimensional Logistic-Sine cascade map(3D-LSCM). Firstly, a three-dimensional Logistic-Sine cascade map was constructed by using the output of the Logistic map as the input of the Sine map. Secondly，this paper designs an image encryption scheme using 3D-LSCM, which consists of two rounds of scrambling and two-way diffusion，and introduces the plaintext as part of initial conditions of 3D-LSCM to enchance the difficulty in deciphering equivalent keys. The feasibility of the algorithm was confirmed throung security analysis and numerical simulation.

In this paper, the digital beamforming algorithm for horizontally polarized directional antenna array in narrowband multipath environment is studied. Because the steering vector of directional antenna arrays is complex and cannot be modeled exactly, blind beamforming algorithm is usually used for directional antenna arrays. As the existing blind beamforming algorithms require much prior information, work poorly for weak signals, have limited applications, and need high computational complexity, this paper designs a multi-polarization heterogeneous array cooperative processing scheme, and proposes a new multipath strengthening beamforming algorithm based on adaptive loading projection. The heterogeneous array consists of vertical polarization omnidirectional antenna array and horizontal polarization directional antenna array. The new algorithm utilizes the respective advantages of the two polarization arrays. This algorithm combines the methods of spatial smoothing, sampling matrix inversion, and spatial projection, and further improves the robustness of waveform estimation through adaptive loading processing. The algorithm requires less computation and less prior information, and has strong adaptability to non-ideal conditions. The implementation process and error analysis of the new algorithm are deduced, and effectiveness of the new algorithm is verified through computer simulations.

Through comparison, algorithm optimization, pipeline accelerated design and structural compatibility extension of a large number of mainstream algorithm, this paper propose an algorithm that is good for modulo operation structure compatibility: improved radix-4 interleaved modular multiplication algorithm. It has advantages of short path and simple structure, and has advantages in compatibility design, can achieve full flow accelerated operation, and achieve the high speed and reconfigurable design purposes. The improved radix-4 interleaved modular multiplication unit proposed in this paper can directly realize plus-minus module division and modular addition without adding extra structure, which solves the problem of the waste of independent resources. The unified modular unit is synthesized under the 65 nm CMOS process. The area is 0.117 mm2. The clock frequency is 526 MHz. A 576 bit modular multiplication and modular division are completed at 0.55 μs and 2.98 μs respectively.

The purpose of this paper is to use genetic algorithm(GA) to process image with bottom noise, and the processing effect is improved through the improvement of GA. Combining with image segmentation, this paper expounds the working mechanism of GA and the design methods of main modules such as fitness calculation, selection, crossover and mutation, and gives the specific values of parameter setting. The key issues such as the relationship between generation gap and excellent individuals, the substitution relationship between individual among different generations, the selection method of intersection points and the selection of mutation positions, and the maintenance of population number are clarified. The image with bottom noise is processed by this algorithm, the results show that the target image can be separated from the background with noise by GA, but the processing time is 7.416 seconds. In order to improve the processing efficiency, the traditional algorithm is improved by adaptively adjusting the crossover probability and mutation probability of the population using evolutionary generation and individual fitness values. The same noise image is segmented by improved GA. The results show that the improved GA has better image segmentation effect, the processing time is only 0.751 seconds, the efficiency is improved by nearly 10 times.

With the development of science and technology, ultrasonic scalpel is widely used in medical treatment. In practical application, in order to make the transducer of the ultrasonic scalpel convert electrical energy into mechanical energy as much as possible, this paper uses parallel impedance matching to realize the impedance matching of the ultrasonic scalpel, which makes the transducer of the ultrasonic scalpel present pure impedance, thus maximizing the output power of the transducer and reducing the transmission of the transducer. Heat has increased the service life. In addition, the parameters of the transducer of the ultrasonic scalpel will change with temperature and environment in the course of using. Therefore, this paper designs a frequency tracking method for the resonant frequency of the transducer, which tracks the phase difference between current and voltage, and then sends it to the FPGA for calculation. A phase difference is obtained. When the phase deviation occurs, a frequency conversion formula is used to adjust the output frequency, so as to achieve the effect of frequency tracking. Through practical verification, the impedance matching and frequency tracking method of the ultrasonic scalpel designed in this paper can achieve the automatic frequency tracking under different loads, and run reliably.

Large-dimensional matrix multiplication is often implemented by submatrix block method. The maximum size of the submatrix determines the speed of the entire matrix multiplication. Concerning the problem that the matrix size directly processed by the classical systolic structure is severely limited by the IO bandwidth, this paper proposes a large-dimensional matrix chain multiplier structure with extremely low IO bandwidth requirements, and completes the hardware design implementation and performance verification. The following is the main work of this thesis. Firstly, optimizing the data organization of matrix multiplication, realizing the input matrix size has nothing to do with IO bandwidth, and make maximum use of the internal logic and storage resources of the device. Secondly, according to the optimized data organization form, the chain multiplier hardware is designed for realizing the source data calculation and transmission overlap operation. Thirdly, the adaptability of the multiplier to the matrix scale is enhanced, and the designed chain multiplier can be configured in real time as multiple independent chains, multiple sets of operations in parallel. Lastly, completing the hardware implementation and performance test of chain multipliers of different sizes on the Xilinx C7V2000T FPGA chip. On this chip, the chain multiplier proposed in this paper supports up to 800 arithmetic units, which is 8 times the size of the classic systolic structure. In the same number of operators, the chain multiplier performance proposed in this paper uses only the classical pulsation structure to calculate the IO bandwidth of 1/8 to obtain equal performance.

This paper interprets the evolution route of network service capability provided by telecom operators in traditional Internet of Vehicles industry chain. On the basis of the development trend of future Internet of Vehicles and intelligent transportation, this paper explains the growth bottlenecks and capacity challenges faced by traditional channel service, and analyses the service capability improvement brought by 5G key technologies such as edge computing and network slicing with Internet of Vehicles. This paper also explores the innovation of Internet of Vehicles service mode and business mode that operators can carry out.

Quartz vibrating beam accelerometer expresses acceleration in the form of frequency output. In the inertial navigation system, it is necessary to convert the frequency value into an acceleration value and then perform attitude calculation. Using software method for floating point calculation requires a lot of computing power of CPU. In order to optimize the calculation speed of frequency conversion, the paper designs a floating-point calculation acceleration unit for frequency conversion applications, which is implemented and verified based on FPGA. The results show that the system from data sampling to frequency conversion, then converts the frequency value into acceleration for the attitude calculation. The angular velocity measured by the gyroscope is integrated, and finally the data fusion is completed. Using the floating-point acceleration unit designed in the paper to achieve frequency conversion, the speed is improved by 2 times.

In this paper, the development trend of intelligent transport system(ITS) and the state art of vehicle-road collaboration are introduced. This paper focuses on the cloud-network-edge-terminal architecture of vehicle-road collaboration, includes central cloud, transport private network/telecommunication network, edge cloud, vehicle/road-side terminal and their collaborations.The open framework of information service capability of multi-source data fusion based on cloud-network-edge-terminal architecture for vehicle-road collaboration is presented. The specific functional requirements, API invocation methods and protocols are discussed in detail.

In order to satisfy the high-performance and low-power dissipation requirement in wireless communication, this paper proposes a low storage capacity and low-power dissipation Turbo decoder architecture based on the reverse recalculation and linear estimation by changing the storage method of the forward state metric, while the FPGA implementation structure is given. The results show that compared with the existing Turbo code decoder, the decoding structure designed in this paper reduces the storage capacity to 65%, and the decoding performance is close to the Log-MAP algorithm. In particular, compared with the traditional decoder architecture, dynamic storage capacity power dissipation is reduced by about 50%, and the overall power dissipation of the decoder architecture is decreased by 4.97%, 8.78%, 11.93%, 14.18% and 14.65% at the frequency of 25 MHz,50 MHz,75 MHz,100 MHz and 125 MHz, respectively.

The increasing functionality and complexity of electronic systems places higher demands on the efficiency of system verification tests. The traditional satellite test platform is customized according to the specific model task. Its design and development cycle is long, and the comprehensive cost is difficult to adapt to the current task requirements. To this end, this paper proposes a general simulation test platform architecture for on-board solid-state storage controller based on System Verilog. It uses a hierarchical model internally, and the signal interface uses APB bus standard uniformly. It can be configured by testing the platform data source and format to adapt to onboard solid-state storage controllers of different capacities, different rates and different configurations. Experiments show that the test platform designed in this paper has certain versatility, which can effectively save test time and improve test coverage compared with traditional test platform.

The computer technology related to unmanned vehicle driving is developing rapidly, and the vehicle identification technology is more and more widely used in intelligent traffic monitoring and intelligent traffic information systems. Vehicle identification includes sports vehicle detection, type classification, vehicle tracking and so on. In this paper, the key technologies such as vehicle identification detection in smart car unmanned technology are deeply explored. Using the image recognition result of the vehicle, this paper chooses the CAM offset algorithm based on the average color difference value as the basis for identification. The algorithm uses the vehicle color as the feature to perform the tracking detection method, and does not follow the light and dark of the vehicle. The shape change is disturbed, and the reliability in the vehicle identification process is high. This algorithm is improved to achieve better recognition of the vehicle.

A miniaturization research of rocket upper stage electric load management center(EMLC) based on SIP concept is presented in this paper, in the background of electron-integrated system on rocket, focusing on the method of design and implementation of a SIP control system in EMLC. The architecture design of SIP control system is completed with the mixed analog-digital FPGA as hardware platform. A design and implementation method of IPs like main protocol engine core, warm backup fiber optic Ethernet core and other main function core has been elaborated. An assistance health management(AHM) system′s design and implementation method in this paper has been simply illustrated. The simulation result of EMLC and experiment result of the SIP control system are also been given.

This paper introduces a high-speed, low-offset sense amplifier with low supply voltage in UMC 28 nm CMOS process. The paper presents a novel structure of the sense amplifier which bases on the traditional differential amplifier, class AB latch and other circuits. It′s designed and verified in Cadence. The simulation results show that the proposed design exhibits 0.2 mV/0.8 mV offset voltage, 63 ps/44 ps delay, 0.37 mW/0.44 mW power dissipation respectively with 1.05 V supply voltage when the clock signal at the 5/10 GHz. Therefore, the proposed sense amplifier is satisfied for the analog-to-digital converter of high-speed interface JESD204B.

Based on the TensorFlow framework, this paper builds convolutional neural networks to recognize the defects in the electroluminescent images of cells. It selects the exposed data set that contains the different types of defects in the cell. Based on the traditional VGGNet network, the full convolution neural network is used for training, and this paper analyzes the training effects of different loss functions and dropout probabilities on data set. Experiments have shown that the algorithm accurately recognizes whether the cell is defective. The study also shows that the compression network structure greatly increases the training rate of the algorithm, which makes the simplified model more portable and provides an effective solution for a wide range of real-time defect recognition.

With the development of computer technology, fire image processing technology combining computer vision, machine learning, deep learning and other technologies has been widely studied and applied. Aiming at the complex preprocessing process and high false positive rate of traditional image processing methods, this paper proposes a method based on deep convolutional neural network model for fire detection, which reduces complex preprocessing links and integrates the whole fire identification process into one single depth neural network for easy training and optimization. In view of the problem of fire detection caused by similar fire scenes in the identification process, this paper uses the motion characteristics of fire to innovatively propose the combination of fire frame position changes before and after the fire video to eliminate the interference of lights and other similar fire scenes. After comparing many open learning open source frameworks, this paper chooses Caffe framework for training and testing. The experimental results show that the method realizes the recognition and localization of fire images. This method is suitable for different fire scenarios and has good generalization ability and anti-interference ability.

With the rapid development of Internet applications and the rapid growth of users , the reviews and opinions of stock market largely reflect the quotation of the stock market,simultaneously it affects the ups and downs of the stock market. Therefore, how to quickly and efficiently analyze the attitudes and opinions of netizens to the stock market,which,this question plays important role in guiding us to predict the stock market. The thesis studies the rising and falling trend of stocks by analyzing the emotional polarity of different professional issuing stocks. This paper proposes a sentiment analysis method based on a dictionary of consistent integrated financial phrases and weighted at the end of paragraph, which can solve the dependency problem of sentiment dictionary on the domain，and it can effectively improve the accuracy of sentiment analysis. In addition, this paper also proposed a windowed stock prediction model, which can be used to analyze the optimal value of the forecast event window. The experimental results shows that it will be better to predict the rising or falling trend of a particular stock just based on the stock market sentiment analysis.

5G Ultra-Dense Network(UDN) allows a large number of high-density low-power transmission points(TPs) to be deployed in heterogeneous networks(HetNets). It greatly improves the system performance whereas introduces inter-cell interference(ICI). To eliminate ICI, CoMP concept has been widely used in UDN. But it also brings new challenges, how to effectively allocate network resources is one of them. This paper proposes a CLA-CoMP algorithm to improve both the rate of the small cell edge users and the system sum rate. In order to deal with the power allocation problem in UDN, this paper adopts P2UPA strategy, which aims to allocate higher power to users with the high probability of association. Simulation results reveal that jointly use CLA-CoMP and P2UPA can significantly improve the system performance, compared with the case of average power allocation.

Unified power flow controller(UPFC) is the most advance flexible ac transmission system(FACTS). Active/reactive power balance for UPFC is analyzed in this paper when it is working in power flow control mode, and an improved real time power coordinate control strategy is proposed in paper. In the parallel converter control loops, an improved real-time dynamic feed forward method is added to reduce over voltage caused by imbalance power. The results of hardware-in-loop simulation based RT-LAB platform show that this new coordinate control strategy，comparing with the conventional one, can decrease excessive voltage excursions of the send terminal ac voltage and DC-link voltage and improve the control performance during power flow changing or system occurring fault.

The defect area located by Faster R-CNN has weak edges. The area would be over-segmented or under-segmented if conventional segmentation algorithm is adopted. This paper made an analysis on precise threshold segmentation algorithm for workpiece defects based on Faster R-CNN location, reconstructing the localization area by morphological opening and closing reconstruction algorithm, processing the reconstructed image by Otsu′s dual threshold method, segmenting transformed images by maximum entropy threshold segmentation method, and finally measuring the area, perimeter and other parameters of the segmented defects. The research shows that the algorithm in this paper has higher segmentation ability regarding workpiece defects(crack, bubble and slag), compared to conventional algorithms. It not only can accurately segment objects with weak edges, but also can effectively remove the interference from the contour background to the segmentation.

Current residual current action protection device cannot detect the type of residual current in the electric shock current. In this paper, an electric shock test platform is built and the residual current data is obtained by experiment. Based on the above data, this paper presents a residual current classification method based on the AdaBoost algorithm, which firstly obtains the characteristic components of different types of residual current components, and then maps these component features to the AdaBoost algorithm. The type of electric shock current in total residual current is detected by AdaBoost algorithm. Compared with SVM and random forests, the results show that the proposed method has some advantages, which can provide theoretical basis and support for the development of the subsequent adaptive residual current action protection device.

In the current anti-jamming communication, the spread spectrum communication has been studied deeply with the advantages of strong anti-jamming ability and good security performance. In this paper, the code index modulation and multi-ary spread spectrum technology are analyzed and compared. Code index modulation technology is based on the development of direct sequence spread spectrum, in the transmitter information bits are mapped to the pseudo noise(PN) code index and the modulation symbol respectively. The in-phase and orthogonal parts of the modulation symbol select PN codes activated to spread spectrum, and then transmitted through an antenna. The difference between multi-ary spread spectrum technology and code index modulation technology is that the multi-ary spread spectrum technology does not have modulation symbols, and only sends out the activated spread spectrum code through the antenna. This paper will introduce the basic models, spread spectrum gain analysis and complexity analysis of the two kinds of spread spectrum technology. The simulation results show that the code index modulation system has higher spectral efficiency and anti-interference ability.

Vehicular Ad hoc Networks(VANETs) have special characteristics on the highway, such as high speed，low density, scarce communication infrastructure, and low vehicle connectivity. These characters make it difficult for vehicles on the highway to access to other vehicles or Road Side Units(RSUs). This paper presents a model which implements fog computing in highway service areas to assist vehicles in acquiring content of interest. In this Vehicle Fog Computing(VFC) model, the highway service area collects various services provided by vehicles from various places, and uses a large number of parked vehicles and slow-moving vehicles as fog devices. Localized forwarding through VFC not only reduces communication latency, but also achieves satisfactory content access and real-time data streaming. Moreover, this paper formulates the relationship between communication energy consumption and system delay, then uses the Outer Approximation(OA) algorithm to optimize its trade-off in fog computing. The simulation results show that the communication delay of the system will be significantly reduced with the increase of energy consumption by using the communication mode and equalization optimization algorithm combined with fog computing and cloud computing.

This paper studies the precoder design in multi-input multi-output(MIMO) interference channels, where it is assumed that interference channels have line-of-sight(LOS) paths and can be modeled as Rice fading channels，and the transmitter can only retrieve channel statistical information—channel mean matrices. Based on maximum signal-to-interference-plus-noise ratio(SINR) and signal-to-interference-plus-noise difference(SIND) criteria, two precoding methods are proposed. For the first method, an auxiliary variable is introduced so that the original non-convex problem is transformed into a convex one. Consequently, this problem is solved by the proposed bisection method that incorporates convex optimization softwares. For the second method, the original problem is decomposed into several sub-problems and the corresponding analytical solution is derived. By computer simulation, this paper demonstrates the convergence of the proposed algorithm and analyzes the effect of Rice factor on the system performance. The proposed methods are compared with a few classical transmission schemes and the results show the superiority of the proposed methods.

In order to improve the regulation capability of the wide range output voltage and light load efficiency, this paper proposed two level control strategy. Asymmetrical Pulse Width Modulation(APWM) was used in the LLC series resonance with voltage double rectifier and a feedback controller was added to get the proper crossover frequency and enough relative stability. Phase-locked loop control(PLL) was used to track the resonant frequency. In addition, considering the high switching frequency of the secondary leakage inductance will affect the input and output voltage gain, this paper added secondary leakage inductance to the model to reduce the reverse blocking voltage of rectifier diode, which can improve the accuracy of modeling. Simulation and experiment have verified the correctness of the theoretical analysis and the feasibility of the control strategy.

Based on the traditional first-order compensation bandgap reference analysis, this paper proposes a novel high-order compensation technique and uses this technique to design a curvature compensation voltage reference with a low temperature coefficient over a wide temperature range. The circuit structure proposed in this paper is designed and simulated based on the VIS 0.15 μm BCD technology. The simulation results show that the average output voltage of the reference circuit is 539 mV and the temperature coefficient is 1.40 ppm/℃ in the wide temperature range(-60 ℃~120 ℃), the line regulation of output voltage at 27 ℃ is 0.001 9% under the power supply from 1.3 V to 2.1 V, and the power supply rejection is greater than 84 dB at 100 Hz.

The real-time and accurate automatic detection technology of road surface damage has important significance for highway management and maintenance. This paper combines the FPGA-based pavement damage detection system to research the positioning accuracy. Due to the influence of factors such as the installation perspective of the system equipment and the speed of the carrier, the accuracy of the road surface positioning may be reduced. This paper proposes a pavement damage location algorithm based on machine vision distance compensation. The experimental results show that when the vehicle′s driving speed is below 72 km/h, the positioning error is reduced from 5 m to 2 m. It meets the requirements of the system design.

There is a problem of poor load regulation in traditional LED power supply. In this paper, a control strategy of inductor current compensation based on the output voltage feedback is used to compensate the output current deviation caused by the load adjustment with the load voltage signal. In this paper, the factors affecting the load regulation of Buck constant current topological are analyzed theoretically. Combined with analysis of engineering application data, the control strategy of load regulation is optimized. Finally, the control method is verified on a LED power supply of 200 watts. The load regulation rate was reduced from 0.03%/V to 0.01%/V, which verified the correctness of the control methodproposed and achieved the effect of optimizing the load regulation rate.

This paper presents a new type of power supply for information network. This scheme has adopted a new topology which can transmit electric energy bidirectional underwater. This function can realize both energy storage and external charging. In this paper, the compensating circuit has been confirmed based on the practical application condition, then the influence of the working frequency to the system has been analyzed, the influence of the eddy current loss in sea water environment has been analyzed and vivificated. At last the experiment of the wireless charging undersea has been completed. This paper has realized 83% of the transfer efficiency when the transmission distance is 15 cm. The result can achieve the technical indicator and verify the feasibility and effectiveness of the proposed strategy.

This paper propose a finger vein recognition algorithm based on the CapsNets(Capsule Network for short) to solve the problem of the information loss of the finger vein in the Convolution Neural Network(CNN). The CapsNets is transferred from the bottom to the high level in the form of capsule in the whole learning process, so that the multidimensional characteristics of the finger vein are encapsulated in the form of vector, and the features will be preserved in the network, but not in the network after the loss is recovered. In this paper, 60 000 images are used as training set, and 10 000 images are used as test set. The experimental results show that the network structure features of CapsNets are more obvious than that of CNN, the accuracy of VGG is increased by 13.6%, and the value of loss converges to 0.01.