repository
Development of Photovoltaic Generation Forecasting and Failure Detection Model
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 정재성 | - |
| dc.contributor.author | JUFRI FAUZAN HANIF | - |
| dc.date.accessioned | 2018-11-08T08:28:54Z | - |
| dc.date.available | 2018-11-08T08:28:54Z | - |
| dc.date.issued | 2018-08 | - |
| dc.identifier.other | 27790 | - |
| dc.identifier.uri | https://dspace.ajou.ac.kr/handle/2018.oak/14195 | - |
| dc.description | 학위논문(석사)--아주대학교 일반대학원 :에너지시스템학과,2018. 8 | - |
| dc.description.tableofcontents | I. Introduction I.1. Previous Studies I.2. Research Motivation II. Solar Irradiance and Photovoltaic System II.1. Solar Irradiance II.2. ASHRAE Clear-Sky Model II.3. Photovoltaic Systems III. PV Generation Forecasting Model III.1. Weather Variables Selection III.2. ANN-based Forecasting Model Development III.3. Sky Conditions III.4. Cross-Validation Technique III.5. PV Generation Forecasting Result IV. PV Failure Detection Model IV.1. Expected PV Generation Model IV.2. SVM-based Failure Detection Model IV.3. Consideration of Daylight Time IV.4. Interaction of Variables IV.5. Multi-stage k-fold Cross-Validation IV.6. PV Failure Detection Result V. Conclusion References | - |
| dc.language.iso | eng | - |
| dc.publisher | The Graduate School, Ajou University | - |
| dc.rights | 아주대학교 논문은 저작권에 의해 보호받습니다. | - |
| dc.title | Development of Photovoltaic Generation Forecasting and Failure Detection Model | - |
| dc.title.alternative | Development of Photovoltaic Generation Forecasting and Failure Detection Model | - |
| dc.type | Thesis | - |
| dc.contributor.affiliation | 아주대학교 일반대학원 | - |
| dc.contributor.department | 일반대학원 에너지시스템학과 | - |
| dc.date.awarded | 2018. 8 | - |
| dc.description.degree | Master | - |
| dc.identifier.localId | 887769 | - |
| dc.identifier.uci | I804:41038-000000027790 | - |
| dc.identifier.url | http://dcoll.ajou.ac.kr:9080/dcollection/common/orgView/000000027790 | - |
| dc.subject.keyword | Photovoltaic Generation Forecasting | - |
| dc.subject.keyword | Photovoltaic Failure Detection | - |
| dc.subject.keyword | Artificial Neural Network | - |
| dc.subject.keyword | Support Vector Machine | - |
| dc.description.alternativeAbstract | A large-scale photovoltaic (PV) system is considered as one of the most promising alternatives to substitute the fossil-based power generation in order to reduce the environmental problem such as greenhouse gas effect. However, challenges arise when the large-scale PV system is integrated into the electricity grid due to its intermittency and uncertainty. Moreover, PV system is also susceptible to the physical disturbances due to the outdoor installation of PV modules which can affect its productivity and reliability. Therefore, it is necessary to develop the PV generation forecasting and failure detection model to resolve the intermittency and uncertainty problems of PV system as well as to improve its productivity and reliability. This study presents the development of the PV generation forecasting and failure detection model with limited data and devices. The PV generation forecasting model is developed without the availability of the solar irradiance as a main independent variable. Whereas, the PV failure detection model is developed without any additional sensory devices such as infrared cameras or voltage and current sensors. This study utilizes the available data which already exist in a PV generation site and then applies the machine learning algorithm to build the models. Thus, the model development algorithm presented in this study can be implemented in any PV generation site with an economic advantage. First, this study designs a PV generation forecasting model by using Artificial Neural Networks (ANN) algorithm. The unavailability of the solar irradiance data is resolved by employing the ASHRAE Clear-Sky model. It also considers the weather information provided by Korea Meteorological Administration (KMA) such as number of clouds, ambient and dew point temperature, relative humidity, rainfall rate, and wind speed. The accuracy of the PV generation forecasting model is improved by applying Pearson Correlation Coefficient and cross-validation technique to select the input variables and the ANN setting parameters. Secondly, this study also develops a PV failure detection model by using Support Vector Machine (SVM) algorithm. It only analyzes the available data on PV generation system such as generated power, voltage, current from the Power Conversion System (PCS), measured solar irradiance from the pyranometer, and temperature from the thermometer. The accuracy of the PV failure detection model is improved by considering the expected power generation and interactions variables through k-fold cross-validation technique. As a result, it is shown that the proposed models can forecast power generation without the historical data of solar irradiance and to detect the failure with a limited number of data and sensory devices with high accuracy. | - |
- Appears in Collections:
- Graduate School of Ajou University > Department of Energy Systems > 3. Theses(Master)
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
-
- License
- STATISTICS
- Total Visit :3,625,514
- Total Download :1,804
- Today View :5,994
