网络化倒立摆系统的偏差攻击及其检测方法

Bias Attack and Detection Method for Networked Inverted Pendulum System

为解决网络化控制系统数据完整性的攻击问题,设计了基于网络化倒立摆平台状态信息的偏差攻击,并提出了对应的检测方法.利用Ettercap工具对平台网络成功入侵,对位置数据进行了偏差攻击;结合支持向量机(SVM)方法,使用LibSVM分类器,对倒立摆系统的4种状态信息进行训练、建模和数据分类,并与K最近邻、决策树方法进行对比;在平台上验证了所提出的方法.仿真和实验结果表明,所设计的攻击方法能够改变系统的稳定状态,与常用的机器学习方法相比,SVM在偏差攻击检测的二分类问题上更加优越,能较好地区分掺杂在数据的虚假数据.

In order to solve the data integrity attack of networked control systems, a bias attack and its detection method based on the networked inverted pendulum platform sensors are designed in this paper. First, the Ettercap tool is utilized to realize network intrusion and inject false data. Next, combined with the support vector machine(SVM) method, the LibSVM classifier is used to train the four kinds of state information in the inverted pendulum system to obtain the model and classify the data. After that, the SVM method is compared with K-nearest neighbor and decision tree methods in the self-built system. Finally, the method proposed is validated on the platform. The simulation and experimental results show that the designed attack method can change the stability of the system. Compared with the commonly used machine learning method, the SVM has more advantages in the binary classification of bias attack detection and can effectively distinguish the false data in the transmission data.