The traditional methods for prediction of oil reservoir utilize the seismic attributes generated when seismic waves passing through the stratum and geologic drilling data to make a comprehensive judgment in combination with the traditional geophysical methods. However, this type of prediction methods has high cost of research and judgement and its accuracy strongly depends on the prior knowledge of the experts. To address the above issues, based on the seismic data of the Subei Basin of Jiangsu Oilfield, and considering the sparseness and randomness of oil-labeling samples, a multi-granularity temporal structure representation based outlier detection algorithm was proposed to perform the prediction by using the post-stack seismic trace data. Firstly, the multi-granularity temporal structures for the single seismic trace data was extracted, and the independent feature representations were formed. Secondly, based on extracting multiple granularity temporal structure representations, feature fusion was carried out to form the fusion representation of seismic trace data. Finally, a cost-sensitive method was utilized for the joint training and judgement to the fused features, so as to obtain the results of oil reservoir prediction for these seismic data. Experiments and simulations of the proposed algorithm were performed on an actual seismic data of Jiangsu Oilfield. Experimental results show that the proposed algorithm is improved by 10% on Area Under Curve (AUC) compared to both of the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) algorithms.

At present, super-resolution reconstruction of a single image achieves a good effect, but most models achieve the good effect by increasing the number of network layers rather than exploring the correlation between channels. In order to solve this problem, an image super-resolution reconstruction method based on Channel Attention mechanism (CA) and Depthwise Separable Convolution (DSC) was proposed. The multi-path global and local residual learning were adopted by the entire model. Firstly, the shallow feature extraction block was used to extract the features of the input image. Then, the channel attention mechanism was introduced in the deep feature extraction block, and the correlation of the channels was increased by adjusting the weights of the feature graphs of different channels to extract the high-frequency feature information. Finally, a high-resolution image was reconstructed. In order to reduce the huge parameter influence brought by the attention mechanism, the depthwise separable convolution technology was used in the local residual block to greatly reduce the training parameters. Meanwhile, the Adaptive moment estimation (Adam) optimizer was used to accelerate the convergence of the model, so as to improve the algorithm performance. The image reconstruction by the proposed method was carried out on Set5 and Set14 datasets. Experimental results show that the images reconstructed by the proposed method have higher Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity index (SSIM), and the parameters of the proposed model are reduced to 1/26 of that of the depth Residual Channel Attention Network (RCAN) model.

The ship scheduling affected by severe weather is a very complex optimization problem, and is also one of the key issues needed to pay attention in liner companies. Therefore, based on the premise of obtaining the latest weather forecasting information in the designed multi-stage rescheduling mechanism period and the real-time positions of all the ships in service in one liner company on a shipping network, the restriction of liner shipping schedule was focused on, and the realistic constraints such as the change of ship's speed between different ports and the ship capacity, a nonlinear mathematical model was built to minimize the total shipping cost of all the ships during the fixed planning period. And an improved genetic algorithm embedded with gene repair operator was designed to solve the built model. Then the optimal multi-stage rescheduling scheme, which was integrated by the solution strategies of charting for direct-transport, dispatching ship across different routes, adjusting port reaching order and goods transfer, was given. Experimental results of examples with large, medium, and small scales show that, compared with the traditional waiting method, multi-stage rescheduling saves more than 15% of the total shipping cost, verifying the effectiveness of the proposed model and scheme; compared with Cplex, the improved genetic algorithm has the calculation efficiency greatly improved and all the deviation values within 5%; and compared with Ant Colony Optimization (ACO) algorithm, Tabu Search (TS) algorithm, Quantum Differential Evolution (QDE) algorithm, the improved genetic algorithm has the cost reduced by about 10% in the effective time, proving that the algorithm is scientific. It can be seen that the proposed method can provide the reference for actual ship scheduling of liner companies.

Recently, deep learning has been a hot research topic in the field of image super-resolution due to the excellent performance of deep convolutional neural networks. Many large-scale models with very deep structures have been proposed. However, in practical applications, the hardware of ordinary personal computers or smart terminals are obviously not suitable for large-scale deep neural network models. A light-weight Network with Automatic Residual Scaling (ARSN) for single image super-resolution was proposed, which has fewer layers and parameters compared with many other deep learning based methods. In addition, the specified residual blocks and skip connections in this network were utilized for residual scaling, global and local residual learning. The results on test datasets show that this model achieves state-of-the-art performance on both reconstruction quality and running speed. The proposed network achieves good results in terms of performance, speed and hardware consumption， and has high practical value.

A Minimum Mean Square Error (MMSE) speech enhancement algorithm based on Laplacian distribution was proposed to solve the problem of noise residual and speech distortion of speech enhanced by the spectral subtraction algorithm based on Gaussian distribution. Firstly, the original noisy speech signal was framed and windowed, and the Fourier transform was performed on the signal of each processed frame to obtain the Discrete-time Fourier Transform (DFT) coefficient of short-term speech. Secondly, the noisy frame detection was performed to update the noise estimation by calculating the logarithmic spectrum energy and spectral flatness of each frame. Thirdly, based on the assumption of Laplace distribution of speech DFT coefficient, the optimal spectral subtraction coefficient was derived under the MMSE criterion, and the spectral subtraction with the obtained coefficient was performed to obtain the enhanced signal spectrum. Finally, the enhanced signal spectrum was subjected to inverse Fourier transform and framing to obtain the enhanced speech. The experimental results show that the Signal-to-Noise Ratio (SNR) of the speech enhanced by the proposed algorithm is increased by 4.3 dB on average, and has 2 dB improvement compared with that of the speech enhanced by the over-subtraction method. In the term of Perceptual Evaluation of Speech Quality (PESQ) score, compared with that of over-subtraction method, the average score of the proposed algorithm has a 10% improvement. The proposed algorithm has better noise suppression and less speech distortion, and has a significant improvement in SNR and PESQ evaluation standards.

To solve the problems that Particle Swarm Optimization (PSO) algorithm is easy to fall into the local extremum due to the lack of interaction between individuals in the search process, the diversity of the population is gradually lost, a Perturbation Particle Swarm Optimization algorithm based on Local Far-neighbor Differential Enhancement (LFDE-PPSO) was proposed. Firstly, in order to enlarge the population search space, the disturbance factor was introduced to make inertia weight and learning factor fluctuate within a small range. Secondly, the reconstruction probability was introduced, and the population with low fitness value was selected to reconstruct intermediate population. Finally, in order to increase the population diversity, the excellent individuals of poor individuals were retained, the irrelevant and far-neighbor individuals were introduced. The far-neighbors with large differences from differential individual genes were used for differential enhancement. The experimental results show that the proposed algorithm can preserve individuals with high fitness in the intermediate population, effectively increase the population diversity, make the population have strong ability to jump out of local extremum, speed up the particle approximation to the global aptimum, and have the advantages of fast convergence and high precision.

Concerning data sparseness, low accuracy and poor real-time performance of traditional collaborative filtering recommendation algorithm in e-commerce system under the background of big data, a new collaborative filtering recommendation algorithm based on improved clustering and matrix decomposition was proposed. Firstly, the dimensionality reduction and data filling of the original data were reliazed by matrix decomposition. Then the time decay function was introduced to deal with user score. The attribute vector of a project was used to characterize the project and the interest vector of user was used to characterize the user, then the projects and users were clustered by k-means clustering algorithm. By using the improved similarity measure method, the nearest neighbors and the project recommendation candidate set in the cluster were searched, thus the recommendation was made. Experimental results show that the proposed algorithm can not only solve the problem of sparse data and cold start caused by new projects, but also can reflect the change of user's interest in multi-dimension, and the accuracy of recommendation algorithm is obviously improved.

Focusing on the issue that the ensemble classifier based on weak classifiers needs to sacrifice a lot of training time to obtain high precision, an ensemble method of strong classifiers based on instances named Fast Strong-classifiers Ensemble (FSE) was proposed. Firstly, the evaluation method was used to eliminate substandard classifier and order the restclassifiers by the accuracy and diversity to obtain a set of classifiers with highest precision and maximal difference. Secondly, the FSE algorithm was used to break the existing sample distribution, to re-sample and make the classifier pay more attention to learn the difficult samples. Finally, the ensemble classifier was completed by determining the weight of each classifier simultaneously. The experiments were conducted on UCI dataset and customized dataset. The accuracy of the Boosting reached 90.2% and 90.4% on both datasets respectively, and the accuracy of the FSE reached 95.6% and 93.9%. The training time of ensemble classifier with FSE was shortened by 75% and 80% compared to the ensemble classifier with Boosting when they reached the same accuracy. The theoretical analysis and simulation results show that FSE ensemble model can effectively improve the recognition accuracy and shorten training time.

Focusing on the problems in bus priority systems, like hysteresis, being not able to fully exploit road capacity, an Adaptive Bus Priority (ABP) model based on the Internet of Vehicles (IOV) was proposed. First, with powerful communication capability of IOV, using time division multiplexing idea, road multiplexed control rules for ordinary bus was designed, and the "space priority" was achieved by setting a virtual bus lane. Secondly, real-time data acquisition of arrival vehicle was used to replace the historical data to solve the problem of hysteresis. Finally, the bus priority signal control model was designed, and bus priority was realized by inserting short phases to make public transit priority. VISSIM software was used to design simulation experiment to compare the ABP model and the traditional model. Simulation results indicate that the ABP model can improve the operation efficiency of bus and intersection pass capacity without causing great impact to social vehicles.

In the cloud computing, the statistical multiplexing is a remarkable character, and the utilization efficiency of physical resource can be improved through the virtual technology. Aiming at the load balancing problem of the resource utilization needed to be discussed in the cloud virtual machine cluster, a Software Defined Network (SDN) based Anti-Saturated Grouping Strategy (ASGS) method was proposed according to the OpenStack cloud platform. The cloud hosts were separated into different groups based on their weights, and then the load information of cloud hosts were obtained by the SDN controller which used the probe with different groups periodically. When a request came, a group was selected randomly using the average weight of each group cloud hosts by the load balancer, and a proper backend was chosen by the polling method within the group. In order to avoid the cloud host downtime caused by the sudden increased requests for too many resources of a backend, the cloud host with higher weight was given a default parameter to increase the weight, and then the host would receive fewer requests on the higher load status. The experimental results show that, whatever the request number changes, the resource utilization standard variance of the proposed ASGS is always smaller than the random and round robin methods when time varies, which is nearly 0. The proposed ASGS has better load balance for the cloud host cluster.

The traditional nonweighted least squares Support Vector Machine (SVM) and weighted least square SVM have a few disadvantages of processing low Signal-to-Noise Ratio (SNR) laser echo in the field of lidar bathymetry, a filtering method named HW-LS-SVM was proposed by combining robust least square and weighted least square SVM. Firstly, strong prior weight function, residual error and mean square error were calculated by elimination weight function, then the weight of least square SVM was computed by weight function. Finally, the echo signal was filtered by iterative computation. The simulation results show that HW-LS-SVM algorithm is more robust than least square SVM, Bayes least square SVM and the traditional weighted least square SVM. The results were satisfactory when the noise rate reached to 45%, and the correct rate of the extracted water surface and bottom was 100%. The extracted water depths from 4 groups of laser echoes in deep area and 4 groups in shallow area all agree with the background data. The proposed method has better anti-noise performance and is more suitable for the filtering processing of the low SNR lidar bathymetry signal.

Focusing on the low accuracy and long detection time of arc detection in engineering drawing vectorization, a fast arc detection algorithm based on tangent lines matching was proposed. Firstly, tangent lines on the circle outer boundary were detected from eight directions (0, π/4, π/2, …, 7π/4) and were added in tangent lines set. Secondly, the tangent lines in the set were paired up, and the center and radius of circles were estimated to obtain circle candidate set. Finally, tracing detection was performed for every candidate circle after merging data of circle candidate set, and every candidate circle was ascertained as a circle or an arc. The paring process was executed during the tangent lines searching, and the number of pairing was effectively reduced by removing the relative tangent lines of the identified candidate circle. In the contrast experiments with RANdom SAmple Consensus (RANSAC) algorithm and Effective Voting Method (EVM), the proposed method reached average detection accuracy of 97.250%, and the average detection time was 12.290 s, which were better than those of the comparison methods. The experimental results illustrate that the proposed method can effectively detect the arc which length is greater than 1/8 circumference in low noise image, improve the accuracy of detection and shorten the detection time.

Source code comments is an important part of the software, so researchers need to use manual or automated methods to generate comments. In the past, the quality assessment of this kind of comments is done manually, which is inefficient and not objective. In order to solve this problem, an assessment criterion was built in which four aspects of the comments including comment format, language form, content and code-related degree were considered. Then a code comments quality assessment method based on an aggregation of classification algorithms was proposed, in which machine learning and natural language processing technology were introduced into comments quality assessment, by using classification algorithms the comments were classified into four levels, including unqualified, qualified, good and excellent ones. The evaluation results were improved by the aggregation of the basic classification algorithms. The precision and F1 measure of the aggregated classification algorithm were improved about 20 percentage points compared with using a single classification algorithm, and all the indexes have reached more than 70% except the macro average F1 measure. The experimental results show that this method can be applied to assess the quality of comments effectively.

To solve the low accuracy and high complexity in feature extraction of stereoscopic vision depth map, a feature extraction algorithm based on Principal Component Analysis and Histogram of Oriented Depth Gradient (PCA-HODG) was proposed. Firstly, disparity computation and depth map extraction were executed for binocular stereoscopic vision image to obtain high quality depth map; secondly, edge detection and gradient calculation of depth map within fixed size window were performed, then the features of region shape histograms were acquired and quantified. Meanwhile, dimensionality reduction by Principal Component Analysis (PCA) was implemented; finally, to realize the accuracy and completeness of feature extraction from depth map, a detection method of sliding window was used for the feature extraction of whole depth map and the dimensionality reduction was implemented once again. In the experiment of feature matching and classification, for the frames of test sequence Street, the average classification accuracy rate of the proposed algorithm increased by 1.15% when compared with the Range-Sample Depth Feature (RSDF) algorithm; while for Tanks, Tunnel, Temple, the average classification accuracy rate increased by 0.69%, 1.95%, 0.49% respectively when compared with the Geodesic Invariant Feature (GIF) algorithm. At the same time, the average running time decreased by 71.65%, 78.05%, 80.06% respectively compared with the Histogram of Oriented Depth (HOD), RSDF, GIF algorithm. The experimental results show that the proposed algorithm can not only detect and extract features of depth map more accurately, but also reduce the running time greatly by dimensionality reduction. Moreover, the proposed algorithm also has better robustness.

For balancing transparency and robustness of watermark, this paper proposed wavelet-domain digital watermarking method based on Fruit Fly Optimization Algorithm (FOA). The algorithm used Discrete Wavelet Transform (DWT) by FOA to watermarking technology and solved the contradiction between transparency and robustness in the watermark by swarm intelligence algorithm. In order to protect the copyright information of digital image, the selected original image was decomposed through a two-dimensional discrete wavelet transform, and watermark image through Arnold transformation was better embedded into wavelet coefficients of vertical sub-band, which guaranteed image quality. In the optimization process, the scaling factor was continuously being trained and updated by FOA. In addition, a new algorithm framework was proposed, which evaluated the scaling factor by prediction feasibility of DWT domain. The experimental results show that, the proposed algorithm has higher transparency and robustness against attacks, with watermarking similarity above 0.95, and 10% higher under geometric attacks such as rotation and shearing compared to some existing watermarking methods based on swarm intelligence.

For the problem that task scheduling program in cloud computing environments usually takes high response time and communication costs, a Honey Bee Behavior inspired Load Balancing (HBB-LB) algorithm was proposed. Firstly, the load was balanced across Virtual Machines (VMs) for maximizing the throughput. Then the priorities of tasks on the machines were balanced. Finally, HBB-LB algorithm was used to improve the overall throughput of processing, and priority based balancing focused on reducing the wait time of tasks on a queue of the VM. The experiments were carried out in cloud computing environments simulated by CloudSim. The experiment results showed that HBB-LB algorithm respectively reduced average response time by 5%, 13%, 17%, 67% and 37% compared with Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Dynamic Load Balancing (DLB), First In First Out (FIFO) and Weighted Round Robin (WRR) algorithms, and reduced maximum completion time by 20%, 23%, 18%, 55% and 46%. The result indicates that HBB-LB algorithm is suitable for cloud computing system and helpful to balancing non-preemptive independent tasks.

The traditional sentiment analysis is coarse-grained and ignores the comment targets, the existing fine-grained sentiment analysis ignores multi-target and multi-opinion sentences. In order to solve these problems, a method of fine-grained sentiment analysis based on Conditional Random Field (CRF) and syntax tree pruning was proposed. A parallel tri-training method based on MapReduce was used to label corpus autonomously. CRF model of integrating various features was used to extract positive/negative opinions and the target of opinions from comment sentences. To deal with the multi-target and multi-opinion sentences, syntax tree pruning was employed through building domain ontology and syntactic path library to eliminate the irrelevant target of opinions and extract the correct appraisal expressions. Finally, a visual product attribute report was generated. After syntax tree pruning, the accuracy of the proposed method on sentiment elements and appraisal expression can reach 89% approximately.The experimental results on two product domains of mobile phone and camera show that the proposed method outperforms the traditional methods on both sentiment analysis accuracy and training performance.

To deal with the output tracking problem of a hypersonic vehicle with parameters uncertainty, an adaptive controller which obtained H _∞ performance was proposed based on hierarchical fuzzy system. In order to solve the problem that the number of rules in a fuzzy controller increases exponentially with the number of variables involved, reduce the number of the parameters to be identified on-line and enhance the real-time performance of the control system, an adaptive controller was designed based on hierarchical fuzzy system. To weak the impact on the stability abused by approximation error of the fuzzy logic system, parameters uncertainty and the external disturbances, the robust compensation terms were introduced to improve the H _∞ performance of the system. The Lyapunov theory was applied to analyze and prove the stability of the system. The simulation results demonstrate that the system can not only track the input exactly, but also possess strong robustness.

To improve the perception ability of indoor mobile robots, the classification method for the commonly structured corridor-scenes, Spiking Neural Network (SNN) and NeuCube, which is a novel computing model based on SNN, were studied. SNN can convey spatio-temporal information by spikes. Besides, SNN is more suitable for analyzing dynamic and time-series data, and for recognizing data of various patterns than traditional Neural Network (NN). It is easy to be implemented by hardware. The principle, learning methods and calculation steps of NeuCube were discussed. Then seven common corridor scenes were recognized by the classification method based on multi-sonar-sensor information and NeuCube. The experimental results show that the proposed method is effective. Additionally, it is helpful for improving autonomy and intelligence of mobile robots.

During the post-earthquake transitional phase, there are relief goods recycling and environmental protection problems. In the premise of meeting the basic demand of people in disaster area, the Location-Routing Problem (LRP) model of emergency logistics facilities with forward and reverse directions was built. First, according to the characteristics that the recycled materials could be partially transported, a mathematical model was established in which the objective function was minimum time of emergency system. Second, a two-phase heuristic algorithm was used to solve the model. Finally, the example analyses verified the feasibility of the model and algorithm. The experimental results show that, compared with the traditional one-way LRP model, the objective function value of the proposed method decreases by 51%. The proposed model can effectively improve the efficiency of emergency logistics system operation and provide auxiliary decision support for emergency management department.

The properties of the measured objects in 3D profile using the grating projection are more and more complex, there are a large number of splits in the extracted refinement grating stripes, and the refinement stripe encoding is very difficult. An automatic coding algorithm based on color structure light was proposed. The paper designed a new model of color structure light, introduced its design principle and implemented a new automatic stripe coding algorithm. First, the algorithm extracted the refinement grating stripe with color information from the color structure grating. Then, orderly encoded the refined stripes of each color by judging the best connected domain. Finally, the article got the stripe coding of the total image through combined coding by using the periodicity of grating model. The simulation experiment results show that the model design of color structure light is simple, the automatic coding algorithm of stripe has high accuracy and the error is decreased to 10 percent. The ideal 3D points cloud data model can be reconstructed through the strip coded data.

Since the traditional centralized architecture Web service registry system suffers from such problems as performance bottleneck, single-point-of-failures, a structure Peer-to-Peer (P2P) based Web service registry system was designed and implemented. The registry system consists of six modules including configuration, schedule and distribution, peer-to-peer communication, rank validation, JUDDI, and network resources monitoring. The pastry based system scheduling and communication algorithms were proposed, and the corresponding Web service registration and discovery process was designed. The Web service registration system was tested and analyzed using SoapUI and LoadRunner. The experimental results show that the system can support large-scale accessing and has dynamic scalability. In the multi-concurrent simulation experiments, the response speeds of Web services registration and discovery are increased 1 times.

When Augmented Reality (AR) browser running in the Point of Interest (POI) dense region, there are some problems like data loading slowly, icon sheltered from the others, low positioning accuracy, etc. To solve above problems, this article proposed a new calculation method of the Global Positioning System (GPS) coordinate mapping which introduced the distance factor, improved the calculating way of coordinates based on the angle projection, and made the icon distinguished effectively after the phone posture changed. Secondly, in order to improve the user experience, a POI labels focus display method which is in more accord with human visual habits was proposed. At the same time, aiming at the low positioning accuracy problem of GPS, the distributed mass scene visual recognition technology was adopted to implement high-precision positioning of scenario.

The cluster number is not generally set by K-means clustering algorithm beforehand, and artificial initial clustering number easily leads to the problem of unstable clustering results. A high-efficient algorithm for determining the K-means optimal clustering number was presented. The algorithm got the upper bound of the number of clustering search range through stratified sample data and designed a new kind of effective clustering indicator to evaluate the clustering degree of similarity between and within class after clustering. Thus the optimal number of clusters was obtained in the search range of the clusters number. The simulation results show that the algorithm can obtain the optimal clustering number fast and accurately, and the dataset clustering effect is good.

Evidence theory is one of the main approaches to implement situation assessment based on rules. But evidence theory can result in paradox problem in conflicting evidence combination. Concerning this problem, by dissimilarity calculation, the importance of evidence was measured and original evidence was modified. A new approach based on the improved evidence theory was proposed. The new approach contained four steps including rules measurement, evidence modification, rules fusion and situation decision. The experiments show that the new approach can avoid paradox problem in the process of fusion based on evidence theory, and it is superior to typical approaches, such as Dempster approach, Yager approach and Leung approach, etc, in efficiency and accuracy of situation assesment.

Considering the complexity and inaccuracy of traditional theoretical modeling for rigid-flexible couple system, the frequency domain subspace method was used to identify the motor's model and piezoelectric ceramic piece's model in the experimental system. Due to the problem of chattering and long reaching time of traditional reaching law, a novel sliding mode control with power reaching law was proposed. Theoretical analysis shows that the reaching time can be shortened and the range of traditional power reaching law's parameter α can be expanded, which will not affect the chattering. Considering the effect of vibration characteristics of flexible beam on system performance, the method of sub-sliding surface was used to design the sliding mode controller. Lastly, experimental results show that the designed controller can track the angle of the center of the rigid body rapidly and suppress the vibration of the flexible beam quickly.

Bad Mouthing and Self-Promoting (BS) collusion attack group and its detection mechanism, called BSCD, were proposed to resolve the security issues of the multiple malicious node collusion attack network nodes and affect their accurate positioning in the Wireless Sensor Network (WSN) reputation system. And the implementation method of the mechanism was given. It detected the abnormal recommended node, analyzed the evaluation behavior similarity between recommended nodes, and effectively detected the existence of collusion attack group, thereby reduced its damage and impact on the reputation of the system. The simulation results show that, BSCD has significant effect on the detection and resisting BS collusion attack group, effectively improves the malicious node detection rate in the reputation system and the capacity of the entire system to resist malicious node.

A new low-power localization algorithm based on the evaluation of distributed reputation was proposed to improve the security and energy consumption of the node positioning for wireless sensor network. The concepts of Trustworthy Node Table (TNT) and the backup cluster head node were introduced to find the reliable beacon nodes quickly, and the backup cluster head node could assist and monitor the cluster head node, reducing the workload of the cluster head and participating in the integration process of the beacon nodes reputation values. The proposed algorithm enhanced the reliability and integrity of the beacon nodes, improved the efficiency and security of the node localization, reduced the systems energy consumption and improved the detection rate of malicious nodes. The simulation results show that in malicious node environment, the algorithm can effectively improve the detection rate of malicious nodes, reduce the positioning error, weaken the malicious nodes damage and influence on the positioning system to achieve the safe positioning of the nodes.

In order to solve low accuracy, large time consumption and limited application range in traditional network traffic classification, a semisupervised network traffic classification method of Support Vector Machine (SVM) was proposed. During the training of SVM, it determined the support vectors from the initial and new sample set by using incremental learning technology, avoided unnecessary repetition training, and improved the situation of original classifiers’ low accuracy and timeconsuming as a result of new samples that appeared. This paper also proposed an improved Tri-training method to train multiple classifiers, and a large number of unlabeled samples and a small amount of labeled samples were used to modify the classifiers, which reduced auxiliary classifier’s noise data and overcame the strict limitation of sample types and traditional Coverification for classification methods. The experimental results show that the proposed algorithm has excellent accuracy and speed in traffic classification.

To deal with the difficulties of lacking the discrimination method of network's traffic self-similarity and producing negative traffic based on classical Fractal Brown Motion (FBM), a discrimination method was proposed based on multiple order moment and a generation method was provided based on modified FBM model. Firstly, the mathematical formula of sample moment was studied. The discrimination method of self-similarity traffic was obtained on account of fractal moment analysis. Secondly, the classical Random Midpoint Displacement (RMD) algorithm was modified. At last, taking account of the real traffic of Bellcore and LBL, the discrimination method and generation method were given. The comparison of the simulation results with the actual experimental data proves that the method is feasible.