Self-trapped excitons (STEs), proved to be the major source of white-light emission in 2D metal halide perovskites (MHPs) van der Waals (vdW) heterostructures, have aroused intense interest in photovoltaic and photoelectric applications. Nevertheless, the intrinsic mechanisms of STEs in these vdW heterostructures are still ambiguous. Herein, the binding energy correction $\Delta E_{\mathrm{B}}$ of a STE stemming from the exciton–phonon coupling in MHPs vdW heterostructures based on the Pollmann–Büttner model is studied. It is found that there are two types of STEs with $\Delta E_{\mathrm{B}}>0$ and $\Delta E_{\mathrm{B}}<0$. The corresponding nuclear coordinate diagrams are given to explain the differences between them and why the STEs with $\Delta E_{\mathrm{B}}<0$ are hard to be observed in experiments. The phase transition between two types of STEs can be achieved by regulating the structural parameters, such as the vertical distance between the encapsulation layers, the position of the monolayer MHP in the heterostructure as well as replacing the encapsulation materials. The theoretical results provide important insights into the analysis and modulation of STEs in 2D vdW heterostructures.