Abstract: To perform the maximized scheduling of load resources that were effectively connected in the same regional substation, a framework for the load network system, in which the joint control of heterogeneous thermal control load cluster was considered, was proposed. Firstly, the physical perception layer of the network system was constituted by the end-side loads with the same electrical connection. Secondly, restricted by regional dispersibility and parameter heterogeneity, the load resources, in which the effective power transfer practically occurred, were limited, thus, the edge data centre was brought in to analyze the generality of the equipment layer of heterogeneous thermal control load and the physically dispersed loads of the same regional substation were virtually aggregated, and a joint control model of multi-load state space coordination was established. Thirdly, the adjustable capacity and operating state of the thermal control load were comprehensively considered in cloud-end power grid data centre, and after the secondary distribution the optimal task participating group and the task load of each group were determined. The fourth, taking the homogeneous aggregation group was taken as the basic control unit, and a collaborative control model of heterogeneous loads was developed on the edge side to issue the unified control instructions to coordinated the end load participating in energy services. Finally, by means of computing example, the effectiveness of both the proposed system and the heterogeneous load joint control model are verified.