Due to the lower efficiency found in traditional metaheuristic algorithms for a solution of a large-scale open workshop scheduling problems (OSSP), a method based on graph convolution network (GCN) has thus been proposed to solve OSSP. Firstly, an open workshop scheduling model based on GCN has been designed, which incorporates the process node features of OSSP into the graph, to be followed by a multi-layer convolution operation on it, thus effectively obtaining the complex dependency relationships between process nodes. Next, in view of an improvement of efficiency and quality of solving large-scale OSSP, an open workshop scheduling algorithm has been proposed based on GCN. The experimental results show that this method can effectively solve OSSP instances of different scales. Compared with metaheuristic algorithms, the proposed method is characterized with a better solution quality and higher efficiency in solving large-scale OSSP instances.