Event-driven sensor/actuator microcontroller using neural network-based parameter reconfiguration method for unknown plant-model control applications

The controller architecture based on event-driven sensor/actuator microcontroller (MCU) is proposed. The specially-design MCU is based on the asymmetric dual-core architecture including parameter-tuning unit based on the artificial neural network (NN) in small battery-operative control applications of unknown plant-model. The operational parameters of the PID controller have to be determined for the optimized accuracy with reasonable operating energy consumption in controlling the output of the controller by sensing the dynamic characteristics of the target plant model. The parameter-tuning operations of the controller are required to provide the optimized accuracy for the plant model described insufficiently about its dynamic behavior, which consume large energy in the complex neural network-based calculation algorithm. In this paper, we adopt the NN-based parameters adaptation approach, which is integrated event-driven sensing/actuating method for dynamically unknown model of the target plant. The event-driven sensor monitoring the output of the controlled target plant enables to allow more time slot in performing the NN-based next parameter reconfiguration. We implement a specially-designed microcontroller with the proposed parameter reconfiguration unit based on the event-driven sensor/actuator architecture. The experimental results are discussed in terms of the energy consumption and accuracy of the event-driven neural network-based parameter tuning method using the implemented microcontroller architecture.