In recent years, traffic flow prediction has played a crucial role in the management of intelligent transportation systems. However, traditional forecasting methods often model non-Euclidean low-dimensional traffic data as a simple graph with single type nodes and edges, failing to capture similar trends among nodes of the same type. This paper proposes MHGNet, a new framework for modeling spatiotemporal multi-heterogeneous graphs, in which the STD Module decouples single-pattern traffic data into multi-pattern traffic data through feature mappings of timestamp embedding matrices and node embedding matrices. Then the Node Clusterer uses the Euclidean distance between nodes and different types of limit points for O(N) time complexity clustering, and the nodes within clusters undergo residual subgraph convolution in the spatiotemporal fusion subgraphs generated by the DSTGG Module before entering the SIE Module for node repositioning and redistribution of weights. This paper conducts a series of ablation and quantitative evaluations on four widely used benchmarks to validate the effectiveness of MHGNet.