Application of Internet of Things (IoT) sensors to support electrical load planning in solar-powered resorts in Kenya

Abstract

Internet of Things (IoT) has evolved has emerged as a powerful new technological paradigm in recent years. It makes it possible to interconnect all types of physical objects and offers real-time collection and processing of sensor data. The rapid development of renewable energy sources, in particular solar energy, has resulted in an increase in availability of clean electrical power that can be efficiently stored and utilized. However, the variability of solar energy based on weather conditions affects its reliability. This makes it difficult for the planning and performing of tasks in resorts, which harness solar energy, such as laundry and food preparation, that require electrical power. This thesis investigates the adoption of Internet of Things (IoT) to optimize the use solar power for meeting electricity demand at solar-powered resorts. This thesis seeks to implement a system that automatically alternates between grid power and solar energy sources, with the aim of optimizing solar energy utilization by performing load planning. This will be achieved by considering the predetermined electrical demand of the resort, employing a predictive model, and using real-time Internet of Things (IoT) sensor data. The prediction model will project solar energy generation which will be compared with real-time IoT sensor readings, for load planning purposes. For the predetermined electrical load of the resort, the system will automatically transition between grid power and solar energy based on whether the two values align or exceed a predefined system threshold. The predictive model will use weather forecast data obtained from publicly accessible sources on the internet, which are usually an approximation, as its input while the IoT sensors will capture real-time measurements relating to weather conditions, solar energy generated and subsequently stored. The overall objective is to optimize the utilization and load planning of solar-generated electrical energy in solar-powered resorts. Ultimately the system will facilitate determination of adequate capacity of expensive solar power storage systems and minimization of the volume of consumed grid electricity. This research adopts an analytical and prototyping methodology. The research will review literature on off-grid solar power installations and their control systems, IoT architectures and Machine Learning techniques which will be adopted in designing the predictive model.

Keywords: solar energy; prediction models; energy management; load planning; Internet of Things (IoT) machine learning; renewable energy forecasting; solar powered resorts