The BH-PREPT ( Bandwidth Preemption) algorithm only concerns with minimizing the waste ofbandwidth without considering the computational complexity and the current development of optical fibercommunication, leading to excessive increase of delay in the networks. An algorithm greatly reducing the delay ofthe networks, named DH-PREPT(Delay and Bandwidth Preemption) is proposed. The DH-PREPT algorithmmakes the computational complexity lower under the premise of minimizing the utilization rate of bandwidth. Itconsiders the delay indicator by the bound of forwarding equivalence class and LSPs(Label Switching Paths) andfast forward the data of custom. Simulation results show that the proposed algorithm significantly outperforms the BH-PREPT algorithm when preemption occurs in the network.

In order to solve the problem of edge blur in image reconstruction using the traditional POCS (Projection onto Convex Set) algorithm, A super-resolution image reconstruction of POCS with edge preserving is proposed. According to the neighborhood of interpolation pixel, the area of the pixel is the edge region or the non edge region. Reference frame is constructed by the improved bilinear interpolation algorithm, reducing the blur of the image edge after the reconstruction of the traditional algorithm. Experiment results demonstrate it can obtain high resolution reconstruction images and image edge isbetter.

For the current limitations of traditional crop's data monitoring, difficult setting up lines, labor-intensive material and other problems, a remote and intelligent crop's low-altitude observation system was designed. Taking Raspberry Pi as the core, the system can achieve the observation on low-altitude and ground through wireless network. For the low-altitude part, DHT11 temperature and humidity sensor was the temperature and humidity acquisition unit, and BH1750 sensor was the light intensity acquisition unit. Temperature-humidity and light intensity acquiring circuit were built through the Raspberry Pi GPIO(General Purpose Input Output). Photos of crops at low-altitude were taken by the Raspberry Pi IP(Internet Protocol) camera. Data and image can be stored in Raspberry Pi. For the ground part, setting STC89C51 microcontroller as the core, temperature-humidity sensors were connected on it. The collected
data then were transferred to the Raspberry Pi through nRF24L01 wireless module. The Raspberry Pi VNC(Virtual Network Computer) access interface was landed on PC(Personal Computer). Upper computer was built on the VNC access interface, on which reaL-time monitoring and analysis of data can be realized. Test result showed that the system implemented a full range of observation of the crop which consisted of temperature, humidity, light intensity, growth and other indicators. A remote and intelligent data transmission and observation of the crops were immplemented. The system used new technologies of UAVs (Unmanned Aerial Vehicle), card-type computer Raspberry Pi, wireless networks and so on, solved the traditional method of stringing, consumption of wealth and other issues. It had a strong advantage over the traditional method.

In order to realize parallel multi-channel high-speed data acquisition and storage, a method which is based on FPGA(Field Programmable Gate Array) and Nios II is proposed. This system can realize the parallel control of multi-channel ADCs(Analog-to-Digital Converter), to achieve parallel high-speed acquisition of multi-channel signal. The data acquired and the FPGA configuration data can share the configuration memory, which can save the extra storage devices, and reduce the system cost. In the experiment, use EP2C35F672C8 as core, AD7980 as ADC, EPCS64 as the storage medium, to achieve entirely parallel acquisition for 15-channel analog signal. The adoption of soft core technology makes the system more flexible and extensible. The results show that it provides new ideas to projects which requires low cost, frequent system upgrade.

In order to maintain robust stability while meeting certain performance index requirements, the problem of guaranteed cost control for time-varyiong time delay singularly perturbed systems with norm bounded uncertainty parameters is discussed. Using Lyapunov stability theory and matrix analysis method, based on a new quadratic L-Y performance index, the sufficient quadratic stability conditions, state feedback guaranteed cost control rate and the guaranteed cost index is presented by showed LMIs, and the minimum performance indication upper bound is derived. The method of the dissertation can be extended to multi-state delay uncertain systems guaranteed cost control problems, the numerical examples are employed to elaborate that this method is effective.

To improve the precision of cam grinding machining, reduce the cam contour error, and improve the robustness of the system, a new error compensation method is adopted. It is named profiling tracking error compensation, in which the actual profiling tracking error compensate the series values of X axis. The two axes linkage feedback system is established by using Matlab, and the fuzzy PID(Proportional-Integral-Derivative) controller is designed to realize online compensation for the system. The cam used in the simulation experiment is difficult to process so as to verify compensation effects and the performance of the controller. The simulation experimental results show that this method can effectively reduce the cam contour error, simplify the calculation process, and improve the response speed of the system. Compared with the traditional PID controller, it has better robustness.

A capacitive three-dimensional force flexible tactile sensor is proposed, which is capable of measuring normal and shear force. The structure design of capacitive three-dimensional force tactile sensor and the working principle are introduced. The tactile perception mechanism is verified by using ANSYS finite element simulation software, and a portable capacitive tactile information perception system was designed based on low power microprocessor CC2530 and AD 7147-1 capacitance-to-digital converter. The sensor has achieved dynamic response time of 226 ms, the normal load measurement is from 0 to 5 N, shear load measurement is from 0 to 3 N and the measured sensitivities are 6. 87 fF/ N and 10. 96 fF/ N. The experiment results indicate that the flexible capacitive sensors array and the capacitive signal extraction system have good stability and sensitivity, providing a design proposal for wearable artificial skin.

In the foggy condition captured image is seriously degraded, we propose a fast image defogging algorithm based on YUV color space. Based on atmospheric scattering model, firstly the luminance is extracted, geting the scattering light after Gauss filter; then the maximum brightness value of the image as the atmospheric airlight; then can obtain a clear image. The experimental results show that the algorithm effectively improves the image quality, and can keep good image details.

In order to solve a common problem of vision drop in teenagers, we design an intelligent instrument for improving vision, which is based on the Physiological characteristics of the eye. The instrument applies computer control technology and dual color LED (Light Emitting Diode) display technology, and uses stm32f103zet6 microprocessor to control LED display screen and man-machine system, and uses RLE(Run Length Encoding) image compression technology to make display flexible, rich and colorful. The tester uses the operating platform (voice prompts and liquid crystal display) to observe various types of training graphics, which can adjust the tension of the eye and relieve vision fatigue. After a period of treatment, the test results show that the students‘ eyesight is generally improved, and the effect is remarkable.

In order to quest for large band gap or complete band gap of two-dimensional photonic crystal, the band gap of the structure in two-dimensional square lattice and triangular lattice with dielectric cylinder using the plane wave method, and the band gap of the compound structure by changing the unite cell are studied. Results show that to the dielectric rod, the middle frequencies of the band gap will reduce with the increase of the radius and refractive index of the dielectric rod. The compound dielectric cylinder structure has big TM(Transverse Magnetic) band gap, and it makes TE(Transverse Electric) band gap and the formation of this make complete band gap be possible. Changing the unite cell to reduce the symmetry of the structure is a feasible method to quest for complete band gap.

The monophyletic Dijkstra shortest path algorithm was analyzed based on GIS(Geographic Information System) platform. A balanced community analysis network segmentation method was proposed, the road network was divided and degraded into balanced community network based on GIS, and the limited layering algorithm was used. Nodes that are unlikely to appear on the optimal path was eliminated to limit the search area and the purpose of reducing the complexity of algorithm is achieved. The experiments show that the optimized results can effectively reduce the number of searching nodes, compared with the classic algorithms, the efficiency is improved.

To avoid traffic accidents caused by abnormal driving status, a method of negative emotional state detection method is proposed. LBP ( Local Binary Patterns) and SVM ( Support Vector Machine) are used combining decision tree algorithm for facial expression recognition and analyzing. The experimental results on the datasets show that the proposed method has high accuracy while maintaining the low cost processes.

Three kinds of range-free localization algorithms including Centroid algorithm, APIT(Approximate Point In Triangulation) algorithm and AIGS(Annulus Intersection and Grid Scan) algorithm are studied. We research on the mathematical relationship between density of anchors, location precision, and energy consumption. Research shows that the three algorithms can all get more accurate location when enhance the density of anchors. However, all of those cause more energy consumption. In the monitoring area with the same proportion of anchors, comparing about positioning accuracy, AIGS algorithm is better than APIT algorithm, but APIT algorithm is better than Centroid algorithm. When density of anchors is low, we propose an iterative scheme which transforms unknown nodes after localization to beacon nodes. The new scheme can increase localization accuracy of WSN(Wireless Sensor Network) with low density of anchors.

To solve dependency of the financial market, we anysis minimum dependency of high frequency between different financial markets. First, we introduced the model method and the model characteristics of Copula GS function. Secondly, the method of estimating the parameters of Copula GS function was studied.
Finally, the tail dependence and negative tail dependency of the model were studied, and mainly did the research based on Eviews and GS Copula function index futures IF1112 Index and SSE 000 001 Index 5 minutes minimum yield sequence dependencies, it is concluded that the data of their yield sequence data are very strong. The result provides theory foundation for weakening financial decision risk.

In order to improve the efficiency of oil field pipe network inspection, reduce the pipeline accident, we present the design and implementation method of the oil field pipeline inspection system based on the Beidou satellite navigation, list the overall design scheme of the system development and describe the realization method of basic data processing and key functions of each functional module in detail. The system replaces the way of building a base station themselves or use GPRS(General Packet Radio Service) in traditional network inspection system with Beidou short message communication technology to solve the problem of current inspection system in the absence of public wireless network coverage area. We choose Dijkstra algorithm which is more suitable for oil field work environment instead of A* to build path navigation model, by practical operation comparison, the processing efficiency of the software is improved more than 30% compared with the same A* algorithm model. Practical application shows that this oil field pipeline inspection system can improve the efficiency of oil field pipe network inspection, reduce the network transmission rate.

Based on the analysis of the interactive process between the OpenStack’s platform structure, generating process, security mechanism and other service components, an improved ID authentication project is proposed to solve the deficiency of fine-grained authentication, low-usage of database and security flaw of data. This project integrates the LDAP(Lightweight Directory Access Protocol), RBAC(Role-Based Access Control), and SSL/ TLS(Secure Sockets Layer/ Transport Layer Security) into Keystone service, which strengthens the performance of Opentack‘s cloud platform in the aspect of the extension and security of ID management.

To solve the problem of incompletement for the captured images, we improved SIFT feature stitching algorithm to get the whole palmprint images. We proposed LoG-SIFT feature points matching stitching algorithm and designed a strategy for mismatching points remove which can improve the accuracy of matching point. The experimental results showed that the improved algorithm enhanced the palm surface ridge characteristics, significantly increased the number of matching feature points and lowered points matching error rate.

The traditional interface performance is not clear, and the programming statements are tedious. We design the interface of facial expression recognition system based on the NI company LabVIEW software, and discuss the key problems in the design of facial expression recognition system, then use efficient Matlab processing software engineering, to extract expression image feature, using the theory of discrete wavelet algorithm and the of flexible template matching algorithm is used to identify the emotion. The experimental results show that the design based on LabVIEW facial expression recognition system uses the LabVIEW software simple statement to design a clear interactive interface, and the recognition rate can reach more than 85%.

A test method is proposed based on k-means average classification thought, which is designed for testing the data from on-load tap changer to help the repair personnel quickly locate the faults. Compared with original k- means algorithm, the experiment shows that using fuzzy k-means algorithm largely improves accuracy and stability of the recognition results, and this classification method has good ability in fault diagnosis, which can make a quick diagnosis according to the data from on-load tap changer, and help the repair personnel quickly locate the faults.

In order to optimize procurement business process of small and middle-small enterprise, to improve the efficiency and effectiveness of procurement, save costs and increase profits for enterprises, facilitate the management of the staff of the procurement process, we develop management information system. We analyzed procurement process of middle-small enterprise, determined design idea and running process of the purchase management information system, using the software of VB, we design a set of purchasing management. The company complex manual business will be converted to computer operation, thereby improving work efficiency.

According to the two characteristics of clothing (the importance of color information and the little variation in local areas) in content-based clothing retrieval application, the paper proposed a clothing retrieval method based on multi-feature index and local constraint. We construct two inverted file index structure with color feature and SIFT feature based on Bag of Words model respectively. Experimental results based on the clothing image database with the various situations demonstrate that our method could work well.