Neuro-Based Adaptive PID Controller for Marine Satellite Tracking Systems

Marine Satellite Tracking Antenna (MSTA) systems are used to communicate between marine vessels and geostationary satellites. These systems enable the use of critical communication channels such as telephone, TV and internet communication in marine vehicles such as boats, yachts and ships. For this reason, improving the performance of systems that can react in real time against the disruptive effects of sea waves is gaining importance. In this study, the design and implementation of an adaptive control system that trains a neural-based proportional integral derivative (PID-NN) controller by modeling with a real-time indirect neural network has been realized. The prominent feature of this work is that it offers an advantageous design in terms of price performance ratio as well as having a unique control design. In this study, all hardware and software components of a high-performance embedded system that, even in 20 degrees/second sea waves, can broadcast continuously were physically implemented with a single-core standard microcontroller unit (MCU) such as Arm Cortex-M4 Core, without the need for special peripherals such as neural processing unit (NPU). In addition, the performance results and implementation methods obtained from this study are discussed in comparison with a commercial product working in a similar field and other academic studies. While the maximum elevation error of the compared commercial product reaches up to 2 degrees in waves of similar intensity, it is observed that the error does not exceed 0.7 degrees under the same conditions in the design presented in this study.