Solar irradiance forecasting using statistical and machine learning methods
The installed capacity of solar photovoltaic (PV) is continues to rise in the world and Malaysia throughout the year. In Malaysia, the average daily solar radiation is 4,000 to 5,000 Wh/m2, with the average daily sunshine duration ranging from 4 to 8 hours. However, the output of solar energy is lac...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2023
|
| Online Access: | http://eprints.utem.edu.my/id/eprint/27144/1/Solar%20irradiance%20forecasting%20using%20statistical%20and%20machine%20learning%20methods.pdf http://eprints.utem.edu.my/id/eprint/27144/2/Solar%20irradiance%20forecasting%20using%20statistical%20and%20machine%20learning%20methods.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The installed capacity of solar photovoltaic (PV) is continues to rise in the world and Malaysia throughout the year. In Malaysia, the average daily solar radiation is 4,000 to 5,000 Wh/m2, with the average daily sunshine duration ranging from 4 to 8 hours. However, the output of solar energy is lack of stability due to weather variation. Solar irradiance forecasting is a crucial component in the effective integration of solar PV systems into the electrical grid. The variability of solar energy and the uncertainties associated with solar irradiance predictions pose significant challenges for grid operators and energy planners. This research project aims to develop advanced forecasting methods and methodologies for accurate and reliable solar irradiance prediction, considering the specific characteristics of local weather conditions. The study begins by analyzing the correlation between weather parameters and solar irradiance in the selected region, identifying the key variables that significantly impact solar irradiance. Quadratic regression methods are developed to forecast solar irradiance by leveraging the relationships between weather parameters. Additionally, artificial neural network (ANN), long-short term memory (LSTM), and seasonality autoregressive integrated moving average (SARIMA) methods are evaluated to determine their suitability for solar irradiance forecasting. Comparative analysis of the developed forecasting methods is conducted using evaluation metrics such as root mean square error (RMSE) and correlation of coefficient (R). The performance and suitability of different statistical and machine learning techniques for solar irradiance forecasting assisting grid operators, energy planners, and policymakers in effectively integrating solar PV systems into the electrical grid and optimizing the utilization of solar energy resources. Overall, this research project aims to advance the field of solar irradiance forecasting, enabling better planning, operation, and management of solar PV systems. By reducing uncertainties in solar energy generation, it contributes to the overall advancement of renewable energy integration and supports the transition towards a sustainable and clean energy future. |
|---|