Research On Industrial Control Safety Of Water Conservancy Pumping Station Based On Deep Neural Network

With the continuous development of Internet of Things technology and the advancement of industrial informatization,the traditional industrial control network has been unable to meet the needs of industrial production.The introduction of "Industrial 4.0" in Germany has strengthened the opening up of industrial control system.A large number of complex network interconnection with the outside world has led to the threat of industrial control network security.In recent years,industrial control safety accidents occur frequently all over the world,which has a great impact on society.In order to realize the safety protection of communication network of industrial control system,this thesis takes Wangyu River pumping station as the background,deeply analyses the communication characteristics of industrial control network,and uses deep neural network to carry out intrusion detection and security protection for industrial control network of water conservancy pumping station.The main research work of this thesis is as follows:(1)Modbus/TCP protocol for industrial control network of water pumping station is analyzed.The characteristics of Modbus/TCP in communication mechanism,message format and function code are discussed emphatically.The security loopholes of Modbus/TCP are summarized.On this basis,the typical abnormal behaviors of Modbus/TCP communication protocol are classified,and four common abnormal behaviors are listed.Then the data stream of Modbus/TCP protocol is pre-filtered,and some data that do not conform to the custom rules are filtered through the white list setting.The request/response mechanism of Modbus/TCP is analyzed.Based on the communication mechanism of Modbus bidirectional function code,the detection features corresponding to the common attack manifestations of the protocol are selected,and the protection against attacks is analyzed.(2)This thesis analyses the advantages of feature learning and related technologies of deep learning,an intrusion detection system based on DBN algorithm is designed,and an algorithm model is constructed by using the communication data flow of industrial control network of water pumping station.The system consists of three modules: data preprocessing,DBN processing and intrusion classification.The experimental results show that the DBN algorithm in this thesis has better intrusion detection effect,completes the identification of legitimate or illegal data streams,performs well in detection rate,false alarm rate and so on,and has better generalization ability and practical innovation.(3)A monitoring system for water pumping station is designed.Users can directly see the protocol audit,DBN algorithm performance and system log information through the interactive interface.The log information of the system is stored in the database for user monitoring.Intrusion detection system generates a large number of statistical charts by association and classification statistics with field equipment,so that users can monitor the status of industrial control network of water pumping station.