Work

Comprehensive Energy Management for Rural Microgrids: Integrating LSTM Forecasting and Kalman-Filtered MPC for Stochastic EV Loads

Year: 2026
Type: conference-paper
Abstract: The integration of electric vehicle (EV) charging infrastructure into rural microgrids introduces significant powerflow volatility. This paper presents a computationally efficient energy management framework integrating long short-term memory (LSTM) forecasting and Kalman-filtered model predictive control (MPC). The proposed multi-tier architecture utilizes a 12-hour predictive horizon to proactively manage storage reserves while mitigating 5% sensor noise through recursive state estimation. Validated via a 35,040-step annual simulation for a rural community, the framework demonstrates superior capability, whereas rule-based strategies struggle with high-power EV surges and sensor degradation. Our approach achieves a 4.2% improvement in system reliability (LPSP) and maintains stability under noisy measurements. By identifying a predictive decay threshold at 12 hours, we demonstrate that tactical foresight outperforms traditional 24-hour windows in stochastic rural environments, achieving an LCOE of $0.189/kWh and bringing microgrids closer to grid parity.
Language: English
Open Access status: hybrid
OpenAlex ID: W7203831004
FWCI:
Cites: 20
Cited by:
Related to: 10