Font Size: a A A

Research On Neural Network Operation Platform Based On Multi-agent SCADA Security Defense Factor

Posted on:2017-10-26Degree:MasterType:Thesis
Country:ChinaCandidate:H B ZhaoFull Text:PDF
GTID:2358330482498966Subject:Control engineering
Abstract/Summary:PDF Full Text Request
SCADA (Supervisory Control and Data Acquisition) system is an important part of the industrial control system, the malicious attack on it is increasing. Traditional security defenses based on common IT system, unable to meet the special requirements of SCADA system security, and exposed many flaws.In order to ensure its information security, this paper take Oil&Gas Gathering and Transferring SCADA System of Petrochina Southwest Oil&Gasfield Company as an example, and analyzes the differences between the SCADA and the traditional information system in the security defense field of SCADA system. At the same time, this paper combined with the mathematical description of knowledge storage, representation and processing, and the simulation of intelligent behavior based on the factor neuron network theory, this paper establish a SCADA system security defense generic model for host computer. The model has a hierarchical and dynamic structure by giving factor neuron a life-cycle.This paper inspired by the BDI (Belief-Desire-Intention) model of Agent and their collaboration in the multi-agent system, and using multi-agent technology to achieve the generic model, finally establish a factor neuron network platform (FNNP), and introduced its structure, advantage and the process of SCADA system security defense tasks.For pragmatism, this paper using Java Agent Development Framework (JADE) accord with the criterion established by Foundation for Intelligent Physical Agents & Institute of Electrical and Electronics Engineers (FIPA-IEEE) to developing a prototype system, and designs an experiment to test its basic functions. The experimental results prove the factor neuron network platform is effective and feasible. In addition, this paper explores the important performances of FNNP and its optimization measures in detail.The SCADA system security defense generic model based on factor neuron network theory as well as factor neuron network platform based on multi-agent technology in this paper have many advantages, such as high robustness, reliability, real-time, flexible, extend easily, can fully satisfy the special demands of SCADA system security defense tasks and so on. In a word, the research of this paper is creative and has great practical application values for building the foundation of SCADA security defense system.
Keywords/Search Tags:SCADA, Security defense, Factor neuron network, Agent, Multi-Agent system
PDF Full Text Request
Related items