Aiming at the nonlinearity and time-varying of high energy consumption load, high-density centralized load forms a high-energy point load situation, causing frequent fluctuations and over-limits of the grid voltage. Due to different time scales, secondary voltage regulation is hard to overcome the problem. In this paper, the optimal setting strategy of excitation system adjustment coefficient of generator set is proposed to improve the adverse effect of high energy consumption point load access. From the perspective of reducing the network losses and voltage fluctuation, a multi-objective function considering the load voltage fluctuation of the network losses and the central node and the high energy consumption point is established. The particle swarm optimization algorithm combined with the improved weight and immune algorithm is used to solve the problem. Taking the IEEE39 node system as an example, the result proves that after the optimization，the situation that the low voltage exceeds the limit can be avoided，the network losses are effectively reduced, and the economic and reliability of the power grid operation is improved.