Metody strojového učení pro optimalizaci řízení monitorovacího systému

Abstract

Wireless sensor networks are constantly evolving. Wireless sensor networks cover a wide area of everyday life, reaching into industrial, commercial and other spheres. Their goal is to make everyday life easier. The expansion of wireless sensor networks is due to the high progress, development and growth of computer technologies. Of course, these facts inherently lead to increasing computing power. The increasing demands on all aspects of the device, such as functionality, sophistication and efficiency of applications are the result of the ever emerging new, more demanding and sophisticated trends in the application area. The concept of digital twin can be used for advanced energy management to set up adaptive duty cycles in Internet of Things devices. Together with~the use of machine learning methods, we will achieve the development of more sophisticated and efficient design models that address the optimization of wireless sensor nodes and~networks as a whole.

The optimization and management of the wireless sensor node will be addressed in this thesis using duty cycle setup and scheduling methods. The optimization design will be later complemented with an algorithm that will apply fuzzy expret rules and differential evolution. The proposed algorithm will be verified through simulation with digital twin technique.