| Networked manufacturing (NM) is a new manufacturing mode which appeared in the past decade. It can integrate the high-quality resources of traditional manufacturing enterprises which are scattered at different geographical locations, and can perform such activities as product research and development (R&D), manufacturing, sales, and management, in order to reach the goals of high efficiency and low cost, and to meet customers’requirements. Among all the challenges NM has to faces, the information security issue is one of the most important.Under the support of related research funding, this dissertation studies the information security issue in the networked manufacturing. In fact, there have been mature researches on the general information security issue; however, there has been no report of research on such an issue in the domain of networked manufacturing. Therefore, this dissertation investigates the following three topics of information security:risk assessment, intrusion detection, and risk control. The main results of research include the followings:(1) Risk assessment:After investigating both international and Chinese standards of information security risk assessment, this dissertation establishes its own information security risk assessment model, based on GB/T20984-2007. First, the key assets, their vulnerabilities, and threats in the aerospace manufacturing enterprise information system are identified; then, the risks are assessed quantitatively by using entropy weight method based on the layered structure obtained with traditional analytical hierarchy process (AHP), and therefore some consistent results are obtained. Since entropy weight method is based on information entropy theory, the subjective factors during the assignment of asset values can be further reduced, hence the final result of risk assessment is more reliable.(2) Intrusion detection:this dissertation proposes the PCA-C4.5, a hybrid feature selection algorithm based on principal component analysis (PCA) and decision tree (C4.5), and then constructs a light-weight intrusion detection system (IDS). By performing experiments on the KDD1999data sets, the PCA-C4.5algorithm is proved to have a fast training time, a fast detection time, a higher detection rate, and a lower false alarm rate, compared to the GA-SVM algorithm (based on genetic algorithm and support vector machine). The PCA-C4.5algorithm is implemented in an IDS, and can help the aerospace manufacturing enterprise to detect potential intrusions more accurately and effectively than ever before, therefore enhancing the security protection level of the enterprise.(3) Risk control:this dissertation proposes a plan of information security risk control based on device correlation method and multiple attribute decision making under uncertainty. A risk control process is studied in detail when the information of attribute weight is known. The device correlation method is used to perform dynamic risk assessment and some preliminary results are obtained. The multiple attribute decision making under uncertainty is studied to propose a research plan for risk control in the networked manufacturing aerospace enterprise.(4) Case study:In the final part of this dissertation, the research results of risk assessment, intrusion detection, and risk control are applied respectively to a real-world information system of the networked manufacturing aerospace enterprise. The results of risk assessment based on entropy weight method are consistent with traditional methods, indicating the overall information security risk is at a relatively low level; the novel intrusion detection algorithm, PCA-C4.5, can help the enterprise to detect more intrusions more accurately and effectively; the risk control plan based on device correlation and multiple attribute decision making under uncertainty helps the enterprise to take effective actions. Furthermore, a security management platform (SMP) is implemented to provide a convenient, effective tool of information system security protection. |
PDF Full Text Request