Pattern Recognition and Prevention Algorithm of Complex Network Attack Based on Genetic Optimization Neural Network

Traditional means of security protection are unable to cope with complex and changeable network attacks. In order to improve the recognition ability of complex network attack patterns, this paper introduces genetic algorithm (GA)to optimize the weights and thresholds of neural networks. In this paper, a model based on multi-layer feedforward neural network (MFNN) is constructed, and the parameters of the neural network are optimized by using the global search and optimization characteristics of GA. GA improves the convergence speed and global search ability by adjusting thecrossover rate and mutation rate adaptively and adopting the strategy of multi-population co-evolution. The experimental part uses KDD Cup 1999 network intrusion detection data set, and carries out data cleaning, feature selection and data normalization. The experimental results show that the recognition accuracy of the neural network model optimizedby GA has been significantly improved, and the recognition accuracy of normal traffic has reached 90.0%, and the recognition accuracy of attack patterns such as DoS, U2R, R2L and Probe has also been improved to 85.0%, 75%, 80.0% and 90.0% respectively. Finally, based on the identification results, this paper puts forward corresponding prevention strategies, including traffic filtering, distributed defense, system update and patchmanagement, minimum authority principle, network isolation, security authentication and intrusion prevention system (IPS), which provide strong support for improving network security and stability. This study not only has theoretical value, but also has urgent practical needs, which provides an effective solution to deal with complex and changeable network attacks.