Estimating Harvestable Solar Energy from Atmospheric Pressure Using Support Vector Regression

Energy neutrality is the desired mode of operation of many sensor networks used for environmental monitoring. Intelligent energy harvesting networks, composed of nodes equipped with solar panels and other types of power-scavenging devices, can plan and manage their operations according to short and long-term predictions of ambient energy availability. This paper introduces a novel method for next-day solar energy prediction based on atmospheric pressure and support vector regression. A location-specific support vector regression model is in this approach created using a collection of geospatially correlated atmospheric pressure and solar intensity measurements. The trained model is used to estimate next day solar energy availability from a time series of recent atmospheric pressure values and their differences. The ability of the proposed system to estimate daily solar energy is compared to a recent evolutionary-fuzzy prediction scheme and traditional analytical estimates.