Document Type : Research Articles

Authors

• Maryam Shahriari ¹
• Seyed Hojat Nourian ²

¹ Shahrekord, Iran

² Electrical Engineering Department, Faculty of Engineering, Shahrekord University

Abstract

This work proposes an adaptive resilient control for uncertain nonlinear cyber-physical systems (CPSs) under deception attacks. It is assumed that attacker injects false data into the commands exchanged between the controller and actuator over the communication channels. The injected false data affects the control input in both additive and multiplicative forms. To deal with the uncertain dynamics of the system and additive term of cyber-attacks, the radial basis function-neural networks (RBF-NNs) are invoked. Also, to handle adverse effects of multiplicative term of cyber-attack, the Nussbaum-type gain function is employed. Then, by integrating the RBF-NN model and Nussbaum function into the command filtered backstepping (CFB) approach, the proposed resilient control scheme is designed. Compared with the existing works, the proposed control eliminates the “explosion of complexity” problem in the conventional backstepping approach, removes the trial and error in choosing time constant of the first order filters in the dynamics surface control (DSC) approach, compensates the filtering error and deals with both additive and multiplicative cyber-attacks in “controller to actuator” channel, simultaneously. Also, it mitigates the effects of the cyber-attack without requiring separate attack estimation unit, controller reconfiguration or readjustment algorithm. Simulation results on the robotic arm under different cyber-attacks verify effective resilient performance of the proposed control scheme.

Keywords

• Deception attack
• False data injection attack
• Nonlinear cyber-physical systems
• Resilient control

Main Subjects

• Control