【摘要】 随着互联网技术在各类工业过程中的广泛应用，工业控制系统遭受恶意网络攻击的可能性不断增加。通过设计网络攻击检测方案，增强工业过程的网络安全防御能力，可以有效减少恶意攻击带来的损失。建立工业过程网络攻击的物理模型，以相应的检测算法，实时检测网络攻击造成的异常状况。以化工生产田纳西-伊斯曼过程模拟网络攻击中的假数据注入攻击，建立一种多层支持向量机方法进行检测。该算法使用递归特征消除方法，对多种攻击类型建立二分类支持向量机模型，并通过融合决策形成多层支持向量机模型。更多还原

【Abstract】 With the widespread application of internet technology in various industrial processes,industrial control systems are increasingly likely to be subject to malicious cyber-attacks.Therefore,a network attack detection scheme can be designed to enhance the network security defense capability in the industrial processes,and to effectively reduce the losses caused by malicious attacks.This study established a physical model of industrial process network attacks and proposed corresponding detection algorithms to detect abnormal conditions caused by cyber-attacks in real-time.This study proposes a multi-layer support vector machine method for detecting false data injection attacks in chemical production Tennessee-Eastman process simulation network attacks.This algorithm uses recursive feature elimination methods to establish a binary support vector machine model for multiple types of attacks,and forms a multi-layer support vector machine model through fusion decision-making.更多还原