<p>Cell-free massive MIMO (CF-mMIMO), as a user-centric architecture, significantly enhances the user experience rate. To maximize the downlink sum rate in CF-mMIMO systems, multi-user joint scheduling and multi-user precoding methods play a pivotal role. However, with numerous RAUs and users, using all network elements can significantly increase computational complexity and data exchange. To address this issue, this paper proposes downlink multi-user joint scheduling system solution incorporating dynamic user clustering. Firstly, we focus on constructing scalable dynamic user clusters by using the spatial sparsity characteristics of the user channel matrix. And the multi-user joint scheduling process is based on dynamic user clusters as the basic unit, with real-time processing within clusters and gradual processing between clusters, which lays the foundation for reducing processing complexity. Secondly, based on the clustering results, we design a multi-user joint scheduling scheme with intra-cluster correlation constraints and inter-cluster unconstrained conditions. This scheme is low in complexity and excellent in performance. Finally, we further optimize the classical zero forcing algorithm to construct real-time interference cancellation weights within clusters and statistical interference cancellation weights between clusters. The concatenated precoding weights demonstrate superior interference suppression performance. The simulation results demonstrate that the proposed multi-user joint scheduling scheme outperforms in sum rate while maintaining comparable fairness among users. Moreover, the optimized algorithm achieves a rate gain of more than 12%.</p>