Integrated Microgrid Scheduling with LSTM-Based Battery Degradation and Cost Minimization

As battery storage becomes integral to modern microgrids, battery degradation increasingly affects operational efficiency and system economics. This study proposes a multi-objective scheduling framework that minimizes both operational and degradation costs. An LSTM neural network is trained on NASA battery data to predict capacity degradation based on voltage, current, and temperature. The predicted degradation is integrated into a Pyomo-based scheduling model using a weighted sum method to balance the two objectives. Simulation results show that the proposed approach improves both cost efficiency and battery lifespan compared to conventional single-objective strategies. The model also outperforms an RNN-based degradation forecast in total cost, demonstrating the advantage of LSTM in capturing battery aging dynamics. This approach provides a scalable and effective solution for intelligent microgrid energy management.