Abstract: In the multi-energy complementary power generation system, the wind-photovoltaic output is unstable and discontinuous, which makes the hydropower units in the system need to change the working condition frequently, which brings challenges to the economy, safety and stability of hydropower station operation. When the unit vibration avoidance strategy is considered in the load distribution of the hydropower plant, the decision variables of the ordinary particle swarm optimization model will become a complex matrix with high dimensionality, mutual coupling and discontinuity. The convergence result of the particle velocity updating process in the model is greatly reduced due to the reduced suitability of the process. In this paper, a bidirectional updating multi-objective particle swarm model is proposed innovatively.Based on the day-ahead hour-level scheduling mode and vibration avoidance strategy of the unit, two objectives of the unit combined crossing number and the sum of the optimal working conditions of the unit throughout the day are optimized. The results show that compared with the ordinary particle swarm optimization algorithm, the calculation speed of the optimization model is improved by 14.7%, and the convergence accuracy range is improved by 4% to 6% on average, which can provide a new theoretical support for the optimization operation of hydropower station with both economy and safety stability.