Based on the dynamic programming, the electromagnetism-like mechanism algorithm is applied to multi-objective dynamic planning. It eliminates the over-dependence of the total force vector on the distance, and by drawing lessons from passive congregation thought, enhances the anti-disturbance and global optimization ability of the population mobility model. In the sectors of weight, local search, and mutation, the adaptive strategy is used to improve the searching speed, and the elitist strategy is used to improve the convergence efficiency. By adding the Markov constraints, the re-dismantling of lines is avoided. Using the weighted method and penalty function method, multiple objectives describing the grid economy, reliability, and static N-1 security are measured in a unified manner to reflect the overall satisfaction of the plan. Applying decimal coding to 19-bus and 46-bus system, the results show that the algorithm can effectively solve the multi-objective dynamic grid planning. Compared with the genetic algorithm, the improved ELM has obvious advantages in searching efficiency and fault-tolerance.