Abstract:In this paper, the design and related technologies of Smart Grid Dispatching Operation Management System of Liaoning Province, which is integrated, is studied. In recent years, with the rapid growth of grid-scale and increasing demand for electric power dispatching, the normative and process-oriented requirements to Dispatching Operation Management System are becoming more prominent. The dispersion of the original independent system has not met the needs of development and construction of Dispatching Operation Management System, so unified updates and re-integration of resources must be enforced.

Abstract:As the traditional method of massive data processing has shortcomings of high cost in hardware and the difficulties in parallel programming, a campus cloud computing system platform to handle massive data is designed based on the theory of cloud computing. This cloud computing system is based on the Hadoop distributed computing framework, using map-reduce programming model achieve parallel processing of the massive data. This system can save cost and reduce the difficulty of parallel programming.

Abstract:Demand is an important index of supply cost optimization system. The most common method of study is using poisson distribution to fit. Through the statistics and analysis, we found the triangle fuzzy function fitting the random demand have better fitting degree and scientific. This paper uses fuzzy theory to model the random demand system, then designs the system algorithm and realizes it.

Abstract:It is the key technology on the automation of fully mechanized caving mining to distinguish between the coal and rock automatically. According to the gray histogram the differences between rock and coal, this paper establishes the CCD sensor digital processing system based on S3C2440A+CPLD platform. It can achieve identification between coal and rock through medium filtering the discrete pulse noise of image by using ARM9 and observing gray histogram obtaibed by caculating. The result obtained by MATLAB simulating is more ideal and pratical in coal industry.

Abstract:As the development of drive's interface and Blu-ray, the main control system has new requirements. This paper is about the design of the main controller. The data bus and control bus of the drive are separate. The main controller is able to control the drives by sending commands to motor board. The main board receives packets from Ethernet, and converts them to Uart commands of the motor board to control all of the optical disc library. The main controller is at the core of embedded microcontroller S3C2440 based on ARM920T architecture, and transplants embedded Linux which can help to improve development efficiency and portability. It is certified that the main controller meets the requirement in practice.

Abstract:To meet the need of real-time information transmission in Smart Grid and interaction between power supplier and users, a response system architecture based on ZigBee was proposed.The system applied CC2430 as nodes and the design of coordinator, router linked to smart meters and demand response terminal was made. The test reslts show that the system is reliable and can meet the communication requirements of Demand Response system.

Abstract:In order to upgrade the existing elevator control system, this paper presents a design of the main Controller in the elevator control system based on Micro Controller Unit of LPC2294. As for the Software design, the preemptive multitasking real-time operating system μC / OS is chosen as platform of the elevator system implementing the input/output, scheduling distribution, CAN communication; The hardware design with LPC2294 to achieve interoperability with the inverter, collection for pulse encoder feedback, input and output, etc.

Abstract:In order to get the vision field of driver of the robot in real time, and to accurate its motion algorithms. A kind of CCD moving image formation system is established based on LABVIEW virtual instrument. Hypogenous microcomputer samples the CCD sensor to get digital image signals and send the signal to the LABVIEW main PC via the RF chip NRF24L01. The image signals could be displayed by intensity pattern of LABVIEW and the real time motion image could be observed. Experiments indicate that with this method the robot's moving thread could be observed clearly and hence lays a foundation for the correction of the motion algorithms.

Abstract:This paper develops a Distributed Platform for Sharing IP Flows (DipSIF) based on NetFlow records between multiple institutions. It is assumed that NetFlow traces collected by each participant are archived on separate storage hosts within their premises and then made available to others using a server that acts as a gateway to the storage. Due to secrecy issues the platform presented here uses a prefix-preserving, cryptography-based, and consistent anonymization algorithm in order to comply to different regulations determining the exchange of traffic traces data.

Abstract:Pro/Engineer is a 3D modeling software based on geometric model of the kernel of ACIS. That own Pro/TOOLKIT development interface and under the support of Top-Down design method. Binding functional surface, under the generalized positioning principle and growth design theory, a reverse engineering system that support innovative design was developed based on dynamic link library. In this paper, the solution process of reverse system was introduced.

Abstract:The paper aims at the insufficient of traditional intrusion detection based on genetic neural network not consider the misclassification cost, integrate the misclassification cost-sensitive features into the network intrusion detection model which based on genetic neural network, to overcome the defect of the traditional model's error classifying result in excessive costs. The experiment results show that after the genetic neural network increased the misclassification cost-sensitive features, it can control the cost caused by the network intrusion detection's false report、 omit report attacks preferably.

Abstract:Based on UML sequence diagram of the Unified Modeling Language (UML, Unified Modeling Language) specification is given a formal definition and syntax descriptions, and further analysis of the object between the sending and receiving messages consistency. Finally, the analysis of case-based reasoning process, the characteristics of UML sequence diagrams for further explanation. This UML sequence diagram based model transformation and model validation provided the necessary precondition for the software system can be used to understand the completeness of the structure and construction of visual modeling tools.

Abstract:Question answering system can answer the users' questionsin the application of the precise and concise language. Question interpretation is the primary work of question answering. The precision of Question analysis has a great effect on the following processing work. In this paper, we present a new method of question interpretation based on the users' question theme and focus. This method by which the interrogative words, syntax analysis, question focus words and the first semenes of the question topic in HowNet have used as classification feature. Experiment result indicate that this method can bring about a better effect on improving the accuracy of the classification.

Abstract:It has a good noise immunity to take the high-order cummulants of the signal as it has characteristic parameters. However, different modulation type may have the same characteristic parameters, which result in partial identified.so, it is necessary to bring in fractal theory. The paper takes the cmbination of the high-order cummulants and the fractal dimension as the characteristic parameters of the signal.it not only avoid the Gaussian noise impact on the modulation recognition, but also solved the partial identified problem exist in the situation that take the order cumulants as the single characteristic parameters. Simulation results show the effectiveness of the algorithm.

Abstract:Mobile Ad hoc Networks (MANET) is a new networking for wireless hosts. Because of self-organization, dynamic topology and openness of wireless communication, it makes them very attractive to attackers. Wormhole attack is one of the most severe threats to ad hoc networks. They create a higher level virtual tunnel between two malicious colluding nodes in the network. In this paper, we proposed a novel statistical analysis mechanism based on RTT (Round Trip Time) to detect and avoid wormhole attacks. It detected wormhole attacks during route setup procedure by statistically analyzing the route information and the transmission time between every two successive nodes along the established path before sending RREP to the source node. The wormhole will be recognize before it can do harm to the network in the route setup. And each node compute the RTT, it only introduces very limited overhead. Simulation results showed that the method can efficiently detect wormhole attacks and have better detection rate.

Abstract:As the feature of the private communication system, group call has significance in practice. After analyzing and researching the function of group call, this paper proposed the structure of group call based on SIP. According to the requirements of dispatch in the private communication system, and the features of group call, improved its implementation by several measures. At last，designed the system and described the flow of group call. The programme has overcomed the shortages and achieved better practicability.

Abstract:The Dual Population Ant Colony Algorithm is used to establish the multicast tree. One group searches from source to destination and the other from destination to source. The pheromone update rule is modified by the effect of QoS parameters and the optimal tree which meets QoS parameters is found. The effect of every QoS parameter is confirmed by orthogonal experiment of statistical. According to the scope of the search paths, a suitable orthogonal table of the orthogonal experiment is implemented. The experimental results show that the proposed algorithm can make full use of QoS resources and get the better result rapidly.

Abstract:A portal is a Web based application that commonly provides personalization, single sign-on and content aggregation. In order to achieve a single sign-on on portal, after analyzing different implementations, the paper finally implement it with the centralized authentication CAS protocol, and improves the CAS defect of single logout, so it progress much on system securities and user access conveniences.

Abstract:MicroBlaze softcore, optimized for Xilinx FPGA devices, is a powerful 32-bit microprocessor, which supports the standard CoreConnect bus peripheral set and has strong compatibility and reusablity. This paper introduces the system design methods based on FPGA cooperating with MicroBlaze embedded soft core and implements a WCDMA evaluation system.

Abstract:6 Sigma is an effective method of process quantization, control and improvement. In consideration of problems which could consist in quality management or process control of agile software development method, the paper analyzes feasibility of 6 Sigma in agile software development method, sets YfX black-box model, white-box model and mathematical model based on 6 Sigma, and analyzes DMAIC process, continual improving the Agile software development process.

Abstract:This paper deals with the hierarchical indexing mechanism for map database based of navigation path optimization. It is based on the road network hierarchical model, researches and designs the G-SDBCScan (Geospatial Database Convergence Scan) indexing algorithm and uses this algorithm for the map database's hierarchical indexing mechanism. Finally, the SQlite embedded database uses in the application of hierarchical indexing mechanism to verify the validity of hierarchical indexing mechanism.

Abstract:We research a bi-directional traffic flows network which two nodes are in each signal coverage area's in this article. Selecting optimum relaying using opportunistic relaying strategy based on instantaneous channel state information, then judging whether needs optimum relaying participates in the cooperation according to the threshold value. Don't need relay for transmission when the direct-transmission channel state is bigger than the threshold value; whereas sources and optimum relaying cooperate together. Comparing with direct transmission model, this strategy improves the network throughput, enhances the performance against fading, and reduces system symbol-error-rate effectively in high signal-to-noise ratio.

Abstract:Forward security and public verifiability are two important security features of signcryption schemes. How to design schemes satisfying the two security features at the same time is a public problem in signcryption study. According to the security defect of the signcryption in the Reference [1] put forward by Zhang Chuanrong et al, a new signcryption scheme is proposed through the introduction of secret information of the signcryptioner in this paper, which satisfies with forward security, public verifiability and resistting the PKG to entrap at the same time. Simultaneously, the new scheme can avoid the attrack of the evil information throngh unsigncrying the ciphertext after the signature verification in the process of unsigncryption.

Abstract:Spatial clustering is a hot issue in the field of spatial data mining. For a spatial object, the spatial location and the thematic attributes of spatial data are the inherent characteristics. However, the existing approaches mostly regard only the distance of spatial location as the similarity metric of spatial clustering, ignoring the thematic attributes of spatial objects. The results of these spatial clustering methods are not reasonable. Thus, a new spatial clustering method, named Concept Lattices Based Spatial Cluster (CLBSC for short) is proposed in this paper. The method considers both the spatial distance and attribute distance, and it simplifies the computation via building multi-dimensional attribute lattices. Furthermore, many concepts about CLBSC are expounded and its algorithm is narrated in detain. Finally, two experiments demonstrate that CLBSC algorithm is able to find more outlier and improve the reliability of spatial clustering using the Same Lattices Number.

Abstract:Because of the large numbers of the students, the server always dies down when the students download or upload their papers simultaneously in large-scale examination. However, the P2P application can share the data with others fast and stably without using server. It is a best choice to solve the problem of max data transmission among the nodes in the large-scale examination.

Abstract:To facilitate the optimization of urban traffic operations, scheduling management, ease congestion, this paper introduced the design of the multifunctional terminal system of public transportation based on GPRS/GPS technology. This system uses S3C2416 which is the ARM chip of SAMSUNG, and developed software in Embedded Linux OS. The system implements the functions such as: automatic phonetic service, real time detection, intelligent scheduling route, wireless transmission of data. And the tests proved that the performance was stable and reliable, it is served in many cities.

Abstract:Based on the ARM microcontroller technology, an effective method to monitor smart home remotely is presented in this paper. In practical applications, to achieve remote data communications monitoring operation, this approach uses Web browser as operating interface. The nRF24L01 Radio Frequency modules which can involve 32 bytes payload, setup the wireless data communication platform for smart home terminal device. By the definition of data frame format, the approach finishes the design of the wireless communication protocols and unifies the application interface of basic devices, which increases the flexibility of smart home development. The remote monitoring system play an important role on monitoring house environment and other distributed sub-system, enrich and extend the operation for house intelligentize.

Abstract:A design of video monitoring system used in the field of smart home is introduced. It is based on the embedded ARM and S3C2440. Video4Linux2 is used to get the video image data captured by USB camera, and displaying the video image in real-time on PC with SDL.

Abstract:Due to the problems of production facilities and productivity effect, traditonal fermentation technology of rice wine is known to be inefficient. In this paper, we address the problems with a new automatic control system. Its hardware platform consists of PLCs, intelligent instruments, and industrial control computers. Its application is programed based on Labview. In the production process of rice wine, it has high reliability and real-time and can monitor the various parameters effectively. Productive results show that it can meet the demand of rice wine fermentation.

Abstract:With the LAMP as development environment of the software,the double-core CPU as the hardware platform, this paper carries out the research and design of online evaluation teaching-supporting system on the platform. It focuses on the design and implementation of on-line evaluation system on a multi-core CPU hardware platform, which will improve the solo-core system and solve the problem of low efficiency in the single-thread or single-process evaluation strategy.

Abstract:This paper describes the specific implementation of device USB driver based on studying USB system architecture and principle of USB device driver in WindowsCE platform, the device is ultrasonic acquisition device that use EZ-USB FX2 series USB controllers maded by Cypress company.Using device driver. The paper acquires ultrasnoic data and control ultrasonic device successfully.

Abstract:This article expounds the designs and implementations of hardware and software for industrial double Ethernet interface based on AT89S52 microcontroller and AX88796L Ethernet controller used for some kind of radar.

Abstract:Accurate and reliable detection of core point for fingerprint classification and matching are important. This paper introduces a good method for core point detection in fingerprint to improve accuracy of the position and reliability of the core point. According to the definition of the core point is that the greatest point of curvature ridge, this paper use the ridge line tracing methods derive the core point. The experimental results show that the method can extract the core point from the fingerprint image more accurately and reliably, and has good robustness.

Abstract:The genetic algorithm convergence slower, easily into the local extremum faults. Through the algorithm is proposed based on hybrid differential evolution of chaos and hybrid genetic algorithm. The algorithm utilizes chaotic motions of the ergodicity and intrinsic stochastic according to create initial group, the evolutionary algorithm for difference strategy, make the chromosomes in solution space more effective search optimal solution. Finally, this hybrid genetic algorithm applied to the multi-processor scheduling problem, the experimental results show that the hybrid algorithm is high efficiency for the optimization, can better optimization results.

Abstract:In order to solve the problem of security of video information, a chaotic video encryption algorithm based on H.264/AVC video coding standard is proposed. Because Stream cipher is simple, fast, etc., The proposed scheme uses discrete chaotic sequence of Logistic to encrypt the stream of CAVLC (Context-based Adaptive Variable Length Coding) entropy coding stage of H.264/AVC, and analyses the security and encryption efficiency of the algorithm. The results showed that the video encryption algorithm ensures the security of information, does not change the structure of the stream and the system has a good real-time and fast speed.

Abstract:A method combining Local Binary Pattern(LBP) with Locality Preserving Projections(LPP) was presented for face expression recognition. First, the algorithm processes facial expression images with LBP operator and integrates local and global characteristics of the face. Then Locality Preserving Projections(LPP) method is used for feature dimension reduction and SVM for classification. The algorithm is experimented in Japanese Female Facial Expression database(JAFFE database). The results show that the proposed algorithm is more effective.

Abstract:As the inspection demands of the transmission, it is necessary to enhance the efficiency of power inspection. Taking all factors which have significant effects on the running status of the transmission lines into consideration, this system establishes a state prediction model of transmission line based on rough set theory and decision tree. The rules which extract from the model apply in the abnormity prediction of the transmission lines. The manager can work out a scientific and reasonable inspection plan in accordance with these rules. This model is proven and have a higher accuracy.

Abstract:The intrusion detection is a new network security technology and Intrusion Detection System forms one of the most important parts of modern computer security. Nowadays most of Intrusion Detection Systems have been using the pattern matching algorithm, this essay introduces the implement principle of the intrusion detection technology based on protocol analysis on the question of the performance bottlenecks of these systems' detect engine at high-speed internet environment. Then we propose the method of detecting attack rapidly by making use of the network protocols' high degree of regularity. In this way, the possibility of false alarm and miscarriage of justice can be reduced. And at the same time, the performance and the efficiency of network Intrusion Detection System can be improved.

Abstract:This paper studies the binarization technique of QR code image and proposes an improved background gray-level based algorithm for uneven illumination QR code image. Firstly, it cuts the QR image as several blocks based on the size of original image, then evaluates the gray-value of each block and set block's gray-value with the bilinear interpolation method. Thirdly, it subtracts the estimated background image from original uneven illumination image get the corrected image, and with this method to eliminate the bad affection for binarization from nonuniform illumination. Experiments show that the algorithm can get a better QR binary image effectively for QR code image which has widespread gray-level and especially for uneven illuminated QR code image.

Abstract:The text aim the currently unsafe transport situation of the symmetric key of symmetric encryption algorithm, presenting the improved key generation algorithm based on the initial keys and key synthesis function matrix.premising the key transmission security effectively without sacrificing computing speed. Finally, the text illustrates the implementation model and implementation process of this improvement dynamic generated symmetric of symmetric key by an example. The text study conclusion is this improved symmetric key dynamic generation algorithm principle is simple, efficiency is high, suitable for the field of encryption research and extension for further use.

Abstract:An implementation of embedded IPv6 stack independent of any platform is proposed. Based on the embedded application, the management of buffer is designed by Mbuf technology. and some functions of ICMPv6 and Neighbor Discovery modules in IPv6 are reduced. TCP and UDP modules are extended. The stack is tested on the s3c2440 platform. All these show a good performance.

Abstract:A new cell-phone-type vehicle traveling data recorder is designed. In the aspect of data processing, information display and data storage, it uses cell phone's processor, LCD and SD card instead of traditional vehicle traveling data recorder's corresponding components. In the aspect of data transmission, it uses Blueteeth to exchange information and order with the data acquisition equipment. It achieves wireless communication. Bluetooth has a special service and an ordinary service. It distinguishes accessing power between system user and guest user. The system could display, record and analyze parameters of vehicles, save and playback data of acquisition. It realizes. It makes the vehicle traveling data recorder portable and wireless.

Abstract:This paper discussed the importance and necessities of formula defined by end users according to their own requirement of a system. It is believed that almost all information management systems should supply such a module during it's developing cycle. After a brief introduction and Comparison of formula define to the common methods currently used by many management systems, it selects one typical of them analyze thoroughly and illustrate data model of the system. Finally, we take the agent Commission management system, for example to illustrate the calculation process, formulas resolution rules and procedures.

Abstract:In the recognition of off-line Chinese char, thinning is an important step. Mathematical morphology algorithm is the most useful thinning method in recent years. The results are heavily influenced by the structuring element we used. With the large amount of the analysis of the thinning algorithm and the character of the Chinese char, we improved the previous structuring element. The objects are altered to unitary char. 23 group of Chinese char have been used to experiment and each group have 100 Chinese char. The results show that this algorithm solved the problem of losing the ending information and reduce the running time.

Abstract:CACHE is a kind of high speed buffer storage device used to connect CPU and main memory, which can improve the speed of system's read and write. The referred CACHE in this article is not real CACHE, we use main memory to implement the function of CACHE.

Abstract:The wide use of UML(Unified Modeling Language) dramatically promotes model status in software development. Model based software development approaches have gradually been accepted. UML, however, is large and complex because it is designed as a generic modeling language. So it is not easy for domain experts to learn. Software designers often fall into the heavy modeling work and can not get out of it. Using DSL(domain specific language) can solve this problem well. Generating DSL by using tools becomes the only choice because the manual design cost of DSL is very high. Meta-modeling is a technique that generates DSL through building domain meta-mode. In this paper, we take LED millipede design as an example to introduce in detail the MetaEdit+ based meta-modeling approach.

Abstract:A cloud classifier based on swarm particle optimization (PSO) is presented, and used in the classification for multi-dimension object. The digital characteristic of cloud model is expected value Ex、entropy and super entropy He, the membership to which every attribute data of classified object belongs to its attribute set center is presents by 1-D cloud model. The digital characteristic of 1-D cloud model is optimized by swarm particle optimization (SPO). The swarm particle optimization cloud classifier (SPOCC) is built from every attribute cloud model, and used in the classification of iris data set, the experiment result is very well.

Abstract:To mine data stream efficiency, a new slide window, combines basic window and time window, adopts a scheme of window based on sector. It could apply to the real data streams and change the size of the windows, and save time and space that no necessary to use. And the algorithm based on frequent itemsets mining-DS-stream could self-adaptively adjust the size of time windows, considering the large number and chronotropic character of the data stream. In the algorithm, transaction list group is adopted as synopsis data structure. The experiments results in Eclipse indicate that the DS-stream algorithm is more effective in term of temporal and spatial performance.

Abstract:Particle, among swam algorithm, is apt to keep its own state. While how to change its state has great influence on the position and the adjustment of the velocity. In this paper presents a new method—an adaptive particle swarm algorithm of periodic random disturbance strategy. And the nonlinear declination as well as Self-adapting inertia improved in the process of particles moving. Better results can be obtained by the new method compared with the former ones and which only adopts linear decline in the particle swarm algorithm.

Abstract:The accurate extraction of fat and muscle from meat image is one of the key technologies in nondestructive test (NDT). To deal with this problem, based on the normalized cut(Ncut) method, in order to reduce the computational complexity, an improved algorithm has been put forward. Firstly, the basedmoment threshold selection method to separate the meat out from the background region. Secondly, the quantitative color grade creates the color histogram for the meat area. Finally, to calculate the similar matrix of color histogram, and exploit the normalized cut as a measurement of spectral clustering on dividing histogram, in the light of the consequence in histogram partition to extract muscle and fat. Experiments illustrate that to compare with the pixel-based spectral clustering algorithm, the improved algorithm does not only reduce the computational complexity, but also the optimum segmentation.

Abstract:After detail analysis the traditional algorithm of 2DPCA, an improved recognition algorithm whose feature extraction applied twice is presented, it can reduce dimensionality. Extensive experiments are performed on the ORL and Yale face databases. The result shows that the improved algorithm has higher feature speed and recognition accuracy than traditional 2DPCA.Finally,compared with PCA, LDA, LPP, etc., the proposed algorithm is superior to other algorithms in recognition rate.

Abstract:In view of small and medium enterprises cannot afford the high cost of CAD software, developed mechanical parts online design platform using Silverlight. CAD elements are organized in tree shape using controls of TreeView. Users modify，edit and operate properties of elements to design, and also can output the designing into picture and AutoCAD format. After running, the platform is easy to use, interface is friendly, will help small and medium enterprises greatly in technology.

Abstract:This paper presents a design and implementation of wireless touch screen ordering system based on Zigbee. The ordering system makes up for PDA ordering. ARM9 processor, wireless transceiver module CC2430 and LCD touch screen ordering system work together to build a terminal node in hardware; the embedded Linux is operation system, QT is used to design the ordering terminal node interface, using vc + + to develop programs of foreground business system for data processing and business process analysis. Experimental results show that the system is easy to use, with low power consumption, high transmission efficiency, with good promotional value.

Abstract:Aiming at lack of SQL Server skills assessment system, this paper proposes a model about skills assessment of SQL Server Based on ADOX. Logic formal method used to Descript control information of automatic marking. Analysis the description structure of logic formal, access the operation information of database and judge rules of operations about create modify and delete. Using the ADOX method to obtain information database objects, using logic formal method to achieve SQL Server database skills assessment automatic marking.

Abstract:According to quantum mechanics, quantum state has many advantages of the coherence and entanglement and the inner parallelism of quantum algorithm is totally different from the classical algorithm. Shor's algorithm shows the advantage of quantum algorithm in solving some classical Non-polynomial problems, then Grover's search algorithm further proves this point. And this brought more and more research on quantum algorithm in many countries and they have got great achievements especially on quantum intelligent algorithm. In this paper, the development and fundamental principles of quantum computing are generalized, then three typical quantum algorithms is introduced, and the superiority of quantum computer is explained. Finally, current research of quantum computation are introduced.

Abstract:Cloud computing as a new business model is developed very fast. Data storage and management is a very important and valuable research field in cloud computing. This paper introduces the concept of cloud storage as well as the advantage and architecture of cloud storage. Then it analyzes the cloud data storage technology-- GFS(Google File System)/HDFS(Hadoop Distributed File System) and the cloud data management system--BigTable/HBase based on the specific cases of enterprises.

Abstract:Variety of crop pests, not only the crop yield and quality has an adverse effect, but also a great threat to human health. This article analyzes research status of crop pests recognition at home and abroad. Different from the previous method of artificial eyes, computer vision has high practical value. Using the image processing and pattern recognition technology to judge disease accurately, in time and quickly. It saves time to take appropriate measures, and can greatly enhance economic efficiency.

Abstract:The safety management of construction enterprises has not only relation to the development of construction industry, but also has a strong impact on harmoniousness and progress of society. This paper constructed the evaluation index system from manager perspective, and explained the relationship of evaluation index briefly. Then, this paper applied the principle component analysis which carries on the pretreatment to the input data of nerve network, and withdrew essential ingredient as the input of the network. Subsequently, genetic algorithm was applied to improve the convergence ratio. Finally, the analysis results was given out based on the example of Tianjin city. The results indicate that the model is not only more fair but also reasonable. It improved the learning efficiency of the neural network model.