This paper designs optimal control polices for networked multiagent pursuit-evasion game (MPEG) problems based on reinforcement learning (RL) technique. Depending on the number of evaders, MPEG is formulated into several simpler multiple-pursuer single-evader games (MPSEGs) by a divide and conquer approach. Then we propose optimal control policies for all the agents in each MPSEG, which constitute a distributed Nash equilibrium, and provide the capturability and Nash equilibrium analysis. Finally, a data-driven RL algorithm is developed to online learn optimal control polices using measurable behavior data. A simulation example is given to verify the effectiveness of the proposed approach.

A co-monitoring problem of state of charge (SOC) and state of health (SOH) under resource-constrained and insecure communication is addressed in this paper. First, an SOC/SOH co-monitoring framework is established to account for battery SOC and SOH dynamics, unknown-but-bounded (UBB) process and measurement noises, dual coupled monitors, and intermittent measurement updates at multi-time scales. Second, a resource-efficient communication mechanism is proposed for determining the measurement updates according to not only the measurement variations but also the realtime network conditions, realizing reasonable utilization of constrained and time-varying communication resources. Third, dual coupled monitors are developed based on state predictors and estimators for SOC/SOH co-monitoring, which provide prediction and estimation ellipsoids enclosing the true states of SOC/SOH dynamics at all times. Then, encountering potential attacks, attack detection strategies are derived by calculating the intersection over union (IOU) of the prediction and estimation ellipsoids. Upon occurrences of attacks, a compensation mechanism is presented to replace the contaminated measurement-based state estimates by the attack influence-free predicted states, suppressing the adverse effects of attacks. Finally, comparative case studies are carried out to verify the efficacy and merit of the proposed methods.

Power cyber-physical systems (PCPSs) are becoming increasingly complex and diverse, benefitting from all aspects of societal issues. A PCPS is a next-generation system that is a tight conjunction of computation, communication, control, and power systems. The integration of Internet of Things and PCPS generates a novel energy system, the Internet of Energy, which is the future of the energy system. In the same way, the connection of bulk components and the extensive communications among them has brought many insecurities, uncertainties, and security challenges. Thus, a study of the analysis and synthesis of the reliability and security of PCPS is presented in this paper. By considering the unsafe uncertainties, models are formulated for a general cyber-attack, and a double-loop architecture for security defense is designed. According to this framework, security control scenarios are obtained from the character of each kind of cyber-attack. Finally, a separately excited dc motor with uncertainty is used as an example to demonstrate the problem.