Support Vector Machine(SVM)-based method for identifying physical attacks in complex oil and gas production systems information
Within data-driven complex oil and gas production systems,the challenge arises from fault data interfering with attack identification,thereby impeding the timely defense against attacks or resolution of faults.Current attack detection methods predominantly concentrate on external attack detection,overlooking the potential influence of internal system fault data on the efficacy of attack detection.Therefore,this paper proposes an innovative method to enhance the accuracy of detecting information-physical attacks in complex oil and gas production systems.The method combines support vector machines and undirected graphs for joint detection,aiming to effectively distinguish between system faults and information attacks.By doing so,the proposed method not only enhances the accuracy of information attack detection but also improves the overall security of the system.Initially,the key sensors within the complex system are topologically organized to construct an undirected graph,which delineates the interconnections among the sensors and captures their data interactions.This forms the basis for subsequent anomaly detection.Subsequently,support vector machines are employed to conduct anomaly detection within the established sensor network.This process involves identifying anomalous data points,discerning between anomalies arising from attacks and sensor faults,and effectively detecting the presence of information attacks.The validation process focuses on the receiving station low-pressure pump and tank system,demonstrating exceptional performance metrics.The accuracy,precision,recall,and F,score for the low-pressure pump surpass 99％,while exceeding 97％for the tank system.Moreover,both systems achieve an F1 score exceeding 99％.This method showcases superior performance,robustness,and accuracy in precisely identifying attacks and anomalies.Its potential for effectively addressing security challenges in oil and gas production systems highlights its significant application prospects.Compared to the traditional k-means clustering algorithm,the innovative method leveraging the joint detection of support vector machines and undirected graphs exhibits superior performance in terms of both detection accuracy and integrity.In summary,the approach integrating support vector machines and undirected graphs for joint detection significantly enhances the accuracy of identifying information attacks within complex oil and gas production systems.This method is anticipated to elevate the security and reliability across various system types,thereby playing a crucial role in preventing attacks and production accidents and ensuring the safety of oil and gas production systems.
safety engineeringcomplex systems for oil and gas productioninformation physical attackexception detectionSupport Vector Machine(SVM)
NETL
NSTL
万方数据
