Research on short-term photovoltaic power generation forecasting model based on multi-strategy improved squirrel...

Area of Science:

• Renewable Energy Systems
• Machine Learning Applications
• Computational Intelligence

Background:

• Solar photovoltaic (PV) power generation faces challenges in prediction accuracy due to environmental factors and redundant data features.
• Existing prediction models can be disrupted by feature complexity, necessitating advanced optimization techniques.
• Accurate PV power forecasting is crucial for grid stability and efficient energy management.

Purpose of the Study:

• To develop a rapid and accurate online prediction method for solar PV power generation.
• To enhance the performance of Support Vector Machine (SVM) models for PV power prediction.
• To address the impact of feature dimensionality and hyperparameter optimization on prediction accuracy.

Main Methods:

• Kernel Principal Component Analysis (KPCA) for feature dimensionality reduction.
• Support Vector Machine (SVM) as the core prediction algorithm.
• A multi-strategy improved Squirrel Search Algorithm (MISSA) for optimizing SVM hyperparameters, incorporating Tent map initialization, nonlinear predator probability, chaotic opposition-based learning, and a selection strategy.

Main Results:

• The MISSA algorithm demonstrated high optimization performance, stability, and significance on CEC2021 test functions.
• The proposed MISSA-SVM prediction model achieved high accuracy and robust prediction stability on two real-world datasets.
• KPCA effectively mitigated the impact of redundant features, improving overall prediction accuracy.

Conclusions:

• The proposed MISSA-SVM method offers a significant advancement in accurate and stable solar PV power prediction.
• Combining KPCA for feature reduction and MISSA for hyperparameter optimization provides a powerful approach for complex prediction tasks.
• This research contributes to more reliable renewable energy forecasting through intelligent optimization techniques.