Search

The requirement of the credibility is analyzed for smart grid data and the traditional theories of data measurement and bad data identification are studied in power system. On the basis of that, this paper mainly studies the credibility between data sources, the construction of the trusted network generation algorithm, and the hierarchical and dynamic trusted network model innovatively which is based on the relationship between data sources. The credibility evaluation of data sources, the credibility evaluation between data sources and the credibility evaluation of data are studied. The credibility between data sources is restricted by the credibility of data sources, the credibility of data sources is restricted both from the credibility of data and the credibility between data sources, the credibility of data is restricted by the credibility between data sources and the credibility of data sources, they are interrelated and restricted by each other and constitute to a whole. The simulation results show that the model can satisfy the requirement of the credibility of smart grid data better, and provide the ideas to solve the problem of the credibility measure for prospective research is feasible.

To evaluate the risk system of power communication transmission network (PCTN) scientifically, this paper puts forward an improved k-means clustering algorithm in allusion to the issue of the power device risk value in the PCNT. Both the rating of the device status and the importance of the device determine the power device risk value, among which the importance of device is decided by the number and type of tasks it bears. Since generated randomly, traditional initial clustering centers is easy to lead the result to a partial best instead of global. Therefore, this text uses an improved Huffman tree method based on adjusted cosine similarity to make sure the initial centre. It is verified that the improved k-means clustering algorithm can do better in clustering data, dividing the risk rank scientifically, which provides a reference frame for risk evaluation.

Data preprocessing plays an essential role in the process of safety equipment linkage system, directly influenced the quality and the results of the later analysis and decision-making. This paper describes the process of data preprocessing in detail, and according to the characteristics of the linkage system data, proposes an improved data cleaning method in order to improve the efficiency of data preprocessing. For each log data format is different, using XML language to describe of the log data, which provides a unified interface for safety equipment linkage system data access, and then solve the problem of heterogeneous data.

Association rules are an important branch of data mining. Aiming at the update maintenance problem of association rules when increase the database data while keep the minimum support unchanged, on the basis of FUP algorithm, this paper proposes a matrix-based incremental association rules mining algorithm, namely MBFUP algorithm. The algorithm just scans the database once, and doesn't produce candidate item sets, greatly reduces the space complexity and time complexity of the algorithm.

The overall objective of the network security situation assessment is to establish security situation assessment system. This system can unify the large-scale network. First, it provides a security policy unified the whole network. It also provides the techniques for wide-area situation assessment. At the same time this system provides situational awareness and decision support tools for the implementation of network security command. Firstly, the information model of security situation was studied based on XML. Then the security situation value of system including three levels: service, host and network was analyzed and calculated. The-support vector regression machine was chosen for predicting the network security situation and determined the parameter values. Finally, the effectiveness of the method was verified through experimental analysis.

In order to effectively finish the dynamic risk assessment of the electricity system, this paper will divide each attack into three distinct phases, The difficulty of attack is assessed by percent of each attack time of distinct stages take up in the total attack time to describe the attack difficulty in order to determine the status of the assets transition matrix, realising the dynamic nature of risk assessment. The real-time dynamic risk assessment methods based on Hidden Markov Model HMM has a strong adaptability and scalability, it can be effectively applied on the network, host, system, service level of risk assessment. This paper designs and implements the dynamic risk assessment examples power system, and then demonstrateds the dynamic assessment model.

Hybrid intrusion detection is a novel kind of model combining the advantages of anomaly detection and misuse detection. We design a new hybrid intrusion system based on Principle Component Analysis and Self Organizing Maps, aiming at establishing an extensible real-time intrusion model with high detection accuracy. Experimental results on the KDD 1999 Cup dataset show that the proposed model is promising in terms of detection accuracy and computational efficiency, thus amenable for real-time intrusion detection.

In order to improve the accuracy of risk assessment of corporate networks, the paper introduces the concept of network shell into the field of security risk assessment. After analyzing the actual needs of corporate networks, we give the risk assessment model of network shell, then based on this model we analyze the application of network shell and give the algorithm to repair the network shell through the attack graph. After that, a calculation method of the integrated value of network shell is given. Finally, an example is given to verify the validity of assessment model.

With the development of information technology and the popularization of the Internet, the amount of information based on the network has raised rapidly, and the emergence and development of network information filtering provide a better choice for people to access to the information quickly, accurately and completely. In this paper, the introduction of PSO uses its simple and efficient to establish accurate user model. It takes the access and update of profile as a process of learning, adjusted, learning again. Model is updated under considering inertia weight, parameters and other factors. Then doing some experiments test it.

In view of the characteristics and the special demands of the electric power information system, this paper designs an assessment model based on variable precision rough set. Then, it uses group decision-making theory, which could make the results of the assessment more accurate and more objective, to calculate the risk value of the target system risk. And then VPRS and AHP are used to determine the index weight involved in the process of the electric power information risk assessment. Finally, it verifies the correctness of the algorithm by analyzing results of the examples of the electric power information system security assessment. We can see that the model could not only find the weakness of the system by analyzing the results got by calculating the current risk value of the electric power system, but also reduce the influence of the subjective factors in some degree.

For evaluation of the risk in information system, variable precision rough set (VPRS) theory is emerging as a powerful tool for reasoning about data. The method for risk rules mining in information system based on VPRS is designed on the basis of related concepts of VPRS. Through the example of the assessment, it is considered that the method can not only reduce greatly the data, but also increase the validity of risk regulations.

In accordance with the characteristics and the special demands of electric power information system, this paper designs a support vector machines (SVM) risk assessment method based on group decision-marking. According to security technology indices of electric power information system, the method chooses the mode of expert scoring, based on the group decision-marking, to calculate integrated value of each index, which is as a training sample used to train SVM, and it forecasts risk level for the system. Finally, it verifies the correctness of the method by analyzing results of the examples of the electric power information system security assessment. The experiment shows that the method can not only forecast the current risk level of the electric power system with a high accuracy rate, but also reduce the influence of the subjective factors in some degree.

This paper use the source of web log data and through AprioriAll algorithm among sequential pattern mining to identify the user pattern in the data pre-processing stage and based on pattern to cluster users’ path pattern and make the similarity users cluster one group in order to reduce pattern. Through testing, the method has been proven to be effective for users in personalized service.