Optimization of distribution network scheduling based on BA and photovoltaic uncertainty

Abstract A novel optimization framework for microgrid (MG) distribution network scheduling has been proposed, which integrates enhanced bat algorithm (BA) with probabilistic photovoltaic (PV) power prediction to address the dual challenges of renewable energy uncertainty and operational efficiency. The method combines genetic algorithm-optimized artificial neural network with Monte Carlo simulation to achieve very low PV output prediction error under variable weather conditions. In addition, the study improves the BA by using a multi-mechanism approach of Halton sequence initialization and dynamic elite retention, which enhances the convergence iteration speed of the algorithm and avoids local optima problems. The experimental results show that the maximum predicted output power (OP) of the PV generator set is 208 kW, and the predicted OP curve is more consistent with the actual OP curve. The enhanced BA has been demonstrated to enhance the optimization of operating costs of MGs. The operating costs of MGs are 125,452 yuan at low loads and 145,562 yuan at high loads. The study provides a replicable template for high-penetration renewable MGs by combining probability prediction with swarm intelligence optimization. The research provides new ideas for the scheduling strategy of MG distribution networks centered on new energy.