Research On Information Security Risk Assessment Based On AHP And BP

With the expansion of human needs demands and the rapid changes in scientific innovation, the state, society and humans are increasingly dependent on information systems. In the new situation, the information system security is an important issue that must be faced. The information security risk assessment is very important to ensure the normal operation of the information system, and the choice of scientific and reasonable risk assessment method is the most important task for effective and scientific information security risk assessment. In terms of the current risk assessment methods, based on the use of qualitative analysis, quantitative analysis of its integration is used again, the combination of the two methods are widely used. More and more people realize that the information security risk assessment is an indispensable step for protecting the information system security and establishing the protection architecture of information system.By comparing the risk assessment method of Analytic Hierarchy Process and BP neural network, the paper concludes that the selected method is insufficient in the actual information security risk assessment process. Aimed at this, the risk factor identification method based on improved analytic hierarchy process is put forward in the stage of risk factor identification. And then, the main element neural network information security risk assessment method optimized by particle swarm optimization is proposed in the stage of risk value establishment. Finally, the improved method is verified by simulation experiment.According to the existing hierarchy analysis method in risk assessment of faults, the structural concept of the fault tree analysis method is introduced. And from the construction of judgment matrix, consistency test, the establishment of the three aspects of risk level to improve the fuzzy comprehensive evaluation method. The improved method can fully identify and risk factors of grade. Aiming at the defect of particle swarm optimization algorithm, such as the network depend on the initial value of strong, easily falling into local minimum, an improved approach based on the improved principal component neural network initial parameters is proposed. And particle swarm optimization of information security risk assessment model of principal component neural networks is established. The simulation experiment shows that the improved method has higher accuracy, and reduce the number of iterations by contrasting with the traditional BP neural network.