OBSERVER BASED DETECTION AND MITIGATION SYSTEM FOR ACTUATOR ATTACKS AND FAULTS FOR HEAVY GROUND VEHICLE

• Main Author: Bin Bashar, Labid
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/81278
• Institution: Institut Teknologi Bandung

A heavy ground vehicle finds applications in various fields, particularly in logistics and freight transportation, involving the movement of containers from ports to local ware-houses and eventual delivery destinations. In certain terminals, specifically in smart ports, the process has been automated, necessitating the vehicle’s autonomous dock-ing and undocking. However, such a cyber-physical system faces the risk of cyber-attacks and internal faults posing hazardous situations within the terminal. Therefore, it becomes crucial to identify and mitigate any cyber-attacks as well as faults during the docking and undocking procedures. Existing literature encompasses detection and mitigation systems for cyber-physical systems, but most of them focus on detecting sensor attacks within linear systems. Since the vehicle’s model consists of coupled nonlinear differential equations, these approaches fail to address the stability of such nonlinear systems. Furthermore, actuator attack detection systems present even greater challenges, especially when dealing with nonlinear models, as the field remains rel-atively unexplored. Consequently, this research aims to develop an observer based detection and mitigation system for actuator attacks and faults for autonomous dock-ing of heavy ground vehicles, employing a point stabilization-based control strategy. The methodology outlined in this study adopts a hybrid approach, involving analyti-cal proof of the asymptotic stability of the detection and mitigation system, comple-mented by numerical simulations and experiments to provide a quantitative evaluation. Initially, the vehicle model is specified, considering potential actuator attacks, which is then translated into the respective actuator attack effects at each channel. Subse-quently, the design problem is formulated by defining the point stabilization control strategy and the autonomous docking strategy. A block diagram is then presented to illustrate the interfaces of the subsystems, comprising the actuator attack detection and iii mitigation system. The proposed system design utilizes the Lyapunov direct method to establish the asymptotic stability of the detection system and subsequently inte-grates it with the point stabilization controller to design the mitigation system, ulti-mately contributing to the asymptotic stability of the desired configuration for docking of the vehicles. The simulation of autonomous docking is performed within a MAT-LAB environment, demonstrating the ability of the detection and mitigation system, in conjunction with the point stabilization controller, to detect simulated actuator at-tacks and mitigate their effects, resulting in a smooth docking process at the desired configuration. The resilient controller with observer-based attack detector and miti-gation system (OADMS) is able to effectively make the vehicle dock with an error of q˜ = (0.0023 m,2 × 10?5 m,1 × 10?7 rad) after 100 seconds. The mean estimation errors (a˜v,a˜? ) are 5.3828 × 10?5 m/s and 0.0638 rad respectively. The standard devi-ations are 0.0054 m/s and 0.1332 rad denoting that the estimate of attack values con-verges to the real attack values. Experimental results validate the effectiveness of the controller, as the vehicle consistently achieves the desired configuration. Although this work primarily focuses on cyberattacks, it is worth noting that, for safety reasons within container ports, the experimental investigation tackles actuator faults instead of actuator attacks, effectively showcasing the system’s ability to detect internal actuator malfunctions. The vehicle can dock at the desired configuration despite the presence of actuator faults with errors of 0.08 m, 0.02 m, and 0.0092 rad. While the average values of faults are 0.0293 m, ?0.5053 m, and ?2.6180 × 10?3 rad, at the end of the motion, the values are zero. The standard deviations of the faults are 0.0392 m, 0.3759 m, and 5.5851 × 10?5 rad respectively. Future endeavors in this area entail reducing system chattering through the application of adaptivity in control and observer gains. Key words: Heavy Ground Vehicle, Cyberattack, Attack Detector, Autonomous Dock-ing, Observer.