Prediction Algorithm Based on Weather Forecast for Energy-Harvesting Wireless Sensor Networks

Solar energy is one of the effective solutions for the perpetual operation of wireless sensor networks (WSNs), but the harvesting of solar energy is highly random, and it is very important to predict the solar energy harvesting in advance. The existing prediction algorithm has better predictive effect in predicting the similar weather conditions in the region, but the accuracy of the prediction algorithm will be reduced when the weather in the prediction area drastic changes. In this paper, the study is about the influence of weather conditions on solar energy harvesting by introducing real-forecast weather into the prediction algorithm. Secondly, based on the Weather Conditions Moving Average (WCMA) algorithm, an efficient and reliable prediction algorithm, Real-Forecast Weather Moving Average (RWMA) is proposed. The algorithm adjusts the prediction result of the following slot according to the error of the harvesting amount of the preceding several slot. Experimental results show that the RWMA algorithm can also predict solar energy when weather drastic changes. Compared with the existing prediction algorithm, RWMA algorithm solves the problem that the weather change affects the prediction result, and the prediction performance is greatly improved.