Energy prediction models for solar harvester

Singapore’s adoption of solar photovoltaic (PV) systems is expected to see strong growth in the coming years. 248 new PV systems were installed during the year 2014 which was more than doubled as compared with year 2013. [1]. However, solar power generation is limited with its inability to generate continuous power due to environmental factors and weather conditions such as the amount of sunlight, cloud cover and shadow. A perpetual operated wireless sensor network relies mostly on solar energy. The inconsistency of solar irradiance poses a problem for grid operators who must compensate for the short fall through additional power source and/or capacity. Therefore, accurate short-term solar irradiance forecasts are crucial for effective and efficient deployment of resources to harvest solar energy. This study seeks to evaluate the accuracy of different forecasting models using data with and without cloud data. Consequently, this study will propose the most suitable forecasting model. The study will be divided into three phases. Where phase one comprises of data pre-processing and modelling of ten forecasting models without factoring cloud data. Followed by phase two, where it involves the collection, processing and classifications of sky imagery. Finally, phase three integrates phase one and two to obtain the forecast accuracy.Bachelor of Engineering (Computer Science