Event-Triggered Adaptive Neural Network Controller in a Cyber–Physical Framework

The importance of remotely placed controller in a cyber space with sensor-controller-actuator network has increased significantly in industrial, defense, and surveillance sector. Such network has large amount of sensor and controller data. A time-triggered control technique may generate redundant control signals and put unnecessary data on network. Therefore, an event-triggered adaptive controller that generates control action at required instants using state- and error-based conditions has been developed in this paper. A data transmission framework has also been designed in this paper that addresses network delay and packet losses. The proposed controller-communication methodology has been validated through two case studies, first, temperature tracking for heating ventilation and air conditioning system and, second, real-time path tracking by automated guided vehicle. The proposed methodology has also been duly compared with its time-triggered counterpart. The control updates are reduced to approximately 41% and 64% in the two case studies, respectively. The experimental results also prove the designed controller to be efficient when compared with event-triggered incremental PID controller using the same data transmission framework.