Solar PV power forecasting considering missing data
Solar power generation has now become a mature technology and is widely used in commercial and civilian applications. In fact, our lives are inseparable from photovoltaic power (PV) generation. But photovoltaic power (PV) generation is sometimes unstable and unpredictable. In some circumstances, pho...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
Nanyang Technological University
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/149619 |
| _version_ | 1826119883950129152 |
|---|---|
| author | Zhai, Chengrui |
| author2 | Xu Yan |
| author_facet | Xu Yan Zhai, Chengrui |
| author_sort | Zhai, Chengrui |
| collection | NTU |
| description | Solar power generation has now become a mature technology and is widely used in commercial and civilian applications. In fact, our lives are inseparable from photovoltaic power (PV) generation. But photovoltaic power (PV) generation is sometimes unstable and unpredictable. In some circumstances, photovoltaic power (PV) generation will be affected by the environment, malfunction, communication and other factors, which usually leads to data instability or even loss. In case of data lost scenarios, the application performance could be dramatically degraded.
This paper proposes a fresh new online training model based on the hybrid Artificial Neuron Network (ANN) machine learning to address the incomplete or missing data issue. A hybrid ensemble learning method of Extreme Learning Machine (ELM) and Random Vector Functional Link (RVFL) networks is designed as the learning algorithm for estimators and simulators. In this proposed method, a set of flawed data input will be used for offline training to restore the PV measurement and the potential imprecisely forecasted features are processed in the data validation stage to improve the overall performance further. The iteratively online training model will eliminate inaccurate predicted values and then compare the optimized value with the original value, thereby maximizing the recovery of lost data in the online mode application. Simulation results show that the new approach achieves low estimation error rate as well as short processing time. |
| first_indexed | 2024-10-01T05:07:25Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/149619 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:07:25Z |
| publishDate | 2021 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1496192023-07-07T18:20:46Z Solar PV power forecasting considering missing data Zhai, Chengrui Xu Yan School of Electrical and Electronic Engineering Power and Clean Energy Design Lab xuyan@ntu.edu.sg Engineering::Electrical and electronic engineering Solar power generation has now become a mature technology and is widely used in commercial and civilian applications. In fact, our lives are inseparable from photovoltaic power (PV) generation. But photovoltaic power (PV) generation is sometimes unstable and unpredictable. In some circumstances, photovoltaic power (PV) generation will be affected by the environment, malfunction, communication and other factors, which usually leads to data instability or even loss. In case of data lost scenarios, the application performance could be dramatically degraded. This paper proposes a fresh new online training model based on the hybrid Artificial Neuron Network (ANN) machine learning to address the incomplete or missing data issue. A hybrid ensemble learning method of Extreme Learning Machine (ELM) and Random Vector Functional Link (RVFL) networks is designed as the learning algorithm for estimators and simulators. In this proposed method, a set of flawed data input will be used for offline training to restore the PV measurement and the potential imprecisely forecasted features are processed in the data validation stage to improve the overall performance further. The iteratively online training model will eliminate inaccurate predicted values and then compare the optimized value with the original value, thereby maximizing the recovery of lost data in the online mode application. Simulation results show that the new approach achieves low estimation error rate as well as short processing time. Bachelor of Engineering (Electrical and Electronic Engineering) 2021-06-05T13:18:16Z 2021-06-05T13:18:16Z 2021 Final Year Project (FYP) Zhai, C. (2021). Solar PV power forecasting considering missing data. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/149619 https://hdl.handle.net/10356/149619 en P1029-192 application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Electrical and electronic engineering Zhai, Chengrui Solar PV power forecasting considering missing data |
| title | Solar PV power forecasting considering missing data |
| title_full | Solar PV power forecasting considering missing data |
| title_fullStr | Solar PV power forecasting considering missing data |
| title_full_unstemmed | Solar PV power forecasting considering missing data |
| title_short | Solar PV power forecasting considering missing data |
| title_sort | solar pv power forecasting considering missing data |
| topic | Engineering::Electrical and electronic engineering |
| url | https://hdl.handle.net/10356/149619 |
| work_keys_str_mv | AT zhaichengrui solarpvpowerforecastingconsideringmissingdata |