The basis for dynamic control and application in intelligent transportation systems (ITS) is traffic flow forecasting, which is essential for reducing traffic congestion. Accurate predictions are still difficult to achieve due to the intricate spatiotemporal correlations of traffic flow. To capture dynamic spatiotemporal information, this research offers a novel spatio-temporal interactive dynamic synchronous graph (STIDSG) convolutional network for traffic flow forecasting. The model is primarily composed of an interactive dynamic graph convolutional network (IDGCN) and a spatio-temporal synchronous graph convolutional layer (STSGCL). IDGCN employs an interactive learning strategy based on the dynamic graph convolution network (DGCN), allowing for feature sharing through interactive learning. Through the integration of DGCN into an interactive educational framework, the model can simultaneously capture temporal dependencies while interactively learning dynamic spatial features. STSGCL is designed with multiple spatio-temporal synchronous graph convolution modules (STSGCMs) across different time periods, effectively capturing complex local dynamic spatiotemporal correlations. According to experimental data, the suggested STIDSG model outperforms widely used baseline techniques in terms of accuracy for prediction and can efficiently extract dynamic spatiotemporal information.