| dc.identifier.uri | http://hdl.handle.net/11401/77229 | |
| dc.description.sponsorship | This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree. | en_US |
| dc.format | Monograph | |
| dc.format.medium | Electronic Resource | en_US |
| dc.language.iso | en_US | |
| dc.publisher | The Graduate School, Stony Brook University: Stony Brook, NY. | |
| dc.type | Dissertation | |
| dcterms.abstract | Accurate prediction of solar energy in short-term and mid-term horizons becomes increasingly important for harvesting solar energy and improving its viability in comparing with fossil fuels. Because clouds are the primary cause of large fluctuations in solar radiation, estimating cloud movements and correlating cloud activities and the variability of solar irradiance are the essential components in short-term or mid-term solar predictions and play a vital role in the subsequent mitigation in responding to potential solar power fluctuations. However cloud detecting and tracking is very challenging due to the volatility and complexity of clouds and expensive meteorological instruments or remote sensing technologies (e.g., satellite imagery) that often have insufficient resolutions and limited availability. Ground-based sky imager offers higher temporal and spatio-resolution than does satellite. However it has a small field of view and lacks of spatial information of clouds, and solar forecast systems utilizing this type of imager are still inefficient and ineffective in cloud identification, motion tracking and as well as short-term irradiance prediction. To fully address the problems of cloud detection and tracking and construct robust solar forecast system, we focus on three major thrusts: robust cloud motion tracking, multi-camera integration, and multi-channel satellite utilization. First, to improve the accuracy and robustness of cloud motion estimation, we designed a hybrid cloud tracking model to incorporate the strength of multiple classic techniques of motion estimation. This new model employ block-wise motion estimation, extracts the dominant motion patterns via histogram statistics. Furthermore, it estimates a dense motion field at a pixel level via customized motion filters and a refined objective function. Compared with state-of-the-art methods, our new model achieved at least 30% and 10% reductions to the errors of motion estimation respectively in simulated and real cloud datasets. Second, to overcome the limitations of ground-based sky images, we implemented a system to integrate multiple sky imagers. The system utilizes a novel method of identifying and tracking clouds in three-dimensional space and constructs an innovative pipeline for forecasting surface solar irradiance based on the image features from clouds. As a result, it robustly detected clouds in multiple layers and outperformed state-of-the-art methods by at least 26% accuracy improvement. Third, to extend the forecast horizons, we devised a mid-term forecast system that integrates the multi-channels of geostationary satellite images and adopts the optical flow approach to estimate cloud motions from noisy satellite images and better subsequent extraction of local cloud features. It employs a robust regression model to combine multiple features together to eliminate outliers and reduce prediction error. The resulting system demonstrates significant improvements over the baseline model in predicting solar irradiance. | |
| dcterms.available | 2017-09-20T16:52:14Z | |
| dcterms.contributor | Huang, Dong | en_US |
| dcterms.contributor | Yu, Dantong | en_US |
| dcterms.contributor | Zhao, Yue. | en_US |
| dcterms.contributor | Doboli, Alex | en_US |
| dcterms.creator | Peng, Zhenzhou | |
| dcterms.dateAccepted | 2017-09-20T16:52:14Z | |
| dcterms.dateSubmitted | 2017-09-20T16:52:14Z | |
| dcterms.description | Department of Computer Engineering. | en_US |
| dcterms.extent | 175 pg. | en_US |
| dcterms.format | Application/PDF | en_US |
| dcterms.format | Monograph | |
| dcterms.identifier | http://hdl.handle.net/11401/77229 | |
| dcterms.issued | 2016-12-01 | |
| dcterms.language | en_US | |
| dcterms.provenance | Made available in DSpace on 2017-09-20T16:52:14Z (GMT). No. of bitstreams: 1
Peng_grad.sunysb_0771E_12793.pdf: 14806029 bytes, checksum: 0947bd1b63e75fcba1ff8eb7e5670a74 (MD5)
Previous issue date: 1 | en |
| dcterms.publisher | The Graduate School, Stony Brook University: Stony Brook, NY. | |
| dcterms.subject | Computer engineering | |
| dcterms.subject | Cloud Detection, Ground-based Sky Imager, Motion Tracking, Multi-source image, Satellite Image, Solar Forecast | |
| dcterms.title | Multi-source Image Integration Towards Solar Forecast | |
| dcterms.type | Dissertation | |
