AI-based solar PV power forecasting
As a kind of renewable energy technology, photovoltaic power generation has been more and more widely used. However, since the output power of photovoltaic power generation is affected by many factors(such as weather conditions, seasonal changes, geographical location), it is quite necessary to accu...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/177282 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | As a kind of renewable energy technology, photovoltaic power generation has been more and more widely used. However, since the output power of photovoltaic power generation is affected by many factors(such as weather conditions, seasonal changes, geographical location), it is quite necessary to accurately predict photovoltaic power generation. This project presents two different methods, long short-term memory network (LSTM) and gradient descent, to forecast photovoltaic power (PV) generation and compared their advantages and disadvantages. Gradient descent algorithm is a common optimization algorithm, which can be used to a variety of problems. I found gradient descent to be very good at predicting PV power, especially in the case of large amount of data and few iterations. However, it is sensitive to initial parameters and requires careful adjustment of the learning rate and other parameters. LSTM has the advantage of processing time series data to capture long-term dependencies. My experimental results show that LSTM performs well in predicting PV power, especially when the amount of data is small and the number of iterations is large. However, LSTM takes longer to train and requires more computing resources. |
|---|