In recent years, satellites have exhibited a trend toward increased intelligence, leading to the need for the on-orbit deployment of artificial intelligence (AI) algorithms. Intelligent applications require heterogeneous computing to achieve accelerated processing. At the same time, employing a centralized onboard computing system can reduce the overhead associated with intra-satellite data interaction and management. In this context, this paper proposes an onboard heterogeneous computing framework compatible with the satellite management system. The design employs a chip-level heterogeneous architecture, utilizing CPU, GPU, and FPGA as the computational units. To manage and utilize the heterogeneous computing resources, the satellite management software structure, based on a multi-core processor, is expanded to form a heterogeneous computing software framework. This framework includes dependency libraries for intelligent applications and introduces an intelligent application management mechanism. The satellite management system, along with a database and intelligent application, is deployed on a hardware platform based on the Yulong810A multi-core heterogeneous processor. Test results demonstrate that the satellite management software, database, and intelligent applications function cooperatively and complete computational tasks within the required timeframes, thereby validating the fundamental capabilities of the proposed onboard heterogeneous computing system.