Research on Intrusion Detection Method of Industrial Internet Based on Machine Learning

Abstract

The mobile Adhoc network (MANET) is being used more and more widely, and the related network security issues have also begun to receive widespread attention. Researching the MANET network's possible attack methods, the paper proposes an intrusion detection performance evaluation model based on machine learning technology and proposes a comprehensive evaluation index. It compares seven machine learning algorithms' performance in MANET network intrusion detection, sufficient for building security. The MANET network is of great significance. Use the GloMoSim simulation tool to simulate the MANET network's normal behavior and the three intrusions of black hole, flood, and packet loss, and analyze the performance of seven machine learning algorithms in various attack situations in various attack situations detail. Our analysis results show that the evaluation model can better reflect the performance of various machine learning algorithms. Multilayer perceptrons, logistic regression, and support vector machines have higher detection rates and lower false alarm rates.