Research On Agent Behavior Monitoring Method For Network Security Of Oil And Gas SCADA System

The SCADA(Supervisory Control And Data Acquisition) system in the industrial control system which based on the advanced computer technology is widely deployed in areas such as water&power grids, oil and gas pipelines etc. Recently, there have been reports on damage behavior of SCADA system. Large oil and gas gathering pipe network automation monitoring and control system, namely the SCADA network system of oil and gas, as a modern industrial control monitoring system, also faces pressing security defense situation. Oil and gas SCADA belongs to large industrial control system, if malicious destruction, will cause huge losses.As the large-scale deployment over a long distance oil and gas pipeline, the classic centralized security defense can’t meet the needs of oil and gas SCADA system security defense. Agent technology is the hotspot of popular technology, the use of the Agent’s autonomy, initiative, adaptability, can better solve the problem of distributed security defense. And the oil and gas SCADA system run produces a large amount of data, using some algorithms of data mining technology can better solve the problem of data processing.Combines the Agent technology and data mining technology, can realize remote heterogeneous environment through a model of oil and gas SCADA monitoring network system security defense and behavior. This article have designed a kind of model which oriented oil and gas SCADA system network defense.And it is applied distributed defense system of multi Agent technology.The model can be implemented to the SCADA system software, hardware and operating behavior defense. Introduced the function of Agents in the model, and design all of the specific function and structure. For Defense Agent in the model of decision making problems, this article selects the classic data mining algorithm combined with the classical data sets, using Weka tools designed to verify the effectiveness of the decision tree in the decision making Agent training, and can realize autonomous decision-making Agent. Finally, combining the above techniques, to achieve the basic prototype of the model, and the prototype system was tested.Compared with the traditional security defense, this system makes full use of the many advantages as a means of Agent architecture, with accurate detection and high detection efficiency, response time, etc.