A low cost weather monitoring, PV and prediction system in East Africa

This paper is about the design of a lowcost system to monitor the weather and resulting PV generation for small-midsized microgrids in Africa. The setups are one 12kWp near-grid and 192-panel 76kWp off-grid systems. The weather parameters collected are wind speed, wind direction, temperature, panel temperature, humidity, and rainfall as well as in-plane, horizontal irradiation. Cloud coverage, a major influence of solar generation, is captured by two 100° to 200° view-field cameras per site. Moreover, we compare this precise data with internet data for variation, particularly with large-grid-area satellite data. A wide and ultra-wide capture of cloud images in multiple resolutions for analysis and development of prediction algorithms. Lowresolution images allow for easier transmission over the internet in remote East Africa. A more accurate analysis of the different cloud types and characteristics is enabled with the dynamic range of two set-ups in addition to a comparison of the consistency and accuracy of the various algorithms to be developed. The ultra-low-cost is achieved across weather stations and sky imaging using a Raspberry Pi-based setup. Beneficial redundancies from multiple sensors per site are expected due to the experimental setup nature.