Development of a system for collecting and processing sky images and meteorological data used for weather prediction
In one sentence. Builds a sky-camera and weather-station system and uses a CNN to classify sky condition with 97.8% accuracy for local short-term prediction.
Abstract
Weather prediction is a crucial element for power management in photovoltaic power plants (PVPP). In this paper, we propose a novel system for collecting essential data used for local short-term weather prediction. Image data consists of all-sky ground-based images obtained by an all-sky camera system with a fish-eye lens. Our proposed weather station collects meteorological data into database. The data include air temperature, humidity, wind speed, relative pressure, and spectrum of solar radiation. First, the whole setup for obtaining all-sky images is described, and setup for weather station is proposed. Then, our all-sky image database is characterized. Finally, to test sky images an experiment was performed to determine sky condition (clear sky, partly cloudy, mostly cloudy, overcast) with the use of a deep convolutional neural network (CNN). The accuracy of this method reached 97,80%.
Labels
- Topic
- Observations & remote sensing, Nowcasting
- Type
- New model or method
- Method
- CNN / U-Net
- Time scale
- 0–6 h
- Domain
- Local / station
- Data
- Station / in-situ, Other
- Status
- Pre-operational / experimental
- Identifiers
- DOI:10.1016/j.trpro.2019.07.214 OpenAlex:W2966730076
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.