Research On Attack Reconstruction And Control For Cyber-Physical Systems Under Deception Attack

With the rapid development of modern control and information technology,the physical model and information transmission of control systems such as power network,transportation and chemical production have become more and more closely linked.This connection can not only improve the performance of the system,but also improve the efficiency and reliability of the system.Cyber-physical system(CPS)can deeply describe the relationship between physical dynamic process and information generation and transmission,so as to better meet people’s diversified needs.CPS is a highly flexible network architecture,which can provide efficient control and management according to different needs and environments.According to the control theory,we must carefully analyze the characteristics and strategies of potential attacks in the system,and adopt a unified method or mode to describe the behavior of these attacks to ensure the security and reliability of the system.In order to ensure the security of cyber-physical systems,it is also very important to study how to carry out effective security control under various attacks.Although a lot of research has been done in this field,there is still much room for improvement.By combining linear system theory and observer theory,the thesis discusses several cyber-physical systems with actuator attack.Some techniques of attack reconstruction and detection based on sliding mode observer are proposed,which solve the problems of attack detection,security state estimation and elastic control.The main work of the thesis is as follows:(1)Different from the existing research results of sensor attacks,a sufficient condition that attack signals can be detected and reconstructed is established,and a new method is proposed,which can deal with linear cyber-physical systems only involving actuator attacks.By combining sliding mode observer with a robust differentiator,this method enables us to better estimate the state of the system and detect attacks more accurately.In this thesis,a continuous sliding mode observer is used to improve the dynamic performance of state estimation.In addition,attack compensation is also considered in the control strategy,and the controller and sliding mode observer are designed separately,which simplifies the design process and makes the system run stably under deception attack.(2)Based on the above work,the research object is extended to a more general class of cyber-physical systems with actuator attacks and unknown external disturbances.The sliding mode observer with disturbance compensation is mainly designed to solve the problems of security state estimation and attack reconstruction at the same time.The estimated disturbance is compensated to the sliding mode observer,which makes the observer have good observation performance.Furthermore,by setting the corresponding observer gain,the attack reconstruction strategy is robust to disturbance observation errors and has good H_∞ performance.The resilient control technology based on attack compensation can greatly improve the security of the system,so that it can still maintain effective and stable operation when it encounters external attacks.