Abstract: In engineering applications, parallel partitioned coupling methods are commonly used to solve multi-physics field coupling problems. The critical step in partitioned coupling is data exchange. However, existing methods suffer from inefficient mapping efficiency due to the improper allocation of coupling surface grids across processes, resulting in a coupling computational time increase. Therefore, this paper proposes a dynamic re-partition-based parallel mapping algorithm. It employs geometric boundary comparison or spatial hashing methods for parallel grid matching, specifically tailored for structured and unstructured grids. When encountering improper grid allocation on the coupling surfaces, the method adopts dynamic loading balancing by spawning threads to dynamically re-partition oversized boundary face grids across processes and performs multi-threaded mapping. Two numerical examples, including fluid impact on a flexible plate and fluid flow through an elastic pipe, demonstrate that this algorithm effectively improves mapping efficiency in fluid-structure coupling cases.