Search

When using international popular algorithms deal with RMB paper currency, there exist three main aspects. First, the size of RMB paper currency is different with other countries. Second, the color of RMB paper currency is different from other countries, and the different face values have different major color, pattern and texture. Third, the location of face value in paper currency of our country is different from other countries. For RMB paper currency image, using edge detection is to separate the paper currency from the image according to boundary. In other words, the background is removed and only paper currency is left. This paper firstly introduces the principles of removing background by gradient operator, LOG algorithm and Canny algorithm. Then it analyzes the edge image results obtained by these three kind algorithms. At last, these three algorithms are evaluated. Results show that the Canny algorithm can satisfy the requirement of removing background and can improve a good foundation for the subsequent of paper currency extraction.

On the basic of multiple populations of immune algorithm and clonal selectionalgorithm, the purpose of this paper is to further improve the detectionefficiency and reducing the false alarm rate. This paper uses the kddcup99 dataset as the experimental data set, and chooses four types of attack data groupof experiment data set as initial population of multiple populations of clonalselection algorithm, through the algorithm to create the optimal model. Basedon the principle of normal data larger than the abnormal data, in turn,experimental data set matched with the normal data set and the optimal model bythe improved R matching algorithm. The results of this paper show that thedetection rate increased significantly

This paper studies the power transformer fault quality diagnosis using rough sets theory and neural network. It is rough sets reduction as the pre-unit of neural network based on reduction algorithm with the attribute significance. The paper describes the reduction algorithm and implementation method detailed. Through the training and testing results with practical data, it is proved that the reduction algorithm with the attribute significance can make the number of input samples shorter, the training speed faster and the diagnostic accuracy higher. The algorithm is feasible and effective for applying to the fault diagnosis system of power transformer.

Teaching evaluation is to process the information of the huge amount of teaching data and to evaluate the teaching activity, process and outcome to provide the basis for the quality and decision of teaching. S-P Chart is a typical evaluating method. The paper introduces both traditional and fuzzy S-P chart. Then, the paper proposes three improved S-P methods, including single-knowledge-point analysis, multi-knowledge-point analysis and cognition analysis. All these three improved methods supplements the shortages of traditional S-P and fuzzy S-P, and are implied in an analysis system of the quality of teaching.

To solve the problems such as randomness user behavior, effective privacy protection and massive data processing shortage, this paper has proposed a household load data processing method with privacy protection, combined data mining and cloud computing. First, it gave the platform architecture and individual protection model. Then, it proposed Mask-k_means database encryption method, algorithm parallelization was implemented by MapReduce. Finally, this paper varied the method was and implemented household load data processing based on statistics and real-measured data respectively. One conclusion is that feasibility of cluster analysis for multi-user was low, while single-user load analysis is high. The other is that this method is simple and practical. This can provide a new way to household load big data under smart electricity consuming.

In order to remedy the defects of traditional methods in the data mining, improving the data mining effect. The paper proposed the improved TFIDF algorithm and applied to the data mining based on the analysis of the flaw and the insufficiency in simple calculation method, minimum value calculation method and the traditional TFIDF algorithm. The paper proved that the improved TFIDF algorithm is better than other traditional methods more scientific, more advantages and practical value based on four data mining results are compared and evaluated.

In order to meet the current needs of the report dealing with large volumes of data, operation difficulties of data sources, fixed form and other issues, The paper raises that basing on principles of modular and hierarchical design thinking, preprocess various data sources in dimension, store in special data base, then dynamically generate report through a combination of different dimensions in the hierarchy. The report system in this paper is designed 4-layer architecture in hierarchical relationship, ensuring that coupling value of each part of the system is a minimum, and this improves system module’s universality and the efficiency of dynamic report system’s deployment and execution.

The evaluation of surface water quality is a problem which relates to whether the environmental function and value do possess or not in the water. A new comprehensive evaluation method which combines Principal Component Analysis (PCA) and the weighted grey correlation method was presented in this paper. The reduction of indicators and the weakening of the multiple correlation among indicators were considered in Principal Component Analysis. According to different functional areas categories of every section, the weighted grey correlation method could determine the evaluation indicator’s weights, strengthen the effect of concerntrations of indicators and consider each indicator’s impact. The Mudanjiang River downstream sections were used as the research object. Based on the water quality monitoring data of typical monitoring section in 2013, the study on the environmental quality assessment of downstream section in mudanjiang was given in this paper. The efficiency of the evaluation process and the accuracy of the results can be improved.

The target of this paper is to introduce the improvement of the technique of image retrieval. At first, it comes up with the concept of image retrieval and shows the importance of this technique. As the techniques of multimedia and Internet are developing rapidly, the resources of images that users obtain are also extended. And then this paper gives the problem about the image retrieval, namely the information of images are disordering. As the result, it is significant to do the effective organization, management and retrieval based on the increasingly extensive image information storage. After that, this paper presents the concept of TBIR and CBIR and gives the definitions of them. It proposes an issue that CBIR is the improvement of TBIR. Based on CBIR, there are also some disadvantages that need to be improved. In terms of the main point of CBIR, the paper raises that the annotation is one of the most difficult techniques that need to be promoted. Then it describes some algorithms about the technique of automatic image annotation. After these algorithms, the paper shows the challenges and developing direction of the technique of image retrieval. At last, it presented the conclusion to emphasize the main points of this paper.

Materialized view is an important topic in data warehouse research, and also affects the query efficiency and maintenance cost. The disk-space view-selection problem is to select a set of materialized views for the purpose of minimizing the total query processing cost and the total maintenance cost. In this paper we introduce evolutionary algorithm using stochastic ranking algorithm, which can enable materialized view selection under disk-space constraint. The algorithm improve the stochastic ranking algorithm, which can find a near-optimal feasible solution. This paper use the algorithm in Police data warehouse.

Data mining is the general methodology for retrieving useful information from big data. Clustering analysis is a mathematical method of classification for unsupervised machine learning. It can be adopted for data classification in Data mining. This paper combines the clustering process by fuzzy way and then deduces a special clustering algorithm with fast fuzzy c-means (FFCM) method. In summary, the paper illustrates the adoption of a series of fuzzy clustering methods in Data Mining. These methods have improved the computational efficiency with learning as the convergence speed is fast. The methodology of this paper presents significantly meaningful for information retrieval of big data.

Analysis of user behavior management control has important significance for network users. This paper uses the clustering algorithm to cluster the user behavior for clicking certain software or Website in different times, through the time context of user behavior to analyze user behavior rules. Additionally, this paper analyzes the clustering result in detail, then, divides user behavior into different types. Finally, according to the clustering result, more targeted pages and applications are recommended to the network users to create huge business value.

An algorithm which named Boosting algorithm occurred in the last ten years. It can raise the learning algorithm accuracy with multiple learning and obviously improve the efficiency of learning algorithm by adopting the principle of “comprehensive optimizing”. In addition, it can effectively develop the “weak learning algorithm” with low efficiency into a “strong learning algorithm” with high efficiency. The boosting, a new learning method to integration machine, is based in learning theory and show its good qualities in many fields. The paper elaborates and summarizes the basic ideas of Boosting algorithm and applies it into Data Mining (DM). In recent years, the DM has been extensively adopted out of labs such as in commerce, technological research and engineering technology. In this background, the paper tries improving traditional DM algorithm to solve these problems in industrial application.

— This paper demonstrates to build a Fuzzy Inference System (FIS) for any model utilizing the Fuzzy Logic Toolbox graphical user interface (GUI) tools. A different conception for decision making process, based on the fuzzy approach, is propounded by authors of the paper.The paper is worked out in two sections. Description about the Fuzzy Logic Tool box is done in the first section.Illustration with an introductory example concludes the second section. Based on various assumptions the authors construct the rule statements which are then converted into fuzzy rules and the GUI tools of the Fuzzy Logic Toolbox built using MATLAB numeric computing environment is used to construct a fuzzy inference system for this process.The output membership functions are expected to be fuzzy sets in Mamdani-type inference.Defuzzification of fuzzy set for each output variable generated after the aggregation process has to be carried out. Application of information technology for Decisions in today's environment which is highly competitive are undeniable principles of organizations and helps managers in making useful decisions meaningfully.

This paper focuses on the correlation analysis method based on vector space model. In the case of dual classification, this paper made a Joint comparison to find the most appropriate method of selecting featured items for the focused crawler; and then made special effort on analysis and verification of LBTF-IDF algorithm in which the weight calculation method has been improved.

This paper mainly discusses the sites of complex mountainous area wireless communication stations which is based on the analysis of the top digital model in order to highlight the advantages and functions of GIS in the practical application. This paper is thought on the theory of visual analysis, which is based on the digital model. According to the characteristic value of terrain data extraction, the obtaining and conversion of interpolation parameters as well as the features of characteristics and hope vector coverage area, the verification of signal coverage appears to be consistent with the shape visibility analysis of the base station position. With the assistance of professional ARCGIS, it can be easily to track the data and complete the 3D analysis and surface analysis of the typical complex mountainous terrain. Finally, the paper presents the prediction, preliminary selection, adjustment, optimization and results of station construction.

In this paper, we present an approach to characterize and model microblog traffic in cellular data network. In contrast to previous methods, our approach is based on the cloud computing platform and the cluster system, including the Hadoop Distributed File System (HDFS) and the parallel processing software framework MapReduce. Whats more, we focus on the contrast of Sina and Tencent microblogs. We analyze the features of microblog traffic in four aspects of increasing details, which are (i) traffic diurnal pattern, (ii) modeling the traffic distribution, (iii) user distribution, (iv) diversity usage of microblogs. This approach of analyzing microblog traffic comprehensively is probably the most important contribution of this paper. Furthermore, our approach has two important features. First, the massive mobile subscriber data we used in our experiments was collected from a commercial Internet Service Provider (ISP) covering an entire province in Southern China. Therefore, it ensures the results indicate the true characteristics of microblog traffic in network. Second, we investigate that the microblog traffic fits with the power law distribution. We demonstrate the electiveness of our approach on three real datasets. Our results are important for cellular network operators to learn user behavior and optimize the future microblog application designs.

Three-dimensional model is high degree of freedom information. In order to realize content-based 3D model retrieval, normalization preprocessing of 3D models is needed. This paper have proposed a normalization method with 3D modeling software based on center of mass, principal axis of inertia and volume parameters to carry out coordinate system normalization of three-dimensional mechanical entity models. In this paper, Specific method is listed. Application examples are given to show the feasibility of this method.

In recent years, the increasing Debris flow has been drawing the geologists' attention for its massive destructive force. At present, the research on the debris flow are mainly based on comprehensive multi-factors evaluation or singlue factor qualitative anlalysis. However, all factors have the impact on the debris flow. At present, the research on the factors that foundamentally affecting the debris flow or its character is quite few. This paper use the method of stepwise discriminant analysis to minimize the correlation and repeatability between factors to ensure the selected factor can be fully represent the character of debris flow. In addition, classification is an important way to illustrate general picture of the debris flow. This paper classified the selected 6 key factors by method of FCM (Fussy C Means) to achieve the transition and evolution from general geological classification to the accurate quantitative analysis. The result of such classification will benefit the debris flow cognition and disaster prevention measures.

FTAe is software tool for Fault Tree Analysis in Microsoft Excel. This paper is focused on short description of created FTAe. A brief description of Fault Tree Analysis is in the first part of paper. The Structure, Logic and Minimal Cut Sets in Fault Tree analysis is described too. Next part of this paper is focused on the choice of algorithm for cut set calculation and brief description of structure of FTAe. Finally, advantages of FTAe software in Microsoft Excel are described in conclusions.

The present paper attempts to clarify which priority rule is appropriate for specific manufacturing environment variations, such as processing time variations, operation additions, and additional job releases. In such manufacturing environments, the performance of scheduling rules should be evaluated with a degree of robustness under dynamic situations and traditional measures, such as makespan, under static situations. The present paper evaluates the robustness of the priority rules FCFS, SPT, MOPNR, and MWKR from the viewpoint of both dynamic and static situations based on the similarity of schedules before and after variations and based on the rate of makespan deterioration. A new measure of the similarity of schedules is proposed. Guidelines for selecting priority rules, such as avoiding rules that directly use the uncertain information in calculating the priorities, are provided.

This paper describes the extending model of Multi-mode Failure Models by using the Weibull and Gamma distribution models presented in a conference [1,2]. Different than the models in the previous papers which are for variable test data, in this paper we will describe the use of attribute test data for our model. In reliability theory, the most important problem is to determine the reliability of a complex system from the reliability of its components. The weakness of most reliability theories is that the systems are described and explained as simply functioning or failed. In many real situations, the failures may be from many causes depending upon the age and the environment of the system and its components. Another problem in reliability theory is one of estimating the parameters of the assumed failure models. The estimation may be based on data collected over censored or uncensored life tests. In many reliability problems, the failure data are simply quantitatively inadequate. The Bayesian analyses are more beneficial than classical analyses in such cases. The Bayesian estimation analyses allow us to combine past knowledge or experience in the form of an apriori distribution with life test data to make inferences of the parameter of interest. In this paper, we have investigated the application of the Bayesian estimation analyses to multi-mode failure systems for attribute test data. The cases are limited to the models with independent causes of failure. We select our investigation by using the Multi-nomial distribution as our model. This distribution is widely used in reliability analysis for attribute test data. This model describes the likelihood function and follows with the description of the posterior function. A Beta prior is used in our analysis for each model and it is followed by the estimation of the point, interval, and reliability estimations.

This paper introduced the ICA theory in the process of analyzing the entrepreneur mental factors which affect the value chain, and analyses the feasibility of applying ICA theory in the extraction of the entrepreneur mental factors. In the empirical study, we separated the main factors influencing value chain by the FastICA algorithm based on ICA theory, and the experimental results are in conformity with theory.It is feasible to sift Entrepreneur mental factors,the research results of this paper also provides a theoretical reference for the application of the ICA theory in value chain management and optimization.

This paper presents an improved GSA-GA algorithm to achieve bad-data detection, identification and correction in power grid. The algorithm combines BP neural network, K-means clustering algorithm, gap statistical algorithm (GSA) and genetic algorithm together. BP neural network preprocesses the data, K-means algorithm clusters the preprocessed data and GSA algorithm determines the optimal clustering number and identifies the presence of bad data. After identifying the bad data, GA-BP algorithm is used to correct the identified data. This paper takes simulation tests to verify the proposed algorithms correctness and effectiveness based on actual grid data considering multiple types of existed bad data.

due to the time sequence has some unique characteristics, problems with the traditional point to point distance method are realized by more and more researchers when measuring the similarity of time series. Therefore, this paper uses a new similarity measure, i.e. global characteristics, extracting values of statistical distribution, seasonal, spectral and other characteristics from time series to construct eigenvector, based on the characteristic vector clustering analysis. By conducting time series clustering of communication base stations in a region, this paper confirms the method is able to deal with global characteristics that has a larger scale of time series, and to calculate effectively large time series.

This paper proposes a dimensionality reduction mathematical model based on feedback constraint for High-dimensional information. It uses feedback restriction technique to construct dimensionality reduction model for multidimensional product data. The data obtained is with high latitudes, where a large number of data are under components involved standardized restrictions. High-dimensional data participating in operation will increase the complexity of operation, and hence, we need to reduce its dimension. In this paper multi-constrained inverse regression model is adopted to reduce the dimension of cloud resource scheduling data in multi-constrained environments. Experimental results show that the proposed method increases the data coverage rate of high-dimensional data mining by 66%, and has great optimizing effect.

In order to effectively increase the incursion features detection accuracy with high-degree camouflage in the network, this paper proposes a high-degree camouflaged incursion features detection algorithm based on BP neural network. This paper utilizes distributed components to collect and analyze the incursion data features with high-degree camouflage, and designs an incursion features detection proxy module based on BP neural network. This paper takes the protocol platform developed by some company in Zhengzhou city as an example to detect the incursion of malicious information and presents the detailed diagnose methods and procedures in which the accuracy can reach up to 93.7%.

This paper proposes an approach to annotate the CAD models based on ontology with the aim of making the design intent understood by computer and applied in follow product development process, such as FEA. Semantic markup can embed the engineering semantic information such as product function, and design principle into the CAD geometry data through annotating, it makes the analyzers reuse design ideas quickly and conveniently to increase efficiency. The paper presents the design domain ontology and FEA domain ontology, and applies feature technologies and the semantic Web to complete annotation. Finally, a case study is carried out.

This paper analyzes how the AS's geographic locations and transmission distances affect on Internet's stability. In contrast to the traditional studies focusing on the future AS network which based on nowadays, this paper emphasizes particularly on the change of AS network's performance under the effect of the economic adjusting or the technology innovation to help make adjusting measures and to lead the technological trend. A double-deck network model of geographic hypergraph and AS network is presented to simulate the geographic locations, which is not reflected in traditional models. The simulate results of our model show that Internets stability ascends with the increase of transmission distances. However, if AS's geographic locations are random, the transmission distances has no obvious effects on Internets stability.

This paper focuses on the issues of real time ETL based on the semi-structured textural data sources. Because the mapping between text data and relational database relies on object-orient programming language, the paper first implements a text-object-relation mapping model for textual data sources based real-time ETL,and then studies the characteristic of real-time and security of purposed model. The cache mechanism aims to solve the performance difference between object-text mapping and object-relation mapping,and furthermore both encryption and signature mechanism aim to guarantee the security of the data in cache. The theoretical analysis and experiments prove the high efficiency and practicality of the mapping model.

Traffic classification is a critical technology in the areas of network management and security monitoring. Traditional port-based and payload-based classification are no longer effective due to the fact that many applications utilize unpredictable port numbers and packet encryption. Researchers tend to apply machine learning (ML) techniques to identify the traffic flows by recognizing statistical features. Unfortunately, looking back upon the related work, most of the ML-based classification algorithms have similar performance, and what really matters now is how to optimize these techniques. In this paper, we analyzed two critical issues (Feature Selection, Configuration of Parameters) of ML classification, and presented the corresponding viable methods to optimize the classification model. This paper also reported the experimental evaluation to assess the performance improvements introduced by our optimized methods; experimental results on real-life datasets and network traffic show that the classification model successfully achieves significant accuracy improvement.

Bayesian network method for system reliability evaluation which is based on a Bayesian network that transformed from a fault tree has gotten much attention these years. After a brief introduction to the method how to transform a fault tree into a Bayesian network, the paper elaborates the Bayesian network inference algorithms. The paper focuses on the way how the inference algorithms can be applied to the practice of system reliability evaluation and designs a systematic flow chart used to evaluate system reliability in a Bayesian network way. The experiment demonstrates the feasibility of the systematic flow chart.

Based on a common Bayesian network that models a system which has a cascade fault, this paper designs a new Bayesian network model and compares the two models according to experiment data. The experimental result shows that the new one is more consistent with the practical situation. The paper also gives a method for constructing a Bayesian network that models a system which has a cascade fault, a method efficient in the field of detecting cascade fault.

In this paper a novel Support vector clustering (SVC) method for outlier detection is proposed. Outlier detection algorithms have application in several tasks such as data mining, data preprocessing, data filter-cleaner, time series analysis and so on. Traditionally outlier detection methods are mostly based on modeling data based on its statistical properties and these approaches are only preferred when large scale set is available. To solve this problem, in this paper we focus on establishing the context of support vector clustering approach for outlier detection. Compared to traditional outlier detection methods , the performance of the SVC is not sensitive to the selection of needed parameters. The experiment results proved the efficiency of our method.

In this paper, in order to improve the precision of the short-term load forecasting, we propose a power load forecasting method combined principal component analysis (PCA) with least squares support vector machine (LS-SVM). Firstly PCA extracts the feature of the influence factors for power load, and then LS-SVM constructs a training model with a new variables extracted by PCA. After using PCA-LS-SVM model this paper proposed to forecast power load of one area, the results show that this method can effectively eliminate the redundant information among influential factors, reduce the input dimension of the prediction model, simplify the structure of the network, increase the learning speed and improve the power load forecasting accuracy. So this method is effectively feasible.

The research on new image segmentation algorithm is a very meaningful work in the processing of image. In the process, it will produce large amount of data redundancy. The efficient algorithm not only can greatly improve the quality of image treatment but also can greatly reduce the time and cost of the treatment. In this context, the paper analyzes several image processing algorithms commonly used in recent years and presents a new computer image processing algorithm--AMT-GA algorithm. In order to verify the effectiveness of AMT-GA algorithm, this paper takes the process of athletics image for example and compares the consistency and time of image segmentation with other literature results and ultimately finds that the consistency of AMT-GA algorithm reaches 0.99. The time in the algorithm execution is only 0.81 which not only achieves effective segmentation of the image but also saves the cost of computing. It also provides a theoretical reference for the research of computer graphics technology.

The Passive Location of Emitter based on Fuzzy Method.[Purpos The use of passive observation station to achieve the precise location has been an important topic of monitoring the electromagnetic spectrum resources and the environment. In order to locate the emitter accurately, a new method called fuzzy method is presented in this papers. [MethoBased on the distribution of points of intersections that are figured out by the LOB(Lines of Bearing), the paper put forward a fuzzy method for the passive location of emitter based on membership function to achieve the location of emitter. We compared this method with Pages method using 100 sets experimental data through the Monte Carlo simulated experiment. [Result The results showed that this method can effectively achieve the location of emitter, and improve the location accuracy. [Conclusion The positioning precision of this algorithm is obviously superior to Pages algorithm.

This paper uses Multi-index to assess regional power quality comprehensively. This paper introduces genetic projection pursuit method on power quality comprehensive evaluation. Without human intervention, this method overcomes the traditional power quality comprehensive evaluation method is influenced by subjective factors. Based on this, we put forward improved genetic projection pursuit algorithm. In the process of projection pursuit model and in the field data, changing the maximum normalized to normalization of national standard limit. According to the field experiment data in QingHai province, we assess the results by analysis and comparison with the original method. The improved method is more accurate on power quality comprehensive assessment.

In this paper, the harm of mycotoxin for grain was analyzed. Then the summarization and comparison of technics and methods for myco-toxin monitoring were given. Meanwhile, this paper discussed the necessity of rapid detection for mycotoxin exists in the grain, which provides the basis to the construction of rapid, accurate and simple detection.

This paper investigates an effective strategy for dynamically adjusting the threshold of Internet information rating. The threshold of Internet information rating is different from generation to generation, people in different period have different standards to the things around. How to adjust the threshold is the focus of this paper. In this paper, we propose an clustering based algorithm to divide the age span and use the cluster centers to represent the standards of this age stage, some improved strategy are also proposed.

Collaborative filtering process is based on the known users of the evaluation of target users to predict the target of interest, and then to the target users to recommend new items. This paper applies the fuzzy clustering technology used in the project of nearest neighbors and the users nearest neighbor search, reduces the project space and user space calculation dimension, to improve the traditional collaborative filtering algorithm scalability. The paper puts forward the novel model of collaborative filtering recommendation based on fuzzy clustering analysis. Compared with the traditional method of item similarity calculation is more accurate, the experiments show that the method improves the accuracy of recommendation.

How to select a target network meeting the needs of Quality of Service (QoS) guarantee in heterogeneous network environment has become a research hotspot. In this paper, a novel differentiate weight algorithm on heterogeneous network selection is proposed. As regard of the different needs of QoS for each class of traffic in heterogeneous network, the QoS parameter weight for each class of traffic is calculated by G1 method, and then the target network is selected by technique for order preference by similarity to ideal solution (TOPSIS). At last, simulation results show that the algorithm proposed in this paper is better than average weight algorithm on selecting target network reasonably and balancing network loads.

This paper first discusses the limitations and shortcomings of data mining under the conditions of massive data. Combined with the advantages of cloud computing, a data mining architecture for cloud computing is designed, and on this basis, the paper discusses the improvement of data mining algorithms for cloud computing. As a theoretical exploration, the paper proposed useful suggestions for the optimization of data mining in face of cloud computing.

Data aggregation is the important method to reduce data traffic and lower energy expenditure in wireless sensor networks (WSN). This paper analyzes the characteristics of data sampled by nodes, and gives the method to decide spatial correlation between neighboring nodes and the criteria to classify and decide data in WSN. On the basis of this, this paper proposes a spatial correlation based data aggregation algorithm for WSN, SCBD. SCBD classifies and decides data according to data criteria and spatial correlation among nodes in normal nodes and cluster heads at the same time, and then aggregates different types of data. The results show that SCBD outperforms RAA in terms of energy consumption, rate of data detection and quality of data aggregation.

The realization of a device test training system requires the use of a lot of domain knowledge, and building knowledge base will play an important role. In view of the uncertainty, inaccuracy and incompleteness of test data in the testing process, this paper makes the data mining algorithms based on rough set as knowledge acquisition algorithm, and proposes an improved algorithm for insufficient of approximate reduction of rough set knowledge based on the tolerance relation of incomplete information system. The paper studies the design and realization of knowledge base system in the developing of device simulation training system on this basis, and validates the method through a design example of knowledge base of a certain device simulation training system.

In object-oriented programming, designing class is one of the most important steps of software development. How to analyze and evaluate the class design is an important topic in the field of software engineering. The paper puts forward a method which is based on information entropy to quantify the coupling of class design. Compared to the conventional methods, this method has a better operability and accuracy. On the above basis, this paper validates the feasibility of the method by the case studies.

Open source software systems are becoming increasingly important these days and will grow stronger in the future. In order to gain more information about them, their structure characteristics and behavior characteristics need to be measured. This paper tests scale free characteristics on open source real-world software empirically. More specially, this paper checks whether Chidamber and Kemerer metrics suite follows power law or not through three versions of Weka. Our empirical results showed that the features of complex networks didn’t exist in Weka.

3D data model is an indispensable component to any 3D GIS, and forms the basis of 3D spatial analysis and representation. At present, plenty of representative 3D data models are proposed. However, existing models neglect the display result and the consumption of storage space. Based on the analysis of existing 3D GIS data model, a 3D surface model is proposed for fast visualization in this paper, which is composed of node, segment and triangle. The data structure and formal representation of the proposed 3D surface model is developed to organize and store data of 3D model. Finally, an experiment is made to compare this 3D surface model with other 3D data model, and the result demonstrates that the 3D surface model proposed in this paper is superior to the existing data model in terms of data volume, moreover, it can acquire fast visualization speed.

With high accuracy, the channel simulator plays an important role in the docking experiment between the ground and the responder beacon. To begin with, this paper introduces the data generation algorithm including the data generation based on simulation technology, the principle of the linear least squares algorithm and then proposes the least squares quadratic spline method to generate highly accurate data in this channel simulator. Secondly, this paper introduces the system design to realize the data generation. Finally, a case which studies the approximation and an error analysis of the data generation algorithm is realized. The algorithm is considered to be accurate and easy to get the source data. The core of the algorithm is using data from Satellite Tool Kit to generate distance and speed sequence, and using the least squares to approximate real data and quadratic spline to fit for obtaining highly accurate data.

Improvement on defining hidden layer’s clustering centers of RBF network is greatly important for the network modeling and prediction, especially for multivariate time series. This paper firstly uses a linear function and a nonlinear function respectively to detect the linear correlations and the nonlinear correlations of the time series. And a small data set which includes effective information of the system is defined. Then, a local search procedure is introduced to optimize K-means clustering algorithm, which is aimed at quickly and effectively adjusting hidden layer’s clustering centers. Finally, input other training samples to determine network weights, LS or its improved method can be selected. Simulation results show that compared with that of conventional RBF network, with the same number of hidden layer’s nodes, the RBF network model in this paper gets better prediction accuracy.