To address the demands of service diversification, operational efficiency improvement, and construction cost optimization in large-scale low earth orbit (LEO) constellations, an autonomous satellite payload operation system based on a modular hierarchical architecture was introduced. A three-dimensional functional framework encompassing autonomous task decomposition, dynamic resource scheduling, fault prediction and tolerance processing, and multi-level control mechanisms was established by the system. Simulation analysis and in-orbit verification demonstrate that in an 880 km altitude, 50° inclination LEO test satellite, the system achieves a 90.3% enhancement in satellite-to-ground communication capacity and a routing re-convergence time of less than 10 seconds during fault scenarios. Through autonomous frequency interference avoidance algorithms and unplanned outage recovery protocols, it mitigates inter-constellation frequency interference and ensures service continuity without increasing ground operational pressure. Notably, the system reduces payload inactive operation duration effectively, significantly enhancing system reliability and flexibility. The research findings provide critical in-orbit validation data for payload component selection and primary-backup redundancy design in LEO constellations, and can be generalized to large-scale LEO constellation engineering implementations.