To address the issues such as curtailed wind and solar power generation, insufficient grid integration, and low resource utilization efficiency in new energy power generation, a virtual power plant (VPP) optimal dispatch strategy based on multi-source flexible resources is proposed. This strategy is aimed at optimizing wind-solar power output, improving energy accommodation capacity and reducing resource waste. First, a mathematical model of multi-source flexible resources for virtual power plant optimal dispatch was constructed. A master-slave game scheme with the virtual power plant operator as the leader and load aggregators as the followers was designed. Second, the hyperbolic sine-cosine algorithm used for model solving was improved. The Tent map was introduced to expand the diversity of population initialization. The elite classification method was adopted to implement hierarchical management of the population, while a bubble predation mechanism strengthens neighborhood exploration of elite populations. The Lévy flight strategy guided by the greedy behavior of excellent populations was improved to direct the potential population to search toward the global optimal population. Third, optimized electricity prices were generated using the improved algorithm. The optimal electricity consumption strategies for various flexible resources were designed in conjunction with the CPLEX solver. The results demonstrate that under the im-proved IEEE 30-node system and three different experimental scenarios, the proposed algorithm and model effectively enhance the accommodation capacity of wind and solar power, improve the resource allocation performance of VPPs, and reduce new energy waste.