Robust Power System Optimization Using LSTM-Based Load Forecasting and AIS Algorithm

Optimizing alternating current (AC) power flow under uncertainty remains a major challenge in modern power systems, particularly with the increasing penetration of variable renewable energy sources. This paper proposes a hybrid two-stage framework that integrates long short-term memory (LSTM) networks for load forecasting with an artificial immune system (AIS)-based optimization approach, embedded within a Monte Carlo simulation scheme to explicitly account for uncertainty. The methodology is validated on the IEEE 30-bus test system. In the first stage, the LSTM model captures temporal dependencies to generate short-term load forecasts, while in the second stage, these forecasts are incorporated into an AIS-based AC optimal power flow (AC-OPF) formulation. Monte Carlo simulations are employed to model stochastic variations and assess system performance across multiple scenarios. The results show that, although the reduction in operational cost is relatively marginal compared to deterministic approaches, the proposed framework significantly enhances the robustness and stability of OPF solutions under forecasting uncertainty, improving the system’s ability to maintain feasible and consistent operating points despite variability in load predictions. However, the forecasting performance of the LSTM model is sensitive to noise and out-of-distribution inputs, which may affect the overall optimization quality. Overall, the main contribution of this work lies in the development of an integrated forecasting–optimization framework that strengthens the reliability and resilience of power system operation under uncertainty.