A machine learning framework enhanced by hybrid optimization for precise solar power generation prediction
<p>Photovoltaic power generation is an efficient method of harnessing solar energy, the best renewable source. However, it is weather-dependent and unstable, affecting electrical systems. Therefore, improving the accuracy of solar forecasts is crucial. This study develops a solar power predict...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Published: |
2025
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | <p>Photovoltaic power generation is an efficient method of harnessing solar energy, the best renewable source. However, it is weather-dependent and unstable, affecting electrical systems. Therefore, improving the accuracy of solar forecasts is crucial. This study develops a solar power prediction model using machine learning, specifically employing a Multi-Layer Perceptron Regression model. Further refinement of the predictive performance of this model has been done by incorporating Tunicate Swarm Algorithm (TSA) and Sooty Tern Optimization Algorithm (STOA). This important combination gives rise to two new hybrid models: MLPR combined with TSA, which is referred to as the MLTS framework, and MLPR combined with STOA, which is referred to as the MLST framework. These frameworks are vital enhancements of the precision of solar power generation forecasts. In Layer 3, during the test phase for the R<sup>2</sup> metric, the MLTS model demonstrated the highest performance with a value of 0.972, while the MLST model ranked second with an R<sup>2</sup> value of 0.952. Similarly, in Layer 2, during the validation phase for the RMSE metric, the MLTS model exhibited superior performance with an RMSE of 1.92E + 04, followed by the MLST model with the second-best performance, achieving an RMSE of 2.17E + 04.</p> |
|---|