ESO-based adaptive sliding mode control for DC–DC buck converters in track vibration energy harvesting

To address the challenges of severe input voltage fluctuations, complex operating conditions, and the difficulty of directly measuring critical states in railway track vibration energy harvesting systems, this study proposes an adaptive sliding mode control strategy based on the extended state observer (ESO) for real-time reconstruction of the input voltage in synchronous rectifier direct current–direct current (DC–DC) buck converters. The proposed strategy regards the input voltage as a quasi-steady disturbance and employs an ESO to estimate the input voltage and inductor current in real time. This approach enables operation using only output voltage feedback, avoiding reliance on additional sensors for inductor current or input voltage measurement. Compared to proportional–integral–derivative control, the proposed strategy reduces the settling time by 68.9%, overshoot by 13%, and output ripple by 86.1% under sinusoidal input voltage disturbances in simulation. Further hardware-in-the-loop experiments conducted on the dSPACE MicroLabBox platform show that this control strategy effectively suppresses the dynamic impact caused by input voltage disturbances and achieves rapid regulation, ensuring stable 5 V output voltage regulation suitable for powering ZigBee wireless sensor nodes. Efficiency tests indicate that the converter maintains high conversion efficiency over a load range from 1 to 100 Ω, exceeding 80% under light-load conditions and achieving a peak efficiency of 90%, attributable to the synergy between synchronous rectification and the proposed adaptive control method. This work presents a high-efficiency DC–DC power converter design that operates using only output voltage feedback, enabling maintenance-free deployment in remote railway vibration energy harvesting systems.