This study presents a dq impedance-decoupled network modelling method of offshore wind power plant (OWPP) for stability analysis. DQ impedance frequency responses of grid-connected inverters (GCIs) and long transmission cables (LTCs) are first measured by frequency scanning method, which are fitted as transfer function matrices using matrix fitting algorithm. Then, the GCIs are modelled as Norton equivalent circuits. In addition, per-unit-length electrical parameters of the LTCs are extracted from the measured dq impedance frequency responses of a specific LTC, based on which dq impedance-decoupled two port network models of the LTCs are established. DQ impedance-decoupled network model of the whole OWPP is then established based on connection relationships of these GCIs and LTCs. Finally, generalised Nyquist criterion (GNC) is performed in all the dq impedance-decoupled subsystems, and the subsystems where the GNC is not satisfied are identified as instability sources. Compared with conventional impedance-based stability analysis methods of OWPP, the proposed dq impedance-decoupled modelling method is able to facilitate the application of GNC and further instability source identification based on partitioning the whole OWPP into several decoupled subsystems. The effectiveness of the network modelling method is validated based on time-domain simulation results in Matlab/Simulink and real-time verification results in OPAL-RT.