[Paper Review] PV Power Forecasting Using Weighted Features for Enhanced Ensemble Method

Mohamed Massaoudi, Inès Chihi+4 more|arXiv (Cornell University)|Oct 21, 2019

Solar Radiation and Photovoltaics32 references4 citations

TL;DR

This paper proposes a weighted feature-enhanced bagging ensemble model for photovoltaic (PV) power forecasting, integrating feature importance analysis to optimize bias-variance trade-offs. By applying a feature weighting vector derived from expert knowledge and statistical analysis, the method improves prediction accuracy, achieving a 5% reduction in RMSE compared to benchmark models.

ABSTRACT

Solar power becomes one of the most promising renewable energy resources in recent years. However, the weather is continuously changing, and this causes a discontinuity of energy generation. PV Power forecasting is a suitable solution to handle sudden disjointedness on energy generation by providing fast dispatching to grid electricity. These methods present a key insight into matchmaking grid electricity and photovoltaic plants. Bootstrap aggregation Ensemble method(Bagging) is classified as one of the most useful machine learning models which are applicable to supervised learning regression tasks. Following this regard, this paper proposes a state-of-art method based on bagging and this method works perfectly for PV power forecasting. The latter had powerful capabilities of tracking the behavior of stochastic problems with good accuracy with the aid of feature importance information. This approach comes to optimize bias/variance using feature weighting vector. Thus, this paper is devoted to present various feature importance techniques for Photovoltaic forecasting parameters. This technique consists of improving the aforementioned ensemble model via contributing the knowledge expertise obtained from features analysis to be directly transformed into the Ensemble model. The proposed model is tested on PV power prediction. Therefore, the benchmarked technique shows an improvement in accuracy in terms of RMSE to 5%.

Motivation & Objective

• To address the challenge of intermittent solar energy generation due to weather variability in photovoltaic systems.
• To improve the accuracy of PV power forecasting using machine learning ensemble methods.
• To integrate feature importance knowledge directly into the bagging ensemble model for better generalization.
• To reduce prediction error, particularly RMSE, in PV power forecasting through feature weighting.
• To validate the proposed method on real-world PV power data with benchmark comparison.

Proposed method

• The method employs bootstrap aggregating (bagging) as the base ensemble framework for regression tasks in PV power forecasting.
• Feature importance is quantified using statistical and domain-specific analysis to generate a feature weighting vector.
• The weighting vector is integrated into the ensemble model to emphasize more informative features during training.
• The model optimizes the bias-variance trade-off by adjusting the contribution of features based on their predictive relevance.
• The approach combines machine learning robustness with expert knowledge from feature analysis to enhance model performance.
• The final model is evaluated using RMSE as the primary metric on real PV power datasets.

Experimental results

Research questions

• RQ1 How can feature importance be systematically incorporated into an ensemble learning model for PV power forecasting?
• RQ2 To what extent does feature weighting improve the accuracy of bagging-based PV forecasting models?
• RQ3 Can the integration of domain knowledge into feature selection enhance model generalization and reduce RMSE?
• RQ4 How does the proposed weighted ensemble model compare to standard bagging and other benchmarks in PV forecasting?
• RQ5 What is the impact of different feature importance techniques on the final prediction performance?

Key findings

• The proposed weighted feature ensemble model achieves a 5% improvement in RMSE compared to the benchmark model.
• Incorporating feature importance through a weighting vector significantly reduces model variance and enhances prediction stability.
• The method effectively tracks stochastic solar power variations due to weather fluctuations with higher accuracy.
• The integration of expert knowledge into feature selection improves the model's ability to generalize across diverse weather conditions.
• The model demonstrates strong performance in handling discontinuous energy generation caused by rapid weather changes.
• The results confirm that feature weighting enhances the effectiveness of bagging in regression tasks for PV power forecasting.