A Random Forest-Based Intelligent Duty-Cycle Prediction Framework for High-Efficiency Dc–Dc Converter Control

Engineering notes

Engineering notes will be added by the Power for Solar editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为光伏效率、钙钛矿太阳能电池、储能技术、太阳能热利用、BIPV、并网技术等高价值论文补充中文说明。

Original abstract

The increasing penetration of renewable energy sources in modern microgrids has created a growing need for intelligent and efficient power converter control strategies capable of handling dynamic operating conditions. This paper presents a machine-learning-based framework for the optimal design and control of a DC–DC converter in renewable-energy-based microgrid applications. The proposed approach utilizes key operational parameters, including photovoltaic (PV) voltage, PV current, PV power, load demand, output voltage, and output current, to predict the optimal converter duty cycle and enhance overall system performance. A Random Forest regression model is developed to capture the nonlinear relationship between converter operating conditions and duty-cycle requirements. The proposed controller is evaluated using comprehensive performance metrics, including duty-cycle prediction accuracy, error distribution analysis, voltage regulation performance, dynamic response characteristics, and feature importance assessment. Furthermore, Explainable Artificial Intelligence (XAI) based on SHAP (SHapley Additive Explanations) is incorporated to improve model transparency and identify the dominant parameters influencing converter control decisions. Simulation results demonstrate excellent agreement between actual and predicted duty-cycle values, with prediction errors concentrated near zero, indicating high accuracy and robustness. The dynamic response analysis confirms stable voltage regulation with fast settling behavior and negligible overshoot. SHAP and feature importance analyses reveal that PV voltage and output voltage are the most influential factors governing duty-cycle prediction. The proposed framework effectively improves voltage regulation, reduces converter losses, enhances energy conversion efficiency, and provides interpretable decision-making capabilities. Therefore, it offers a reliable and intelligent solution for next-generation renewable-energy microgrids and smart power management systems. Keywords— DC–DC Converter, Renewable Energy Systems, Microgrid, Machine Learning, Random Forest Regression, Duty-Cycle Prediction, Voltage Regulation, Converter Efficiency, Power Loss Minimization, Explainable Artificial Intelligence (XAI), SHAP Analysis, Intelligent Control, Photovoltaic Systems, Energy Management, Smart Grid.

Links and sources

• PDF from original source

Comments