Improved gravitational search algorithm and novel power flow prediction network for multi-objective optimal active dispatching problems

Abstract

Multi-objective optimal active dispatching (MO-OAD) is a high-profile topic in power grid optimizations. Traditional methods are insufficient for the MO-OAD problem due to its nonconvexity and strict constraints. Therefore, an improved multi-objective gravitational search algorithm (IMGSA) with better population diversity and search capability is put forward in this paper. Five MO-OAD experiments considering various user requirements demonstrate that the IMGSA achieves preferable Pareto-front (PF) over the non-dominated sorting genetic algorithm-II (NSGA-II) and other recently published algorithms. More importantly, to further explore potential elite schemes after obtaining the best compromise scheme (BCS) of IMGSA, an efficient power flow prediction model based on the radial basis function (RBF) network is proposed for the first time. By accurately mapping the nonlinear relationship between the control variables of MO-OAD and optimization objectives, more than five elite dispatching schemes are determined by the suggested RBF prediction model. These elite schemes greatly reduce the fuel cost, power loss and exhaust emission of different IEEE power systems. Quantitative evaluations such as generational distance (GD) and hyper-volume (HV) indicators prove that the proposed IMGSA with RBF prediction model (IMGSA-RBF) is more competitive than many existing algorithms. Detailed MO-OAD experiments show that IMGSA-RBF method has significant advantages in PF-uniformity, PF-diversity, and the quality of optimal dispatching schemes. In general, the innovative IMGSA-RBF method provides a valuable technology to realize desirable power grid operation with less carbon emission and better economy.