Research On Network Intrusion Detection Technology Of Industrial Control System In A Laboratory

Industrial control systems are widely used in electric power,water conservancy,metallurgy,oil and other industries.It consists of various automation control equipment components and real-time data acquisition,monitoring network systems,is the core of national key infrastructure,once they suffer,it is easy to cause National economic losses,and even threaten the safety of people's lives and property.With the gradual integration of informationization and industrialization,the internal communication network of industrial control system gradually communicates with the Internet,inevitably breaks the original software and hardware closure of the industrial control system,which is easy to suffer more attacks,so the industrial control system is subject to Network security threats and risks are constantly increasing,so research on industrial control system security is very urgent to ensure the safe operation of the industrial control system.Intrusion detection technology is one of the most important security precautions in the network security of industrial control system,which effectively detects known attacks,improves the ability of industrial control systems to identify attacks and warning.However,due to the real-time requirements of the industrial control system and the finite to device resources,existing industrial control system intrusion detection methods still have some shortcomings,such as low detection efficiency,and unknown attacks cannot be quickly and efficient.Therefore,it is necessary to design a reasonable intrusion detection method to increase the intrusion detection rate of the industrial control system,reduce the leakage rate and false positive rate.Based on the existing research results,the network security characteristics of industrial control system are analyzed,and the following main tasks are conducted for intrusion detection based on network traffic.1)A feature selection method based on a plurality of two-point algorithm is proposed,and the industrial control network data is reduced.According to the characteristics of industrial control network data multi-source,complexity,and high dimensional characteristics,the feature selection method suitable for industrial control system information data is proposed.This method combines a variety of two-point algorithms to filter through wrapped feature,which is new and a characteristic subset of primary features.2)After the characteristics of the data,new data set training is based on AdamW optimized neural network intrusion detection model.First,select the characteristic subset,train for intrusion detection model,the training is obtained,and the parameter setting is saved;then the parameter setting is saved;then the industrial control system information intrusion detection characteristics are used to verify the design.The effectiveness of the intrusion detection model;the experimental results show that the neural network intrusion detection model based on AdamW is increased than the traditional intrusion detection model,which has declined significantly in the leakage rate and false positive rate.Dimensional data and original downtime data were carried out,and experiments show that the reduction test is increased in terms of detection,increased by 0.91%,and the false positive rate is reduced by 36.93%,and the leakage rate is reduced by 27.74%.Overall,dove data can effectively improve the performance of the model.At the same time,the generalization between Adam and Adan W.The results show that the improved AdamW generalization is more excellent than the Adam before the improvement.3)The data visual design and implementation process of the industrial control network intrusion detection system has been described briefly,and combines system configuration diagrams and function graphs to introduce the function of the system one by one.