Solar Power Forecasting Using Hybrid Deep Learning: Performance Enhancement with Random Forest-BiLSTM and Ensemble Modeling

Summary

This study advances concentrated solar power plant performance through comprehensive data analysis and error correction methodologies. By integrating spectrum analysis, thermal efficiency optimization, and hybrid machine learning models, the research provides actionable strategies for enhancing operational efficiency and reliability, thereby supporting the role of solar energy as a sustainable power source.

Abstract

Accurate solar power forecasting is critical for grid integration and operational stability of renewable energy systems. This study presents a hybrid deep learning ensemble approach to predict solar generation by capturing complex temporal dependencies in irradiance data. Five hybrid architectures were evaluated: RF-BiLSTM, CNN-LSTM, CNN-BiLSTM, CNN-GRU, and CNN-Transformer, each combining convolutional or recurrent components to extract spatial and sequential features from historical time series. The RF-BiLSTM model achieved the best individual performance with R² = 0.6568, MAE = 30,728 W, and MSE = 1.81 × 10⁹ W². An ensemble model integrating the top three architectures using inverse MAE-weighted averaging demonstrated superior performance with R² = 0.6933, MAE = 28,809.89 W, and MSE = 1.53 × 10⁹ W², reducing prediction error by 6.2% compared to the best individual model. The proposed ensemble framework effectively balances model strengths, enhances forecast robustness, and provides a scalable, data-driven solution for renewable energy forecasting in smart grid and energy management systems.

Introduction

The accelerating global transition toward renewable energy has positioned solar power as a pivotal source in the sustainable energy mix. As countries increasingly commit to decarbonizing their energy systems, solar photovoltaic (PV) technology has witnessed exponential growth due to its scalability, declining costs, and environmental benefits. However, the widespread integration of solar energy into national and regional power grids presents significant challenges, primarily due to its intermittent and weather-dependent nature. Solar irradiance is influenced by a variety of environmental factors, including cloud cover, atmospheric conditions, seasonal shifts, and diur....

Protocol

Dataset collection and description
The dataset (Figure 1) used in this research comprises historical records crucial for solar power forecasting. The dataset comprises daily operational data from a 50 MW concentrated solar thermal plant operated by Megha Engineering and Infrastructures Limited (MEIL), located near Anantapur, Andhra Pradesh, India, utilizing parabolic trough Concentrating Solar Power(CSP) technology that captures Direct Normal Irradiance (DNI) and transfers heat via a Heat Transfer Fluid (HTF) to generate electricity. The dataset was collected from 01 January 2015 to 03 October 2025 and contains seve....

Results

Individual model performance evaluation
The performance evaluation of five hybrid deep learning (DL) models RF-BiLSTM, CNN-GRU, CNN-BiLSTM, CNN-LSTM, and CNN-transformer was conducted using a comprehensive set of standard regression metrics, including R² (coefficient of determination), mean absolute error (MAE), and mean squared error (MSE), to rigorously assess their capability in forecasting solar power generation under varying meteorological conditions and temporal dependencies.

Discussion

The proposed methodology follows a structured workflow as shown in Figure 12. Initially, the dataset undergoes comprehensive preprocessing, including missing value imputation, normalization, and feature engineering, to ensure data quality and enhance model learning^3,6. The processed dataset is then partitioned into training (70%), validation (15%), and testing (15%) sets to enable robust model development and performance evaluation

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
BiLSTM	TensorFlow/Keras	TensorFlow 2.10.0
CNN layers	TensorFlow/Keras	TensorFlow 2.10.0
Google Colab	Google LLC	Cloud Platform
GRU	TensorFlow/Keras	TensorFlow 2.10.0
Matplotlib	Matplotlib Dev Team	3.7.1
NumPy	NumFOCUS	1.25.2
NVIDIA T4 GPU	NVIDIA Corporation	Tesla T4
Pandas	NumFOCUS	2.0.3
Pyrheliometer for DNI measurement	Kipp & Zonen	CH1-DL
Python	Python Software Foundation	3.10.12
Random Forest	Scikit-learn Developers	1.2.2
Scikit-learn	Scikit-learn Developers	1.2.2
Temperature sensors	Vaisala	HMP155
TensorFlow/Keras	Google	Version 2.10.0
Transformer	TensorFlow/Keras	TensorFlow 2.10.0
Weather station	Davis Instruments	Vantage Pro2