Sky Imager Data Reduction Using Autoencoder and Internet of Things Computing

Abstract

Forecasting the output power of renewable energy sources is of an utmost importance for maintaining the stability of the electricity grid and ensuring uninterrupted service. Internet-of-Things (IoT) devices equipped with sky imagers are often used to estimate solar irradiance whose value is essential for forecasting solar energy. This paper considers distributed sky imagers whose data are transmitted through the Internet to a cloud for possible storage, forecasting, and/or performing other signal processing operations. The data comprise a collection of images, and this makes their streaming to the cloud a burden, i.e., they require a large bandwidth. In an attempt to deal with that, an IoT system is proposed in this paper such that it collects, compresses, transmits, and reconstructs those images with as little bandwidth as possible. More specifically, an IoT device multiplexes the collected images and compresses them using a deep learning algorithm, the encoder of a convolutional autoencoder (CAE) in particular. It, then, transmits the compressed images to a cloud server where they could be decompressed with high-fidelity using the decoder of the CAE. As the core to the proposed system, the performance of the CAE algorithm is evaluated using a publicly available dataset of sky images. Results indicate that the images can be compressed to 2% of their original size (i.e., compression ratio of 98%) and reconstructed again with high fidelity, i.e., average structural similarity index measure (SSIM) of 99%. Hardware implementation of the CAE algorithm has also been attempted using a Raspberry Pi platform representing the IoT device with an average processing speed of 6.6681 seconds per frame. This speed well meets practical requirements given the low frame-rate of sky imagers.INDEX TERMS Convolutional auto-encoder, image compression, sky imager, raspberry pi, Internet of Things.

Image Insertion Using Depth- and Topology- Constrained Minimum Spanning Tree for Compressed Image Sets

“…smart phones and tablets. With their wide uses, a great number of digital photos have been taken by people anytime and anywhere, which are usually delivered to and then stored in cloud servers [1], [2], [3], [4], [5]. For any image in the enormous amount of photos, there may be some other images with similar contents, which thus can be called similar images.…”

Real-Time Microgrid Centralized Control For Consuming Water Pump

“…The monitoring system produces good performance in seeing the potential of solar energy for electrical energy needs. Author [17]- [19] IoT devices equipped with weather monitors for prediction of solar radiation intensity. One method is using the internet to the cloud for forecasting and other signal processing.…”